leaf_core/doc.rs
1//! The document model: a `twig::Editor` plus a byte-offset caret and selection.
2//!
3//! Where bough moves a selection through the *tree*, leaf moves a *caret*
4//! through the *characters* — a normal text editor's model — and expresses
5//! every mutation as one of twig's offset-addressed ops:
6//!
7//! - typing / delete → `edit_range(start, end, text)` (P0)
8//! - re-anchoring → the returned `Change` (P1)
9//! - cursor context → `node_at` / `ancestors_at` (P3)
10//! - the toolbar → `wrap_range`/`toggle_inline`/`set_block`,
11//! `toggle_block_container`/`insert_link` (P5)
12//!
13//! twig reparses after every edit and leaves everything outside the splice
14//! byte-for-byte untouched, so the document stays a live, navigable AST while
15//! you type into it.
16
17// `PathBuf` names the `path` field and the untitled marker on every build;
18// `Path` is only touched by the filesystem I/O gated behind the `fs` feature.
19// The docs in this file lay their `- key → meaning` lists out in aligned
20// columns, which puts a continuation line further right than clippy's
21// list-indent rule likes. A lazy continuation renders as the same paragraph
22// either way, and the alignment is what makes those tables readable, so the
23// layout wins over the lint.
24#![allow(clippy::doc_overindented_list_items)]
25
26use std::collections::HashMap;
27use std::ops::Range;
28#[cfg(feature = "fs")]
29use std::path::Path;
30use std::path::PathBuf;
31
32#[cfg(feature = "fs")]
33use anyhow::Context;
34use anyhow::{Result, anyhow};
35use twig::{
36 Alignment, BlockContainerKind, BlockKind, Change, Editor, FlatNode, Format, Gesture,
37 InlineKind, Kind, MarkdownExtensions, NodeId, QueryMatch,
38};
39use unicode_segmentation::GraphemeCursor;
40
41use crate::html;
42use crate::source::{self, SourceMap};
43use crate::style::{Align, FontFamily, LineSpacing, MarkColor, SizeStep};
44use crate::wysiwyg::{self, MediaKind, MediaStop, VisualMap};
45
46/// Which view the body shows.
47#[derive(Clone, Copy, PartialEq, Eq, Debug)]
48pub enum View {
49 /// The raw document with a caret in source bytes.
50 Source,
51 /// Markup resolved to real styles, caret riding the rendered glyphs.
52 Wysiwyg,
53}
54
55/// How much of the source markup the WYSIWYG view exposes — a per-editor
56/// preference, orthogonal to [`View`]. Named for markup rather than for Markdown
57/// because leaf is grammar-agnostic: twig hands it Djot, HTML and XML on the same
58/// terms, and every rung below is about *delimiters*, whatever grammar spells
59/// them. The examples are Markdown only because that is what most documents are.
60///
61/// A single ladder over two underlying axes, because only three of their four
62/// combinations are coherent:
63///
64/// | | authoring off | authoring on |
65/// |---|---|---|
66/// | delimiters hidden | [`None`](Self::None) | [`Shortcuts`](Self::Shortcuts) |
67/// | caret line revealed | *incoherent* | [`Full`](Self::Full) |
68///
69/// The empty quadrant would show delimiters on the caret's line and then escape
70/// the ones you type — a surface that displays a syntax it refuses to accept.
71/// Someone who wants to read raw markup without authoring it has
72/// [`View::Source`], which is the better tool for it.
73///
74/// The two axes are read separately by the code that cares — see
75/// [`reveals_caret_line`](Self::reveals_caret_line) and
76/// [`authors`](Self::authors) — so neither behaviour has to know it's spelled
77/// as a ladder.
78#[derive(Clone, Copy, PartialEq, Eq, Debug, Default)]
79pub enum MarkupMode {
80 /// Delimiters stay hidden even on the caret's line, and typed syntax stays
81 /// literal — twig escapes anything that would open markup, so formatting
82 /// comes from commands (⌘b, the toolbar) instead of from spelling. The clean
83 /// reading surface for people who don't write markup by hand; the default,
84 /// and what Diaryx ships.
85 #[default]
86 None,
87 /// Delimiters stay hidden, but typing them authors real markup: `*x*`
88 /// becomes italic and the asterisks disappear into the styling
89 /// (Typora/Bear-shaped). For someone who knows the syntax but wants the
90 /// clean surface back once it has been applied.
91 Shortcuts,
92 /// The caret's line shows its raw markup while every other line renders
93 /// resolved (Obsidian live-preview-shaped), and typed syntax authors markup
94 /// — for people fluent in the document's grammar who want to see and edit
95 /// the delimiters they type.
96 Full,
97}
98
99impl MarkupMode {
100 /// Whether the rich view shows raw delimiters on the line holding the caret.
101 /// The rendering axis — read by [`Doc::reveal_line`] and threaded into the
102 /// WYSIWYG builder.
103 pub fn reveals_caret_line(self) -> bool {
104 matches!(self, MarkupMode::Full)
105 }
106
107 /// Whether typed markup characters author real formatting. The editing axis
108 /// — read by [`Doc::insert`], which escapes typed syntax when this is false.
109 pub fn authors(self) -> bool {
110 !matches!(self, MarkupMode::None)
111 }
112}
113
114/// How the WYSIWYG view treats a *soft break* — a bare newline inside a
115/// paragraph. An axis of its own, orthogonal to [`MarkupMode`] (which governs
116/// inline-markup delimiters) and to [`View`]: any reveal preference pairs with
117/// either flow. The renderer consults it when it lays a block's inline content
118/// into visual rows.
119#[derive(Clone, Copy, PartialEq, Eq, Debug, Default)]
120pub enum LineFlow {
121 /// A soft break folds into a space and the paragraph reflows to the
122 /// viewport width — flowing prose, where the source's line wrapping is
123 /// insignificant. The default, and what Diaryx ships.
124 #[default]
125 Fold,
126 /// A soft break renders as a line break exactly where it was written, so
127 /// the author's source line structure shows on screen unchanged — the mode
128 /// for people who lay out their prose deliberately (one sentence or clause
129 /// per line, semantic line breaks). The break is still a soft break in the
130 /// source; only its rendering changes.
131 Preserve,
132}
133
134/// What the file behind a document looks like right now, against the bytes leaf
135/// last read from it or wrote to it — the question a frontend asks before it
136/// saves (a `Changed` file plus a `dirty` document is an overwrite about to
137/// happen) or when its window regains focus. See [`Doc::disk_state`].
138#[derive(Clone, Copy, Debug, PartialEq, Eq)]
139pub enum DiskState {
140 /// The file holds exactly the bytes leaf last read or wrote.
141 Unchanged,
142 /// Someone else wrote the file since. Saving overwrites their work; see
143 /// [`Doc::reload`] for the other direction.
144 Changed,
145 /// The file is gone — deleted or renamed away. A save recreates it.
146 Missing,
147 /// There is a path, but the file couldn't be read (permissions, a directory
148 /// in the way): leaf can't tell, and won't guess.
149 Unreadable,
150 /// No file behind this document yet — see [`Doc::blank`]. Nothing can have
151 /// changed under a document that was never on disk.
152 Untitled,
153}
154
155/// The inline marks in force at a point in the document — what a toolbar
156/// lights up. A `Copy` bitset rather than a `HashSet`, because
157/// [`Doc::active_inline_marks`] is called on every frame that draws a toolbar
158/// and a set that allocates to answer "is Bold on?" is a set that shouldn't.
159#[derive(Clone, Copy, Debug, Default, PartialEq, Eq)]
160pub struct InlineMarks(u8);
161
162impl InlineMarks {
163 /// Every kind, in the order [`InlineMarks::iter`] yields them.
164 const ALL: [InlineKind; 8] = [
165 InlineKind::Strong,
166 InlineKind::Emph,
167 InlineKind::Verbatim,
168 InlineKind::Mark,
169 InlineKind::Superscript,
170 InlineKind::Subscript,
171 InlineKind::Insert,
172 InlineKind::Delete,
173 ];
174
175 pub const fn empty() -> Self {
176 InlineMarks(0)
177 }
178
179 /// Private: the set is an *answer*, and adding a mark to it doesn't mark
180 /// anything ([`Doc::toggle`] does that). `FromIterator` is the way in.
181 fn insert(&mut self, kind: InlineKind) {
182 self.0 |= Self::bit(kind);
183 }
184
185 /// Flip `kind` in the set — the sticky-marks toggle at a collapsed caret.
186 fn flip(&mut self, kind: InlineKind) {
187 self.0 ^= Self::bit(kind);
188 }
189
190 /// The symmetric difference: which marks differ between the two sets. Used
191 /// to resolve the marks already in force at the caret against the pending
192 /// delta — a bit set in the delta flips the base mark for the next keystroke.
193 fn xor(self, other: InlineMarks) -> InlineMarks {
194 InlineMarks(self.0 ^ other.0)
195 }
196
197 /// Whether `kind` is in force — the toolbar's "is Bold active?".
198 pub fn contains(self, kind: InlineKind) -> bool {
199 self.0 & Self::bit(kind) != 0
200 }
201
202 pub fn is_empty(self) -> bool {
203 self.0 == 0
204 }
205
206 /// The marks in force, for a frontend that renders whatever is on rather
207 /// than asking after a fixed list.
208 pub fn iter(self) -> impl Iterator<Item = InlineKind> {
209 Self::ALL.into_iter().filter(move |&k| self.contains(k))
210 }
211
212 fn bit(kind: InlineKind) -> u8 {
213 1 << match kind {
214 InlineKind::Strong => 0,
215 InlineKind::Emph => 1,
216 InlineKind::Verbatim => 2,
217 InlineKind::Mark => 3,
218 InlineKind::Superscript => 4,
219 InlineKind::Subscript => 5,
220 InlineKind::Insert => 6,
221 InlineKind::Delete => 7,
222 }
223 }
224}
225
226impl FromIterator<InlineKind> for InlineMarks {
227 fn from_iter<I: IntoIterator<Item = InlineKind>>(iter: I) -> Self {
228 let mut m = InlineMarks::empty();
229 for k in iter {
230 m.insert(k);
231 }
232 m
233 }
234}
235
236/// What kind of edit produced an undo group. Same-kind edits in a row coalesce
237/// into one undo step (a run of typed characters undoes together); `Other` never
238/// coalesces, so a paste, format toggle, or block change is always its own step.
239#[derive(Clone, Copy, PartialEq, Eq)]
240enum EditKind {
241 Insert,
242 Delete,
243 /// One step of an IME composition — see [`Doc::edit_composing`]. Its own kind
244 /// rather than `Insert`'s because a composition is not typing: each step
245 /// *replaces* the last (`か` → `かん` → `感`), so the run has to coalesce even
246 /// though no two steps insert the same bytes, and it must not fold into the
247 /// typed characters on either side of it.
248 Compose,
249 Other,
250}
251
252/// Which side of the caret a delete looks for an in-cell `<br>` break to swallow
253/// whole — see [`Doc::cell_break_at`]. `Backward` is Backspace (a break ending at
254/// the caret), `Forward` is Delete (one starting at it).
255#[derive(Clone, Copy)]
256enum BreakEdge {
257 Backward,
258 Forward,
259}
260
261/// A re-spelling of one inline mark run, held ready in case the edit about to
262/// happen breaks it — see [`Doc::mark_edge_fix`] and [`Doc::repair_mark_edges`].
263/// Every offset in it is in the coordinates the document will have *after* the
264/// plain edit, since that is when it may be applied.
265struct MarkEdgeFix {
266 /// The run's kind, and an offset inside what was its content: together they
267 /// answer "did the plain edit actually break this mark?" — the question that
268 /// decides whether any of this is applied at all.
269 kind: InlineKind,
270 probe: usize,
271 /// The byte range to re-spell (the run's delimiters included) and its new
272 /// spelling, with the edge whitespace moved outside the delimiters.
273 start: usize,
274 end: usize,
275 text: String,
276 /// Where the caret belongs afterwards — the same place on screen it would
277 /// have had, which is now on the other side of a delimiter.
278 caret: usize,
279 /// The marks in force for text typed at that caret. The caret can land
280 /// outside a run it was inside, and the marks have to survive the move or
281 /// the toolbar goes dark mid-word.
282 want: InlineMarks,
283}
284
285/// The caret and selection at one moment — the part of a history step twig's
286/// `Change` cannot carry, because the caret is leaf's state and twig only knows
287/// about bytes. leaf serializes it into the opaque per-state blob twig now
288/// stores in its own undo history (see `record_caret`), so undo and redo hand
289/// back the caret that matches the source they restore.
290#[derive(Clone, Copy)]
291struct CaretState {
292 caret: usize,
293 anchor: Option<usize>,
294}
295
296impl CaretState {
297 /// Pack into the fixed 17-byte blob leaf hands twig: the caret as a u64,
298 /// then an anchor-present flag and the anchor. twig copies these bytes and
299 /// never reads them.
300 fn to_blob(self) -> [u8; 17] {
301 let mut b = [0u8; 17];
302 b[..8].copy_from_slice(&(self.caret as u64).to_le_bytes());
303 if let Some(a) = self.anchor {
304 b[8] = 1;
305 b[9..].copy_from_slice(&(a as u64).to_le_bytes());
306 }
307 b
308 }
309
310 /// Recover a state from twig's blob, or `None` when it is empty or the wrong
311 /// length — a state twig restored that never had a caret set on it, which
312 /// leaves the caller to fall back to the edit site.
313 fn from_blob(b: &[u8]) -> Option<Self> {
314 let b: &[u8; 17] = b.try_into().ok()?;
315 let caret = u64::from_le_bytes(b[..8].try_into().unwrap()) as usize;
316 let anchor = (b[8] != 0).then(|| u64::from_le_bytes(b[9..].try_into().unwrap()) as usize);
317 Some(CaretState { caret, anchor })
318 }
319}
320
321/// A footnote reference and the note it names — the answer to
322/// [`Doc::footnote_at`].
323///
324/// The two `Option`s move together: a reference whose definition is missing has
325/// neither a body to show nor a place to jump to, and one that resolved has
326/// both.
327#[derive(Clone, PartialEq, Eq, Debug)]
328pub struct FootnoteRef {
329 /// The reference's label — the `1` of `[^1]`, with neither the `^` that
330 /// spells it a footnote nor the brackets around it.
331 pub label: String,
332 /// The note's body as source bytes (see
333 /// [`wysiwyg::footnote_body_span`](crate::wysiwyg)), or `None` when the
334 /// document defines no `[^label]:` to read one from.
335 pub text: Option<String>,
336 /// Where the note's *body* starts, for a "go to note" that moves the caret
337 /// there. `None` alongside a `None` `text`.
338 ///
339 /// The body rather than the definition, because this is an offset to put a
340 /// caret on and the `[^1]:` marker is decoration the caret can't occupy —
341 /// aiming at the definition's first byte snaps to the nearest real stop,
342 /// which is up in the paragraph above the note. It is also simply where a
343 /// reader following a reference wants to land: at the note's first word,
344 /// ready to read or amend it.
345 pub offset: Option<usize>,
346 /// Where the note's body ends, exclusive — so a frontend can ask which
347 /// *rendered rows* the note occupies and draw those instead of [`text`](Self::text).
348 ///
349 /// The rows are the note with its markup resolved: `see *later*` reaches a
350 /// frontend as an italic run, not as asterisks. `text` is the source bytes
351 /// and stays the honest answer for anything that wants the note as written
352 /// (a search index, a copy); this pair of offsets is for anything that wants
353 /// it as *read*. `None` alongside a `None` `offset`.
354 pub end: Option<usize>,
355}
356
357/// A footnote definition and the reference that sends a reader to it — the
358/// answer to [`Doc::footnote_definition_at`], and the other half of the round
359/// trip [`FootnoteRef`] starts.
360///
361/// A note is a place a reader *arrives*, so the useful thing to know while
362/// standing in one is the way back. Without this the jump to a note is a
363/// one-way door: the definitions sit at the foot of the document, so returning
364/// by hand means scrolling back up and finding the sentence again.
365#[derive(Clone, PartialEq, Eq, Debug)]
366pub struct FootnoteDef {
367 /// The definition's label — the `1` of `[^1]: …`, marker and colon stripped,
368 /// spelled exactly as [`FootnoteRef::label`] spells the same footnote's.
369 pub label: String,
370 /// Where the reference's *label* is, for a "back to reference" that moves
371 /// the caret there. `None` for a note nothing refers to — an orphan, which
372 /// is worth being able to say rather than silently doing nothing.
373 ///
374 /// The label rather than the reference's first byte, for
375 /// [`FootnoteRef::offset`]'s reason: a reference's brackets are decoration
376 /// and its label is the only part of it the caret can rest on.
377 ///
378 /// The *first* reference, when a label is cited more than once: a repeated
379 /// citation has no one true home, and the first is both the one a reader
380 /// most likely came from and the only choice that doesn't depend on how
381 /// they got here.
382 pub offset: Option<usize>,
383}
384
385/// Where a locator lands — the answer to [`Doc::locate`].
386///
387/// A locator (the `v2` of a `chapter.dj#v2`) names a *place* rather than a
388/// document, and a place is a span rather than a point: a reader following one
389/// wants the caret at its first byte, and a reader merely *peeking* at one wants
390/// the block it covers drawn. Both are served by carrying the whole span, and
391/// only one of the two can be recovered from an offset alone.
392#[derive(Clone, PartialEq, Eq, Debug)]
393pub struct Landing {
394 /// The first byte of the block the locator names — where a caret goes.
395 pub start: usize,
396 /// One past its last byte, so a frontend can map the pair through
397 /// [`VisualMap::row_range_for`](crate::wysiwyg::VisualMap::row_range_for) to the rendered rows the block occupies and draw
398 /// those, the way a footnote peek draws a note ([`FootnoteRef::end`]).
399 pub end: usize,
400}
401
402/// A selection cited out of the source: the text itself, up to a requested
403/// number of characters either side, and the byte range it came from. See
404/// [`Doc::selection_quote`].
405///
406/// The prefix and suffix are what make the quote *re-findable*: the same text
407/// can occur twice, and a little of what surrounded it is how a later reader —
408/// or the same document after an edit — tells the occurrences apart. The Web
409/// Annotation model calls this a `TextQuoteSelector`; the shape is older than
410/// the name.
411#[derive(Debug, Clone, PartialEq, Eq)]
412pub struct Quote {
413 /// The selected source, verbatim.
414 pub exact: String,
415 /// What immediately preceded it — possibly empty, at the document's start.
416 pub prefix: String,
417 /// What immediately followed it — possibly empty, at the document's end.
418 pub suffix: String,
419 /// Byte offset in the source where the selection begins.
420 pub start: usize,
421 /// Byte offset where it ends (exclusive).
422 pub end: usize,
423}
424
425/// A host-painted range of the source — an annotation's footprint, a search
426/// hit, a reviewer's mark. Leaf renders it (a background wash behind the
427/// glyphs whose source falls inside it) and hands back the `id` when the
428/// reader activates it; what the range *means* is entirely the host's.
429///
430/// Ranges are source bytes, like the caret and the selection, so a host that
431/// anchors quotes against the source ([`Doc::selection_quote`] is the other
432/// half of that loop) can paint what it found without any coordinate
433/// conversion. A range that drifts off the text it meant is the host's to
434/// re-anchor; leaf draws what it is told.
435#[derive(Debug, Clone, PartialEq, Eq)]
436pub struct Highlight {
437 /// Byte offset in the source where the wash begins.
438 pub start: usize,
439 /// Byte offset where it ends (exclusive).
440 pub end: usize,
441 /// The host's name for it, handed back on activation. Opaque to leaf.
442 pub id: String,
443 /// A rendering hint the frontend maps — a `#RRGGBB` hex string, or
444 /// nothing for the theme's default wash.
445 pub color: Option<String>,
446 /// A margin glyph's name, or nothing for wash-only ink. A highlight with
447 /// a marker gets a small glyph in the margin beside its first line, and
448 /// the glyph — not the wash — is what activates it: the wash is ink, the
449 /// marker is the control, which is what lets a reader put a caret in (or
450 /// copy from) annotated text without a card leaping at them. The name is
451 /// opaque to leaf; an Apple frontend reads it as an SF Symbol, a web one
452 /// as a class.
453 pub marker: Option<String>,
454}
455
456impl Highlight {
457 /// The range covering source `offset` in a list [`Doc::set_highlights`]
458 /// sorted, first by start where several overlap — the one place that
459 /// question is answered, for the frontends that paint by asking it as well
460 /// as for [`Doc::highlight_at`].
461 ///
462 /// The list is sorted by `(start, end)`, so the scan can stop at the first
463 /// range starting past `offset` rather than running to the end. A painter
464 /// asking once per glyph wants [`HighlightCursor`] instead; this is the
465 /// one-shot form, for the host asking what the reader just activated.
466 pub fn covering(highlights: &[Highlight], offset: usize) -> Option<&Highlight> {
467 highlights
468 .iter()
469 .take_while(|h| h.start <= offset)
470 .find(|h| offset < h.end)
471 }
472}
473
474/// [`Highlight::covering`] for a caller walking the document in order — which
475/// is every painter, since a frontend draws rows top to bottom and glyphs left
476/// to right.
477///
478/// The one-shot form is a scan from the front of the list per glyph, and a
479/// document with two hundred search hits pays that two hundred times a row. A
480/// range that ends at or before an offset can never cover that offset *or any
481/// later one*, so the cursor retires those permanently and each glyph costs the
482/// ranges that actually reach it. The answer is identical to
483/// [`Highlight::covering`]'s, offset for offset — this is the same scan with
484/// the part that was being redone dropped, not a cheaper approximation.
485///
486/// Offsets are expected to arrive non-decreasing. One that goes backwards is
487/// still answered correctly: the cursor re-seats to the front, since a painter
488/// that revisits a row is asking a question the retired ranges may own again.
489pub struct HighlightCursor<'a> {
490 highlights: &'a [Highlight],
491 /// The first range not yet retired.
492 at: usize,
493 /// The last offset asked about, to notice a caller going backwards.
494 last: usize,
495}
496
497impl<'a> HighlightCursor<'a> {
498 pub fn new(highlights: &'a [Highlight]) -> Self {
499 HighlightCursor {
500 highlights,
501 at: 0,
502 last: 0,
503 }
504 }
505
506 /// The range covering `offset`, advancing the cursor past every range that
507 /// can no longer cover anything.
508 pub fn at(&mut self, offset: usize) -> Option<&'a Highlight> {
509 if offset < self.last {
510 self.at = 0;
511 }
512 self.last = offset;
513 while self
514 .highlights
515 .get(self.at)
516 .is_some_and(|h| h.end <= offset)
517 {
518 self.at += 1;
519 }
520 Highlight::covering(&self.highlights[self.at..], offset)
521 }
522}
523
524/// The identity of a built [`VisualMap`] — see [`Doc::visual_key`]. Opaque on
525/// purpose: the only useful question is whether two of them are the same map,
526/// and what is behind it — which `Doc` built it, and the (revision, wrap,
527/// reveal line) it was built from — is core's business.
528///
529/// The document is part of it because the rest is not unique to one: two
530/// documents opened at the same width are both at revision zero with no reveal
531/// line, and a frontend holding one copy of a map across the two would take
532/// the second's key for the first's and paint the wrong document.
533#[derive(Clone, PartialEq, Eq, Debug)]
534pub struct VisualKey(u64, Option<(u64, Option<usize>, Option<Range<usize>>)>);
535
536pub struct Doc {
537 editor: Editor,
538 pub format: Format,
539 pub path: PathBuf,
540 /// Current source, refreshed from the editor after every successful edit.
541 pub source: String,
542 /// The caret, as a byte offset into `source` (always on a char boundary).
543 pub caret: usize,
544 /// The selection's fixed end, if a selection is active; the moving end is
545 /// the caret. `None` means no selection.
546 pub anchor: Option<usize>,
547 pub dirty: bool,
548 pub status: Option<String>,
549 pub view: View,
550 /// Whether the document refuses to change — a *reading* surface over the
551 /// same rendering, selection, and navigation the editor has.
552 ///
553 /// Enforced here rather than by each frontend hiding its input paths,
554 /// because every mutation funnels through a few doors —
555 /// [`splice_exact`](Self::splice_exact), [`undo`](Self::undo),
556 /// [`redo`](Self::redo), and the handful of inserts that go to twig's own
557 /// verbs directly rather than through the splice (a typed literal, a link,
558 /// an image, a rule, a footnote, a cell's line break) — and guarded doors
559 /// are a guarantee where a frontend's suppressed keyboard is a hope. A
560 /// gated door reports exactly like a rolled-back splice, a path every
561 /// caller already handles. `a_read_only_document_refuses_every_door` is
562 /// the list; a new `self.editor.insert_*` call belongs on it.
563 read_only: bool,
564 /// The host-painted ranges, kept sorted by start — see [`Highlight`].
565 /// State like the selection rather than like the text: no edit history,
566 /// no dirty bit, redrawn from whatever the host last set.
567 highlights: Vec<Highlight>,
568 /// How much of the source markup the rich view exposes — a frontend preference (see
569 /// [`MarkupMode`]). Its two axes are read apart: the rendering one by
570 /// [`reveal_line`](Self::reveal_line), the editing one by
571 /// [`insert`](Self::insert).
572 markup_mode: MarkupMode,
573 /// Whether soft breaks fold into the reflowed paragraph or render where
574 /// they were written (see [`LineFlow`]) — an independent frontend
575 /// preference the WYSIWYG builder consults when it lays out a block.
576 line_flow: LineFlow,
577 /// The kind of the last edit, for coalescing: twig owns the undo *history*
578 /// (see `undo`/`redo`), but "what counts as one undo step" is a frontend-UX
579 /// call, so leaf decides when a run continues and tells twig to coalesce.
580 last_edit_kind: Option<EditKind>,
581 /// The inline marks the user has toggled *at a collapsed caret* with no
582 /// selection — "start typing bold here". Held as the XOR delta from the marks
583 /// already in force at [`pending_at`](Self::pending_at): a set bit means
584 /// "flip this kind for the next typed text", so it both turns a mark on where
585 /// none is (type into bold) and off where one already covers the caret (type
586 /// past the bold you're standing in). [`Doc::insert`] realises it onto the
587 /// freshly typed text and then clears it — a mark once realised is carried by
588 /// the caret sitting inside the run, not by this delta.
589 pending_marks: InlineMarks,
590 /// The caret offset [`pending_marks`](Self::pending_marks) applies to. The
591 /// delta is live only while the caret still stands here with no selection;
592 /// any motion or edit ([`move_to`](Self::move_to), a splice, a click) drops
593 /// it, so a toggled-but-never-typed format doesn't leak onto text elsewhere.
594 pending_at: Option<usize>,
595 /// The source as of the last open/save — `dirty` is `source != clean_source`,
596 /// so undoing back to the saved state correctly clears the modified flag.
597 clean_source: String,
598 /// A hash of the bytes leaf last read from `path` or wrote to it; `None`
599 /// while the document has no file behind it. [`Doc::disk_state`] compares
600 /// the file against this to catch an edit made *outside* leaf before a save
601 /// silently overwrites it — `clean_source` only knows what leaf itself did.
602 ///
603 /// A hash, not an mtime: mtime is the cheap answer and the wrong one — two
604 /// writes inside one filesystem timestamp tick are indistinguishable, a
605 /// clock that steps backwards (or a writer that restores an mtime) hides a
606 /// real change, and a `touch` invents one. The whole point of the watermark
607 /// is to not clobber someone's work, so it reads the bytes and compares what
608 /// is actually there. That costs a file read per question, which is why the
609 /// question is asked on a user event (focus, save) and not every frame.
610 disk_hash: Option<u64>,
611 /// The "sticky" display column vertical motion aims for, in the active
612 /// view's grid. Set on the first `move_up`/`move_down` of a run and
613 /// reused by every subsequent one in that run, so passing through a
614 /// shorter line doesn't permanently forget the original column. Any
615 /// horizontal motion or edit clears it.
616 ///
617 /// A column, not a character index: dropping down a line of `你好` onto one
618 /// of ASCII has to land under the glyph the caret was drawn beneath, which
619 /// is the only thing the user can see to aim by. Where the goal falls inside
620 /// a wide character on the target line, the mapping resolves it to that
621 /// character — the caret lands on it rather than between its cells.
622 goal_col: Option<usize>,
623 /// The rendered map for the WYSIWYG view; empty in the source view. Movement
624 /// and clicks read it to stay in visible space.
625 pub vmap: VisualMap,
626 /// The syntax map for the source view; empty in the WYSIWYG view, which
627 /// styles resolved glyphs instead. Built by [`Doc::build_source`] — a
628 /// frontend that never calls it paints raw source unstyled, which is what
629 /// every frontend did before this map existed.
630 pub smap: SourceMap,
631 /// The revision `smap` was built from, or `None` before the first build.
632 /// The map is a pure function of the text alone — no width, no caret, no
633 /// reveal line — so unlike [`vmap_key`](Self::vmap_key) the revision is the
634 /// whole key.
635 smap_key: Option<u64>,
636 /// Everything the map is built from, as one number: bumped whenever the
637 /// document's text changes, and never by a motion, a selection, or a save.
638 /// A frontend can hold work against it — see [`Doc::revision`].
639 revision: u64,
640 /// How many history steps stand behind the caret, and how many ahead of
641 /// it — the answer to a native Edit menu's "may Undo be enabled?", which
642 /// twig's history does not ask itself. Counted at [`refresh`](Self::refresh),
643 /// the funnel every edit comes through, and moved back and forth by
644 /// [`undo`](Self::undo)/[`redo`](Self::redo). An upper bound rather than
645 /// an exact depth: a coalesced run of typing is one of twig's steps but
646 /// several of these, and twig's own cap on history is not mirrored here.
647 /// Neither error can make `can_undo` false while a step remains, which is
648 /// the only property a menu needs; the one place the bound can be wrong the
649 /// other way — the cap has retired every step — is reconciled the moment
650 /// twig reports nothing to undo.
651 undo_steps: usize,
652 redo_steps: usize,
653 /// What `vmap` was built from, or `None` before the first build. The map is
654 /// a pure function of `(revision, wrap, reveal line)`, so when those haven't
655 /// moved, rebuilding it produces the identical map — see
656 /// [`Doc::build_visual`].
657 ///
658 /// The reveal line ([`Doc::reveal_line`]) is the caret's, and is `None` in
659 /// every mode but [`MarkupMode::Full`] — so outside that mode the key is
660 /// text and width alone, and a caret motion still rebuilds nothing.
661 vmap_key: Option<(u64, Option<usize>, Option<Range<usize>>)>,
662 /// Which `Doc` this is, distinct from every other one built in this
663 /// process. Folded into [`VisualKey`] so that a map stashed by a frontend
664 /// can never be mistaken for another document's — see
665 /// [`Doc::visual_key`]. Nothing else reads it.
666 identity: u64,
667 /// Per-block row cache backing the incremental rebuild: when the text
668 /// changes, only the top-level blocks whose bytes moved are re-rendered and
669 /// the rest are reused shifted (see [`wysiwyg::BlockCache`]). Persists across
670 /// builds; a pure accelerator, so it's never read for correctness.
671 block_cache: wysiwyg::BlockCache,
672 /// How many visual rows each block image reserves, keyed by its destination —
673 /// set by the frontend through [`Doc::set_media_rows`] once it has decoded and
674 /// measured the pictures. Core does no image I/O, so this is the only way it
675 /// learns a picture's height; a destination not in the map reserves the bare
676 /// one-row placeholder. Threaded into the builder so [`wysiwyg::build_cached`]
677 /// sizes each placeholder, and folded into `vmap_key` so a height change
678 /// rebuilds the map.
679 media_rows: HashMap<String, usize>,
680
681 // View geometry the renderer stamps each frame, so mouse events can map a
682 // screen cell back to a byte offset.
683 pub scroll: usize,
684 pub body_origin: (u16, u16),
685 /// Width of the body rectangle last painted by the frontend. Zero means
686 /// unknown (used by tests or a frontend that has not drawn yet).
687 pub body_width: u16,
688 pub body_height: u16,
689 /// The caret as of the last frame drawn, or `None` before the first.
690 ///
691 /// Scrolling is the viewport's business, not the caret's: the view follows
692 /// the caret when the caret *moves*, but a wheel that doesn't touch the
693 /// caret has to be free to scroll away from it — otherwise the view is
694 /// pinned to the caret and stops dead at the edge of the document you can
695 /// see. Comparing against this is what tells the two apart, and it catches a
696 /// caret set by any route, including a frontend assigning the field itself.
697 pub drawn_caret: Option<usize>,
698}
699
700/// The Markdown extensions every leaf document is parsed with — four of them,
701/// each departing from twig's defaults for a reason leaf can state.
702///
703/// `html_elements` promotes embedded raw HTML (`<img>`, `<picture>`,
704/// `<source>`, …) into semantic AST nodes, so a picture becomes a real `image`
705/// node the frontends can frame and rasterize instead of opaque `raw_block`
706/// text. `directives` turns on generic `:::name{.class}` fenced-div containers
707/// (`directive` nodes), which a host app uses for its own semantics (diaryx's
708/// `:::vis{.audience}` visibility blocks) — core renders any directive as a
709/// plain tinted container, agnostic of `name`.
710///
711/// `highlight` and `highlight_colors` are the pair that makes Markdown read
712/// `==text==` as a `mark` node, and `==🔴 text==` as one carrying a
713/// `data-color`. leaf already had somewhere to put both: the
714/// [`Mark`](crate::Role::Mark) role and the ⌘⇧M highlight button predate them,
715/// and until twig 3.3 a Markdown document could only ever *receive* a highlight
716/// from a Djot one it was converted from — the button wrote `==…==` and the
717/// reparse read it straight back as text.
718/// They are on together because a colour is inert without the highlight itself,
719/// and a document that writes `==🔴 x==` means the colour by it.
720///
721/// Every flag is inert for non-Markdown formats, so it's safe to pass them
722/// unconditionally. Threading this through every constructor (not just `open`)
723/// keeps `from_source`, `blank`, and `reload` parsing the same document the same
724/// way — twig reparses with these same flags after each edit.
725pub(crate) fn parse_extensions() -> MarkdownExtensions {
726 MarkdownExtensions {
727 html_elements: true,
728 directives: true,
729 highlight: true,
730 highlight_colors: true,
731 ..Default::default()
732 }
733}
734
735/// Build an editor over `bytes` in `format` with leaf's [`parse_extensions`],
736/// mapping twig's error into the `anyhow` context every constructor shares.
737fn new_editor(bytes: &[u8], format: Format) -> Result<Editor> {
738 Editor::new_ext(bytes, format, parse_extensions()).map_err(|e| anyhow!("twig parse: {e}"))
739}
740
741/// Does `format` spell a table as a **pipe table** — the one grid twig's table
742/// editor knows how to emit?
743///
744/// This is the single capability leaf still has to answer for itself, and the
745/// only hand-maintained format list left in this file. Every other gesture is
746/// [`Format::supports`], which is twig's own answer read across the C ABI — but
747/// twig deliberately leaves the table ops out of that query, because they read
748/// no `Syntax` table at all. They rewrite a grid that is already in the source
749/// and refuse on *position*, never on format. Handed a caret inside an HTML
750/// `<table>`, `table_insert_row` therefore re-emits the whole element as
751/// `| a | b |` and reports success — a real splice, a clean reparse, an honest
752/// `dirty` flag, and nothing downstream able to tell it from a good edit.
753///
754/// So the list is narrow on purpose. `Format` is `#[non_exhaustive]`, and the
755/// wildcard answers "no" for a format leaf has never heard of: a new twig
756/// language that *does* spell pipe tables loses its grid controls until this
757/// line is updated, which shows up as a missing button. The other default hands
758/// it to [`Doc::table_op`], which rewrites documents it cannot spell.
759fn spells_pipe_tables(format: Format) -> bool {
760 matches!(format, Format::Markdown | Format::Djot)
761}
762
763/// Which of leaf's authoring controls this document's format can actually
764/// spell — one flag per toolbar button, resolved once so a frontend can build
765/// its chrome instead of discovering each refusal on a click.
766///
767/// Every field but [`table`](Self::table) is `Format::supports_with` on the
768/// gesture the matching [`Doc`] method calls, so this record cannot drift from
769/// what the ops do; `table` is [`spells_pipe_tables`], the one answer twig
770/// doesn't export.
771///
772/// `supports_with` rather than `supports` because two of these are facts about
773/// the *parse options* as much as about the format. `Format::supports` answers
774/// for twig's defaults, and leaf never parses with those — it parses with
775/// [`parse_extensions`], and a document's toolbar has to describe the document
776/// it is over. A Markdown editor holding `highlight` authors `==text==`; one
777/// without it would mint bytes its own reparse hands back as plain text, which
778/// is why twig asks before it writes.
779///
780/// **The formats are ragged, and that is the point.** A single per-document
781/// boolean was enough while the two authorable formats were Markdown and djot
782/// and everything else spelled nothing. HTML is neither: it writes seven of the
783/// eight inline marks as a tag pair, plus `<code>`, `<hr>`, an in-cell
784/// `<br>`, and — since twig 3.4 — a heading or paragraph rebuilt as its tag
785/// pair, and since 3.5 a quote, a list, a code block, a link and an image
786/// printed as fresh nodes; it spells no task box (a form control there) and
787/// no footnote, and its `<table>` is one twig reads but will not write. So
788/// ⌘B, ⌘1 and the quote button work in an HTML document and the task and
789/// table buttons do not, and no one flag can say that. Markdown and djot
790/// differ from each other too:
791/// `^superscript^` is djot-only, and an in-cell `<br>` is Markdown-only.
792#[derive(Clone, Copy, Debug, Eq, PartialEq)]
793pub struct Capabilities {
794 /// ⌘B — `InlineKind::Strong`.
795 pub bold: bool,
796 /// ⌘I — `InlineKind::Emph`.
797 pub italic: bool,
798 /// Inline code — `InlineKind::Verbatim`.
799 pub code: bool,
800 /// Highlight — `InlineKind::Mark`. Djot spells it, and so does Markdown
801 /// under the `highlight` extension [`parse_extensions`] turns on: the
802 /// button writes `==text==`, which is what the reparse reads back.
803 pub mark: bool,
804 /// ⌘U — `InlineKind::Insert`, which every format that marks at all spells.
805 pub underline: bool,
806 /// Strikethrough — `InlineKind::Delete`. Markdown spells GFM's `~~text~~`
807 /// out of the box, since twig parses it out of the box.
808 pub strike: bool,
809 /// The highlight *palette* — [`Doc::set_mark_color`]. Narrower than
810 /// [`mark`](Self::mark) and deliberately its own flag: Markdown spells a
811 /// colour on a highlight (`==🔴 text==`) and djot spells only the highlight,
812 /// so a toolbar offering the swatches wherever the button lights would offer
813 /// them in a document that cannot write one. Pair with
814 /// [`Doc::caret_in_mark`], which asks the other question — the palette needs
815 /// a highlight to colour as much as a format that spells one.
816 pub mark_color: bool,
817 pub superscript: bool,
818 pub subscript: bool,
819 /// Heading levels and "make this a paragraph" — [`Doc::set_block`].
820 pub heading: bool,
821 pub blockquote: bool,
822 pub bullet_list: bool,
823 pub ordered_list: bool,
824 /// The checkbox controls: giving an item a box, and ticking one.
825 pub task: bool,
826 pub link: bool,
827 /// Covers [`Doc::insert_media`] too — see the note there on why the three
828 /// media kinds stand or fall together.
829 pub image: bool,
830 /// The horizontal-rule button. HTML spells this one (`<hr>`).
831 pub thematic_break: bool,
832 /// The footnote button — [`Doc::insert_footnote`]. Markdown and djot spell
833 /// the pair; HTML has no footnote of its own, so the button goes away rather
834 /// than writing brackets that would render as brackets.
835 pub footnote: bool,
836 /// Setting a fenced block's language — a control only ever offered with the
837 /// caret already in a fence.
838 pub code_language: bool,
839 /// The grid controls: insert/delete/move a row or column, set a column's
840 /// alignment. Pair with [`Doc::caret_in_table`], which asks the other
841 /// question — an HTML `<table>` holds the caret and still can't be edited.
842 pub table: bool,
843 /// Shift+Return inside a cell. Markdown and HTML spell it; djot has no
844 /// idiomatic in-cell break.
845 pub cell_line_break: bool,
846 /// The alignment control — [`Doc::set_alignment`], twig's
847 /// `Gesture::SetBlockAttrs`. Every format leaf opens but XML spells a
848 /// block's attributes, Markdown under the `html_elements`
849 /// [`parse_extensions`] turns on (a `<div>` around the block) and AsciiDoc
850 /// through its `[…]` line.
851 pub alignment: bool,
852 /// The line-spacing menu — [`Doc::set_line_spacing`]. The same gesture as
853 /// [`alignment`](Self::alignment) and so the same answer, and its own flag
854 /// because a toolbar dims controls one at a time and the pair may yet
855 /// diverge.
856 pub line_spacing: bool,
857 /// The size menu — [`Doc::set_font_size`], twig's `Gesture::WrapRangeAttrs`
858 /// over a selection. **Narrower than the block pair**: AsciiDoc's
859 /// `[#id.role]#text#` keeps an id and a role and has no slot for a
860 /// `data-` key, so twig refuses the span there and this is `false` while
861 /// [`alignment`](Self::alignment) is `true`. The block-level form of the
862 /// same property — the caret in a paragraph, no selection — goes through
863 /// `SetBlockAttrs` and still works, which is why the flag describes the
864 /// control rather than the caret.
865 pub font_size: bool,
866 /// The face menu — [`Doc::set_font_family`]. `WrapRangeAttrs`, as
867 /// [`font_size`](Self::font_size) is.
868 pub font_family: bool,
869 /// The text-colour swatches — [`Doc::set_text_color`]. `WrapRangeAttrs`,
870 /// and not to be confused with [`mark_color`](Self::mark_color): that is a
871 /// highlight's background and rides the `mark` node twig already owns,
872 /// this is a run's foreground and rides an attributed span.
873 pub text_color: bool,
874 /// The page-break button — [`Doc::insert_page_break`], twig's
875 /// `Gesture::InsertDirective`. Markdown under the `directives` extension
876 /// [`parse_extensions`] turns on (`::page-break`) and djot, which spells
877 /// it as an empty `::: page-break` fence.
878 ///
879 /// **Those two and no others**, though twig spells the gesture in HTML and
880 /// AsciiDoc as well — see [`Capabilities::of`].
881 pub page_break: bool,
882}
883
884impl Capabilities {
885 /// Resolve every flag for `format`, as leaf parses it. Pure and cheap —
886 /// twig computes each from a static table — but a frontend that wants to
887 /// hold them can.
888 ///
889 /// The extensions are not a parameter because they are not a choice a
890 /// caller makes: every leaf document is parsed with [`parse_extensions`],
891 /// so the format is the whole of what varies.
892 pub fn of(format: Format) -> Self {
893 let exts = parse_extensions();
894 let supports = |g| format.supports_with(exts, g);
895 let inline = |k| supports(Gesture::ToggleInline(k));
896 let container = |k| supports(Gesture::ToggleBlockContainer(k));
897 Self {
898 bold: inline(InlineKind::Strong),
899 italic: inline(InlineKind::Emph),
900 code: inline(InlineKind::Verbatim),
901 mark: inline(InlineKind::Mark),
902 underline: inline(InlineKind::Insert),
903 strike: inline(InlineKind::Delete),
904 mark_color: supports(Gesture::SetMarkColor),
905 superscript: inline(InlineKind::Superscript),
906 subscript: inline(InlineKind::Subscript),
907 heading: supports(Gesture::SetBlock),
908 blockquote: container(BlockContainerKind::BlockQuote),
909 bullet_list: container(BlockContainerKind::BulletList),
910 ordered_list: container(BlockContainerKind::OrderedList),
911 // Both halves of the checkbox story, and leaf offers no control that
912 // needs only one: the item gesture mints the box, the checked one
913 // ticks it, and a format spelling a `task_marker` spells both.
914 task: supports(Gesture::ToggleTaskItem) && supports(Gesture::ToggleTaskChecked),
915 link: supports(Gesture::InsertLink),
916 image: supports(Gesture::InsertImage),
917 thematic_break: supports(Gesture::InsertThematicBreak),
918 footnote: supports(Gesture::InsertFootnote),
919 code_language: supports(Gesture::SetCodeLanguage),
920 table: spells_pipe_tables(format),
921 cell_line_break: supports(Gesture::InsertLineBreak),
922 // The presentation vocabulary, one gesture per level: the two
923 // line-level properties are a block's attributes and the three
924 // run-level ones a span's. They are asked separately because the
925 // formats answer differently — AsciiDoc spells the block and not
926 // the span — and a toolbar that dimmed all five together would dim
927 // three controls that work.
928 alignment: supports(Gesture::SetBlockAttrs),
929 line_spacing: supports(Gesture::SetBlockAttrs),
930 font_size: supports(Gesture::WrapRangeAttrs),
931 font_family: supports(Gesture::WrapRangeAttrs),
932 text_color: supports(Gesture::WrapRangeAttrs),
933 // Narrower than the gesture, on purpose. Twig spells
934 // `InsertDirective` in HTML and AsciiDoc too, and spells it
935 // *differently* there — `<page-break></page-break>` and `<<<` —
936 // and the walker reads only the two spellings above. An HTML page
937 // break draws as nothing at all (no row, no caret home) and an
938 // AsciiDoc one as an empty unlabelled row, so the button would
939 // write a break the author cannot see and cannot get back to.
940 // The proposal claims Markdown and djot, and this is that claim.
941 // Widening it is the walker's work, not this line's — see
942 // `docs/tasks/page-break-in-html-and-asciidoc.md`.
943 page_break: supports(Gesture::InsertDirective)
944 && matches!(format, Format::Markdown | Format::Djot),
945 }
946 }
947}
948
949/// The source of [`Doc::identity`], one per document ever built.
950static NEXT_IDENTITY: std::sync::atomic::AtomicU64 = std::sync::atomic::AtomicU64::new(0);
951
952impl Doc {
953 #[cfg(feature = "fs")]
954 pub fn open(path: PathBuf) -> Result<Self> {
955 let bytes = std::fs::read(&path).with_context(|| format!("reading {}", path.display()))?;
956 Self::from_disk_bytes(path, bytes)
957 }
958
959 /// An empty document *named* `path`, for a file that isn't there yet — what
960 /// every other terminal editor gives you when you name a file that doesn't
961 /// exist. It is a real named document, not a [`Doc::blank`]: `is_untitled`
962 /// is false, so ⌘S writes straight to `path` with no Save As detour, and
963 /// the header shows the name the user asked for.
964 ///
965 /// The format comes from the extension, exactly as [`Doc::open`] reads it —
966 /// so `leaf notes.dj` starts a djot buffer rather than the Markdown
967 /// [`Doc::blank`] has to assume for want of a name. An extension leaf can't
968 /// parse is still an error: a mistyped flag or a stray argument should say
969 /// so, not open a buffer promising to save somewhere.
970 ///
971 /// The watermark is the hash of *no bytes*, not `None`, and that is the
972 /// whole trick: `None` means untitled, and would leave [`Doc::disk_state`]
973 /// answering [`DiskState::Untitled`] for a document that has a path and
974 /// intends to write to it. Hashing `""` instead makes the answers the true
975 /// ones — [`DiskState::Missing`] while the file still isn't there (a save
976 /// recreates it, which is exactly what this is for), and
977 /// [`DiskState::Changed`] if somebody creates it underneath us between
978 /// launch and save, so the frontend's overwrite prompt guards a new file as
979 /// it guards an opened one.
980 ///
981 /// Nothing is written here. A buffer that is never typed into never touches
982 /// the filesystem, and a `path` whose directory doesn't exist is allowed to
983 /// open — the write is where that fails, and it says so then.
984 #[cfg(feature = "fs")]
985 pub fn create(path: PathBuf) -> Result<Self> {
986 Self::from_disk_bytes(path, Vec::new())
987 }
988
989 /// [`Doc::open`] when the file is there, [`Doc::create`] when it isn't —
990 /// the call a CLI frontend wants for its path argument.
991 ///
992 /// The decision is made from the failed read itself rather than a `exists()`
993 /// check first, so there is no window between the two for the file to appear
994 /// or vanish in. Only `NotFound` opens a new buffer: a permissions error or
995 /// a directory in the way is still an error, because pretending those are
996 /// "no file yet" would offer to save over something leaf couldn't read.
997 #[cfg(feature = "fs")]
998 pub fn open_or_create(path: PathBuf) -> Result<Self> {
999 match std::fs::read(&path) {
1000 Ok(bytes) => Self::from_disk_bytes(path, bytes),
1001 Err(e) if e.kind() == std::io::ErrorKind::NotFound => Self::create(path),
1002 Err(e) => Err(e).with_context(|| format!("reading {}", path.display())),
1003 }
1004 }
1005
1006 /// The shared body of [`Doc::open`] and [`Doc::create`]: bytes that are (or
1007 /// stand in for) the file at `path`, parsed as the format its extension
1008 /// names. Keeping the two on one path is what makes a new file's document
1009 /// identical in every respect to an opened one but its contents.
1010 #[cfg(feature = "fs")]
1011 fn from_disk_bytes(path: PathBuf, bytes: Vec<u8>) -> Result<Self> {
1012 let format = detect_format(&path)?;
1013 let editor = new_editor(&bytes, format)?;
1014 let source = String::from_utf8(bytes).map_err(|_| anyhow!("document is not UTF-8"))?;
1015 let disk_hash = Some(hash_bytes(source.as_bytes()));
1016 // Store the document's *absolute* path. A relative one (`leaf README.md`)
1017 // has an empty parent, so a frontend can't resolve a relative image
1018 // destination (``) against the document's directory and the
1019 // picture silently falls back to its text placeholder. `absolute` is
1020 // purely lexical — it prefixes the current directory and normalizes, but
1021 // reads nothing and resolves no symlinks — so `file_name` and save are
1022 // unchanged; it only gives `path.parent()` something to join against.
1023 let path = std::path::absolute(&path).unwrap_or(path);
1024 Ok(Doc::from_parts(editor, format, path, source, disk_hash))
1025 }
1026
1027 /// Build a document from an in-memory string, the format named explicitly —
1028 /// the portable, filesystem-free counterpart to [`Doc::open`] (which reads a
1029 /// path and sniffs the format from its extension). A wasm or FFI host, which
1030 /// has no path to read, uses this: it hands over bytes it fetched however it
1031 /// could, and later persists [`Doc::source`] however it can (a browser
1032 /// download, `localStorage`, a backend `PUT`) and calls [`Doc::mark_saved`].
1033 ///
1034 /// No file backs the result, so it starts untitled ([`Doc::is_untitled`] is
1035 /// true) exactly like a [`Doc::blank`] that has been given content.
1036 pub fn from_source(source: String, format: Format) -> Result<Self> {
1037 let editor = new_editor(source.as_bytes(), format)?;
1038 Ok(Doc::from_parts(
1039 editor,
1040 format,
1041 PathBuf::new(),
1042 source,
1043 None,
1044 ))
1045 }
1046
1047 /// An untitled, empty document — the `+` button and a `leaf` launched with
1048 /// no file argument. Nothing on disk backs it until a [`Doc::save_as`].
1049 ///
1050 /// It is Markdown, because a format has to be chosen before a name exists to
1051 /// read one from: `detect_format` reads the extension and an untitled
1052 /// document has neither. Markdown is what leaf's own files are, what its
1053 /// block markers are already written for (`insert_block_prefix`), and the
1054 /// extension a Save As will overwhelmingly pick — a wrong guess here would
1055 /// mean typing djot into a buffer parsing it as Markdown. Note that Save As
1056 /// *doesn't* revisit this: see [`Doc::save_as`].
1057 pub fn blank() -> Result<Self> {
1058 let format = Format::Markdown;
1059 let editor = new_editor(b"", format)?;
1060 // An empty `path` is the untitled marker (`path` is a public `PathBuf`
1061 // field two frontends already read; making it an `Option` to say this
1062 // would break both). `is_untitled` is the question to ask, not the
1063 // representation to copy.
1064 Ok(Doc::from_parts(
1065 editor,
1066 format,
1067 PathBuf::new(),
1068 String::new(),
1069 None,
1070 ))
1071 }
1072
1073 /// The fields every constructor agrees on, so `open` and `blank` can't drift
1074 /// apart in the ones neither of them has an opinion about.
1075 // `identity` is taken from a counter rather than from the `Doc`'s address,
1076 // which moves — a session that holds one is moved into and out of
1077 // containers freely, and an identity that changed with it would defeat the
1078 // one comparison it exists for.
1079 fn from_parts(
1080 editor: Editor,
1081 format: Format,
1082 path: PathBuf,
1083 source: String,
1084 disk_hash: Option<u64>,
1085 ) -> Self {
1086 Doc {
1087 editor,
1088 format,
1089 path,
1090 disk_hash,
1091 clean_source: source.clone(),
1092 source,
1093 caret: 0,
1094 anchor: None,
1095 dirty: false,
1096 status: None,
1097 read_only: false,
1098 highlights: Vec::new(),
1099 // leaf opens in the rich-text (WYSIWYG) view by default — the
1100 // markup-resolved surface is leaf's differentiator. Frontends can
1101 // still start in source view explicitly (e.g. a CLI flag), and ⌘e/⌥w
1102 // toggles at runtime.
1103 view: View::Wysiwyg,
1104 // `None` by default — the clean surface Diaryx ships, with typed
1105 // syntax kept literal; a markup-fluent frontend can climb the
1106 // ladder to `Shortcuts` or `Full`.
1107 markup_mode: MarkupMode::default(),
1108 // Fold by default — flowing prose that reflows to the viewport, the
1109 // behaviour every frontend had before this preference existed.
1110 line_flow: LineFlow::default(),
1111 last_edit_kind: None,
1112 pending_marks: InlineMarks::empty(),
1113 pending_at: None,
1114 goal_col: None,
1115 vmap: VisualMap::default(),
1116 smap: SourceMap::default(),
1117 // No map yet — the first `build_source` always builds.
1118 smap_key: None,
1119 revision: 0,
1120 undo_steps: 0,
1121 redo_steps: 0,
1122 // No map yet — the first `build_visual` always builds.
1123 vmap_key: None,
1124 identity: NEXT_IDENTITY.fetch_add(1, std::sync::atomic::Ordering::Relaxed),
1125 block_cache: wysiwyg::BlockCache::default(),
1126 media_rows: HashMap::new(),
1127 scroll: 0,
1128 body_origin: (0, 0),
1129 body_width: 0,
1130 body_height: 0,
1131 drawn_caret: None,
1132 }
1133 }
1134
1135 /// Whether this document has no file behind it yet — a [`Doc::blank`] that
1136 /// has never been saved. The question a ⌘S handler asks to know it should
1137 /// open a Save As picker instead ([`Doc::save`] won't guess a name), and the
1138 /// header asks to know the name it shows is a placeholder.
1139 pub fn is_untitled(&self) -> bool {
1140 self.path.as_os_str().is_empty()
1141 }
1142
1143 pub fn toggle_view(&mut self) {
1144 self.view = match self.view {
1145 View::Source => View::Wysiwyg,
1146 View::Wysiwyg => View::Source,
1147 };
1148 self.scroll = 0;
1149 self.status = None;
1150 // Entering WYSIWYG, the caret may be sitting in now-hidden frontmatter;
1151 // lift it to the first rendered offset.
1152 self.clamp_caret();
1153 }
1154
1155 /// The current markup-exposure preference (see [`MarkupMode`]).
1156 pub fn markup_mode(&self) -> MarkupMode {
1157 self.markup_mode
1158 }
1159
1160 /// Set the markup-exposure preference. Both of its axes take effect at
1161 /// once: the editing one on the next [`insert`](Self::insert), and the
1162 /// rendering one on the next build — which is why this drops the cached
1163 /// visual map and the per-block render cache, exactly as
1164 /// [`set_line_flow`](Self::set_line_flow) does.
1165 pub fn set_markup_mode(&mut self, mode: MarkupMode) {
1166 if self.markup_mode == mode {
1167 return;
1168 }
1169 self.markup_mode = mode;
1170 // Neither cache is keyed on the mode, and moving between `Full` and the
1171 // hidden modes changes every row the caret's line renders to — so
1172 // invalidate both explicitly.
1173 self.vmap_key = None;
1174 self.block_cache = wysiwyg::BlockCache::default();
1175 }
1176
1177 /// The source byte range of the line the caret sits on, when that line
1178 /// should render its raw delimiters — `None` in every mode and view that
1179 /// hides them, which is what the builder reads as "reveal nothing".
1180 ///
1181 /// A *source* line (newline to newline), not a visual row: a wrapped
1182 /// paragraph and a `LineFlow::Preserve` soft break both split one source
1183 /// line across several rows, and revealing half a delimiter pair because the
1184 /// other half wrapped would be worse than revealing neither. The range
1185 /// excludes the terminating newline and is empty-but-present on a blank
1186 /// line, which reveals nothing but still keys the caches correctly.
1187 ///
1188 /// Only in [`View::Wysiwyg`]: source view already shows every byte, so
1189 /// there is nothing there to reveal.
1190 pub(crate) fn reveal_line(&self) -> Option<Range<usize>> {
1191 if !self.markup_mode.reveals_caret_line() || self.view != View::Wysiwyg {
1192 return None;
1193 }
1194 Some(source_line_range(&self.source, self.caret))
1195 }
1196
1197 /// The current soft-break flow preference (see [`LineFlow`]).
1198 pub fn line_flow(&self) -> LineFlow {
1199 self.line_flow
1200 }
1201
1202 /// Set the soft-break flow preference. The mode changes how every block lays
1203 /// out, so a change drops the cached visual map and the per-block render
1204 /// cache, forcing the next [`build_visual`] to rebuild under the new flow.
1205 ///
1206 /// [`build_visual`]: Self::build_visual
1207 pub fn set_line_flow(&mut self, mode: LineFlow) {
1208 if self.line_flow == mode {
1209 return;
1210 }
1211 self.line_flow = mode;
1212 // Both caches are keyed on `(revision, wrap)`, neither of which moved —
1213 // so invalidate them explicitly, or the next build would reuse rows laid
1214 // out under the old flow.
1215 self.vmap_key = None;
1216 self.block_cache = wysiwyg::BlockCache::default();
1217 }
1218
1219 pub fn view_name(&self) -> &'static str {
1220 match self.view {
1221 View::Source => "source",
1222 View::Wysiwyg => "wysiwyg",
1223 }
1224 }
1225
1226 /// Rebuild the WYSIWYG visual map for the current tree at `width` columns
1227 /// (called by the renderer each frame it's in the WYSIWYG view).
1228 /// Build the WYSIWYG map, wrapped at `width` display columns.
1229 ///
1230 /// Cheap to call every frame, which is what both frontends do: the map is a
1231 /// pure function of the document and the wrap width, so a call that would
1232 /// rebuild the same map returns the one already built. Only an edit (or a
1233 /// resize) pays.
1234 ///
1235 /// That isn't a micro-optimisation. A frontend repaints for reasons that have
1236 /// nothing to do with the text — a blinking caret, a scroll, a focus change —
1237 /// and rebuilding here is O(document): 23 ms on a 1 MB file, of which 5 ms is
1238 /// marshalling twig's AST across the C ABI. Paid twice a second by the GUI's
1239 /// blink timer, that was 14% of a core spent redrawing an unchanged document.
1240 /// (`cargo run --release -p leaf-core --example bench` for the numbers.)
1241 pub fn build_visual(&mut self, width: usize) {
1242 self.build_map(Some(width));
1243 }
1244
1245 /// Build the WYSIWYG map with each block as a single unwrapped row — for a
1246 /// frontend (the GUI) that wraps at its own proportional pixel width rather
1247 /// than a fixed character column.
1248 pub fn build_visual_unwrapped(&mut self) {
1249 self.build_map(None);
1250 }
1251
1252 /// Build the source view's syntax map ([`Doc::smap`]) — the styling for
1253 /// [`View::Source`], the way [`build_visual`](Self::build_visual) is the
1254 /// styling for [`View::Wysiwyg`].
1255 ///
1256 /// A frontend calls this before painting raw source. One that doesn't gets
1257 /// an empty map and paints unstyled text, so this is additive: nothing
1258 /// breaks by not calling it.
1259 ///
1260 /// Built at most once per revision, and the revision is the whole key — the
1261 /// map has no width and no caret in it, so it survives every resize, every
1262 /// motion, and every selection change.
1263 ///
1264 /// The builds it does do cost a whole-arena marshal, which is precisely what
1265 /// the WYSIWYG path works to avoid, so this has no incremental path where
1266 /// that one has two. From `cargo run --release -p leaf-core --example
1267 /// bench`, per keystroke, against the WYSIWYG build the source view is
1268 /// *not* doing:
1269 ///
1270 /// | size | nodes | marshal | `source::build` | (`wysiwyg::build`) |
1271 /// |------:|-------:|--------:|----------------:|-------------------:|
1272 /// | 10 KB| 613 | 0.16 ms| 0.07 ms | 0.28 ms |
1273 /// | 100 KB| 6 097 | 0.84 ms| 0.38 ms | 2.43 ms |
1274 /// | 1 MB| 60 601 | 5.67 ms| 3.12 ms | 23.39 ms |
1275 ///
1276 /// Linear, two thirds of it the marshal, and the build itself five to seven
1277 /// times cheaper than the one it stands in for at every size. Comfortable
1278 /// well past any document a person edits in a terminal — a megabyte is where
1279 /// it would want [`Editor::dirty_range`] and the same splice treatment
1280 /// `build_spliced` gives the other map. The door is open; nothing has needed
1281 /// it yet.
1282 pub fn build_source(&mut self) {
1283 if self.smap_key == Some(self.revision) {
1284 return;
1285 }
1286 let nodes = self.nodes();
1287 self.smap = source::build(&nodes, &self.source);
1288 self.smap_key = Some(self.revision);
1289 }
1290
1291 /// Tell the model how many visual rows each block image should reserve, keyed
1292 /// by the image's destination. A terminal frontend calls this once it has
1293 /// decoded and measured its pictures — core does no image I/O, so this is the
1294 /// only way it learns a height — and the next [`Doc::build_visual`] lays each
1295 /// placeholder out that tall (the label row plus blank filler rows the
1296 /// frontend paints the raster over). A destination left out of the map falls
1297 /// back to the bare one-row placeholder, which is also what a frontend that
1298 /// can't draw pictures (or lays them out in its own units, like the GUI) gets
1299 /// by never calling this.
1300 ///
1301 /// Cheap to call every frame with the same map: only a *change* invalidates
1302 /// the built map (and the block-row cache, since a height isn't part of a
1303 /// block's bytes and so wouldn't otherwise re-render it). Steady state is a
1304 /// no-op, so a frontend can just hand over its current measurements each frame.
1305 pub fn set_media_rows(&mut self, rows: HashMap<String, usize>) {
1306 if self.media_rows == rows {
1307 return;
1308 }
1309 self.media_rows = rows;
1310 // A height lives outside the block's source bytes, so the content-keyed
1311 // block cache would hand back the old-height rows on a hit. Drop it (and
1312 // the splice layout it carries) so the next build re-renders every block
1313 // at the new heights, and force that build by clearing the map key.
1314 self.block_cache = wysiwyg::BlockCache::default();
1315 self.vmap_key = None;
1316 }
1317
1318 /// The revision the document's text is at — bumped by every edit, undo,
1319 /// redo, and reload, and by nothing else. A frontend caches against this to
1320 /// tell a repaint that needs new work from one that doesn't.
1321 ///
1322 /// It counts *edits*, not distinct texts: typing `x` and deleting it again
1323 /// lands on the same text two revisions later. Work is only ever rebuilt
1324 /// needlessly, never wrongly reused.
1325 pub fn revision(&self) -> u64 {
1326 self.revision
1327 }
1328
1329 /// The identity of the map presently in [`vmap`](Self::vmap) — what the last
1330 /// [`build_visual`](Self::build_visual) built it from, or the identity of an
1331 /// unbuilt map before the first one.
1332 ///
1333 /// This is *not* [`revision`](Self::revision). The revision says where the
1334 /// text is; this says where the map is, and the two part company the moment
1335 /// an edit lands, until something rebuilds. A frontend that keeps its own
1336 /// copy of the map — leaf-ratatui stashes core's before splicing filler rows
1337 /// under an oversized heading — compares this against the value it held when
1338 /// it took the copy, and learns whether `vmap` is still the map it stashed
1339 /// or one somebody else has since rebuilt. Restoring a copy over a newer
1340 /// map would paint a stale document; restoring nothing hands core's
1341 /// incremental rebuild a map it never built.
1342 ///
1343 /// "Somebody else" includes another document. The key names the `Doc`
1344 /// as well as the build, so a frontend that draws two documents through
1345 /// one stash — a host with several buffers, or one that opens the next
1346 /// document where the last one stood — never has the copy it took of one
1347 /// accepted by the other, however alike their builds are.
1348 pub fn visual_key(&self) -> VisualKey {
1349 VisualKey(self.identity, self.vmap_key.clone())
1350 }
1351
1352 /// The map, built at most once per `(revision, wrap)`. `clamp_caret` still
1353 /// runs on every call: the caret moves without the document changing, and
1354 /// keeping it on a legal stop is this function's job either way.
1355 fn build_map(&mut self, wrap: Option<usize>) {
1356 // Under `MarkupMode::Full` the map is a function of the caret's *line*
1357 // as well as the text, so the line joins the key: moving within a line
1358 // still reuses the map, and crossing into another one rebuilds it. In
1359 // every other mode `reveal_line` is `None` and the key is what it was,
1360 // so caret motion goes on costing nothing.
1361 let reveal = self.reveal_line();
1362 let key = (self.revision, wrap, reveal.clone());
1363 if self.vmap_key.as_ref() != Some(&key) {
1364 // Enumerate the top-level blocks cheaply — no whole-arena marshal.
1365 // A subtree is pulled only for the block(s) that actually changed, so
1366 // the FFI marshal shrinks from O(document) to O(edited block).
1367 let top = self.top_blocks();
1368
1369 // Fast path: when twig reports a dirty byte range, try to patch the
1370 // previous map in place — a single-block edit moves the prefix,
1371 // shifts the suffix, and re-renders only one block. `build_spliced`
1372 // returns `None` (and we fall back to the always-correct full rebuild)
1373 // whenever the edit reshaped the block structure, hit a table, or
1374 // there's no previous map to patch.
1375 // Preserve soft breaks as written when the flow preference asks for
1376 // it — the builder renders each as its own visual row instead of
1377 // folding it into the reflowed paragraph.
1378 let preserve_soft = self.line_flow == LineFlow::Preserve;
1379 let spliced = match self.editor.dirty_range() {
1380 Some(dirty) => {
1381 let prev = std::mem::take(&mut self.vmap);
1382 let source = &self.source;
1383 let cache = &mut self.block_cache;
1384 let media_rows = &self.media_rows;
1385 let editor = &mut self.editor;
1386 wysiwyg::build_spliced(
1387 prev,
1388 source,
1389 wrap,
1390 preserve_soft,
1391 &top,
1392 dirty,
1393 media_rows,
1394 reveal.clone(),
1395 cache,
1396 |id| editor.subtree(NodeId(id)).unwrap_or_default(),
1397 )
1398 }
1399 None => None,
1400 };
1401 self.vmap = spliced.unwrap_or_else(|| {
1402 let source = &self.source;
1403 let cache = &mut self.block_cache;
1404 let media_rows = &self.media_rows;
1405 let editor = &mut self.editor;
1406 wysiwyg::build_cached(
1407 &top,
1408 source,
1409 wrap,
1410 preserve_soft,
1411 media_rows,
1412 reveal,
1413 cache,
1414 |id| editor.subtree(NodeId(id)).unwrap_or_default(),
1415 )
1416 });
1417 // Acknowledge the dirty range so the next edit's range starts fresh.
1418 self.editor.clear_dirty();
1419 self.vmap_key = Some(key);
1420 }
1421 self.clamp_caret();
1422 }
1423
1424 fn nodes(&mut self) -> Vec<FlatNode> {
1425 self.editor.nodes().unwrap_or_default()
1426 }
1427
1428 /// The document's top-level blocks for the incremental render. See
1429 /// [`wysiwyg::top_blocks`] for why this isn't simply `child_spans(None)`.
1430 fn top_blocks(&mut self) -> Vec<QueryMatch> {
1431 wysiwyg::top_blocks(&mut self.editor)
1432 }
1433
1434 pub fn format_name(&self) -> &'static str {
1435 // `Format` is `#[non_exhaustive]` as of twig 3.0, so the wildcard is
1436 // required. It also covers `Asciidoc`, which twig parses but cannot
1437 // serialize — leaf never opens a document in it (see `Doc::open`).
1438 match self.format {
1439 Format::Djot => "djot",
1440 Format::Markdown => "markdown",
1441 Format::Xml => "xml",
1442 Format::Html => "html",
1443 _ => "unknown",
1444 }
1445 }
1446
1447 /// Whether this document's format offers *any* door in — `false` only for a
1448 /// wholly parse-only format (XML, AsciiDoc), where every gesture refuses and
1449 /// a frontend may as well open the file read-only.
1450 ///
1451 /// This is a much weaker claim than the name suggests, and driving per-button
1452 /// state from it is exactly the mistake to avoid: HTML answers `true` because
1453 /// it spells the inline marks with a tag pair (`<strong>`, `<em>`, `<code>`)
1454 /// while a heading, a quote, a list, a task box, a link and a code fence all
1455 /// remain unspellable there. Ask [`capabilities`](Self::capabilities) — or
1456 /// [`supports`](Self::supports) — per control.
1457 pub fn authorable(&self) -> bool {
1458 self.format.is_authorable()
1459 }
1460
1461 /// Whether this document can spell `gesture`, which is twig's own answer
1462 /// rather than a copy of it: `Format::supports_with` reads the same
1463 /// `Syntax` table the `Editor` method consults before refusing, chosen by
1464 /// the very [`parse_extensions`] this document's editor reparses with — so
1465 /// what the toolbar offers and what the splice will accept are one table.
1466 ///
1467 /// It is a fact about the *document*, not about the caret. `true` does not
1468 /// promise the gesture succeeds where it is standing — a link over a table
1469 /// border still fails — only that it will not fail with
1470 /// `UnsupportedFormat`. Gray out on `false`; don't read `true` as "this
1471 /// will work here".
1472 pub fn supports(&self, gesture: Gesture) -> bool {
1473 self.format.supports_with(parse_extensions(), gesture)
1474 }
1475
1476 /// Every control's enabled state in one read — what a toolbar builds itself
1477 /// from when a document opens or its format changes. See [`Capabilities`].
1478 pub fn capabilities(&self) -> Capabilities {
1479 Capabilities::of(self.format)
1480 }
1481
1482 /// Refuse a gesture this document's format cannot spell, saying so in the
1483 /// status line. `true` means the caller must return without calling twig.
1484 ///
1485 /// Most of these refusals duplicate one twig would make anyway, and they are
1486 /// made here regardless because a message naming the *document's* format
1487 /// reads better than one naming twig's internals. Two of them are not
1488 /// duplicates and are the reason this is a guard rather than an error
1489 /// translation:
1490 ///
1491 /// - The table family (see [`table_op`](Self::table_op)) consults no
1492 /// `Syntax` table, so twig does not refuse it at all.
1493 /// - [`toggle`](Self::toggle) at a collapsed caret never reaches twig — it
1494 /// arms a sticky mark for text not yet typed, which is a promise `insert`
1495 /// could not keep.
1496 fn refuse_unsupported(&mut self, what: &str, gesture: Gesture) -> bool {
1497 self.refuse_unless(what, self.supports(gesture))
1498 }
1499
1500 /// [`refuse_unsupported`](Self::refuse_unsupported) against a capability leaf
1501 /// answers itself — today only [`spells_pipe_tables`].
1502 fn refuse_unless(&mut self, what: &str, supported: bool) -> bool {
1503 if supported {
1504 return false;
1505 }
1506 self.status = Some(format!("{what}: not supported in {}", self.format_name()));
1507 true
1508 }
1509
1510 /// The name to show for this document. An untitled one has no file to name
1511 /// it, and both frontends put this straight on screen — an empty path
1512 /// renders as an empty header, so it says so instead.
1513 pub fn file_name(&self) -> String {
1514 if self.is_untitled() {
1515 return "untitled".into();
1516 }
1517 self.path
1518 .file_name()
1519 .map(|s| s.to_string_lossy().into_owned())
1520 .unwrap_or_else(|| self.path.display().to_string())
1521 }
1522
1523 /// The selection as an ordered `[start, end)` byte range, or `None` when the
1524 /// caret and anchor coincide (an empty selection is no selection).
1525 pub fn selection(&self) -> Option<(usize, usize)> {
1526 self.anchor
1527 .map(|a| (a.min(self.caret), a.max(self.caret)))
1528 .filter(|(s, e)| s != e)
1529 }
1530
1531 /// The selected text, or `None` when there's no selection — the source
1532 /// slice a copy/cut hands to the system clipboard.
1533 pub fn selected_text(&self) -> Option<&str> {
1534 self.selection().map(|(s, e)| &self.source[s..e])
1535 }
1536
1537 /// The selection as a quote with a little of what surrounds it — the shape
1538 /// a host that cites, annotates, or searches for a passage wants, cut from
1539 /// the **source** rather than from anything rendered, so the quote is
1540 /// findable in the document again by plain string search.
1541 ///
1542 /// `context` is a count of characters (not bytes) on each side, clipped at
1543 /// the document's edges; the slices land on char boundaries by
1544 /// construction. `None` when nothing is selected.
1545 pub fn selection_quote(&self, context: usize) -> Option<Quote> {
1546 let (start, end) = self.selection()?;
1547 let mut before = start;
1548 for _ in 0..context {
1549 match self.source[..before].chars().next_back() {
1550 Some(c) => before -= c.len_utf8(),
1551 None => break,
1552 }
1553 }
1554 let mut after = end;
1555 for _ in 0..context {
1556 match self.source[after..].chars().next() {
1557 Some(c) => after += c.len_utf8(),
1558 None => break,
1559 }
1560 }
1561 Some(Quote {
1562 exact: self.source[start..end].to_string(),
1563 prefix: self.source[before..start].to_string(),
1564 suffix: self.source[end..after].to_string(),
1565 start,
1566 end,
1567 })
1568 }
1569
1570 /// Whether the document refuses to change — see the field.
1571 pub fn read_only(&self) -> bool {
1572 self.read_only
1573 }
1574
1575 /// Turn the read-only gate on or off. A frontend preference like
1576 /// [`set_markup_mode`](Self::set_markup_mode): nothing about the document
1577 /// itself changes, only what may be done to it from here on.
1578 pub fn set_read_only(&mut self, on: bool) {
1579 self.read_only = on;
1580 }
1581
1582 /// The host-painted ranges, sorted by start — see [`Highlight`].
1583 pub fn highlights(&self) -> &[Highlight] {
1584 &self.highlights
1585 }
1586
1587 /// Replace the host-painted ranges wholesale. The whole set each time,
1588 /// rather than add/remove verbs: the host owns the list (it derives it
1589 /// from its own state — annotations, search hits), and a replace can
1590 /// never leave the two disagreeing about what should be on screen.
1591 pub fn set_highlights(&mut self, mut highlights: Vec<Highlight>) {
1592 highlights.retain(|h| h.start < h.end);
1593 highlights.sort_by_key(|h| (h.start, h.end));
1594 self.highlights = highlights;
1595 }
1596
1597 /// The highlight covering source `offset`, if one does — first by start
1598 /// when several overlap, which makes overlapping washes resolvable rather
1599 /// than undefined. What a frontend asks when the reader activates a spot.
1600 ///
1601 /// [`Highlight::covering`] is the whole of it: the frontends paint by
1602 /// asking the same question per glyph, against a slice they were handed
1603 /// rather than against a `Doc`, and one answer for both is what keeps a
1604 /// wash and an activation agreeing about which range a spot is in.
1605 pub fn highlight_at(&self, offset: usize) -> Option<&Highlight> {
1606 Highlight::covering(&self.highlights, offset)
1607 }
1608
1609 /// The AST breadcrumb at the caret (root → deepest), e.g.
1610 /// `doc › para › strong`. Read live from twig via `ancestors_at`.
1611 pub fn breadcrumb(&mut self) -> String {
1612 match self.editor.ancestors_at(self.caret) {
1613 Ok(chain) => chain
1614 .iter()
1615 .map(|m| m.kind.as_str())
1616 .collect::<Vec<_>>()
1617 .join(" › "),
1618 Err(_) => String::new(),
1619 }
1620 }
1621
1622 // ── editing ──────────────────────────────────────────────────────────────
1623
1624 /// Replace the byte range `[start, end)` with `text`, re-anchoring the caret
1625 /// after it. The public form of the internal splice — a pixel frontend that
1626 /// hit-tests to a byte offset (or an IME that hands back an explicit range)
1627 /// edits through this, the same twig `edit_range` the caret ops use.
1628 pub fn edit(&mut self, start: usize, end: usize, text: &str) {
1629 self.splice(start, end, text, EditKind::Other);
1630 }
1631
1632 /// Insert typed `text` at the caret, replacing the selection if there is one.
1633 /// A single typed character coalesces with the run of typing before it; a
1634 /// newline or a multi-character insert is its own undo step.
1635 ///
1636 /// Typed input only — clipboard text goes through [`paste`](Self::paste).
1637 pub fn insert(&mut self, text: &str) {
1638 // The read-only gate, up front: the paths below reach twig by several
1639 // verbs, not all of them through the splice — see the field.
1640 if self.read_only {
1641 return;
1642 }
1643 // Typing against a block picture would dissolve it, and typing past a
1644 // table would grow it a row — see `open_paragraph_at_block_edge`. Give
1645 // the text a paragraph first, so what the caret was standing beside
1646 // stays what it was.
1647 self.open_paragraph_at_block_edge(text);
1648 // Armed sticky marks (⌘b with no selection) turn the next typed text
1649 // bold/italic/… and then retire — see `insert_with_marks`. Whitespace is
1650 // the exception: it takes no mark of its own and keeps the delta armed
1651 // for the character behind it — see `insert_space_with_marks`.
1652 let pending = self.pending_here();
1653 if !pending.is_empty() && self.selection().is_none() && !text.is_empty() {
1654 if text.trim().is_empty() {
1655 self.insert_space_with_marks(self.caret, text, pending);
1656 } else {
1657 self.insert_with_marks(self.caret, text, pending);
1658 }
1659 return;
1660 }
1661 // `MarkupMode::None`: typed syntax stays literal — twig escapes
1662 // anything that would open markup, so a Diaryx user never mints
1663 // formatting by keyboard (it comes from commands instead). The other two
1664 // rungs of the ladder author markup from what you type, which is the
1665 // whole difference between them and this one. Only in the rendered view
1666 // (source view is for typing raw markup) and only where the format has a
1667 // literal spelling at all: escaping is a backslash before a byte from the
1668 // format's own alphabet, and a format with no such alphabet (HTML escapes
1669 // with entities, XML spells nothing) would have `\&` written into it,
1670 // which is two literal characters and not an escape. Marks (⌘b) still
1671 // format — that path returned above; and leaf's own structural inserts go
1672 // through `insert_raw`, never here, so a list marker or quote gutter is
1673 // written as the markup it is.
1674 if !self.markup_mode.authors()
1675 && self.view == View::Wysiwyg
1676 && !text.is_empty()
1677 && self.supports(Gesture::InsertLiteral)
1678 {
1679 self.insert_literal_typed(text);
1680 return;
1681 }
1682 self.insert_raw(text);
1683 }
1684
1685 /// Insert `text` verbatim at the caret (replacing any selection) — the plain
1686 /// path with no Hidden-mode literal escaping. leaf's own structural inserts
1687 /// (a list marker, a quote gutter, an in-cell `<br>`) call this: they ARE
1688 /// markup by design and must not be escaped.
1689 fn insert_raw(&mut self, text: &str) {
1690 let (s, e) = self.selection().unwrap_or((self.caret, self.caret));
1691 self.splice(s, e, text, typed_edit_kind(text));
1692 }
1693
1694 /// Open a paragraph for text about to be inserted at one of a block media's
1695 /// two caret stops, or at a table's trailing stop, and leave the caret
1696 /// standing in it.
1697 ///
1698 /// A block image is a paragraph whose entire content is the picture, and the
1699 /// caret's only homes on it are in front of it and just past it (see
1700 /// [`VisualMap::block_media_stop`]). Text inserted at either offset joins
1701 /// *that* paragraph — and a paragraph holding anything besides the image is
1702 /// no longer a block image but a line of text with an inline one in it. The
1703 /// frontend that was painting a photo there paints a text run instead; the
1704 /// picture is still in the file, and nothing said a word. Those two offsets
1705 /// are also exactly where a click on the picture lands, so the whole accident
1706 /// is one tap and one keystroke.
1707 ///
1708 /// So the break goes in first and the text lands in the new empty paragraph —
1709 /// what pressing Return before typing would have done, which is a habit no
1710 /// one should have to learn from losing a photo. A no-op everywhere else, and
1711 /// over a selection (which is replaced, not joined into).
1712 ///
1713 /// A picture inside a quote or a list leaves its container, because `\n\n`
1714 /// ends the block. The alternative is worse: the `\n> ` / next-item
1715 /// continuation [`newline`](Self::newline) writes stays in the same
1716 /// *paragraph*, which is the thing being prevented.
1717 ///
1718 /// A table's trailing stop ([`VisualMap::table_end_stop`]) is the same
1719 /// accident from the other side of a different block: the stop sits at the
1720 /// end of the table's last source line, and a line glued under a table is
1721 /// a row of it — `| 1 | 2 |x` is a three-cell row, not a paragraph. So the
1722 /// break goes in there too, and the text lands under the table.
1723 ///
1724 /// Only in the rendered view. Source view is for typing raw markup, where
1725 /// putting a character against an image is exactly what it looks like.
1726 fn open_paragraph_at_block_edge(&mut self, text: &str) {
1727 if self.view != View::Wysiwyg || text.is_empty() || text == "\n" {
1728 return;
1729 }
1730 if self.selection().is_some() {
1731 return;
1732 }
1733 // The map may be a revision behind (nothing has drawn since the last
1734 // edit), and this asks it about offsets — a stale answer would splice a
1735 // break into the wrong place. Free when it is already current, which it
1736 // is whenever a frontend drew a frame between keystrokes.
1737 self.rebuild_map();
1738 let at = self.caret;
1739 let side = match self.vmap.block_media_stop(at) {
1740 Some((side, _)) => side,
1741 None if self.vmap.table_end_stop(at) => MediaStop::After,
1742 None => return,
1743 };
1744 if !self.splice(at, at, "\n\n", EditKind::Other) {
1745 return;
1746 }
1747 // The break is part of the keystroke, not an edit of its own: leave the
1748 // run marked as typing so the character about to arrive folds into it and
1749 // one undo puts the document back the way it was found. (A paste, or a
1750 // multi-character insert, is `EditKind::Other` and stays its own step —
1751 // as it would have been anywhere else in the document.)
1752 self.last_edit_kind = Some(EditKind::Insert);
1753 if side == MediaStop::Before {
1754 // The break went in above the picture and the caret rode to the end
1755 // of it — which is still hard against the picture. Step back onto the
1756 // blank line it opened, so the text lands above rather than in front.
1757 self.caret = at;
1758 }
1759 }
1760
1761 /// A delete key pressed at one of a block picture's two caret stops, handled
1762 /// as the picture being an *atom* rather than a run of bytes. Returns whether
1763 /// the key was consumed.
1764 ///
1765 /// The caret rests in front of a block image and just past it, never inside
1766 /// its markup — which the rendered view doesn't show. So the byte a delete
1767 /// key nominally takes there is one the writer cannot see, and taking it
1768 /// leaves the picture as broken markup rather than as anything anyone asked
1769 /// for: Backspace at the stop past `` removes the closing paren, and
1770 /// a photo becomes the literal text `
1773 /// prevents from the typing side, and it cost this repository's own test vault
1774 /// a photo before it was found.
1775 ///
1776 /// So the key aimed *at* the picture deletes the picture, whole — Backspace
1777 /// when it is behind the caret, Delete when it is in front — which is what
1778 /// every editor does with an embed, and one undo away. The key aimed *away*
1779 /// from it would otherwise delete the paragraph break and merge a neighbour
1780 /// into the picture's own paragraph, which dissolves it just as surely; it
1781 /// steps the caret over the boundary instead and leaves the
1782 /// next press to delete in the block it has reached — the same "first press
1783 /// steps out of the atom, second press deletes" every delete key here gets,
1784 /// word-deletes included (⌥⌫ in front of a picture is aimed at the prose
1785 /// above, and reaches it on the second press rather than taking the break and
1786 /// the picture with it on the first).
1787 fn delete_around_block_media(&mut self, forward: bool) -> bool {
1788 // The map answers about offsets, so it has to be this revision's — see
1789 // the same call in `open_paragraph_at_block_edge`.
1790 self.rebuild_map();
1791 let Some((side, span)) = self.vmap.block_media_stop(self.caret) else {
1792 return false;
1793 };
1794 let aimed_at_it = side
1795 == if forward {
1796 MediaStop::Before
1797 } else {
1798 MediaStop::After
1799 };
1800 if !aimed_at_it {
1801 let over = if forward {
1802 self.vmap.stop_after(self.caret)
1803 } else {
1804 self.vmap.stop_before(self.caret)
1805 };
1806 if let Some(off) = over.filter(|&o| o >= self.caret_floor()) {
1807 self.caret = off;
1808 self.anchor = None;
1809 self.goal_col = None;
1810 }
1811 return true;
1812 }
1813 // Take the break that held the picture apart from its neighbour with it,
1814 // so the delete doesn't leave a blank paragraph standing where the
1815 // picture was. The last arm is a picture that is the whole document.
1816 let (from, to) = if self.source[..span.start].ends_with("\n\n") {
1817 (span.start - 2, span.end)
1818 } else if self.source[span.end..].starts_with("\n\n") {
1819 (span.start, span.end + 2)
1820 } else {
1821 (span.start, span.end)
1822 };
1823 self.splice(from.max(self.caret_floor()), to, "", EditKind::Other);
1824 true
1825 }
1826
1827 /// The Hidden-mode typing path: replace any selection, then insert `text`
1828 /// escaped so it stays literal. When it replaces a selection the two edits
1829 /// fold into one undo step, so an overwrite undoes atomically (and restores
1830 /// the selection) exactly as a plain one does.
1831 fn insert_literal_typed(&mut self, text: &str) {
1832 let kind = typed_edit_kind(text);
1833 match self.selection() {
1834 Some((s, e)) => {
1835 if !self.splice(s, e, "", EditKind::Other) {
1836 return;
1837 }
1838 // Typing over a whole marked run takes its delimiters with it
1839 // (the empty content couldn't hold them — see
1840 // `repair_mark_edges`) and leaves its marks armed at the caret.
1841 // The text taking the run's place inherits them, exactly as it
1842 // would have by landing inside a run that survived.
1843 let pending = self.pending_here();
1844 if !pending.is_empty() && !text.trim().is_empty() {
1845 self.insert_with_marks(self.caret, text, pending);
1846 return;
1847 }
1848 self.insert_literal_at(self.caret, text, kind, true);
1849 }
1850 None => {
1851 self.insert_literal_at(self.caret, text, kind, false);
1852 }
1853 }
1854 }
1855
1856 /// The sticky-mark delta that is live right now: the marks armed by [`toggle`]
1857 /// at a collapsed caret, but only while the caret still stands where they
1858 /// were armed and nothing is selected. Empty otherwise, so a stale delta
1859 /// never styles text it wasn't meant for.
1860 fn pending_here(&self) -> InlineMarks {
1861 if self.anchor.is_none() && self.pending_at == Some(self.caret) {
1862 self.pending_marks
1863 } else {
1864 InlineMarks::empty()
1865 }
1866 }
1867
1868 /// Drop the armed sticky marks — any caret motion, selection, or edit does
1869 /// this, so "start bold here" only ever applies at the exact spot it was
1870 /// asked for.
1871 fn clear_pending(&mut self) {
1872 self.pending_marks = InlineMarks::empty();
1873 self.pending_at = None;
1874 }
1875
1876 /// Insert `text` at `at` carrying the armed sticky `marks`: a mark not yet in
1877 /// force is wrapped around the freshly typed text; a mark the caret already
1878 /// stands inside is *shed* — the text is inserted past the run's end so it
1879 /// lands unmarked ("type normally again"). The caret comes to rest inside any
1880 /// added runs, so continued typing inherits the marks with no re-wrapping,
1881 /// and the delta is cleared: the marks now live in the document, not here.
1882 fn insert_with_marks(&mut self, at: usize, text: &str, marks: InlineMarks) {
1883 let base = self.mark_spans_at(at);
1884 let base_set: InlineMarks = base.iter().map(|(k, _)| *k).collect();
1885 // Nothing to shed, and a run of exactly these marks standing just behind
1886 // the caret: carry on writing *that* run rather than opening a second
1887 // one beside it.
1888 if base_set.is_empty() && self.rejoin_run(at, text, marks) {
1889 return;
1890 }
1891 // Shed the marks we're turning off: step the insertion point past the
1892 // end of each run the caret sits in, so the new text falls outside it.
1893 let mut ins_at = at;
1894 for (kind, span) in &base {
1895 if marks.contains(*kind) {
1896 ins_at = ins_at.max(span.end);
1897 }
1898 }
1899 if !self.splice_exact(ins_at, ins_at, text, EditKind::Other) {
1900 return;
1901 }
1902 // The plain splice inserted exactly `text` at `ins_at`; that byte range
1903 // is the content every added mark wraps.
1904 let (mut cs, mut ce) = (ins_at, ins_at + text.len());
1905 for kind in marks.iter() {
1906 if !base_set.contains(kind) {
1907 let (ncs, nce) = self.wrap_span(cs, ce, kind);
1908 cs = ncs;
1909 ce = nce;
1910 }
1911 }
1912 self.caret = ce.min(self.source.len());
1913 self.anchor = None;
1914 self.last_edit_kind = None;
1915 // Realised: the marks are in the document now, and the caret sits inside
1916 // them, so there is no delta left to carry. Arm nothing, but remember the
1917 // spot so a *further* toggle before typing starts a clean delta here.
1918 self.pending_marks = InlineMarks::empty();
1919 self.pending_at = Some(self.caret);
1920 self.clamp_caret();
1921 self.record_caret();
1922 }
1923
1924 /// Carry on the marked run just behind `at` — moving its closing delimiters
1925 /// out past the new text — instead of opening a second run of the same marks
1926 /// beside it. Returns whether it did.
1927 ///
1928 /// This is the far half of the mark-edge rule (see [`splice`](Self::splice)).
1929 /// A space typed after a bold word steps the caret out of the run, because
1930 /// `**bold **` is not bold; the next character has to step back *in*, or the
1931 /// writer who typed one bold phrase is left with `**bold** **and**` — two
1932 /// runs that read the same to a reader but spell the file in a way nobody
1933 /// wrote. Only whitespace may stand in the gap (a run doesn't reach across
1934 /// words it isn't marking), and the marks behind it must be exactly the ones
1935 /// armed — a run of *some* other kind is a neighbour, not this phrase.
1936 fn rejoin_run(&mut self, at: usize, text: &str, marks: InlineMarks) -> bool {
1937 if text.is_empty() || text.trim() != text {
1938 return false;
1939 }
1940 let gap_at = self.source[..at].trim_end_matches([' ', '\t']).len();
1941 // Walk in through the delimiters stacked at that point, innermost last:
1942 // `***both*** ` closes two runs with one `***`, and rejoining means
1943 // getting behind all of them.
1944 let (mut cut, mut kinds) = (gap_at, InlineMarks::empty());
1945 while let Some((kind, content_end)) = self
1946 .editor
1947 .ancestors_at(prev_boundary(&self.source, cut))
1948 .unwrap_or_default()
1949 .into_iter()
1950 .filter(|m| m.span.end == cut)
1951 .find_map(|m| Some((inline_kind(&m.kind)?, m.content_span.clone()?.end)))
1952 {
1953 if content_end >= cut {
1954 break; // a mark with no closing delimiter to step behind
1955 }
1956 kinds.insert(kind);
1957 cut = content_end;
1958 }
1959 if cut == gap_at || kinds != marks {
1960 return false;
1961 }
1962 // Re-spell the tail: the gap, then the new text, then the delimiters that
1963 // used to close in front of them — read out of the document rather than
1964 // written from a table, so whatever twig spells them with is what moves.
1965 let tail = format!(
1966 "{}{text}{}",
1967 &self.source[gap_at..at],
1968 &self.source[cut..gap_at]
1969 );
1970 if !self.splice_exact(cut, at, &tail, EditKind::Other) {
1971 return false;
1972 }
1973 self.caret = (cut + (at - gap_at) + text.len()).min(self.source.len());
1974 self.anchor = None;
1975 self.last_edit_kind = None;
1976 self.pending_marks = InlineMarks::empty();
1977 self.pending_at = Some(self.caret);
1978 self.clamp_caret();
1979 self.record_caret();
1980 true
1981 }
1982
1983 /// Insert typed whitespace at a caret with sticky marks armed. Whitespace is
1984 /// never itself wrapped: a mark around a space draws nothing a reader can
1985 /// see, and in Markdown and Djot it draws its own delimiters instead
1986 /// (`** **`). So the space goes in unmarked — outside any run the armed
1987 /// marks are shedding — and the marks stay armed for the character after it,
1988 /// which rejoins the run (see [`rejoin_run`](Self::rejoin_run)).
1989 fn insert_space_with_marks(&mut self, at: usize, text: &str, marks: InlineMarks) {
1990 let base = self.mark_spans_at(at);
1991 // What the *next* character carries: the armed delta resolved against the
1992 // marks in force here, which the space must not quietly drop.
1993 let want = base
1994 .iter()
1995 .map(|(k, _)| *k)
1996 .collect::<InlineMarks>()
1997 .xor(marks);
1998 let mut ins_at = at;
1999 for (kind, span) in &base {
2000 if marks.contains(*kind) {
2001 ins_at = ins_at.max(span.end);
2002 }
2003 }
2004 if !self.splice(ins_at, ins_at, text, typed_edit_kind(text)) {
2005 return;
2006 }
2007 self.rearm(want);
2008 self.record_caret();
2009 }
2010
2011 /// Wrap `[s, e)` in `kind` via twig and return the byte span the *content*
2012 /// (not the delimiters) occupies afterwards. Markdown/Djot inline delimiters
2013 /// are symmetric (`**`…`**`, `_`…`_`, `` ` ``…`` ` ``), so the bytes twig
2014 /// added split evenly around the content — half the growth on each side.
2015 fn wrap_span(&mut self, s: usize, e: usize, kind: InlineKind) -> (usize, usize) {
2016 // The read-only gate — this door reaches twig without the splice.
2017 if self.read_only {
2018 return (s, e);
2019 }
2020 match self.editor.toggle_inline(s, e, kind) {
2021 Ok(change) => {
2022 self.last_edit_kind = None;
2023 self.refresh();
2024 self.dirty = self.source != self.clean_source;
2025 let added = (change.new.end - change.new.start).saturating_sub(e - s);
2026 let half = added / 2;
2027 (change.new.start + half, change.new.end - half)
2028 }
2029 // Unsupported here (e.g. mark on Markdown): leave the text unwrapped
2030 // rather than lose the keystroke.
2031 Err(e2) => {
2032 self.status = Some(format!("{kind:?}: {e2}"));
2033 (s, e)
2034 }
2035 }
2036 }
2037
2038 /// The safe offset to splice a block-level break at, given a caret that may
2039 /// sit exactly between an inline mark's content and its own closing
2040 /// delimiter (`content_span.end == off < span.end` for some enclosing mark
2041 /// — the WYSIWYG caret's natural resting place at the end of `**bold**`
2042 /// with nothing following it on the line: the closing `**` renders no
2043 /// glyph of its own, so the caret's "end of line" offset lands right
2044 /// before it). Splicing a paragraph/list/quote break at `off` itself would
2045 /// sever the delimiter from its content, stranding it alone on the new
2046 /// line. Walks out to the *outermost* such mark's `span.end` instead, so
2047 /// nested marks closing at the same point (`**_x_**`) all clear together.
2048 /// A no-op everywhere else — mid-run, or past real trailing content, no
2049 /// mark's `content_span` ends exactly at `off`.
2050 fn skip_trailing_close_delims(&mut self, off: usize) -> usize {
2051 let off = off.min(self.source.len());
2052 self.editor
2053 .ancestors_at(off)
2054 .unwrap_or_default()
2055 .into_iter()
2056 .filter(|m| inline_kind(&m.kind).is_some())
2057 .filter(|m| off < m.span.end && m.content_span.as_ref().is_some_and(|c| c.end == off))
2058 .map(|m| m.span.end)
2059 .max()
2060 .unwrap_or(off)
2061 }
2062
2063 /// The offset a *delete* aimed at the character before `off` should stop at,
2064 /// when `off` is the start of a run's text and the bytes behind it are that
2065 /// run's opening delimiter. The rich view draws no glyph for a `**`, so the
2066 /// byte behind the caret at the start of a bold word is not a character the
2067 /// writer can see, let alone one they aimed Backspace at: taking it leaves
2068 /// `a *bold** c` — the styling gone and a literal asterisk in its place. The
2069 /// delete steps over the whole delimiter to the visible character in front of
2070 /// it instead. Walks out to the *outermost* mark opening there, so
2071 /// `**_x_**` clears every delimiter at once, and is a no-op anywhere else.
2072 fn skip_leading_open_delims(&mut self, off: usize) -> usize {
2073 let off = off.min(self.source.len());
2074 self.editor
2075 .ancestors_at(off)
2076 .unwrap_or_default()
2077 .into_iter()
2078 .filter(|m| inline_kind(&m.kind).is_some())
2079 .filter(|m| {
2080 m.span.start < off && m.content_span.as_ref().is_some_and(|c| c.start == off)
2081 })
2082 .map(|m| m.span.start)
2083 .min()
2084 .unwrap_or(off)
2085 }
2086
2087 /// `off` moved *inside* the run whose closing delimiters end there — the
2088 /// other offset the rich view draws in the same place, since a `**` renders
2089 /// no glyph of its own. `**bold**` has a caret home on each side of its
2090 /// closing delimiter, one column apart on screen and eight bytes and a whole
2091 /// run apart in the file, and a plain ← lands on the outer one whenever a
2092 /// space follows the phrase. The inner one is what the writer is pointing at
2093 /// there: the end of their bold word. Walks in through every mark closing at
2094 /// that point, innermost last, so `***both***` lands inside both. A no-op
2095 /// anywhere else — mid-run, or in prose, no mark's span ends at `off`.
2096 fn step_inside_close_delims(&mut self, off: usize) -> usize {
2097 let mut off = off.min(self.source.len());
2098 loop {
2099 let inner = self
2100 .editor
2101 .ancestors_at(prev_boundary(&self.source, off))
2102 .unwrap_or_default()
2103 .into_iter()
2104 .filter(|m| inline_kind(&m.kind).is_some() && m.span.end == off)
2105 .filter_map(|m| m.content_span.clone().map(|c| c.end))
2106 .filter(|&end| end < off)
2107 .max();
2108 match inner {
2109 Some(end) => off = end,
2110 None => return off,
2111 }
2112 }
2113 }
2114
2115 /// The mirror at the opening edge: `off` moved inside the run whose
2116 /// delimiters *start* there, onto the first character of its text. See
2117 /// [`step_inside_close_delims`](Self::step_inside_close_delims).
2118 fn step_inside_open_delims(&mut self, off: usize) -> usize {
2119 let mut off = off.min(self.source.len());
2120 loop {
2121 let inner = self
2122 .editor
2123 .ancestors_at(off)
2124 .unwrap_or_default()
2125 .into_iter()
2126 .filter(|m| inline_kind(&m.kind).is_some() && m.span.start == off)
2127 .filter_map(|m| m.content_span.clone().map(|c| c.start))
2128 .filter(|&start| start > off)
2129 .min();
2130 match inner {
2131 Some(start) => off = start,
2132 None => return off,
2133 }
2134 }
2135 }
2136
2137 /// The inline mark kinds whose span covers `off`, each with that span — the
2138 /// span-carrying sibling of [`marks_at`](Self::marks_at), which reports node
2139 /// ids instead. Used to shed a mark by stepping past the end of its run.
2140 fn mark_spans_at(&mut self, off: usize) -> Vec<(InlineKind, std::ops::Range<usize>)> {
2141 let off = off.min(self.source.len());
2142 self.editor
2143 .ancestors_at(off)
2144 .unwrap_or_default()
2145 .into_iter()
2146 .filter(|m| off < m.span.end)
2147 .filter_map(|m| inline_kind(&m.kind).map(|k| (k, m.span.clone())))
2148 .collect()
2149 }
2150
2151 /// Insert clipboard `text` at the caret, replacing the selection if there is
2152 /// one — always its own undo step, whatever its length.
2153 ///
2154 /// Provenance is the whole point, and only the caller has it. `insert` reads
2155 /// a lone character as a keystroke and folds it into the run around it,
2156 /// which is right for typing and wrong for a one-character paste: that paste
2157 /// would vanish mid-run on an undo it was never part of, and the characters
2158 /// the user actually typed would go with it. Length can't tell the two
2159 /// apart — `⌘V` of `x` and typing `x` are the same string — so the door the
2160 /// caller comes through is what says which happened.
2161 pub fn paste(&mut self, text: &str) {
2162 // Pasting against a block picture or a table's end joins the block
2163 // exactly as typing does, and for the same reason — see
2164 // `open_paragraph_at_block_edge`.
2165 self.open_paragraph_at_block_edge(text);
2166 let (s, e) = self.selection().unwrap_or((self.caret, self.caret));
2167 self.splice(s, e, text, EditKind::Other);
2168 }
2169
2170 /// Replace `[start, end)` with `text` as one step of an IME composition —
2171 /// the same splice as [`edit`](Self::edit), but marked so the run of steps
2172 /// folds into a single undo.
2173 ///
2174 /// A composition is *one* act of writing. Typing `かんじ` and picking 感じ is a
2175 /// dozen calls here, each replacing the last one's provisional bytes, and an
2176 /// undo step per call means undoing a word means pressing ⌘Z until the reading
2177 /// unspools backwards through kana — the intermediate states were never text
2178 /// the user wrote. Only the frontend knows a call is provisional (the bytes
2179 /// look like any other edit), so the door the caller comes through is what
2180 /// says so, exactly as it is for [`paste`](Self::paste) versus
2181 /// [`insert`](Self::insert).
2182 ///
2183 /// Pair with [`end_composition`](Self::end_composition), or the *next*
2184 /// composition folds into this one.
2185 pub fn edit_composing(&mut self, start: usize, end: usize, text: &str) {
2186 self.splice(start, end, text, EditKind::Compose);
2187 }
2188
2189 /// Close the open composition run, so the next one is its own undo step.
2190 /// Call when the IME commits or withdraws a composition.
2191 ///
2192 /// Only clears a *composition* run: a frontend that reports an end it never
2193 /// began (some IMEs unmark unprompted) would otherwise split the run of
2194 /// typing around it into two undo steps for no reason the user can see.
2195 pub fn end_composition(&mut self) {
2196 if self.last_edit_kind == Some(EditKind::Compose) {
2197 self.last_edit_kind = None;
2198 }
2199 }
2200
2201 // ── the clipboard's rich flavor ──────────────────────────────────────────
2202
2203 /// The selection rendered as HTML, for the clipboard's `text/html` flavor —
2204 /// what lets a paste into Docs/Mail/Slack keep its formatting. `None` when
2205 /// nothing is selected, or when the selection doesn't render (the caller
2206 /// still has [`selected_text`](Self::selected_text), which is what to publish
2207 /// as `text/plain` either way).
2208 ///
2209 /// **The fragment is a source substring, and that is the honest limit here.**
2210 /// It's parsed standalone, so a selection whose meaning depends on its
2211 /// surroundings converts as what it literally says rather than what it looks
2212 /// like on screen: half a list item is a paragraph, a row torn out of a table
2213 /// is the text of a row, the `**` of a bold run selected without its closing
2214 /// `**` is two asterisks. Every one of those still *renders* — there's no
2215 /// error to report — it just renders as the fragment and not as the document.
2216 /// Widening the range to whole blocks would publish text the user didn't
2217 /// select, which is a worse lie than a fragment being a fragment; the plain
2218 /// flavor has the same substring, so the two flavors at least agree.
2219 pub fn selection_html(&mut self) -> Option<String> {
2220 let (start, end) = self.selection()?;
2221 let inline = self.selection_is_inline(start, end);
2222 let html = html::render_fragment(&self.source[start..end], self.format)?;
2223 Some(match inline {
2224 true => html::strip_sole_paragraph(html),
2225 false => html,
2226 })
2227 }
2228
2229 /// Paste the clipboard's `text/html` flavor, converting it to this document's
2230 /// format first. Its own undo step, like any [`paste`](Self::paste).
2231 ///
2232 /// Returns whether it landed. `false` means the HTML didn't convert to
2233 /// anything worth pasting — the caller should fall back to the plain flavor
2234 /// rather than treat it as an error. The `html` module has the full list of
2235 /// what that covers: a table twig won't build, markup it doesn't recognise,
2236 /// an empty result.
2237 pub fn paste_html(&mut self, html: &str) -> bool {
2238 match html::parse_fragment(html, self.format) {
2239 Some(source) => {
2240 self.paste(&source);
2241 true
2242 }
2243 None => false,
2244 }
2245 }
2246
2247 /// Does the selection live *inside* a single top-level block?
2248 ///
2249 /// The question [`selection_html`](Self::selection_html) needs and the
2250 /// fragment can't answer: `**bold**` renders as `<p><strong>bold</strong></p>`
2251 /// whether the user selected one word of a sentence or a whole paragraph, and
2252 /// only the document knows which. Selecting a word and pasting into Docs
2253 /// should extend the line you paste into; selecting the paragraph should make
2254 /// a paragraph. So a selection strictly within one block is inline (its `<p>`
2255 /// is an artifact of standalone parsing), and one that covers a whole block —
2256 /// or spans two — keeps its structure.
2257 ///
2258 /// Reads the block from twig rather than guessing from the bytes:
2259 /// `ancestors_at` is `[doc, block, …inline]`, so index 1 is the top-level
2260 /// block containing an offset, and two ends inside the same one cannot have
2261 /// crossed a block boundary.
2262 fn selection_is_inline(&mut self, start: usize, end: usize) -> bool {
2263 // The last *character*, not `end - 1`: the selection's end is exclusive
2264 // and may sit mid-codepoint's-worth of bytes past the last char.
2265 let Some((off, _)) = self.source[start..end].char_indices().next_back() else {
2266 return false;
2267 };
2268 let (Some(head), Some(tail)) =
2269 (self.top_block_span(start), self.top_block_span(start + off))
2270 else {
2271 return false;
2272 };
2273 head == tail && !(start <= head.start && end >= head.end)
2274 }
2275
2276 /// The byte span of the top-level block containing `offset`, or `None` at an
2277 /// offset that belongs to no block (the blank line between two of them).
2278 fn top_block_span(&mut self, offset: usize) -> Option<std::ops::Range<usize>> {
2279 self.editor
2280 .ancestors_at(offset)
2281 .ok()?
2282 .get(1)
2283 .map(|m| m.span.clone())
2284 }
2285
2286 // ── indentation ──────────────────────────────────────────────────────────
2287
2288 /// One indent level.
2289 ///
2290 /// Two spaces, not the four both frontends type for Tab today, because in a
2291 /// markdown document four columns isn't a width — it's a *meaning*. Four
2292 /// spaces at the head of a line is markdown's indented-code-block marker, so
2293 /// one Tab on a paragraph would reparse it into code and style it as such;
2294 /// two cannot, and the line stays the prose it was. Two is also exactly
2295 /// where a `- ` bullet's content starts, so an indented line lands under its
2296 /// parent item's text instead of beside it — the column a list-aware indent
2297 /// has to hit anyway, which keeps this width from being relitigated later.
2298 const INDENT: &'static str = " ";
2299
2300 /// Indent the selected lines — or the caret's line, with no selection — by
2301 /// one level (Tab).
2302 pub fn indent(&mut self) {
2303 self.reindent(true);
2304 // Nesting changes an ordered list's numbering (the nested item restarts,
2305 // its old siblings resume) — keep the source markers in step.
2306 self.renumber_here();
2307 // Nesting an empty `-` item under a text line reparses that text as a
2308 // setext heading; swap the dash for a `*` before it can (a no-op unless
2309 // the collapse actually happened).
2310 self.avoid_setext_collapse();
2311 }
2312
2313 /// Take one indent level back off the selected lines, or the caret's line
2314 /// (Shift+Tab). A line with no indentation is left exactly as it is.
2315 ///
2316 /// A line with *less* than a full level gives back what it has rather than
2317 /// refusing: outdent's job is to walk a line left, and real documents — hand
2318 /// written, or reflowed by some other editor — are full of indentation that
2319 /// was never a clean multiple of anything. Refusing there would strand the
2320 /// line at a depth Shift+Tab couldn't undo.
2321 pub fn outdent(&mut self) {
2322 self.reindent(false);
2323 self.renumber_here();
2324 }
2325
2326 /// The body of [`indent`](Self::indent) / [`outdent`](Self::outdent).
2327 ///
2328 /// One splice across the whole line range, never one per line: a Tab is one
2329 /// thing the user did, so it has to be one undo step and one reparse. Per
2330 /// line, twig would reparse the document once per line and leave a stack of
2331 /// steps that Shift+⌘Z walks back one line at a time.
2332 fn reindent(&mut self, add: bool) {
2333 let (sel_start, sel_end) = self.selection().unwrap_or((self.caret, self.caret));
2334 let start = source_line_range(&self.source, sel_start).start;
2335 let end = source_line_range(&self.source, sel_end).end;
2336 let region = self.source[start..end].to_string();
2337 let lines: Vec<&str> = region.split('\n').collect();
2338 // A blank line has no text to move, and padding it would leave nothing
2339 // but trailing whitespace — but Tab on a blank line *is* a request for
2340 // indentation to type into, so the skip only applies where the op has
2341 // other lines to do real work on.
2342 let skip_blank = add && lines.len() > 1;
2343
2344 let mut out = String::with_capacity(region.len() + lines.len() * Self::INDENT.len());
2345 let mut deltas: Vec<isize> = Vec::with_capacity(lines.len());
2346 let mut line_off = start;
2347 for (i, full) in lines.iter().enumerate() {
2348 if i > 0 {
2349 out.push('\n');
2350 }
2351 // A list item moves by having its whole leading prefix *replaced*,
2352 // never by having spaces pushed in front of the line. twig spells
2353 // both prefixes, so the quote markers, the parent's indent and an
2354 // ordered marker's extra column all come out right without leaf
2355 // measuring any of them — and a line that only looks like an item
2356 // (a Djot continuation) reports no marker and is left to the plain
2357 // path, where a Tab is just a Tab.
2358 let marker = self.list_marker_on_line(line_off);
2359 let own = marker
2360 .as_ref()
2361 .map(|m| m.marker_start - m.line_start)
2362 .unwrap_or(0);
2363 let delta = if add {
2364 if skip_blank && full.trim().is_empty() {
2365 out.push_str(full);
2366 0
2367 } else if marker.is_some() && self.first_item_of_list(line_off) {
2368 // The first item of a list has no preceding sibling to nest
2369 // under, so a Tab here can't spell a sub-list — twig would
2370 // reparse the shoved-over marker as the same list, only
2371 // indented, which Shift+Tab then can't cleanly undo. Leave the
2372 // item where it is, the way every list editor refuses to
2373 // over-indent a list's first line.
2374 out.push_str(full);
2375 0
2376 } else if marker.is_some() {
2377 // Nesting means standing where a *continuation* of this line
2378 // would stand: past the parent's marker, inside its content
2379 // column. That is `continuation_prefix`, less a checkbox.
2380 let new = self.nesting_prefix_at(line_off);
2381 let delta = new.len() as isize - own as isize;
2382 out.push_str(&new);
2383 out.push_str(&full[own..]);
2384 delta
2385 } else {
2386 out.push_str(Self::INDENT);
2387 out.push_str(full);
2388 Self::INDENT.len() as isize
2389 }
2390 } else if marker.is_some() {
2391 // Unnesting is the mirror: stand where the parent item's own
2392 // line starts, which drops exactly the level it contributed.
2393 let new = self.outdent_prefix_at(line_off);
2394 let delta = new.len() as isize - own as isize;
2395 out.push_str(&new);
2396 out.push_str(&full[own..]);
2397 delta
2398 } else {
2399 // A plain line gives back the ordinary step.
2400 let strip = outdent_width(full, Self::INDENT.len());
2401 out.push_str(&full[strip..]);
2402 -(strip as isize)
2403 };
2404 deltas.push(delta);
2405 line_off += full.len() + 1;
2406 }
2407 // Nothing to give back. Returning before the splice keeps an outdent at
2408 // column zero from spending an undo step on a document it never changed.
2409 if deltas.iter().all(|d| *d == 0) {
2410 return;
2411 }
2412
2413 // Every line's text keeps its offset *within the line*, so the caret is
2414 // remapped by its column, not by its byte offset — which the prefixes on
2415 // the lines above it have already invalidated.
2416 let remap = |off: usize| -> usize {
2417 let (mut old_ls, mut new_ls) = (start, start);
2418 for (line, delta) in lines.iter().zip(&deltas) {
2419 let old_le = old_ls + line.len();
2420 let new_len = (line.len() as isize + delta) as usize;
2421 if off <= old_le {
2422 let col = (off - old_ls) as isize;
2423 return new_ls + ((col + delta).max(0) as usize).min(new_len);
2424 }
2425 old_ls = old_le + 1;
2426 new_ls += new_len + 1;
2427 }
2428 start + out.len()
2429 };
2430 let placed = match self.selection() {
2431 // Keep the rewritten region selected, the way a container toggle
2432 // keeps its own: it leaves a second Tab aimed at the same lines
2433 // rather than at whatever the shifted offsets now happen to cover.
2434 Some(_) => (start + out.len(), Some(start)),
2435 None => (remap(self.caret), None),
2436 };
2437
2438 // A rolled-back splice leaves the old source in place, where every offset
2439 // computed above addresses text that was never written.
2440 if !self.splice(start, end, &out, EditKind::Other) {
2441 return;
2442 }
2443 // `splice` re-anchors to the end of the `Change`, which for a whole-region
2444 // rewrite is the last line's end — nowhere the caret was. Place it, then
2445 // re-record the caret so this is the state redo restores, not the one
2446 // `splice` left behind from the `Change`.
2447 self.caret = placed.0.min(self.source.len());
2448 self.anchor = placed.1;
2449 self.clamp_caret();
2450 self.record_caret();
2451 }
2452
2453 /// The Enter key.
2454 ///
2455 /// In source view it's a literal newline. In WYSIWYG it's **AST-aware**: a
2456 /// bare `\n` is only a markdown soft break (same paragraph), so the block the
2457 /// caret is in decides what actually gets written.
2458 ///
2459 /// - paragraph → twig's [`Editor::split_block`], which parts the
2460 /// block at the caret and reopens its container
2461 /// - list item → likewise: the next item, its indent, quote
2462 /// prefix and `[ ]` box all reproduced by twig —
2463 /// except an *empty* item, which exits the list
2464 /// - block quote → likewise: a new paragraph inside the quote
2465 /// - heading → a new *paragraph*, not another heading
2466 /// - code block → a literal newline (stay in the block)
2467 /// - blank line → a literal newline (one Backspace undoes it)
2468 /// - [`LineFlow::Preserve`] → a single soft break, which renders as a
2469 /// visible line
2470 ///
2471 /// Where `split_block` is used it replaces markup leaf used to spell by hand,
2472 /// and it is better at it: it drops the whitespace the caret was sitting in
2473 /// front of instead of stranding it at the head of the second half, and it
2474 /// knows continuations leaf's marker scan never covered — a checklist item
2475 /// continues as an *unchecked* checklist item rather than a plain bullet.
2476 ///
2477 /// The exceptions above are exceptions because `split_block` is either wrong
2478 /// there or refuses: parting a fence yields two fences with the code split
2479 /// between them, parting a heading yields a second heading where every editor
2480 /// gives a paragraph, and a blank line, an empty item, a setext heading and a
2481 /// table all report an error rather than a split.
2482 pub fn newline(&mut self) {
2483 if self.view == View::Source {
2484 self.insert_raw("\n");
2485 return;
2486 }
2487 // Enter over a selection replaces it with a paragraph break.
2488 if let Some((s, e)) = self.selection() {
2489 self.splice(s, e, "\n\n", EditKind::Other);
2490 return;
2491 }
2492 // A caret resting exactly between an inline mark's content and its own
2493 // closing delimiter (`**bold**` with nothing after it on the line —
2494 // the WYSIWYG caret's natural end-of-line position) must not splice a
2495 // block break there: every path below eventually does via
2496 // `insert_raw`/`self.caret`, and splicing before the hidden closing
2497 // delimiter would strand it alone on the new line.
2498 self.caret = self.skip_trailing_close_delims(self.caret);
2499 // The block the caret is in. `block_offset_for_caret` nudges off a line
2500 // end (where the caret sits at the doc level); on a bare line (e.g. an
2501 // empty list item) fall back to the caret so the enclosing list/quote is
2502 // still visible in the ancestors.
2503 let off = self.block_offset_for_caret().unwrap_or(self.caret);
2504 let kinds: Vec<Kind> = self
2505 .editor
2506 .ancestors_at(off)
2507 .map(|c| c.into_iter().map(|m| m.kind).collect())
2508 .unwrap_or_default();
2509 let has = |k: Kind| kinds.contains(&k);
2510
2511 if has(Kind::CodeBlock) {
2512 self.insert_raw("\n");
2513 return;
2514 }
2515 // An *empty* list item exits the list — the standard double-Enter — which
2516 // `split_block` reports as an error rather than a split (there is no
2517 // content to part), so it stays leaf's. `list_marker_on_line` is itself
2518 // the AST gate — it answers from the tree, so a `- ` that reads as a
2519 // marker byte-for-byte but opens no item (a setext underline, a Djot
2520 // continuation line) never reaches here.
2521 if let Some(marker) = self.list_marker_on_line(self.caret)
2522 && self.item_is_empty(&marker)
2523 {
2524 self.exit_list(&marker);
2525 return;
2526 }
2527 // On an *empty* paragraph line, a lone Enter should add a single blank line,
2528 // not another full paragraph break — so it moves down one line and one
2529 // Backspace undoes it, not two. (`split_block` errors here too.)
2530 let line_start = self.source[..self.caret].rfind('\n').map_or(0, |i| i + 1);
2531 let line_end = self.source[self.caret..]
2532 .find('\n')
2533 .map_or(self.source.len(), |i| self.caret + i);
2534 if self.source[line_start..line_end].trim().is_empty() {
2535 self.insert_raw("\n");
2536 return;
2537 }
2538 // In `Preserve` flow a soft break is a *visible* line the author means to
2539 // make, so Enter writes a single `\n` and typing continues the same
2540 // paragraph on the next line — the behaviour of an ordinary text editor.
2541 // A second Enter then lands on the blank line above and takes the
2542 // empty-line branch, so double-Enter still promotes to a full paragraph
2543 // break; and Backspace, which deletes a lone `\n` over a soft break,
2544 // undoes a single Enter symmetrically. In `Fold` flow a lone `\n` would
2545 // render as an invisible space, so Enter keeps making the paragraph break
2546 // that actually shows.
2547 //
2548 // Only in running prose. A list or a quote has a continuation of its own
2549 // to write, and a `\n` there is not a soft line but a lost container.
2550 let in_container = has(Kind::ListItem) || has(Kind::TaskListItem) || has(Kind::BlockQuote);
2551 if self.line_flow == LineFlow::Preserve && !in_container {
2552 self.insert_raw("\n");
2553 return;
2554 }
2555 // A heading gets a *paragraph*, never a second heading: Enter at the end
2556 // of a title is how every editor is asked for the body under it, and
2557 // `split_block` would repeat the `#` instead. Whitespace at the split
2558 // point goes with the break rather than opening the new paragraph, which
2559 // is what `split_block` does everywhere else.
2560 if has(Kind::Heading) {
2561 let mut end = self.caret;
2562 while self.source.as_bytes().get(end) == Some(&b' ') {
2563 end += 1;
2564 }
2565 self.splice(self.caret, end, "\n\n", EditKind::Other);
2566 return;
2567 }
2568 self.split_block_here();
2569 }
2570
2571 /// Part the block at the caret with twig's [`Editor::split_block`], leaving
2572 /// the caret in the second half.
2573 ///
2574 /// twig reopens whatever the first half was inside of — the bullet with its
2575 /// indent, the quote's `>`, a checklist item's `[ ]` — which is the whole
2576 /// reason this replaced the markup leaf used to spell from the line's bytes.
2577 /// It renumbers nothing, though: a new item mid-list is written with its
2578 /// neighbour's number, so [`renumber_here`](Self::renumber_here) still runs
2579 /// behind it, folded into the same undo step.
2580 ///
2581 /// Falls back to a plain paragraph break if twig declines, so an unhandled
2582 /// shape still moves the caret down rather than swallowing the keystroke.
2583 fn split_block_here(&mut self) {
2584 // The read-only gate — this door reaches twig without the splice.
2585 if self.read_only {
2586 return;
2587 }
2588 match self.editor.split_block(self.caret) {
2589 Ok(change) => {
2590 self.last_edit_kind = None;
2591 self.refresh();
2592 self.anchor = None;
2593 self.caret = change.new.end;
2594 self.dirty = self.source != self.clean_source;
2595 self.status = None;
2596 self.clamp_caret();
2597 self.record_caret();
2598 // Aimed at the new block's *start*: the caret twig leaves is one
2599 // past the marker it wrote, where there is no list in reach.
2600 self.renumber_at(change.new.start);
2601 }
2602 Err(_) => self.insert_raw("\n\n"),
2603 }
2604 }
2605
2606 /// Whether the item on the marker's line carries no content — the shape
2607 /// double-Enter reads as "I'm done with this list."
2608 fn item_is_empty(&self, line: &ListMarker) -> bool {
2609 let content_start = line.content_start().min(self.source.len());
2610 let line_end = self.source[self.caret..]
2611 .find('\n')
2612 .map(|i| self.caret + i)
2613 .unwrap_or(self.source.len());
2614 self.source[content_start..line_end.max(content_start)]
2615 .trim()
2616 .is_empty()
2617 }
2618
2619 /// Leave the list: replace the empty item's marker with a blank line, so the
2620 /// caret lands in a fresh paragraph below it.
2621 ///
2622 /// Inside a quote the blank line has to stay quoted (a bare one would end the
2623 /// quote), and the caret's new line keeps the `> ` it was already behind —
2624 /// leaving the list without also leaving the quote.
2625 fn exit_list(&mut self, line: &ListMarker) {
2626 let prefix = self.quote_prefix_at(line.marker_start);
2627 let blank = prefix.trim_end();
2628 self.splice(
2629 line.line_start,
2630 self.caret,
2631 &format!("{blank}\n{prefix}"),
2632 EditKind::Other,
2633 );
2634 }
2635
2636 /// What a line continuing the containers at `off` has to open with — the
2637 /// quote markers reproduced, each enclosing item's marker as its width in
2638 /// spaces. Also the column a nested item's marker stands in, which is what
2639 /// makes it Tab's answer.
2640 fn continuation_prefix_at(&mut self, off: usize) -> String {
2641 self.editor
2642 .document()
2643 .and_then(|mut d| d.continuation_prefix(off))
2644 .map(|p| p.text)
2645 .unwrap_or_default()
2646 }
2647
2648 /// The column a *nested list* may open at inside the item at `off` — which
2649 /// is not always where the item's own text continues.
2650 ///
2651 /// twig counts a task item's `[ ] ` box as part of its marker, correctly:
2652 /// it is markup a rich view hides, and the item's own wrapped text does
2653 /// stand past it. But a nested list may only open at the *list* marker's
2654 /// column, and four columns further in is an indented continuation of the
2655 /// paragraph instead — `- [ ] a` + ` - [ ] b` is one item, not two.
2656 /// So the box's own width goes back.
2657 ///
2658 /// The one place leaf still reads a checkbox's spelling. It goes when twig
2659 /// reports the list marker's column apart from the box; `checked` is what
2660 /// says a box is there at all, so only its width is being measured here.
2661 fn nesting_prefix_at(&mut self, off: usize) -> String {
2662 let cont = self.continuation_prefix_at(off);
2663 let Some(item) = self.innermost_list_item(off) else {
2664 return cont;
2665 };
2666 if item.checked.is_none() {
2667 return cont;
2668 }
2669 let box_width = item
2670 .marker_span
2671 .and_then(|m| self.source.get(m))
2672 .and_then(|marker| marker.rfind('[').map(|i| marker.len() - i))
2673 .unwrap_or(0);
2674 // The trailing columns are the ones the item's own marker contributed,
2675 // so trimming from the end leaves any quote prefix standing.
2676 cont[..cont.len().saturating_sub(box_width)].to_string()
2677 }
2678
2679 /// Where the line of the item *containing* the item at `off` begins — the
2680 /// prefix Shift+Tab moves back to, which gives up exactly the level the
2681 /// parent contributed. The quote prefix alone for a top-level item, which
2682 /// has no level left to give.
2683 fn outdent_prefix_at(&mut self, off: usize) -> String {
2684 let items: Vec<usize> = self
2685 .editor
2686 .document()
2687 .and_then(|mut d| d.ancestors_at_caret(off))
2688 .map(|c| {
2689 c.into_iter()
2690 .filter(|m| m.kind == Kind::ListItem || m.kind == Kind::TaskListItem)
2691 .map(|m| m.span.start)
2692 .collect()
2693 })
2694 .unwrap_or_default();
2695 // The second-innermost item is the parent; its own line's indent is the
2696 // target. `list_marker_on_line` gives that line's prefix directly.
2697 let parent = items.len().checked_sub(2).map(|i| items[i]);
2698 match parent.and_then(|p| self.list_marker_on_line(p)) {
2699 Some(m) => self.source[m.line_start..m.marker_start].to_string(),
2700 None => self.quote_prefix_at(off),
2701 }
2702 }
2703
2704 /// The block-quote prefix in force at `off` — `""` outside a quote, `"> "`
2705 /// inside one, `"> > "` inside two.
2706 ///
2707 /// Assembled from each enclosing quote's own [`FlatNode::marker_span`], so
2708 /// the `>` and the space after it are twig's spelling rather than leaf's.
2709 /// The whole line prefix can't answer this: it also carries the indent of
2710 /// whatever the quote holds, which a blank separator line must *not* repeat.
2711 fn quote_prefix_at(&mut self, off: usize) -> String {
2712 let Ok(chain) = self
2713 .editor
2714 .document()
2715 .and_then(|mut d| d.ancestors_at_caret(off))
2716 else {
2717 return String::new();
2718 };
2719 let quotes: Vec<usize> = chain
2720 .iter()
2721 .filter(|m| m.kind == Kind::BlockQuote)
2722 .map(|m| m.node_id as usize)
2723 .collect();
2724 let Ok(nodes) = self.editor.nodes() else {
2725 return String::new();
2726 };
2727 quotes
2728 .iter()
2729 .filter_map(|id| nodes.get(*id)?.marker_span.clone())
2730 .filter_map(|s| self.source.get(s))
2731 .collect()
2732 }
2733
2734 /// Whether the item at `off` sits inside another one — the test Backspace
2735 /// uses to choose between outdenting and dropping the marker.
2736 ///
2737 /// Counted from the AST rather than from the line's leading whitespace,
2738 /// which is indentation in Markdown and, in Djot, may be nothing at all.
2739 fn item_is_nested(&mut self, off: usize) -> bool {
2740 self.editor
2741 .document()
2742 .and_then(|mut d| d.ancestors_at_caret(off))
2743 .map(|c| {
2744 c.into_iter()
2745 .filter(|m| m.kind == Kind::ListItem || m.kind == Kind::TaskListItem)
2746 .count()
2747 > 1
2748 })
2749 .unwrap_or(false)
2750 }
2751
2752 /// The innermost list item containing `probe`, under twig's **caret**
2753 /// containment rule — a block's end is inside it.
2754 ///
2755 /// Half-open containment can't answer this. An empty item's span is exactly
2756 /// its marker, so the caret sitting after `- ` is one past the end and the
2757 /// item it is plainly in tests as out of reach; that is the shape
2758 /// double-Enter has to recognise to leave the list.
2759 fn innermost_list_item(&mut self, probe: usize) -> Option<FlatNode> {
2760 let chain = self
2761 .editor
2762 .document()
2763 .and_then(|mut d| d.ancestors_at_caret(probe))
2764 .ok()?;
2765 let id = chain
2766 .iter()
2767 .rev()
2768 .find(|m| m.kind == Kind::ListItem || m.kind == Kind::TaskListItem)?
2769 .node_id as usize;
2770 self.editor.nodes().ok()?.get(id).cloned()
2771 }
2772
2773 /// The list marker opening `off`'s line, per twig — `None` when that line
2774 /// opens no list item.
2775 ///
2776 /// [`Document::line_prefix`] is the whole hidden run from the line start:
2777 /// `> 1. ` is a quote's marker, an indent, and an item's marker together,
2778 /// and it is `None` on a *continuation* line, which opens nothing. That last
2779 /// case is the one leaf could never get right by reading bytes. `- a\n - b`
2780 /// is two items in Markdown and one in Djot, where a marker cannot interrupt
2781 /// a paragraph and ` - b` is literal text — identical bytes, and only the
2782 /// parser knows which document it is looking at.
2783 ///
2784 /// The item's own marker is separated out via its
2785 /// [`FlatNode::marker_span`], so `marker_start` splits the prefix into what
2786 /// the containers around it contribute and what the item does.
2787 fn list_marker_on_line(&mut self, off: usize) -> Option<ListMarker> {
2788 let off = off.min(self.source.len());
2789 let prefix = self.editor.document().ok()?.line_prefix(off).ok()??;
2790 // The prefix belongs to a list only when an item's marker closes it —
2791 // a heading's `# ` or a bare quote's `> ` is a prefix too.
2792 let item = self.innermost_list_item(prefix.end.min(self.source.len()))?;
2793 let marker = item.marker_span.clone()?;
2794 if marker.end != prefix.end {
2795 return None;
2796 }
2797 Some(ListMarker {
2798 line_start: prefix.start,
2799 marker_start: marker.start,
2800 text: self.source.get(prefix)?.to_string(),
2801 })
2802 }
2803
2804 /// Whether the list item on `line_start`'s line is the **first item** of its
2805 /// list — the one Tab must not nest, because nesting needs a preceding
2806 /// sibling to become the new parent and a first item has none. `false` for a
2807 /// line that isn't a list item, and for an item with a sibling above it (the
2808 /// one Tab *can* nest). Gated on the AST, not the marker bytes: `- ` reads
2809 /// the same in a setext underline that opens no list at all.
2810 fn first_item_of_list(&mut self, line_start: usize) -> bool {
2811 let Some(marker) = self.list_marker_on_line(line_start) else {
2812 return false;
2813 };
2814 // Probe just inside the marker, where the item's own node is in reach —
2815 // the marker offset itself can resolve to the enclosing list, not the
2816 // `list_item`, whose span starts at the marker.
2817 let probe = marker.content_start().min(self.source.len());
2818 let Some(item) = self.innermost_list_item(probe) else {
2819 return false;
2820 };
2821 let Ok(nodes) = self.editor.nodes() else {
2822 return false;
2823 };
2824 match item.parent {
2825 // First when the parent list opens with this very item.
2826 Some(pid) => nodes
2827 .get(pid.0 as usize)
2828 .is_some_and(|p| p.first_child == Some(item.id)),
2829 // A parentless item is trivially the first (and only) one.
2830 None => true,
2831 }
2832 }
2833
2834 pub fn backspace(&mut self) {
2835 if let Some((s, e)) = self.selection() {
2836 self.splice(s, e, "", EditKind::Other);
2837 return;
2838 }
2839 // WYSIWYG: Backspace at the very start of a list item's content is a
2840 // structural key, not a character delete — it walks the "un-indent, then
2841 // un-list" ladder every list editor gives that keystroke (outdent a
2842 // nested item, strip a top-level one's marker to a paragraph). In source
2843 // view the `- ` is visible text the user is deleting a byte of, so it
2844 // keeps its literal meaning there, like Enter does.
2845 if self.view != View::Source && self.backspace_list_start() {
2846 return;
2847 }
2848 // WYSIWYG: and the same at the start of a heading's content — the `# `
2849 // there is markup the rich view hides, not text the user typed.
2850 if self.view != View::Source && self.backspace_heading_start() {
2851 return;
2852 }
2853 // WYSIWYG: at a block picture's stops, a byte-at-a-time delete would take
2854 // the markup apart under a caret that cannot see it — see
2855 // `delete_around_block_media`.
2856 if self.view != View::Source && self.delete_around_block_media(false) {
2857 return;
2858 }
2859 // WYSIWYG: Backspace at a table's trailing stop steps back into its last
2860 // cell rather than taking the byte behind the caret — the row's closing
2861 // `|`, which the rich view never drew, so the key would have looked like
2862 // it did nothing. The stop before is the last cell's end.
2863 if self.view != View::Source && self.backspace_at_table_end() {
2864 return;
2865 }
2866 // WYSIWYG: Backspace on a *blank line* deletes back to the previous caret
2867 // stop, not a single newline. On a line with no text of its own, the byte
2868 // before the caret is a `\n` that spells part of a block boundary — the gap
2869 // between two blocks, drawn but never a caret home. Removing just it strands
2870 // the caret in that gap and leaves an odd blank line the eye reads as one
2871 // separator but the caret can't land on: the "extra newline" left behind
2872 // after leaving a list (Enter, Enter) or a paragraph and pressing Backspace.
2873 // Deleting to the previous stop instead collapses the whole break at once,
2874 // landing the caret at the end of the block above. Two blank lines in a row
2875 // are one stop apart, so this still removes exactly one — the lone-Enter /
2876 // lone-Backspace symmetry the empty-line case is built on is untouched.
2877 if self.view != View::Source
2878 && self.caret > self.caret_floor()
2879 && self.caret_on_blank_line()
2880 && let Some(stop) = self.vmap.stop_before(self.caret)
2881 {
2882 let stop = stop.max(self.caret_floor());
2883 if stop < self.caret {
2884 self.splice(stop, self.caret, "", EditKind::Delete);
2885 return;
2886 }
2887 }
2888 if self.caret > self.caret_floor() {
2889 // An in-cell `<br>` draws as one newline glyph, so Backspace over it
2890 // takes the whole tag — a single-byte step would leave a broken `<br`
2891 // showing in the cell. Rich view only (source view edits the literal).
2892 if self.view != View::Source
2893 && let Some((start, end)) = self.cell_break_at(BreakEdge::Backward)
2894 {
2895 let start = start.max(self.caret_floor());
2896 if start < end {
2897 self.splice(start, end, "", EditKind::Delete);
2898 return;
2899 }
2900 }
2901 // Aim the delete at the character the writer can *see* behind the
2902 // caret, never at a delimiter the rich view drew nothing for. Two
2903 // steps, and either can apply: from the far side of a run's closing
2904 // `**` step back into the run (the caret is drawn at the end of its
2905 // word), and at the start of a run's text step out past its opening
2906 // `**` to the character in front of it, leaving the run standing.
2907 // Without them a plain Backspace unspells the phrase it is editing
2908 // and leaves a literal asterisk on screen.
2909 let end = if self.view == View::Source {
2910 self.caret
2911 } else {
2912 let inside = self.step_inside_close_delims(self.caret);
2913 self.skip_leading_open_delims(inside)
2914 .max(self.caret_floor())
2915 };
2916 // Never delete back across the floor — that would eat hidden
2917 // frontmatter the WYSIWYG caret can't even see.
2918 let mut prev = prev_boundary(&self.source, end).max(self.caret_floor());
2919 // Take a hidden escape backslash with the char it escapes: the rich
2920 // view draws `\*` as a single `*`, so Backspace over it must delete
2921 // both bytes, never strand the `\` as a lone visible backslash (the
2922 // mirror of the Hidden-mode typing that wrote the escape). Source view
2923 // shows the `\`, so there it is an ordinary character.
2924 if self.view != View::Source
2925 && prev > self.caret_floor()
2926 && self.is_hidden_escape(prev - 1)
2927 {
2928 prev -= 1;
2929 }
2930 if prev < end {
2931 self.splice(prev, end, "", EditKind::Delete);
2932 }
2933 }
2934 }
2935
2936 /// Backspace at a table's trailing stop: move onto the stop before it (the
2937 /// last cell's end) and consume the key. `false` anywhere else. See
2938 /// [`VisualMap::table_end_stop`] for why the byte behind the caret there is
2939 /// not one to delete.
2940 fn backspace_at_table_end(&mut self) -> bool {
2941 // The map answers about offsets, so it has to be this revision's — see
2942 // `open_paragraph_at_block_edge`.
2943 self.rebuild_map();
2944 if !self.vmap.table_end_stop(self.caret) {
2945 return false;
2946 }
2947 if let Some(off) = self
2948 .vmap
2949 .stop_before(self.caret)
2950 .filter(|&o| o >= self.caret_floor())
2951 {
2952 self.caret = off;
2953 self.anchor = None;
2954 self.goal_col = None;
2955 }
2956 true
2957 }
2958
2959 /// Whether the caret's own source line holds nothing but whitespace — an
2960 /// empty paragraph, or the blank line a block boundary is spelled with. The
2961 /// test for [`backspace`](Self::backspace)'s stop-wise delete: such a line has
2962 /// no text of its own, so the newline before the caret belongs to the gap
2963 /// between blocks rather than to any word the caret is editing.
2964 fn caret_on_blank_line(&self) -> bool {
2965 let line_start = self.source[..self.caret].rfind('\n').map_or(0, |i| i + 1);
2966 let line_end = self.source[self.caret..]
2967 .find('\n')
2968 .map_or(self.source.len(), |i| self.caret + i);
2969 self.source[line_start..line_end].trim().is_empty()
2970 }
2971
2972 /// The source span of an in-cell hard break (`<br>`) touching the caret on the
2973 /// `edge` side — the byte range to delete whole. A table row is one source
2974 /// line, so its break is spelled `<br>` yet drawn as a single newline glyph
2975 /// (see `wysiwyg.rs`); a delete over it must take every byte, or a one-byte
2976 /// step strands a broken `<br` in the cell. `Backward` matches a break ending
2977 /// at the caret (Backspace), `Forward` one starting at it (Delete). `None`
2978 /// when no such break is adjacent. Only the in-cell break is spelled `<br>`
2979 /// (an ordinary hard break is ` \n`), so the leading `<` alone tells them
2980 /// apart — no ancestor walk needed. Rich view only; source view shows the
2981 /// literal tag and deletes it a byte at a time.
2982 fn cell_break_at(&mut self, edge: BreakEdge) -> Option<(usize, usize)> {
2983 let caret = self.caret;
2984 let nodes = self.nodes();
2985 let src = self.source.as_bytes();
2986 nodes
2987 .iter()
2988 .find(|n| {
2989 n.kind == Kind::HardBreak
2990 && n.span.start < n.span.end
2991 && src.get(n.span.start) == Some(&b'<')
2992 && match edge {
2993 BreakEdge::Backward => n.span.end == caret,
2994 BreakEdge::Forward => n.span.start == caret,
2995 }
2996 })
2997 .map(|n| (n.span.start, n.span.end))
2998 }
2999
3000 /// Whether the source byte at `off` is a backslash twig consumed as an escape
3001 /// (hidden in the rich view), as against a literal backslash (drawn). A
3002 /// backslash escapes exactly an ASCII-punctuation character (the CommonMark /
3003 /// Djot rule twig follows), so `\` + punctuation is the whole test — no AST
3004 /// round-trip needed.
3005 fn is_hidden_escape(&self, off: usize) -> bool {
3006 let b = self.source.as_bytes();
3007 b.get(off) == Some(&b'\\') && b.get(off + 1).is_some_and(u8::is_ascii_punctuation)
3008 }
3009
3010 /// Backspace's list behaviour: when the caret sits exactly at the start of a
3011 /// list item's content (right after its marker), outdent the item if it's
3012 /// nested, else strip the marker so it becomes a paragraph. Returns whether
3013 /// it acted — `false` leaves Backspace its ordinary character delete.
3014 fn backspace_list_start(&mut self) -> bool {
3015 let Some(marker) = self.list_marker_on_line(self.caret) else {
3016 return false;
3017 };
3018 // Only right after the marker. That the line opens a real item is
3019 // already settled: `list_marker_on_line` answers from the tree.
3020 if self.caret != marker.content_start() {
3021 return false;
3022 }
3023 if self.item_is_nested(marker.marker_start) {
3024 // Nested: give back one level, keeping the marker and carrying the
3025 // caret with it.
3026 self.outdent();
3027 } else {
3028 // Top level: drop the marker, leaving a paragraph, then renumber the
3029 // siblings the removed item was counted among. Only the marker goes —
3030 // a quote prefix in front of it still has a quote to hold up.
3031 self.splice(marker.marker_start, self.caret, "", EditKind::Other);
3032 self.renumber_here();
3033 }
3034 true
3035 }
3036
3037 /// Backspace's heading behaviour: with the caret exactly at the start of an
3038 /// ATX heading's content — right after the `#` marker the rich view hides —
3039 /// strip the marker so the line becomes a paragraph. The peer of
3040 /// [`backspace_list_start`](Self::backspace_list_start)'s ladder, and the same
3041 /// reasoning: hidden block markup is structure, so the keystroke over it is
3042 /// structural.
3043 ///
3044 /// Without this the ordinary delete takes the space out of `# Title` and
3045 /// leaves `#Title`, which is no longer a heading at all — the hash the view
3046 /// had been hiding surfaces as literal text the user has to delete a second
3047 /// time, having never typed it. A closing sequence (`# Title #`, hidden at the
3048 /// other end) goes with the marker for the same reason.
3049 ///
3050 /// Returns whether it acted; `false` leaves Backspace its character delete.
3051 fn backspace_heading_start(&mut self) -> bool {
3052 let caret = self.caret;
3053 // The heading whose content opens exactly at the caret. A bare `#` has no
3054 // content span at all — its content starts (and ends) where the line does.
3055 let Some((span, content_end, marker)) = self.nodes().iter().find_map(|n| {
3056 let (start, end) = match &n.content_span {
3057 Some(c) => (c.start, c.end),
3058 None => (n.span.end, n.span.end),
3059 };
3060 (n.kind == Kind::Heading && start == caret)
3061 .then(|| (n.span.clone(), end, n.marker_span.clone()))
3062 }) else {
3063 return false;
3064 };
3065 // twig reports the marker's own extent, so there is nothing to walk back
3066 // over and no `#` in this file. A setext heading has no marker — its
3067 // content opens the line — so it falls through to the ordinary delete,
3068 // as does anything else sitting at a content start.
3069 // `m.end == caret` is what excludes a setext heading, whose marker is the
3070 // underline *after* the content rather than a prefix before it.
3071 let Some(marker) = marker.filter(|m| m.end == caret) else {
3072 return false;
3073 };
3074 let start = marker.start;
3075 // A closing `#` sequence is hidden too, so it can't be left behind. Only
3076 // when the tail really is one: trailing spaces alone are nothing to strip.
3077 let tail = &self.source[content_end..span.end];
3078 if tail.contains('#') && tail.chars().all(|c| c == '#' || c.is_whitespace()) {
3079 let kept = self.source[caret..content_end].to_string();
3080 self.splice(start, span.end, &kept, EditKind::Other);
3081 // The splice leaves the caret past the text it re-wrote; the caret
3082 // belongs where the content now starts, which is where it already was.
3083 self.caret = start;
3084 self.record_caret();
3085 } else {
3086 self.splice(start, caret, "", EditKind::Other);
3087 }
3088 true
3089 }
3090
3091 pub fn delete_forward(&mut self) {
3092 if let Some((s, e)) = self.selection() {
3093 self.splice(s, e, "", EditKind::Other);
3094 } else if self.caret < self.source.len() {
3095 // The mirror of Backspace's: forward-delete in front of a picture
3096 // would eat the `!` off its markup and leave a link where a photo was.
3097 if self.view != View::Source && self.delete_around_block_media(true) {
3098 return;
3099 }
3100 // Delete forward over an in-cell `<br>` takes the whole tag, the mirror
3101 // of Backspace's swallow (see `cell_break_at`) — else a byte-step
3102 // strands a broken `<br` in the cell.
3103 if self.view != View::Source
3104 && let Some((start, end)) = self.cell_break_at(BreakEdge::Forward)
3105 {
3106 self.splice(start, end, "", EditKind::Delete);
3107 return;
3108 }
3109 // The mirror of Backspace's two steps: from in front of a run's
3110 // opening `**` step into it, onto the first letter of its text, and
3111 // at the end of a run's text step out past its closing `**` to the
3112 // character beyond. Either way Delete takes the character it looks
3113 // like it is pointing at, and never a delimiter drawn as nothing.
3114 // The caret then settles back inside the run it was standing in —
3115 // see `settle_inside_close_delims`.
3116 let from = if self.view == View::Source {
3117 self.caret
3118 } else {
3119 let inside = self.step_inside_open_delims(self.caret);
3120 self.skip_trailing_close_delims(inside)
3121 };
3122 let next = next_boundary(&self.source, from);
3123 if from < next {
3124 self.splice(from, next, "", EditKind::Delete);
3125 }
3126 }
3127 }
3128
3129 /// Delete from the caret back to the start of the previous word (⌥⌫ /
3130 /// Ctrl+⌫). Deletes the selection instead when one is active.
3131 pub fn delete_word_back(&mut self) {
3132 if let Some((s, e)) = self.selection() {
3133 self.splice(s, e, "", EditKind::Other);
3134 } else {
3135 // A word back from just past a picture is a word *of its markup*, and
3136 // a word back from in front of one runs through the paragraph break
3137 // into the prose above — dissolving the picture either way. See
3138 // `delete_around_block_media`.
3139 if self.view != View::Source && self.delete_around_block_media(false) {
3140 return;
3141 }
3142 let start = self.word_left_from(self.caret).max(self.caret_floor());
3143 if start < self.caret {
3144 let (s, e) = self.widen_over_emptied_inlines(start, self.caret);
3145 self.splice(s, e, "", EditKind::Delete);
3146 }
3147 }
3148 }
3149
3150 /// Delete from the caret forward to the end of the next word (⌥⌦ /
3151 /// Ctrl+Del). Deletes the selection instead when one is active.
3152 pub fn delete_word_forward(&mut self) {
3153 if let Some((s, e)) = self.selection() {
3154 self.splice(s, e, "", EditKind::Other);
3155 } else {
3156 // The mirror: a word forward from in front of a picture is its markup.
3157 if self.view != View::Source && self.delete_around_block_media(true) {
3158 return;
3159 }
3160 let end = self.word_right_from(self.caret);
3161 if end > self.caret {
3162 let (s, e) = self.widen_over_emptied_inlines(self.caret, end);
3163 self.splice(s, e, "", EditKind::Delete);
3164 }
3165 }
3166 }
3167
3168 /// Delete from the caret back to the start of its line (⌘⌫). Deletes the
3169 /// selection instead when one is active, as every other delete here does.
3170 ///
3171 /// The line is the view's own — the one Home and End work on, so in WYSIWYG
3172 /// a soft-wrapped row is a line. It is not Home's *target*, though: Home
3173 /// stops at the first character and this takes the indentation with it, the
3174 /// way Cocoa's `deleteToBeginningOfLine:` does. Stopping at the text would
3175 /// leave an indent behind that nothing can then ask to delete, where a caret
3176 /// left at column 0 is one press of Home away from either.
3177 pub fn delete_to_line_start(&mut self) {
3178 if let Some((s, e)) = self.selection() {
3179 self.splice(s, e, "", EditKind::Other);
3180 return;
3181 }
3182 // Never back across the floor: hidden frontmatter isn't on this line, or
3183 // on any line the WYSIWYG caret can see.
3184 let (start, _) = self.line_span();
3185 let start = start.max(self.caret_floor());
3186 if start < self.caret {
3187 let (s, e) = self.widen_over_emptied_inlines(start, self.caret);
3188 self.splice(s, e, "", EditKind::Delete);
3189 }
3190 }
3191
3192 /// Kill from the caret to the end of its line (^K). Deletes the selection
3193 /// instead when one is active.
3194 ///
3195 /// At the end of the line it does nothing, rather than pulling the line
3196 /// below up into this one. Joining has no meaning to give it in both views
3197 /// at once: a WYSIWYG line ends at a soft wrap as often as at a newline, and
3198 /// there is nothing there to delete, while the newline a *source* line ends
3199 /// with is only half of the blank line that separates two paragraphs —
3200 /// deleting one leaves a soft break, which is not the join it looks like.
3201 /// The views agreeing is worth more than emacs' second press, and Delete is
3202 /// already the key that joins.
3203 pub fn delete_to_line_end(&mut self) {
3204 if let Some((s, e)) = self.selection() {
3205 self.splice(s, e, "", EditKind::Other);
3206 return;
3207 }
3208 let (_, end) = self.line_span();
3209 if end > self.caret {
3210 let (s, e) = self.widen_over_emptied_inlines(self.caret, end);
3211 self.splice(s, e, "", EditKind::Delete);
3212 }
3213 }
3214
3215 /// Grow a WYSIWYG word-delete to swallow any inline node it empties.
3216 ///
3217 /// A glyph-space range covers what the user can see, which for `**bold**` is
3218 /// the word and never the delimiters around it — so deleting the word on its
3219 /// own leaves `a **** c`, markup wrapped around nothing. They asked for the
3220 /// word, and the styling was the word's; the two go together. Only the
3221 /// node's delimiters are taken, and those are hidden here anyway, so nothing
3222 /// visible outside the range is lost.
3223 ///
3224 /// Repeated to a fixed point: emptying `***bold***` empties the emph inside
3225 /// the strong, and only then is the strong empty too.
3226 fn widen_over_emptied_inlines(&mut self, start: usize, end: usize) -> (usize, usize) {
3227 if self.view == View::Source {
3228 return (start, end);
3229 }
3230 let nodes = self.nodes();
3231 let (mut s, mut e) = (start, end);
3232 loop {
3233 let mut grew = false;
3234 for n in nodes.iter().filter(|n| wysiwyg::is_inline(n)) {
3235 let Some(text) = inline_content_span(n, &self.source) else {
3236 continue;
3237 };
3238 // Some of its text survives, so the node still has a job.
3239 if text.start < s || text.end > e {
3240 continue;
3241 }
3242 if n.span.start < s || n.span.end > e {
3243 s = s.min(n.span.start);
3244 e = e.max(n.span.end);
3245 grew = true;
3246 }
3247 }
3248 if !grew {
3249 return (s, e);
3250 }
3251 }
3252 }
3253
3254 /// One splice of document text, keeping the **mark-edge rule**: an inline
3255 /// mark's content never begins or ends with whitespace. In Markdown and Djot
3256 /// a delimiter standing against a space is not a delimiter at all — `**bold **`
3257 /// is four literal asterisks around a word, and a rich view drawing the
3258 /// document faithfully has no choice but to show them. That is correct
3259 /// rendering of what the file says, and nobody typing a space after a bold
3260 /// word meant to say it.
3261 ///
3262 /// So the space goes *outside* the run instead — `**bold** ` — which is the
3263 /// same document to a reader and a live one to a parser. The caret follows it
3264 /// out and keeps the marks armed (see [`rearm`](Self::rearm)), so the next
3265 /// character rejoins the run (see [`rejoin_run`](Self::rejoin_run)) and the
3266 /// writer sees one unbroken bold phrase, never a flash of raw syntax.
3267 ///
3268 /// Every ordinary edit — typing, deleting, pasting, an IME step — comes
3269 /// through here, so the rule holds however the whitespace arrives at the
3270 /// edge. The repair is decided *after* the plain edit, by asking whether the
3271 /// mark actually died: a code span's backticks aren't whitespace-sensitive
3272 /// (`` `code ` `` is still code), and nothing is re-spelled when nothing broke.
3273 fn splice(&mut self, start: usize, end: usize, text: &str, kind: EditKind) -> bool {
3274 let fix = self.mark_edge_fix(start, end, text);
3275 if !self.splice_exact(start, end, text, kind) {
3276 return false;
3277 }
3278 if let Some(fix) = fix {
3279 self.repair_mark_edges(fix);
3280 }
3281 if text.is_empty() && end > start {
3282 self.settle_inside_close_delims();
3283 }
3284 true
3285 }
3286
3287 /// After a delete, take a caret left standing past a run's closing delimiters
3288 /// back inside the run.
3289 ///
3290 /// A delete leaves the caret where the deleted bytes began, and when those
3291 /// bytes were the last thing after a marked phrase — the space the mark-edge
3292 /// rule pushed out of `**bold** `, say — that spot is the far side of the
3293 /// closing `**`. The rich view has nothing to draw there: the delimiters are
3294 /// hidden, so the caret shows at the end of the word either way, and the two
3295 /// offsets are one place on screen with two different meanings. Typing at the
3296 /// outer one lands past the run, so the writer who backspaced a space out of
3297 /// their bold phrase watches the next character come out plain, and the
3298 /// toolbar button go dark, with the caret never appearing to move.
3299 ///
3300 /// The end of the run's text is the caret's home there — a delete that took
3301 /// away everything after a phrase leaves the caret at the end of that phrase,
3302 /// which is inside it — so it settles onto that
3303 /// ([`step_inside_close_delims`](Self::step_inside_close_delims) does the
3304 /// walk, through every mark closing at the point): the word stays bold, the
3305 /// button stays lit, and the next character carries on the phrase.
3306 ///
3307 /// Rich view only, and only where a mark really closes at the caret — mid-run
3308 /// or in plain prose no span ends there and the caret stays put. The opening
3309 /// edge is left alone on purpose: a caret in front of a run inherits from the
3310 /// text on its left, which is the plain text outside.
3311 fn settle_inside_close_delims(&mut self) {
3312 if self.view != View::Wysiwyg {
3313 return;
3314 }
3315 let at = self.step_inside_close_delims(self.caret);
3316 if at != self.caret {
3317 self.caret = at;
3318 self.clear_pending();
3319 self.record_caret();
3320 }
3321 }
3322
3323 /// The splice exactly as asked, with no mark-edge repair — for the callers
3324 /// that are *writing* the delimiters themselves ([`insert_with_marks`](Self::insert_with_marks)
3325 /// and [`rejoin_run`](Self::rejoin_run)) and place their own offsets around
3326 /// the bytes they inserted.
3327 ///
3328 /// One `edit_range` through twig, then re-anchor the caret from the returned
3329 /// `Change` and refresh the cached source. A reparse-breaking edit (rare for
3330 /// Markdown/Djot) leaves the document untouched and reports.
3331 ///
3332 /// Returns whether the edit landed — for a caller that has offsets of its
3333 /// own to place afterwards, which a rolled-back splice would leave pointing
3334 /// into text that never came to exist.
3335 fn splice_exact(&mut self, start: usize, end: usize, text: &str, kind: EditKind) -> bool {
3336 // The read-only gate, for every edit at once — see the field.
3337 if self.read_only {
3338 return false;
3339 }
3340 // twig records an undo step for every edit; when this one continues a
3341 // run of the same kind (typing, deleting), tell twig to fold it into the
3342 // step before it so the whole run undoes at once.
3343 let coalesce = kind != EditKind::Other && self.last_edit_kind == Some(kind);
3344 // Hand twig the pre-edit caret before the splice, so the undo step it
3345 // retires carries where the caret was standing.
3346 self.record_caret();
3347 match self.editor.edit_range(start, end, text) {
3348 Ok(change) => {
3349 if coalesce {
3350 let _ = self.editor.coalesce_last_undo();
3351 }
3352 self.last_edit_kind = Some(kind);
3353 self.refresh();
3354 self.caret = change.new.end;
3355 self.anchor = None;
3356 self.goal_col = None;
3357 self.clear_pending();
3358 self.dirty = self.source != self.clean_source;
3359 self.status = None;
3360 // And the post-edit caret, so a later redo restores it.
3361 self.record_caret();
3362 true
3363 }
3364 // The edit was rolled back, so twig's history did not move and
3365 // neither may ours: pushing here would leave a step with no edit
3366 // under it and shift every later undo onto the wrong caret.
3367 Err(e) => {
3368 self.status = Some(format!("edit: {e}"));
3369 false
3370 }
3371 }
3372 }
3373
3374 /// The re-spelling that would keep the mark-edge rule for the edit
3375 /// `[start, end)` → `text`, or `None` when the edit leaves no whitespace
3376 /// against a delimiter and the plain splice is already right. Computed
3377 /// *before* the edit, while the run's spans and delimiters can still be read
3378 /// off the document; applied afterwards, and only if the mark really died —
3379 /// see [`repair_mark_edges`](Self::repair_mark_edges).
3380 ///
3381 /// Rich view only. Source view is for typing raw markup, where a space put
3382 /// against a `**` is exactly the character it looks like.
3383 fn mark_edge_fix(&mut self, start: usize, end: usize, text: &str) -> Option<MarkEdgeFix> {
3384 if self.view != View::Wysiwyg || start > end || end > self.source.len() {
3385 return None;
3386 }
3387 // Every inline mark standing over the edit, outermost first, with the
3388 // content span that says where its delimiters are.
3389 let chain: Vec<(InlineKind, std::ops::Range<usize>, std::ops::Range<usize>)> = self
3390 .editor
3391 .ancestors_at(start)
3392 .unwrap_or_default()
3393 .into_iter()
3394 .filter_map(|m| {
3395 let kind = inline_kind(&m.kind)?;
3396 let content = m.content_span.clone()?;
3397 Some((kind, m.span.clone(), content))
3398 })
3399 .collect();
3400 // The innermost run whose *content* holds the whole edit: the one whose
3401 // text is being changed, rather than one the edit merely sits under.
3402 let (kind, span, content) = chain
3403 .iter()
3404 .rev()
3405 .find(|(_, _, c)| c.start <= start && end <= c.end)?
3406 .clone();
3407 // What that content becomes. Whitespace at either end of it is what
3408 // would put out the mark.
3409 let body = format!(
3410 "{}{text}{}",
3411 &self.source[content.start..start],
3412 &self.source[end..content.end]
3413 );
3414 let (lead, trail) = if body.trim().is_empty() {
3415 // Nothing but whitespace left: there is no content to mark at all,
3416 // and the delimiters go with it rather than closing on a space.
3417 (body.len(), 0)
3418 } else {
3419 (
3420 body.len() - body.trim_start().len(),
3421 body.len() - body.trim_end().len(),
3422 )
3423 };
3424 // Nothing against a delimiter, and something still between them: the
3425 // plain edit stands. An emptied run is broken just as surely (`**b**`
3426 // with the `b` deleted is the literal `****`) and is re-spelt as the
3427 // nothing it now says.
3428 if lead == 0 && trail == 0 && !body.is_empty() {
3429 return None;
3430 }
3431 // Marks that open or close exactly where this one does — `***both***` is
3432 // two runs sharing an edge — spell their delimiters as one run of bytes,
3433 // so the whitespace has to clear all of them together.
3434 let (mut open_at, mut close_at) = (span.start, span.end);
3435 for _ in 0..chain.len() {
3436 match chain.iter().find(|(_, _, c)| c.start == open_at) {
3437 Some((_, s, _)) => open_at = s.start,
3438 None => break,
3439 }
3440 }
3441 for _ in 0..chain.len() {
3442 match chain.iter().find(|(_, _, c)| c.end == close_at) {
3443 Some((_, s, _)) => close_at = s.end,
3444 None => break,
3445 }
3446 }
3447 let open = &self.source[open_at..content.start];
3448 let close = &self.source[content.end..close_at];
3449 let core = &body[lead..body.len() - trail];
3450 let respelt = if core.is_empty() {
3451 body.clone()
3452 } else {
3453 format!(
3454 "{}{open}{core}{close}{}",
3455 &body[..lead],
3456 &body[body.len() - trail..]
3457 )
3458 };
3459 // The caret sits just past the inserted text within the new content —
3460 // which, when that lands in the whitespace, is now outside the delimiters.
3461 let pos = (start - content.start) + text.len();
3462 let caret = if core.is_empty() || pos <= lead {
3463 open_at + pos
3464 } else if pos >= lead + core.len() {
3465 open_at + lead + open.len() + core.len() + close.len() + (pos - lead - core.len())
3466 } else {
3467 open_at + lead + open.len() + (pos - lead)
3468 };
3469 Some(MarkEdgeFix {
3470 kind,
3471 probe: content.start,
3472 start: open_at,
3473 end: close_at + text.len() - (end - start),
3474 text: respelt,
3475 caret,
3476 // The marks in force here, resolved against any armed sticky delta —
3477 // what the writer is typing in, and so what has to still be true on
3478 // the far side of the delimiter the caret just stepped over.
3479 want: chain
3480 .iter()
3481 .filter(|(_, s, _)| start < s.end)
3482 .map(|(k, _, _)| *k)
3483 .collect::<InlineMarks>()
3484 .xor(self.pending_here()),
3485 })
3486 }
3487
3488 /// Apply a [`MarkEdgeFix`] — but only if the edit it was computed for really
3489 /// did break the mark. Whether whitespace at a delimiter is fatal is the
3490 /// format's business, not leaf's: `**bold **` is no longer strong, while
3491 /// `` `code ` `` is still perfectly good verbatim, and Djot's braced spellings
3492 /// don't care either. Asking the parser afterwards settles it for every kind
3493 /// and format at once, and costs a re-spelling only where one is due.
3494 ///
3495 /// The repair rides along with the edit that caused it — one undo step puts
3496 /// back what the writer typed, not a delimiter shuffle they never saw.
3497 fn repair_mark_edges(&mut self, fix: MarkEdgeFix) {
3498 if fix.end > self.source.len() {
3499 return;
3500 }
3501 if self.marks_at(fix.probe).iter().any(|(k, _)| *k == fix.kind) {
3502 return; // still a mark: these delimiters don't mind the whitespace
3503 }
3504 let resumed = self.last_edit_kind;
3505 if !self.splice_exact(fix.start, fix.end, &fix.text, EditKind::Other) {
3506 return;
3507 }
3508 let _ = self.editor.coalesce_last_undo();
3509 // The keystroke owns the undo step, so the run of typing it belongs to
3510 // keeps coalescing over the repair rather than breaking in two here.
3511 self.last_edit_kind = resumed;
3512 self.caret = fix.caret.min(self.source.len());
3513 self.anchor = None;
3514 self.goal_col = None;
3515 self.rearm(fix.want);
3516 self.clamp_caret();
3517 self.record_caret();
3518 }
3519
3520 /// Arm whatever sticky delta reproduces `want` at the caret — the marks the
3521 /// writer is typing in, carried across an edit that moved the caret out of
3522 /// the run holding them. Arms nothing when the caret already stands in
3523 /// exactly those marks, but still remembers the spot, so a further ⌘b starts
3524 /// a clean delta here (see [`toggle`](Self::toggle)).
3525 fn rearm(&mut self, want: InlineMarks) {
3526 let here: InlineMarks = self
3527 .marks_at(self.caret)
3528 .into_iter()
3529 .map(|(k, _)| k)
3530 .collect();
3531 self.pending_marks = want.xor(here);
3532 self.pending_at = Some(self.caret);
3533 }
3534
3535 /// Insert `text` at `at` as a *literal* run via twig's `insert_literal`,
3536 /// which backslash-escapes any character that would otherwise open markup in
3537 /// this format and position (`*` → `\*`, a line-start `#` → `\#`). The mirror
3538 /// of [`splice`](Self::splice) for the Hidden reveal mode's typing path, with
3539 /// the same caret re-anchor, coalescing, and rollback contract. `at` must be
3540 /// a collapsed point — a selection is deleted by the caller first, since
3541 /// `insert_literal` inserts rather than replaces.
3542 fn insert_literal_at(
3543 &mut self,
3544 at: usize,
3545 text: &str,
3546 kind: EditKind,
3547 force_coalesce: bool,
3548 ) -> bool {
3549 // The read-only gate: this door goes to twig directly, not through
3550 // `splice_exact`, so it guards itself — see the field.
3551 if self.read_only {
3552 return false;
3553 }
3554 // `force_coalesce` folds this into the immediately preceding edit (the
3555 // selection-delete of an overwrite) so the pair is one undo step; else it
3556 // coalesces only when it continues a run of the same-kind typing.
3557 let coalesce =
3558 force_coalesce || (kind != EditKind::Other && self.last_edit_kind == Some(kind));
3559 // The mark-edge rule holds for typed text however it is spelled — see
3560 // `splice`. Only an insert twig passed through unchanged can use it,
3561 // since a fix is measured in the bytes that actually land, and an escape
3562 // adds bytes this couldn't have counted.
3563 let fix = self.mark_edge_fix(at, at, text);
3564 self.record_caret();
3565 match self.editor.insert_literal(at, text) {
3566 Ok(change) => {
3567 if coalesce {
3568 let _ = self.editor.coalesce_last_undo();
3569 }
3570 self.last_edit_kind = Some(kind);
3571 self.refresh();
3572 self.caret = change.new.end;
3573 self.anchor = None;
3574 self.goal_col = None;
3575 self.clear_pending();
3576 self.dirty = self.source != self.clean_source;
3577 self.status = None;
3578 self.record_caret();
3579 if let Some(fix) = fix.filter(|_| change.new.end - change.new.start == text.len()) {
3580 self.repair_mark_edges(fix);
3581 }
3582 true
3583 }
3584 Err(e) => {
3585 self.status = Some(format!("edit: {e}"));
3586 false
3587 }
3588 }
3589 }
3590
3591 /// After a structural list edit (a new item, a nest/unnest), renumber the
3592 /// ordered list the caret sits in so its source markers run `1, 2, 3, …`
3593 /// again — a raw splice leaves them stale (`1. 2. 2. 3.`). twig does the
3594 /// renumber as its own edit; fold it into the edit that triggered it so the
3595 /// two undo as one, and only when it actually changed the source (a no-op or
3596 /// a caret outside any ordered list must not coalesce the real edit into the
3597 /// step before it).
3598 fn renumber_here(&mut self) {
3599 self.renumber_at(self.caret);
3600 }
3601
3602 /// [`renumber_here`](Self::renumber_here) aimed somewhere other than the
3603 /// caret — for an edit that leaves the caret one past the item it just wrote,
3604 /// where twig resolves no list to renumber.
3605 fn renumber_at(&mut self, off: usize) {
3606 // The read-only gate — this door reaches twig without the splice.
3607 if self.read_only {
3608 return;
3609 }
3610 let before = self.source.clone();
3611 if self.editor.renumber_ordered_lists(off).is_err() {
3612 return; // not inside an ordered list — nothing to renumber
3613 }
3614 self.refresh();
3615 if self.source != before {
3616 let _ = self.editor.coalesce_last_undo();
3617 self.dirty = self.source != self.clean_source;
3618 self.clamp_caret();
3619 self.record_caret();
3620 }
3621 }
3622
3623 /// Repair the one trap a list edit can spring on itself. An *empty* `-`
3624 /// sub-item written directly beneath a text line reparses that text as a
3625 /// setext heading — `- hello\n - ` is `<h2>hello</h2>`, because a lone `-`
3626 /// is also a setext-H2 underline (twig is right; pandoc agrees). `*` and `+`
3627 /// bullets can't underline anything, so swap the dash for a `*`: the item
3628 /// stays an empty nested bullet, the parent stays prose, and the source
3629 /// round-trips instead of hiding a heading the user never asked for. Folded
3630 /// into the triggering edit's undo step, the way renumbering is.
3631 ///
3632 /// Gated on the collapse having actually happened (the swapped dash was
3633 /// swallowed into a `heading`), so a real setext heading the author wrote —
3634 /// or a `- x` with content, which can't underline anything — is never
3635 /// touched. This has to live in the *edit*, not the renderer: leaving the
3636 /// hazardous bytes on disk and only painting over them would ship a file
3637 /// every other CommonMark tool reads as a heading.
3638 ///
3639 /// This one keeps its own byte scan, and has to: the hazard is precisely
3640 /// that the dash stopped being a list marker, so [`list_marker_on_line`] —
3641 /// which asks twig which lines open an item — reports nothing here. There is
3642 /// no node to ask about. It is also the last Markdown spelling leaf writes on
3643 /// purpose rather than for want of an answer; once twig spells continuations
3644 /// itself, avoiding the trap becomes twig's, and this goes.
3645 ///
3646 /// [`list_marker_on_line`]: Self::list_marker_on_line
3647 fn avoid_setext_collapse(&mut self) {
3648 let caret = self.caret.min(self.source.len());
3649 let line_start = self.source[..caret].rfind('\n').map_or(0, |i| i + 1);
3650 let bytes = self.source.as_bytes();
3651 let mut dash = line_start;
3652 while matches!(bytes.get(dash), Some(b' ' | b'\t')) {
3653 dash += 1;
3654 }
3655 // A dash bullet is the only marker that doubles as a setext underline.
3656 if bytes.get(dash) != Some(&b'-') {
3657 return;
3658 }
3659 // Only an *empty* item is a bare underline; `- x` carries content and
3660 // can't fold the line above into a heading.
3661 let line_end = self.source[dash..]
3662 .find('\n')
3663 .map_or(self.source.len(), |i| dash + i);
3664 if !self.source[dash + 1..line_end].trim().is_empty() {
3665 return;
3666 }
3667 // The tell: that dash was swallowed into a `heading`. A properly nested
3668 // empty item sits under a `list_item`, with no heading in reach. Probe
3669 // the dash byte itself (well inside the heading), not the caret, whose
3670 // end-of-line offset can fall on the half-open span boundary.
3671 let collapsed = self
3672 .editor
3673 .ancestors_at(dash)
3674 .map(|c| c.into_iter().any(|m| m.kind == Kind::Heading))
3675 .unwrap_or(false);
3676 if !collapsed {
3677 return;
3678 }
3679 let caret = self.caret;
3680 if self.splice(dash, dash + 1, "*", EditKind::Other) {
3681 // Same width, so the caret keeps its column; fold into the edit that
3682 // triggered this so Tab stays one undo step.
3683 let _ = self.editor.coalesce_last_undo();
3684 self.caret = caret.min(self.source.len());
3685 self.clamp_caret();
3686 self.record_caret();
3687 }
3688 }
3689
3690 fn snapshot(&self) -> CaretState {
3691 CaretState {
3692 caret: self.caret,
3693 anchor: self.anchor,
3694 }
3695 }
3696
3697 /// Hand twig the current caret and selection as the blob for the live
3698 /// document state. Called before an edit — so the step twig retires records
3699 /// where the caret was, and undo can restore it — and again once the op has
3700 /// placed the caret, so redo restores where the edit left it.
3701 ///
3702 /// This is the whole of leaf's undo-caret bookkeeping now. twig carries the
3703 /// caret through its own history, so coalescing falls out for free (folding
3704 /// two twig steps into one drops the intermediate blob, keeping the run's
3705 /// first) and the parallel stacks that had to march in lockstep — and could
3706 /// silently drift out of it — are gone.
3707 fn record_caret(&mut self) {
3708 let _ = self.editor.set_caret_blob(&self.snapshot().to_blob());
3709 }
3710
3711 /// Toggle an inline mark over the selection (Bold / Italic / Code / …). Keeps
3712 /// the toggled region selected so a second press cleanly reverses it.
3713 pub fn toggle(&mut self, kind: InlineKind) {
3714 // The read-only gate — this door reaches twig without the splice.
3715 if self.read_only {
3716 return;
3717 }
3718 // Ahead of the no-selection branch below: arming a mark for text not yet
3719 // typed is a promise `insert` cannot keep in a format with no delimiters
3720 // to spell it with. Per *kind*, not per format — Markdown spells five
3721 // of the eight marks (highlight among them, under the `highlight`
3722 // extension leaf parses with), djot all eight, HTML seven.
3723 if self.refuse_unsupported(&format!("{kind:?}"), Gesture::ToggleInline(kind)) {
3724 return;
3725 }
3726 let Some((s, e)) = self.selection() else {
3727 // No selection: arm the mark for the next text typed here, the way a
3728 // word processor does. `⌘b`, type, `⌘b` again toggles bold on and off
3729 // in the flow of typing without ever selecting anything — the delta
3730 // is realised onto the freshly typed text by `insert`. A fresh caret
3731 // position starts the delta over from the marks actually in force.
3732 if self.pending_at != Some(self.caret) {
3733 self.pending_marks = InlineMarks::empty();
3734 self.pending_at = Some(self.caret);
3735 }
3736 self.pending_marks.flip(kind);
3737 self.status = None;
3738 return;
3739 };
3740 // Whitespace at the edge of a selection is not part of what was chosen —
3741 // a double-click takes the space after the word with it — and a mark
3742 // cannot close against one anyway: `**word **` is four literal asterisks
3743 // (the mark-edge rule, see `splice`). Mark the words, leave the spaces.
3744 let picked = &self.source[s..e];
3745 let (s, e) = (
3746 s + (picked.len() - picked.trim_start().len()),
3747 e - (picked.len() - picked.trim_end().len()),
3748 );
3749 if s >= e {
3750 self.status = Some(format!("{kind:?}: nothing selected to mark"));
3751 return;
3752 }
3753 // Styling a selection is a one-shot act, not a sticky mode.
3754 self.clear_pending();
3755 self.record_caret();
3756 match self.editor.toggle_inline(s, e, kind) {
3757 Ok(change) => {
3758 self.last_edit_kind = None; // structural edit is its own undo step
3759 self.refresh();
3760 self.anchor = Some(change.new.start);
3761 self.caret = change.new.end;
3762 self.dirty = self.source != self.clean_source;
3763 self.status = None;
3764 self.record_caret();
3765 }
3766 Err(e) => self.status = Some(format!("{kind:?}: {e}")),
3767 }
3768 }
3769
3770 /// Whether the caret stands in a highlight — what a frontend asks to enable
3771 /// or disable its highlight-colour controls, the way
3772 /// [`caret_in_table`](Self::caret_in_table) gates the grid ones.
3773 ///
3774 /// A fact about the *caret*, and the other half of
3775 /// [`Capabilities::mark_color`], which is the fact about the format. A
3776 /// frontend needs both: djot spells a highlight and no colour for it, so a
3777 /// caret standing in `{=word=}` answers `true` here and still has no palette
3778 /// to offer.
3779 ///
3780 /// The rule is [`active_inline_marks`](Self::active_inline_marks)' rule, so
3781 /// the palette appears exactly where the Highlight button is lit — with one
3782 /// deliberate exception: a mark *armed* at a bare caret and not yet typed
3783 /// into lights the button and answers `false` here, because there is no node
3784 /// to colour until the text exists.
3785 pub fn caret_in_mark(&mut self) -> bool {
3786 self.mark_offset().is_some()
3787 }
3788
3789 /// The offset [`set_mark_color`](Self::set_mark_color) speaks for — the one
3790 /// standing in the highlight the gesture means — or `None` when neither end
3791 /// of what is selected is in one.
3792 ///
3793 /// The caret first, and the selection's *start* after it, because of what
3794 /// [`toggle`](Self::toggle) leaves behind: a fresh `==word==` is selected
3795 /// whole, with the caret at its far edge, one past the closing `==` and so
3796 /// (by `marks_at`' half-open rule) not in the mark at all. Highlight a word
3797 /// and colour it — the two presses a coloured highlight is made of — would
3798 /// otherwise refuse on the second, having just written the highlight the
3799 /// author is pointing at.
3800 fn mark_offset(&mut self) -> Option<usize> {
3801 let in_mark = |d: &mut Self, off: usize| {
3802 d.marks_at(off)
3803 .into_iter()
3804 .any(|(k, _)| k == InlineKind::Mark)
3805 .then_some(off)
3806 };
3807 let caret = self.caret.min(self.source.len());
3808 in_mark(self, caret).or_else(|| {
3809 let start = self.selection()?.0;
3810 in_mark(self, start)
3811 })
3812 }
3813
3814 /// The colour of the highlight at the caret — `None` both when the caret is
3815 /// in no highlight and when the highlight it is in names no colour, which
3816 /// are the same answer to "which swatch is lit".
3817 ///
3818 /// The innermost mark, by span, for the same reason
3819 /// [`current_heading_level`](Self::current_heading_level) walks the tree:
3820 /// what the caret is *in* is the deepest node containing it. A `data-color`
3821 /// naming a colour this build has no variant for reads as `None` — the
3822 /// renderer already draws that as a plain highlight rather than guessing,
3823 /// and the toolbar agrees with the renderer.
3824 pub fn mark_color_at_caret(&mut self) -> Option<MarkColor> {
3825 let at = self.mark_offset()?;
3826 self.mark_color_at(at)
3827 }
3828
3829 /// [`mark_color_at_caret`](Self::mark_color_at_caret) at a given offset —
3830 /// the innermost `mark` covering it, and the colour it names.
3831 fn mark_color_at(&mut self, off: usize) -> Option<MarkColor> {
3832 self.nodes()
3833 .into_iter()
3834 .filter(|n| n.kind == Kind::Mark)
3835 .filter(|n| n.span.start <= off && off < n.span.end)
3836 .min_by_key(|n| n.span.end - n.span.start)
3837 .and_then(|n| MarkColor::from_attrs(&n.attrs))
3838 }
3839
3840 /// Colour the highlight at the caret, or clear its colour with `None` — the
3841 /// palette behind a toolbar's Highlight button.
3842 ///
3843 /// Markdown only, and the one gesture whose availability is a fact about the
3844 /// *parse extensions* rather than about the format alone: the colour is
3845 /// spelled `==🔴 text==`, an emoji twig reads back out of the content and
3846 /// records as the mark's `data-color`, and only an editor parsing with
3847 /// `highlight_colors` (which [`parse_extensions`] turns on for every leaf
3848 /// document) reads it back that way. Djot spells the highlight and no colour
3849 /// for it, so this refuses there — see [`Capabilities::mark_color`].
3850 ///
3851 /// **A colour is a property of a highlight that already exists.** There is
3852 /// no "highlight this in red" here, because that is two splices and would be
3853 /// two undo steps under one press; a frontend that wants it calls
3854 /// [`toggle`](Self::toggle) with [`InlineKind::Mark`] first, which is the
3855 /// order the two buttons already sit in. With no highlight at the caret this
3856 /// says so in the status line and writes nothing.
3857 ///
3858 /// The caret keeps its place in the *text*: the splice is entirely in the
3859 /// prefix between the opening `==` and the first word, so an offset past it
3860 /// rides the emoji's width, and one standing on the prefix itself lands
3861 /// where the prefix now ends.
3862 pub fn set_mark_color(&mut self, color: Option<MarkColor>) {
3863 // The read-only gate — this door reaches twig without the splice.
3864 if self.read_only {
3865 return;
3866 }
3867 if self.refuse_unsupported("highlight colour", Gesture::SetMarkColor) {
3868 return;
3869 }
3870 let Some(at) = self.mark_offset() else {
3871 self.status = Some("highlight colour: no highlight at the caret".into());
3872 return;
3873 };
3874 // Clearing a colour a highlight hasn't got is twig's one *successful*
3875 // no-op, and the `Change` it hands back then describes whatever edit came
3876 // before it — a stale span that would drag the caret somewhere it never
3877 // was. Answer it here, where the question is cheap, rather than trusting
3878 // a change that isn't one.
3879 if color.is_none() && self.mark_color_at(at).is_none() {
3880 self.status = None;
3881 return;
3882 }
3883 self.record_caret();
3884 match self.editor.set_mark_color(at, color.map(twig_mark_color)) {
3885 Ok(change) => {
3886 // Re-anchored from the offsets as they were, *before* `refresh`
3887 // sees the new bytes: the caret it clamps is one standing inside
3888 // a prefix that didn't exist a moment ago, and walking it back to
3889 // a char boundary of the emoji loses the place this is restoring.
3890 let caret = reanchor(self.caret, &change);
3891 let anchor = self.anchor.map(|a| reanchor(a, &change));
3892 self.last_edit_kind = None; // structural edit is its own undo step
3893 self.refresh();
3894 self.caret = caret;
3895 self.anchor = anchor;
3896 self.dirty = self.source != self.clean_source;
3897 self.status = None;
3898 self.clamp_caret();
3899 self.record_caret();
3900 }
3901 Err(e) => self.status = Some(format!("highlight colour: {e}")),
3902 }
3903 }
3904
3905 /// One press of a colour swatch: colour the highlight at the caret, or —
3906 /// over a selection that isn't highlighted yet — highlight it and colour it,
3907 /// as **one** undo step.
3908 ///
3909 /// [`set_mark_color`](Self::set_mark_color) is the exact gesture and stays
3910 /// one splice; this is the compound every toolbar actually presses, and it
3911 /// lives here rather than in each frontend because the rule it encodes —
3912 /// what a swatch means when there is no highlight under it yet — is one
3913 /// answer, not one per frontend. The two splices are folded into a single
3914 /// history step, so the press that made a red highlight is taken back by a
3915 /// single undo rather than leaving an uncoloured one behind.
3916 ///
3917 /// `None` clears the colour, and over an unhighlighted selection means
3918 /// simply "highlight this" — the same thing the Highlight button does.
3919 /// A bare caret in no highlight is left alone with a status line, because
3920 /// [`toggle`](Self::toggle) there arms a mark for text not yet typed and a
3921 /// colour cannot be armed with it.
3922 pub fn highlight(&mut self, color: Option<MarkColor>) {
3923 if self.caret_in_mark() || self.selection().is_none() {
3924 self.set_mark_color(color);
3925 return;
3926 }
3927 self.toggle(InlineKind::Mark);
3928 // The format may not spell a highlight at all (`toggle` said so), and
3929 // there is nothing to colour if it doesn't.
3930 if self.status.is_some() {
3931 return;
3932 }
3933 let before = self.revision;
3934 self.set_mark_color(color);
3935 // Only fold when the colour really spliced. `highlight(None)` over a
3936 // fresh highlight is a no-op by design, and coalescing there would eat
3937 // the *previous* edit into the toggle instead.
3938 if self.revision != before {
3939 let _ = self.editor.coalesce_last_undo();
3940 }
3941 }
3942
3943 // ── the presentation vocabulary ─────────────────────────────────────────
3944 //
3945 // Six gestures and five queries over twig's two attribute ops. Each gesture
3946 // edits **one key and keeps the rest**: it reads the node's attributes,
3947 // removes its own key (and, for alignment, its own tokens out of `class`),
3948 // adds the new value or nothing, and passes the list back whole — twig's
3949 // contract is replace-not-merge, so the read is the caller's job. A
3950 // paragraph that came in as `class="lead center" id="intro"
3951 // data-line-height="1.5"` and is right-aligned goes out as `class="lead
3952 // right" id="intro" data-line-height="1.5"`. Nothing leaf did not write is
3953 // touched, which is what lets a document from elsewhere pass through the
3954 // editor unharmed.
3955 //
3956 // Clearing is the same gesture with `None`: the key goes, and an empty list
3957 // at the end unwraps the span or the Markdown div, which twig does.
3958
3959 /// Set — or with `None` clear — the alignment of the block the caret is in.
3960 ///
3961 /// A block property, so the gesture is `set_block_attrs` on the caret's
3962 /// block **whatever is selected**: a line is a block's, and "centre this"
3963 /// with three words selected means the paragraph, not the words. The
3964 /// vocabulary is [`Align`], written as `class` tokens; other tokens on the
3965 /// same `class` are kept.
3966 ///
3967 /// In Markdown the attributes live on a `<div>` around the block — twig has
3968 /// no paragraph attribute syntax to write — and this reads them back off
3969 /// that div when the block is its sole child, so a second press rewrites
3970 /// the div rather than nesting a second one.
3971 pub fn set_alignment(&mut self, align: Option<Align>) {
3972 let attrs = self.block_attrs_at_caret();
3973 if align.is_none()
3974 && self.refuse_clear_from_div("alignment", &attrs, |a| Align::from_attrs(a).is_some())
3975 {
3976 return;
3977 }
3978 let attrs = with_class_token(
3979 &attrs,
3980 |t| Align::from_token(t).is_some(),
3981 align.map(Align::name),
3982 );
3983 self.write_block_attrs("alignment", attrs);
3984 }
3985
3986 /// Set — or with `None` clear — the line spacing of the block the caret is
3987 /// in. [`set_alignment`](Self::set_alignment)'s peer in every respect but
3988 /// the key: [`LineSpacing`] under `data-line-height`.
3989 pub fn set_line_spacing(&mut self, spacing: Option<LineSpacing>) {
3990 let attrs = self.block_attrs_at_caret();
3991 if spacing.is_none()
3992 && self.refuse_clear_from_div("line spacing", &attrs, |a| {
3993 LineSpacing::from_attrs(a).is_some()
3994 })
3995 {
3996 return;
3997 }
3998 let attrs = with_attr(&attrs, "data-line-height", spacing.map(LineSpacing::name));
3999 self.write_block_attrs("line spacing", attrs);
4000 }
4001
4002 /// Set — or with `None` clear — the size of the selected run, or of the
4003 /// caret's whole block when nothing is selected.
4004 ///
4005 /// Size, face and colour are the *run's*, and the block's when no run is
4006 /// chosen. With a selection the gesture is `wrap_range_attrs`, which wraps
4007 /// the range in an attributed span or re-styles the span it already lies in
4008 /// (never nesting a second, and unwrapping it when the last key goes). With
4009 /// no selection it is `set_block_attrs` on the caret's block, so that "make
4010 /// this paragraph larger" is a click with the caret in it rather than a
4011 /// select-all first.
4012 ///
4013 /// The walker reads the key at both levels with the nearer winning, so a
4014 /// span's `data-size` inside a block carrying its own applies to the span.
4015 pub fn set_font_size(&mut self, size: Option<SizeStep>) {
4016 self.set_run_attr("size", "data-size", size.map(SizeStep::name));
4017 }
4018
4019 /// Set — or with `None` clear — the face of the selected run, or of the
4020 /// caret's whole block. [`set_font_size`](Self::set_font_size)'s peer, with
4021 /// [`FontFamily`] under `data-font`.
4022 pub fn set_font_family(&mut self, font: Option<FontFamily>) {
4023 self.set_run_attr("font", "data-font", font.map(FontFamily::name));
4024 }
4025
4026 /// Set — or with `None` clear — the *text* colour of the selected run, or of
4027 /// the caret's whole block. [`set_font_size`](Self::set_font_size)'s peer,
4028 /// with [`MarkColor`] under `data-color`.
4029 ///
4030 /// The same key and the same seven names [`set_mark_color`](Self::set_mark_color)
4031 /// writes, and a different thing: that one colours a highlight's
4032 /// *background* and rides the `mark` node twig owns the spelling of, this
4033 /// one colours the letters and rides an attributed span. The two never
4034 /// collide, because a `mark` is a `mark` and a span is a span — and they
4035 /// share a vocabulary on purpose, so that a frontend with a red for a
4036 /// highlight has a red for text and both are *that* red.
4037 pub fn set_text_color(&mut self, color: Option<MarkColor>) {
4038 self.set_run_attr("text colour", "data-color", color.map(MarkColor::name));
4039 }
4040
4041 /// Insert a page break at the caret — `::page-break`, a leaf directive with
4042 /// no label and no attributes, which twig spells in every format that names
4043 /// a leaf container (Markdown under the `directives` extension
4044 /// [`parse_extensions`] turns on, and djot, where it is an empty `:::
4045 /// page-break` fence).
4046 ///
4047 /// Placed exactly as [`insert_thematic_break`](Self::insert_thematic_break)
4048 /// places a rule, and for the same reason: a directive is a block, so twig
4049 /// alone has nowhere to put one mid-paragraph and lands it after the
4050 /// caret's whole block. A bare paragraph is therefore parted at the caret
4051 /// first and the break aimed at the *first* half. See that method for the
4052 /// whole of the rule, including why a code block, a list item, a table and
4053 /// a setext heading are left unsplit.
4054 ///
4055 /// The frontends that paginate read the row's
4056 /// [`DirectiveMark`](crate::wysiwyg::DirectiveMark) and open a page there;
4057 /// the ones that do not draw the `⧉ page-break` placeholder every leaf
4058 /// directive gets.
4059 pub fn insert_page_break(&mut self) {
4060 if self.read_only || self.refuse_unsupported("page break", Gesture::InsertDirective) {
4061 return;
4062 }
4063 self.caret = self.skip_trailing_close_delims(self.caret);
4064 // A selection is replaced by the break, as a rule replaces one.
4065 if let Some((s, e)) = self.selection() {
4066 self.splice(s, e, "", EditKind::Other);
4067 }
4068 self.anchor = None;
4069 self.record_caret();
4070 let at = self.caret;
4071 if self.caret_parts_bare_paragraph() {
4072 // A failure here is not fatal: the break still lands after the
4073 // block, which is what this call was trying to improve on.
4074 let _ = self.editor.split_block(at);
4075 }
4076 match self.editor.insert_directive(at, PAGE_BREAK, None, &[]) {
4077 Ok(change) => {
4078 self.last_edit_kind = None;
4079 self.refresh();
4080 self.anchor = None;
4081 self.caret = change.new.end;
4082 self.dirty = self.source != self.clean_source;
4083 self.status = None;
4084 self.clamp_caret();
4085 self.record_caret();
4086 }
4087 Err(e) => self.status = Some(format!("page break: {e}")),
4088 }
4089 }
4090
4091 /// The alignment in force at the caret, or `None` for the theme's default —
4092 /// which swatch of an alignment control is lit.
4093 ///
4094 /// Read off the nearest node that names one: the block the caret is in, and
4095 /// the `div`s around it after that. [`mark_color_at_caret`](Self::mark_color_at_caret)'s
4096 /// shape, one property along.
4097 pub fn alignment_at_caret(&mut self) -> Option<Align> {
4098 self.presentation_chain()
4099 .iter()
4100 .find_map(|attrs| Align::from_attrs(attrs))
4101 }
4102
4103 /// The line spacing in force at the caret, or `None` for the theme's own.
4104 /// [`alignment_at_caret`](Self::alignment_at_caret)'s peer.
4105 pub fn line_spacing_at_caret(&mut self) -> Option<LineSpacing> {
4106 self.presentation_chain()
4107 .iter()
4108 .find_map(|attrs| LineSpacing::from_attrs(attrs))
4109 }
4110
4111 /// The size in force at the caret, or `None` for the theme's own — the
4112 /// entry a size menu shows ticked.
4113 ///
4114 /// Run-level, so the chain starts one node deeper: the attributed span the
4115 /// caret stands in, then its block, then the `div`s around it. The nearest
4116 /// wins, which is the rule the walker draws by.
4117 pub fn font_size_at_caret(&mut self) -> Option<SizeStep> {
4118 self.presentation_chain()
4119 .iter()
4120 .find_map(|attrs| SizeStep::from_attrs(attrs))
4121 }
4122
4123 /// The face in force at the caret, or `None` for the theme's body face.
4124 /// [`font_size_at_caret`](Self::font_size_at_caret)'s peer.
4125 pub fn font_family_at_caret(&mut self) -> Option<FontFamily> {
4126 self.presentation_chain()
4127 .iter()
4128 .find_map(|attrs| FontFamily::from_attrs(attrs))
4129 }
4130
4131 /// The *text* colour in force at the caret, or `None` for the theme's.
4132 /// [`font_size_at_caret`](Self::font_size_at_caret)'s peer, and not
4133 /// [`mark_color_at_caret`](Self::mark_color_at_caret) — that one reads a
4134 /// highlight's background off a `mark`, and a `mark` is never in this chain.
4135 pub fn text_color_at_caret(&mut self) -> Option<MarkColor> {
4136 self.presentation_chain()
4137 .iter()
4138 .find_map(|attrs| MarkColor::from_attrs(attrs))
4139 }
4140
4141 /// The selection-or-caret half of the three run-level gestures: a span over
4142 /// a real selection, the caret's block over none.
4143 fn set_run_attr(&mut self, what: &str, key: &str, value: Option<&str>) {
4144 match self.selection() {
4145 Some((start, end)) => {
4146 let attrs = with_attr(&self.run_attrs_over(start, end), key, value);
4147 self.write_run_attrs(what, start, end, attrs);
4148 }
4149 None => {
4150 let own = self.block_attrs_at_caret();
4151 if value.is_none()
4152 && self.refuse_clear_from_div(what, &own, |a| a.iter().any(|(k, _)| k == key))
4153 {
4154 return;
4155 }
4156 let attrs = with_attr(&own, key, value);
4157 self.write_block_attrs(what, attrs);
4158 }
4159 }
4160 }
4161
4162 /// A clear this gesture cannot carry out, said out loud instead of written:
4163 /// the node it rewrites — the caret's block, or the `<div>` around it that
4164 /// [`block_attrs_at_caret`](Self::block_attrs_at_caret) folds to in Markdown
4165 /// — does not name the property at all, and a `div` further out does.
4166 ///
4167 /// Handing twig the block's attributes with the key already absent changes
4168 /// no byte, and the query goes on answering `Some` off the div: the menu
4169 /// entry the author pressed stays unticked, and nothing says why. Twig's
4170 /// `set_block_attrs` reaches one node, so leaf cannot clear a key it did not
4171 /// write on a node it is not rewriting — the honest answer is the status
4172 /// line, in the voice the other refusals use.
4173 ///
4174 /// `names` is the property's own reading of an attribute list, because
4175 /// alignment lives in a `class` token rather than a key of its own. Spans
4176 /// are skipped: one inside the block is not what a *block* gesture writes
4177 /// either, but neither is it "the div around the block", and the run-level
4178 /// gestures reach it through a selection.
4179 fn refuse_clear_from_div(
4180 &mut self,
4181 what: &str,
4182 own: &Attrs,
4183 names: impl Fn(&Attrs) -> bool,
4184 ) -> bool {
4185 if names(own) {
4186 return false;
4187 }
4188 let caret = self.caret.min(self.source.len());
4189 if !self
4190 .attr_chain_at(caret)
4191 .iter()
4192 .any(|(span, attrs)| !span && names(attrs))
4193 {
4194 return false;
4195 }
4196 self.status = Some(format!("{what}: set on the div around the block"));
4197 true
4198 }
4199
4200 /// Hand `attrs` to twig as the caret's block's whole attribute set, with the
4201 /// status, undo and caret plumbing [`set_mark_color`](Self::set_mark_color)
4202 /// has.
4203 ///
4204 /// **The caret keeps its place in the text, not its byte offset.** How a
4205 /// format spells a block's attributes is markup written *around* the block
4206 /// — djot's `{…}` line above it, a `<div …>` and two blank lines in front of
4207 /// it in Markdown, a longer opening tag in HTML — and every one of those
4208 /// grows or shrinks above the author's own bytes. Where twig's change
4209 /// rewrites the block whole (Markdown's div is spliced as one region, block
4210 /// included) the plain arithmetic of [`reanchor`] has nothing to shift by
4211 /// and parks the caret at the end of the splice, past the closing `</div>`:
4212 /// the caret is then in no block at all, so a second press of the same menu
4213 /// answers "no block at the caret" and the toolbar's queries read nothing.
4214 /// [`reanchor_in_block`] is what carries it across instead — the block's
4215 /// content span before and after, which is the one thing the respelling
4216 /// leaves alone.
4217 ///
4218 /// Read *before* the splice and applied *after* `refresh`, because both
4219 /// halves of that mapping are facts about a tree twig is between: the
4220 /// block's old bytes are gone once the edit lands, and its new ones are not
4221 /// in `self.source` until the refresh puts them there.
4222 fn write_block_attrs(&mut self, what: &str, attrs: Attrs) {
4223 if self.read_only || self.refuse_unsupported(what, Gesture::SetBlockAttrs) {
4224 return;
4225 }
4226 // A blank line has no block to carry an attribute, and twig answers
4227 // `NotFound` there — say so in leaf's own words instead.
4228 let Some(at) = self.block_offset_for_caret() else {
4229 self.status = Some(format!("{what}: no block at the caret"));
4230 return;
4231 };
4232 self.record_caret();
4233 let pairs = attr_pairs(&attrs);
4234 let was = self.block_content_at(at);
4235 let text = was.clone().map(|s| self.source[s].to_string());
4236 match self.editor.set_block_attrs(at, &pairs) {
4237 Ok(change) => {
4238 let (caret, anchor) = (self.caret, self.anchor);
4239 self.last_edit_kind = None; // structural edit is its own undo step
4240 self.refresh();
4241 let now = self.block_content_in(&change.new, text.as_deref());
4242 // A block the two halves cannot both name — a code block, a
4243 // caret in a list's marker — takes the plain arithmetic, which
4244 // is what it had before.
4245 let block = was.as_ref().zip(now.as_ref());
4246 self.caret = reanchor_in_block(caret, &change, block);
4247 self.anchor = anchor.map(|a| reanchor_in_block(a, &change, block));
4248 self.dirty = self.source != self.clean_source;
4249 self.status = None;
4250 self.clamp_caret();
4251 self.record_caret();
4252 }
4253 Err(e) => self.status = Some(format!("{what}: {e}")),
4254 }
4255 }
4256
4257 /// The content span of the innermost paragraph or heading covering `off` —
4258 /// the author's own bytes, without the `# ` or the `<p>` that spells the
4259 /// block around them.
4260 ///
4261 /// The same two kinds [`block_attrs_at_caret`](Self::block_attrs_at_caret)
4262 /// reads, so that what a gesture re-anchors by is the block it wrote to.
4263 fn block_content_at(&mut self, off: usize) -> Option<Range<usize>> {
4264 self.nodes()
4265 .into_iter()
4266 .filter(|n| matches!(n.kind, Kind::Para | Kind::Heading))
4267 .filter(|n| n.span.start <= off && off <= n.span.end)
4268 .min_by_key(|n| n.span.end - n.span.start)
4269 .map(|n| n.content_span.unwrap_or(n.span))
4270 }
4271
4272 /// [`block_content_at`](Self::block_content_at)'s other half: the content
4273 /// span of the block `region` holds now, found by the bytes it held before.
4274 ///
4275 /// Matched on the text rather than taken as the first block in the region,
4276 /// because a rewritten region is markup and all — `<div class="center">`
4277 /// carries words of its own — and because the block this gesture moved is
4278 /// the one whose content the respelling did not touch. `None` where the
4279 /// region holds no block at all, which is djot's every case: the `{…}` line
4280 /// is spliced above the block and the block itself never moves through the
4281 /// change at all, only past it.
4282 fn block_content_in(
4283 &mut self,
4284 region: &Range<usize>,
4285 text: Option<&str>,
4286 ) -> Option<Range<usize>> {
4287 let text = text?;
4288 let spans: Vec<Range<usize>> = self
4289 .nodes()
4290 .into_iter()
4291 .filter(|n| matches!(n.kind, Kind::Para | Kind::Heading))
4292 .filter(|n| region.start <= n.span.start && n.span.end <= region.end)
4293 .map(|n| n.content_span.unwrap_or(n.span))
4294 .collect();
4295 spans
4296 .into_iter()
4297 .find(|s| self.source.get(s.clone()) == Some(text))
4298 }
4299
4300 /// Hand `attrs` to twig as the attribute set of the span over `[start,
4301 /// end)` — wrapping one, or re-styling the one the range already lies in,
4302 /// or unwrapping it when `attrs` is empty.
4303 ///
4304 /// What the splice leaves selected is the span's **content** — the author's
4305 /// words — and not the whole of `change.new`, which is markup and all:
4306 /// `[big]{data-size="large"}` in djot, `<span …>big</span>` in Markdown. A
4307 /// selection reaching past the node's own span lies in no span at all, so a
4308 /// second press of the menu would nest a fresh one instead of re-styling
4309 /// the one just written.
4310 fn write_run_attrs(&mut self, what: &str, start: usize, end: usize, attrs: Attrs) {
4311 if self.read_only || self.refuse_unsupported(what, Gesture::WrapRangeAttrs) {
4312 return;
4313 }
4314 self.record_caret();
4315 let pairs = attr_pairs(&attrs);
4316 match self.editor.wrap_range_attrs(start, end, &pairs) {
4317 Ok(change) => {
4318 self.last_edit_kind = None;
4319 self.refresh();
4320 let content = self.span_content_in(&change.new);
4321 self.anchor = Some(content.start);
4322 self.caret = content.end;
4323 self.dirty = self.source != self.clean_source;
4324 self.status = None;
4325 self.clamp_caret();
4326 self.record_caret();
4327 }
4328 Err(e) => self.status = Some(format!("{what}: {e}")),
4329 }
4330 }
4331
4332 /// The content range of the attributed span `spliced` now holds — the
4333 /// outermost one inside it, since that is the one just written — or
4334 /// `spliced` itself where the splice left no span, which is what an unwrap
4335 /// leaves behind.
4336 fn span_content_in(&mut self, spliced: &Range<usize>) -> Range<usize> {
4337 self.nodes()
4338 .into_iter()
4339 .filter(wysiwyg::is_run_span)
4340 .filter(|n| spliced.start <= n.span.start && n.span.end <= spliced.end)
4341 .max_by_key(|n| n.span.end - n.span.start)
4342 .and_then(|n| n.content_span)
4343 .unwrap_or_else(|| spliced.clone())
4344 }
4345
4346 /// The attribute set `set_block_attrs` is about to **replace** at the caret
4347 /// — which is the block's own, except in Markdown, where twig writes a
4348 /// block's attributes onto a `<div>` around it and rewrites that div when
4349 /// the block is its sole child. Reading the paragraph there would hand back
4350 /// an empty list and quietly drop everything the div said.
4351 ///
4352 /// Empty when the caret is in no block at all, which is the same list a
4353 /// block carrying no attributes gives — and the right one either way, since
4354 /// the gesture then refuses on its own.
4355 fn block_attrs_at_caret(&mut self) -> Attrs {
4356 let Some(off) = self.block_offset_for_caret() else {
4357 return Vec::new();
4358 };
4359 let nodes = self.nodes();
4360 let Some(block) = nodes
4361 .iter()
4362 .filter(|n| matches!(n.kind, Kind::Para | Kind::Heading))
4363 .filter(|n| n.span.start <= off && off <= n.span.end)
4364 .min_by_key(|n| n.span.end - n.span.start)
4365 else {
4366 return Vec::new();
4367 };
4368 if self.format == Format::Markdown
4369 && let Some(parent) = block.parent.and_then(|p| nodes.iter().find(|n| n.id == p))
4370 && wysiwyg::element_tag(parent) == Some("div")
4371 && nodes.iter().filter(|n| n.parent == Some(parent.id)).count() == 1
4372 {
4373 return parent.attrs.clone();
4374 }
4375 block.attrs.clone()
4376 }
4377
4378 /// The attribute set `wrap_range_attrs` is about to **replace** over
4379 /// `[start, end)` — the innermost attributed span the range lies inside,
4380 /// which twig re-styles rather than nesting a second one in. Empty when the
4381 /// range lies in no span, where the gesture mints a fresh one.
4382 fn run_attrs_over(&mut self, start: usize, end: usize) -> Attrs {
4383 self.nodes()
4384 .into_iter()
4385 .filter(wysiwyg::is_run_span)
4386 .filter(|n| n.span.start <= start && end <= n.span.end)
4387 .min_by_key(|n| n.span.end - n.span.start)
4388 .map(|n| n.attrs)
4389 .unwrap_or_default()
4390 }
4391
4392 /// The attribute lists that bear on a presentation query, **nearest first**:
4393 /// the attributed spans the caret stands in (innermost first), then its
4394 /// block, then the `div`s around it. A `find_map` down this is the whole of
4395 /// each query, and the order is the rule the walker draws by.
4396 ///
4397 /// Read at the caret, and at the selection's *start* when the caret stands
4398 /// in no span there. [`write_run_attrs`](Self::write_run_attrs) leaves the
4399 /// caret one past the span it just wrote — `toggle`'s convention — so
4400 /// asking the menu which entry that press just ticked must not answer
4401 /// `None`. Exactly the reason [`mark_offset`](Self::mark_offset) tries both.
4402 fn presentation_chain(&mut self) -> Vec<Attrs> {
4403 let caret = self.caret.min(self.source.len());
4404 let mut chain = self.attr_chain_at(caret);
4405 if !chain.iter().any(|(span, _)| *span)
4406 && let Some((start, _)) = self.selection()
4407 {
4408 let alt = self.attr_chain_at(start);
4409 if alt.iter().any(|(span, _)| *span) {
4410 chain = alt;
4411 }
4412 }
4413 chain.into_iter().map(|(_, attrs)| attrs).collect()
4414 }
4415
4416 /// [`presentation_chain`](Self::presentation_chain) at one offset — every
4417 /// node bearing the vocabulary that covers it, innermost first, each paired
4418 /// with whether it is an attributed span (which is what tells the caller
4419 /// its run-level answer came from a run).
4420 ///
4421 /// Sorted by span length, which *is* the nesting order: a span lies inside
4422 /// its block and a block inside its div, so shortest-first is
4423 /// nearest-first without a second tree walk.
4424 fn attr_chain_at(&mut self, off: usize) -> Vec<(bool, Attrs)> {
4425 let off = off.min(self.source.len());
4426 let mut hits: Vec<(usize, bool, Attrs)> = Vec::new();
4427 for n in self.nodes() {
4428 let span = wysiwyg::is_run_span(&n);
4429 let block = matches!(n.kind, Kind::Para | Kind::Heading);
4430 let div = wysiwyg::element_tag(&n) == Some("div");
4431 if !(span || block || div) {
4432 continue;
4433 }
4434 // A span is half-open, the way a mark is: the offset one past it is
4435 // the text after it. A block and a div claim their end too, so a
4436 // caret resting at the end of a line still reads its paragraph.
4437 let inside = if span {
4438 n.span.start <= off && off < n.span.end
4439 } else {
4440 n.span.start <= off && off <= n.span.end
4441 };
4442 if !inside {
4443 continue;
4444 }
4445 hits.push((n.span.end - n.span.start, span, n.attrs));
4446 }
4447 hits.sort_by_key(|(len, _, _)| *len);
4448 hits.into_iter()
4449 .map(|(_, span, attrs)| (span, attrs))
4450 .collect()
4451 }
4452
4453 /// Convert the block at the caret to a heading level or paragraph.
4454 pub fn set_block(&mut self, kind: BlockKind) {
4455 // The read-only gate — this door reaches twig without the splice.
4456 if self.read_only {
4457 return;
4458 }
4459 if self.refuse_unsupported(&format!("{kind:?}"), Gesture::SetBlock) {
4460 return;
4461 }
4462 self.record_caret();
4463 // A blank line has no node to convert, and twig opens a block there
4464 // rather than declining — so the caret's own offset is the right thing
4465 // to hand it when `block_offset_for_caret` finds nothing.
4466 let offset = self.block_offset_for_caret().unwrap_or(self.caret);
4467 match self.editor.set_block(offset, kind) {
4468 Ok(change) => {
4469 self.last_edit_kind = None;
4470 self.refresh();
4471 // Opening a block on a blank line writes a marker the caret
4472 // belongs *after*; converting an existing one moves nothing.
4473 self.caret = self.caret.max(change.new.end);
4474 self.clamp_caret();
4475 self.anchor = None;
4476 self.dirty = self.source != self.clean_source;
4477 self.status = None;
4478 self.record_caret();
4479 }
4480 Err(e) => self.status = Some(format!("{kind:?}: {e}")),
4481 }
4482 }
4483
4484 /// Whether `off` is inside a text block (paragraph, heading, code block…).
4485 fn has_block_at(&mut self, off: usize) -> bool {
4486 self.editor.ancestors_at(off).ok().is_some_and(|chain| {
4487 chain
4488 .iter()
4489 .any(|m| !wysiwyg::is_inline_kind(&m.kind) && !is_block_container(&m.kind))
4490 })
4491 }
4492
4493 /// The offset to hand twig's `set_block`: the caret when it is already inside
4494 /// a block, otherwise nudged onto the previous character (a caret at a line
4495 /// end sits at the doc level, outside the block). `None` when the caret is on
4496 /// a blank line — a new paragraph with no block node to convert.
4497 fn block_offset_for_caret(&mut self) -> Option<usize> {
4498 let caret = self.caret.min(self.source.len());
4499 if self.has_block_at(caret) {
4500 return Some(caret);
4501 }
4502 // Nudge to the previous character — but never across a newline: that would
4503 // target the previous block, and a blank line genuinely has no block.
4504 if let Some((i, ch)) = self.source[..caret].char_indices().next_back()
4505 && ch != '\n'
4506 && self.has_block_at(i)
4507 {
4508 return Some(i);
4509 }
4510 None
4511 }
4512
4513 /// The heading level of the text block at the caret, or `None` when that
4514 /// block is not a heading.
4515 pub fn current_heading_level(&mut self) -> Option<u32> {
4516 let caret = self.caret;
4517 self.nodes()
4518 .into_iter()
4519 .filter(|n| n.kind == Kind::Heading)
4520 .find(|n| n.span.start <= caret && caret <= n.span.end)
4521 .and_then(|n| n.level)
4522 }
4523
4524 /// The inline marks in force at the caret (or over the selection) — what a
4525 /// toolbar draws lit, and the block-level [`Doc::current_heading_level`]'s
4526 /// inline counterpart. Cheap enough to call every frame: one twig
4527 /// `ancestors_at` query per caret (two with a selection), each walking root
4528 /// → deepest node at one offset. It never snapshots the tree the way
4529 /// `current_heading_level` does, and the returned set is a `Copy` bitset, so
4530 /// the only allocation is twig's own small ancestor `Vec`.
4531 ///
4532 /// **A selection reports a mark only when the mark covers *all* of it.**
4533 /// That's what every real toolbar means by an active button — Bold lit over
4534 /// a half-bold selection would claim a press turns bold *off*, when
4535 /// [`Doc::toggle`] hands the range to twig and gets the whole thing bolded.
4536 /// Whole-coverage is asked as "is the same mark node standing over both the
4537 /// first and the last character?": inline nodes are contiguous, so one node
4538 /// covering both ends covers every byte between them. Two touching runs
4539 /// (`**a****b**`) are two nodes, and correctly light nothing.
4540 ///
4541 /// At a bare caret a mark is active when the caret stands inside the mark's
4542 /// span — `span.start <= caret < span.end`, delimiters included, which is
4543 /// what makes the boundaries behave. In `a **bold** b` the offsets from the
4544 /// opening `*` (2) through the last byte of the closing `**` (9) are all
4545 /// bold, so the WYSIWYG caret both before `b` and after `d` (the delimiters
4546 /// are hidden, and those offsets are 4 and 8) reports bold — matching where
4547 /// typing would actually land inside the marked run. The offset one past the
4548 /// mark (10) is the text after it and reports nothing, at the end of the
4549 /// buffer exactly as in the middle.
4550 pub fn active_inline_marks(&mut self) -> InlineMarks {
4551 let Some((start, end)) = self.selection() else {
4552 // The marks actually in force at the caret, flipped by any armed
4553 // sticky delta — so `⌘b` at a bare caret lights the Bold button
4554 // immediately, before a single character is typed.
4555 let base: InlineMarks = self
4556 .marks_at(self.caret)
4557 .into_iter()
4558 .map(|(k, _)| k)
4559 .collect();
4560 return base.xor(self.pending_here());
4561 };
4562 // The selection's *last character*, not its exclusive end: `end` is the
4563 // offset one past the selection, which for a selection ending exactly at
4564 // a mark's close is already outside it (`[4,10)` of `a **bold** b` is
4565 // entirely bold, but offset 10 is the space after).
4566 let last = prev_boundary(&self.source, end);
4567 let head = self.marks_at(start);
4568 let tail = self.marks_at(last);
4569 head.into_iter()
4570 .filter(|m| tail.contains(m))
4571 .map(|(k, _)| k)
4572 .collect()
4573 }
4574
4575 /// The inline marks whose span covers `off`, each with the id of the node
4576 /// carrying it — the id is what lets a selection tell one mark node from
4577 /// another of the same kind.
4578 fn marks_at(&mut self, off: usize) -> Vec<(InlineKind, u32)> {
4579 let off = off.min(self.source.len());
4580 self.editor
4581 .ancestors_at(off)
4582 .unwrap_or_default()
4583 .into_iter()
4584 // `span.end` is the offset one *past* the mark, so it isn't in it.
4585 // twig already resolves a boundary to whatever starts there — in
4586 // `**bold** x` offset 8 is the following text, not the strong — but
4587 // when nothing follows, the tie has nobody to break for and the
4588 // chain still ends at the mark. That would make the answer at the
4589 // last offset of the document depend on whether the file happens to
4590 // end in a newline; the rule is `span.start <= off < span.end`, and
4591 // it's the same rule at the end of a buffer as in the middle.
4592 .filter(|m| off < m.span.end)
4593 .filter_map(|m| inline_kind(&m.kind).map(|k| (k, m.node_id)))
4594 .collect()
4595 }
4596
4597 /// Toggle a heading at the caret: if the block is already this heading level,
4598 /// revert it to a paragraph; otherwise convert it to this heading level.
4599 /// This gives the heading commands the same toggle feel as bold/italic/code —
4600 /// re-applying a heading a line already has turns it back into body text.
4601 pub fn toggle_heading(&mut self, level: u32) {
4602 if self.current_heading_level() == Some(level) {
4603 self.set_block(BlockKind::Paragraph);
4604 } else {
4605 self.set_block(BlockKind::Heading(level));
4606 }
4607 }
4608
4609 /// Toggle a block quote around the selection, or around the block at the
4610 /// caret — the toolbar's Quote button.
4611 pub fn toggle_blockquote(&mut self) {
4612 self.toggle_container(BlockContainerKind::BlockQuote);
4613 }
4614
4615 /// Toggle a numbered (`ordered`) or bulleted list over the selection, or
4616 /// over the block at the caret — one op with the kind as a flag, the way
4617 /// `toggle_heading` takes its level, so a frontend needs no twig type to
4618 /// name the two buttons.
4619 ///
4620 /// Pressing the *other* list's button while in a list converts in place
4621 /// rather than nesting, so the pair reads as one three-state control
4622 /// (bulleted / numbered / neither) rather than two independent wrappers.
4623 pub fn toggle_list(&mut self, ordered: bool) {
4624 self.toggle_container(if ordered {
4625 BlockContainerKind::OrderedList
4626 } else {
4627 BlockContainerKind::BulletList
4628 });
4629 }
4630
4631 // ── Task list items ──────────────────────────────────────────────────────
4632 // The checkbox in `- [x] done`. twig owns all three gestures: the box is
4633 // inline content of the item's first paragraph rather than part of its
4634 // marker, so adding or removing one must leave the item's continuation
4635 // indentation alone, and an item inside a quote is found past the quote
4636 // markers. leaf names the gesture and the offset; the spelling is twig's.
4637
4638 /// Whether the list item at the caret carries a checkbox, and which way it
4639 /// faces — `Some(true)` ticked, `Some(false)` empty, `None` for a plain list
4640 /// item or no item at all. What a toolbar reads to light its checkbox button.
4641 pub fn task_checked_at_caret(&mut self) -> Option<bool> {
4642 self.task_checked_at(self.caret)
4643 }
4644
4645 /// [`task_checked_at_caret`](Self::task_checked_at_caret) for an arbitrary
4646 /// offset — what a frontend asks before deciding a click landed on a box.
4647 pub fn task_checked_at(&mut self, offset: usize) -> Option<bool> {
4648 self.innermost_list_item(offset.min(self.source.len()))?
4649 .checked
4650 }
4651
4652 /// Tick or untick the task item at the caret (the checkbox's keyboard half).
4653 /// A no-op with a reported reason when the caret is in no task item — minting
4654 /// a box here is [`toggle_task_item`](Self::toggle_task_item)'s job.
4655 pub fn toggle_task_checked(&mut self) {
4656 self.toggle_task_at(self.caret);
4657 }
4658
4659 /// Tick or untick the task item covering `offset` — what a *click* on a
4660 /// rendered checkbox is. Separate from the caret form because a click carries
4661 /// its own offset and must not first move the caret there: ticking a box
4662 /// three paragraphs away should not take the cursor with it.
4663 pub fn toggle_task_at(&mut self, offset: usize) {
4664 // The read-only gate — this door reaches twig without the splice.
4665 if self.read_only {
4666 return;
4667 }
4668 if self.refuse_unsupported("task", Gesture::ToggleTaskChecked) {
4669 return;
4670 }
4671 let offset = offset.min(self.source.len());
4672 self.record_caret();
4673 match self.editor.toggle_task_checked(offset) {
4674 Ok(_) => self.after_task_edit(),
4675 Err(e) => self.status = Some(format!("task: {e}")),
4676 }
4677 }
4678
4679 /// Give the list item at the caret a checkbox, or take its checkbox away —
4680 /// the gesture that converts between a plain bullet and a task. A new box
4681 /// arrives unticked.
4682 pub fn toggle_task_item(&mut self) {
4683 // The read-only gate — this door reaches twig without the splice.
4684 if self.read_only {
4685 return;
4686 }
4687 if self.refuse_unsupported("task", Gesture::ToggleTaskItem) {
4688 return;
4689 }
4690 let caret = self.caret.min(self.source.len());
4691 self.record_caret();
4692 match self.editor.toggle_task_item(caret) {
4693 Ok(_) => self.after_task_edit(),
4694 Err(e) => self.status = Some(format!("task: {e}")),
4695 }
4696 }
4697
4698 /// Settle after a task gesture. The caret rides its old byte offset and is
4699 /// clamped back in: a box is three or four bytes on the item's first line, so
4700 /// text after it shifts by that much at most, and `clamp_caret` lands it on a
4701 /// real stop either way.
4702 fn after_task_edit(&mut self) {
4703 self.last_edit_kind = None;
4704 self.refresh();
4705 self.anchor = None;
4706 self.dirty = self.source != self.clean_source;
4707 self.status = None;
4708 self.clamp_caret();
4709 self.record_caret();
4710 }
4711
4712 // ── Tables ───────────────────────────────────────────────────────────────
4713 // A table is a grid, and twig edits it as one — add/remove/move a row or
4714 // column, set a column's alignment — re-spelling the whole table in a single
4715 // splice. Every gesture is anchored at the caret's cell. leaf just names the
4716 // gesture and re-reads the result; the whole table's numbering, borders, and
4717 // delimiter are twig's to keep straight.
4718
4719 /// Whether the caret is inside a table — what a frontend asks to enable or
4720 /// disable its table controls.
4721 ///
4722 /// An HTML `<table>` still answers `true`: the caret really is in a table,
4723 /// and the reason the grid controls stay dark there is
4724 /// [`Capabilities::table`], which is a fact about the document's format
4725 /// rather than about the caret. A frontend needs both.
4726 pub fn caret_in_table(&mut self) -> bool {
4727 let caret = self.caret.min(self.source.len());
4728 self.editor
4729 .ancestors_at(caret)
4730 .map(|c| c.into_iter().any(|m| m.kind == Kind::Table))
4731 .unwrap_or(false)
4732 }
4733
4734 /// One grid op, guarded and settled — the shared body of the seven below.
4735 ///
4736 /// The guard is why this exists rather than seven copies of the same three
4737 /// lines, and it is the one guard leaf cannot delegate to twig. The table
4738 /// editor is the gesture family that consults no `Syntax` table (it spells a
4739 /// grid, not a delimiter) and therefore the one twig's `Format::supports`
4740 /// deliberately has no variant for: handed an HTML `<table>` it rebuilds the
4741 /// grid as a *pipe table* and reports success, swapping the element out for
4742 /// `| a | b |` and taking the rest of the document's markup with it. Nothing
4743 /// downstream could tell that from a successful edit — the splice is real,
4744 /// the reparse succeeds, `dirty` is honest — which is what makes it worth
4745 /// stopping at the door rather than detecting after the fact. See
4746 /// [`spells_pipe_tables`].
4747 fn table_op(
4748 &mut self,
4749 what: &str,
4750 op: impl FnOnce(&mut Editor, usize) -> Result<(), twig::Error>,
4751 ) {
4752 if self.refuse_unless(what, spells_pipe_tables(self.format)) {
4753 return;
4754 }
4755 self.record_caret();
4756 let at = self.caret;
4757 let r = op(&mut self.editor, at);
4758 self.apply_table(r, what);
4759 }
4760
4761 /// Insert an empty row below (`below`) or above the caret's row.
4762 pub fn table_insert_row(&mut self, below: bool) {
4763 self.table_op("table row", |e, at| e.table_insert_row(at, below));
4764 }
4765
4766 /// Delete the caret's row (not the header, not the last body row).
4767 pub fn table_delete_row(&mut self) {
4768 self.table_op("table row", |e, at| e.table_delete_row(at));
4769 }
4770
4771 /// Insert an empty column right (`right`) or left of the caret's column.
4772 pub fn table_insert_column(&mut self, right: bool) {
4773 self.table_op("table column", |e, at| e.table_insert_column(at, right));
4774 }
4775
4776 /// Delete the caret's column (unless it is the only one).
4777 pub fn table_delete_column(&mut self) {
4778 self.table_op("table column", |e, at| e.table_delete_column(at));
4779 }
4780
4781 /// Set the caret's column to `alignment`.
4782 pub fn table_set_alignment(&mut self, alignment: Alignment) {
4783 self.table_op("table alignment", |e, at| {
4784 e.table_set_alignment(at, alignment)
4785 });
4786 }
4787
4788 /// Move the caret's row one place down (`down`) or up, within the body rows.
4789 pub fn table_move_row(&mut self, down: bool) {
4790 self.table_op("table row", |e, at| e.table_move_row(at, down));
4791 }
4792
4793 /// Move the caret's column one place right (`right`) or left.
4794 pub fn table_move_column(&mut self, right: bool) {
4795 self.table_op("table column", |e, at| e.table_move_column(at, right));
4796 }
4797
4798 /// Settle the caret and document flags after a table op (or report its
4799 /// error). twig re-spells the whole table, so the caret rides its old byte
4800 /// offset and is clamped back into the rebuilt bytes — near enough to where
4801 /// it was, since the op preserves the cells' content and order around it.
4802 fn apply_table(&mut self, result: Result<(), twig::Error>, what: &str) {
4803 match result {
4804 Ok(()) => {
4805 self.last_edit_kind = None;
4806 self.refresh();
4807 self.anchor = None;
4808 self.clamp_caret();
4809 self.dirty = self.source != self.clean_source;
4810 self.status = None;
4811 self.record_caret();
4812 }
4813 Err(e) => self.status = Some(format!("{what}: {e}")),
4814 }
4815 }
4816
4817 /// One `toggle_block_container` over the block-level target.
4818 ///
4819 /// leaf says *where*; twig decides everything else — which blocks the range
4820 /// covers, whether that means wrapping, unwrapping, nesting or converting,
4821 /// and how this document's format spells the prefix. The rule that a
4822 /// container only comes off when the range covers every block it holds is
4823 /// what the re-anchoring below is built around.
4824 fn toggle_container(&mut self, kind: BlockContainerKind) {
4825 // The read-only gate — this door reaches twig without the splice.
4826 if self.read_only {
4827 return;
4828 }
4829 if self.refuse_unsupported(&format!("{kind:?}"), Gesture::ToggleBlockContainer(kind)) {
4830 return;
4831 }
4832 let selected = self.selection();
4833 // A blank line holds no block, and twig opens an *empty* container on one
4834 // — since 3.2.0; it used to decline the range with `NotFound`, which is
4835 // why this used to lend it a scratch paragraph to wrap. Worth knowing
4836 // here because the line-for-line caret mapping below cannot describe it:
4837 // opening one under a paragraph writes the blank line the format needs
4838 // above the marker too, so the rewritten region has a line the old one
4839 // didn't, and "the same line, the same distance from its end" lands on
4840 // that new blank instead of in the container.
4841 let opened_empty = selected.is_none() && self.block_offset_for_caret().is_none();
4842 // Without a selection the target is the caret's own block, resolved the
4843 // way `set_block` resolves it — a caret at a line end sits at the doc
4844 // level and has to be nudged back onto the block it looks like it's in.
4845 // An empty range is enough: twig widens to the whole lines it touches.
4846 let (start, end) = match selected {
4847 Some(range) => range,
4848 None => {
4849 let off = self.block_offset_for_caret().unwrap_or(self.caret);
4850 (off, off)
4851 }
4852 };
4853 self.record_caret();
4854 match self.editor.toggle_block_container(start, end, kind) {
4855 Ok(change) => {
4856 // Read the caret's place out of the *pre-edit* source, before
4857 // `refresh` swaps that source out from under it.
4858 let place = (selected.is_none() && !opened_empty)
4859 .then(|| self.caret_line_tail(&change.old));
4860 self.last_edit_kind = None; // structural edit is its own undo step
4861 self.refresh();
4862 match place {
4863 // Both land the caret at the far end of what twig wrote, and
4864 // differ only in what they leave selected.
4865 //
4866 // From a selection: select what the container now holds, the
4867 // way `toggle` keeps its marked region selected — and for a
4868 // stronger reason than symmetry: a container comes *off* only
4869 // a range covering every block it holds, so a selection left
4870 // on its old bytes (now short by a prefix per line) would nest
4871 // on the second press instead of reversing the first.
4872 //
4873 // From a blank line: nothing to select, and the end of the
4874 // region is exactly past the bare `> ` / `- ` twig wrote —
4875 // the caret standing inside the container that was asked for.
4876 None => {
4877 self.anchor = (!opened_empty).then_some(change.new.start);
4878 self.caret = change.new.end;
4879 }
4880 Some(place) => {
4881 self.anchor = None;
4882 self.caret = self.line_tail_offset(&change.new, place);
4883 }
4884 }
4885 self.dirty = self.source != self.clean_source;
4886 self.status = None;
4887 self.clamp_caret();
4888 self.record_caret();
4889 }
4890 Err(e) => self.status = Some(format!("{kind:?}: {e}")),
4891 }
4892 }
4893
4894 /// The caret's place inside the region a container toggle is rewriting, in
4895 /// the only terms the rewrite preserves: which of the region's lines it sits
4896 /// on, and how many bytes of that line lie ahead of it.
4897 ///
4898 /// A container's markup goes in at column 0 and never touches what follows
4899 /// on the line, so that pair survives the edit exactly where a byte offset
4900 /// does not — a caret left on its old offset slides back by one prefix per
4901 /// line above it, which on a hard-wrapped paragraph parks it *inside* the
4902 /// `> ` it just asked for.
4903 fn caret_line_tail(&self, old: &std::ops::Range<usize>) -> (usize, usize) {
4904 let caret = self.caret.clamp(old.start, old.end);
4905 let line = self.source[old.start..caret].matches('\n').count();
4906 let end = self.source[caret..old.end]
4907 .find('\n')
4908 .map_or(old.end, |i| caret + i);
4909 (line, end - caret)
4910 }
4911
4912 /// [`caret_line_tail`](Self::caret_line_tail) undone against the rewritten
4913 /// region: the offset `tail` bytes back from the end of the region's `line`.
4914 ///
4915 /// Both walks are clamped rather than trusted, because the one op that does
4916 /// *not* keep a region's lines one-to-one is stripping a list — twig blows
4917 /// the items back apart with blank lines between them — and a caret landing
4918 /// on the nearest line of the right item beats one landing out of the region
4919 /// entirely.
4920 fn line_tail_offset(
4921 &self,
4922 new: &std::ops::Range<usize>,
4923 (line, tail): (usize, usize),
4924 ) -> usize {
4925 let region = &self.source[new.start.min(self.source.len())..new.end.min(self.source.len())];
4926 let mut start = 0;
4927 for _ in 0..line {
4928 match region[start..].find('\n') {
4929 Some(i) => start += i + 1,
4930 None => break,
4931 }
4932 }
4933 let end = region[start..]
4934 .find('\n')
4935 .map_or(region.len(), |i| start + i);
4936 new.start + end.saturating_sub(tail).max(start)
4937 }
4938
4939 /// Link the selection to `destination` — the toolbar's Link button. With no
4940 /// selection it acts at the caret, which re-points a link the caret is
4941 /// already standing in (twig replaces an existing link's destination and
4942 /// keeps its text) and otherwise spells a link that has no text of its own:
4943 /// an autolink (`<https://x.dev>`) where the destination is one, and
4944 /// `[destination](destination)` where it isn't.
4945 ///
4946 /// `destination` reaches twig raw. Escaping it is format knowledge and the
4947 /// two formats genuinely disagree — Markdown ends a destination at the first
4948 /// space and moves it into `<…>`, djot reads that `<…>` as part of the URL
4949 /// itself — so the side holding the document is the side that gets to spell
4950 /// it. A destination twig can't carry at all (one with a newline) comes back
4951 /// as an error rather than a quietly rewritten URL.
4952 pub fn insert_link(&mut self, destination: &str) {
4953 if self.read_only || self.refuse_unsupported("link", Gesture::InsertLink) {
4954 return;
4955 }
4956 let (start, end) = self.selection().unwrap_or((self.caret, self.caret));
4957 self.record_caret();
4958 match self.editor.insert_link(start, end, destination) {
4959 Ok(change) => {
4960 self.last_edit_kind = None;
4961 self.refresh();
4962 match self.link_text_span(change.new.start) {
4963 // A link with text of its own: select it, so typing replaces
4964 // a `[dest](dest)`'s stand-in label and a second press
4965 // re-points what the first one linked.
4966 Some(text) => {
4967 self.anchor = (text.start != text.end).then_some(text.start);
4968 self.caret = text.end;
4969 }
4970 // An autolink is finished the moment it's written — its text
4971 // *is* the URL. Leaving it selected would aim the next press
4972 // at the one shape twig still wraps instead of re-points.
4973 None => {
4974 self.anchor = None;
4975 self.caret = change.new.end;
4976 }
4977 }
4978 self.dirty = self.source != self.clean_source;
4979 self.status = None;
4980 self.clamp_caret();
4981 self.record_caret();
4982 }
4983 Err(e) => self.status = Some(format!("link: {e}")),
4984 }
4985 }
4986
4987 /// Insert a block-level image at the caret: ``. Any
4988 /// selection becomes the alt text (so "select a caption, insert image" labels
4989 /// it); with no selection, `alt` is used — empty for none. The caret lands
4990 /// just past the inserted image.
4991 ///
4992 /// Both halves go through twig (`insert_literal` for the alt text,
4993 /// `insert_image` for the image), so neither is spelled here. That used to be a
4994 /// `format!`, and it was wrong the first time an app inserted a real filename:
4995 /// Markdown ends a destination at the first space, so `` is
4996 /// not an image at all — and the fix is per-format, since moving into the
4997 /// `<…>` form is exactly wrong for Djot, where `<…>` becomes the URL itself.
4998 pub fn insert_image(&mut self, destination: &str, alt: &str) {
4999 if self.read_only || self.refuse_unsupported("image", Gesture::InsertImage) {
5000 return;
5001 }
5002 let (start, end) = self.selection().unwrap_or((self.caret, self.caret));
5003 self.record_caret();
5004 // With no selection and an explicit `alt`, the alt text has to exist in the
5005 // document before it can be the image's — and it is raw caller input, so
5006 // it goes in through `insert_literal`, which escapes it for the format
5007 // rather than letting a `]` in someone's caption close the image early.
5008 let (start, end) = if start == end && !alt.is_empty() {
5009 match self.editor.insert_literal(start, alt) {
5010 Ok(change) => (change.new.start, change.new.end),
5011 Err(e) => {
5012 self.status = Some(format!("image: {e}"));
5013 return;
5014 }
5015 }
5016 } else {
5017 (start, end)
5018 };
5019 match self.editor.insert_image(start, end, destination) {
5020 Ok(change) => {
5021 self.last_edit_kind = None;
5022 self.refresh();
5023 // Just past the image, nothing selected — where a caret belongs
5024 // after inserting one.
5025 self.anchor = None;
5026 self.caret = change.new.end;
5027 self.dirty = self.source != self.clean_source;
5028 self.status = None;
5029 self.clamp_caret();
5030 self.record_caret();
5031 }
5032 Err(e) => self.status = Some(format!("image: {e}")),
5033 }
5034 }
5035
5036 /// Insert a block-level image, video, or audio at the caret. The image case
5037 /// is [`insert_image`](Self::insert_image); video and audio are spelled as
5038 /// HTML elements, which is the only spelling Markdown and Djot have for them:
5039 ///
5040 /// ```text
5041 /// <video src="clip.mp4" controls>alt</video>
5042 /// <audio src="take.mp3" controls>alt</audio>
5043 /// ```
5044 ///
5045 /// HTML rather than a `::video{…}` directive deliberately. A directive means
5046 /// something only to an app that knows the vocabulary, so the document would
5047 /// read as literal punctuation everywhere else; `<video>` is what every other
5048 /// renderer already understands, and what leaf's own reader picks back up
5049 /// through `html_elements` promotion (see [`parse_extensions`]).
5050 ///
5051 /// The one-line spelling needs twig ≥ 2.5.1, which widened CommonMark's
5052 /// HTML-block tag list to cover `<video>`/`<audio>`/`<picture>` under
5053 /// `html_elements`. Before that only the multi-line form parsed as a block at
5054 /// all, and this wrote three lines to work around it.
5055 ///
5056 /// `controls` is always written: a player with no transport is a still frame
5057 /// the reader can't do anything with. Any selection becomes the element's
5058 /// fallback text, exactly as it becomes an image's alt.
5059 ///
5060 /// The same verbatim-insertion caveat as [`insert_image`](Self::insert_image)
5061 /// applies, and bites harder here: a `"` in `destination` closes the
5062 /// attribute. A frontend taking these from a file picker is fine; one taking
5063 /// them from free text should keep them tame.
5064 ///
5065 /// [`MediaInfo`]: crate::MediaInfo
5066 pub fn insert_media(&mut self, kind: MediaKind, destination: &str, alt: &str) {
5067 if kind == MediaKind::Image {
5068 return self.insert_image(destination, alt);
5069 }
5070 // Gated on the *image* gesture, not on one of its own — there isn't one,
5071 // since the bytes below are spelled here rather than by twig, and an HTML
5072 // document would in fact parse them. The button is one control with three
5073 // kinds behind it, and two of them working in a format where the third
5074 // cannot is a worse surface than three that agree — especially as
5075 // `insert_image` is the kind anyone reaches for first.
5076 if self.refuse_unsupported("media", Gesture::InsertImage) {
5077 return;
5078 }
5079 let (start, end) = self.selection().unwrap_or((self.caret, self.caret));
5080 let alt_text = self
5081 .selected_text()
5082 .map(str::to_string)
5083 .unwrap_or_else(|| alt.to_string());
5084 let tag = match kind {
5085 MediaKind::Audio => "audio",
5086 _ => "video",
5087 };
5088 let markup = format!("<{tag} src=\"{destination}\" controls>{alt_text}</{tag}>");
5089 self.edit(start, end, &markup);
5090 }
5091
5092 /// Insert a thematic break at the caret — the toolbar's Horizontal Rule
5093 /// button. Spelling and placement are both twig's; leaf used to write `---`
5094 /// itself, which was the Markdown spelling in a djot document too.
5095 ///
5096 /// A rule is a block, so `insert_thematic_break` alone has nowhere to put one
5097 /// mid-paragraph and lands it after the caret's whole block. To get a rule
5098 /// *at* the caret — the paragraph parted in two around it, which is what a
5099 /// rule button is understood to do — the paragraph is first divided with
5100 /// `split_block` and the rule then aimed at the **first** half. Aiming it at
5101 /// the offset `split_block` returns puts the rule after the *second* half
5102 /// instead, which is a rule in the right document and the wrong place.
5103 ///
5104 /// Only a plain paragraph is split, and only where there is something to
5105 /// part: at the paragraph's end the split has no second half to mint and
5106 /// would write the separator anyway — a blank line and the empty slot Enter
5107 /// leaves for the next paragraph, which the rule then lands above and
5108 /// nothing fills — so there the rule goes straight after the paragraph,
5109 /// which is where the split-and-aim was sending it regardless. At the
5110 /// paragraph's *start* the split is kept, though it parts nothing either:
5111 /// `|para` becomes `\npara` with the caret on the new blank line, and a
5112 /// rule aimed at a blank line is written on it (twig ≥ 3.5.2), which is how
5113 /// "before the paragraph" is said through a gesture that only knows
5114 /// "after" — `---\n\npara`, and `prev\n\n---\n\npara` mid-document. Everywhere
5115 /// else the rule simply lands after the block, which is both twig's own
5116 /// answer and the better one: splitting a fenced code block would leave two
5117 /// fences with a rule between them, and splitting a list item would mint an
5118 /// item nobody asked for on the way to a rule that lands after the list
5119 /// regardless. A table and a setext heading refuse the split outright, so
5120 /// they take the same path by themselves.
5121 pub fn insert_thematic_break(&mut self) {
5122 if self.read_only || self.refuse_unsupported("thematic break", Gesture::InsertThematicBreak)
5123 {
5124 return;
5125 }
5126 self.caret = self.skip_trailing_close_delims(self.caret);
5127 // A selection is replaced by the rule, so collapse it first and let the
5128 // split-and-rule below run from the caret it leaves behind.
5129 if let Some((s, e)) = self.selection() {
5130 self.splice(s, e, "", EditKind::Other);
5131 }
5132 self.anchor = None;
5133 self.record_caret();
5134 let at = self.caret;
5135 if self.caret_parts_bare_paragraph() {
5136 // A failure here is not fatal: the rule still lands after the block,
5137 // which is exactly what this call was trying to improve on.
5138 let _ = self.editor.split_block(at);
5139 }
5140 match self.editor.insert_thematic_break(at) {
5141 Ok(change) => {
5142 self.last_edit_kind = None;
5143 self.refresh();
5144 self.anchor = None;
5145 self.caret = change.new.end;
5146 self.dirty = self.source != self.clean_source;
5147 self.status = None;
5148 self.clamp_caret();
5149 self.record_caret();
5150 }
5151 Err(e) => self.status = Some(format!("thematic break: {e}")),
5152 }
5153 }
5154
5155 /// Insert a fresh table at the caret — the toolbar's Table button. One
5156 /// header row, `rows` empty body rows, `cols` columns, spelled by twig in
5157 /// the document's own dialect and placed the way its thematic break is:
5158 /// after the caret's block, blank-separated. A bare paragraph is parted
5159 /// around the caret first, exactly as
5160 /// [`insert_thematic_break`](Self::insert_thematic_break) parts it, so the
5161 /// table lands *at* the caret rather than after everything the caret's
5162 /// paragraph says.
5163 ///
5164 /// The caret ends in the first header cell, selected the way Tab selects
5165 /// a cell — the natural next act is to type the heading, and Tab then
5166 /// walks the grid. That cell is read back from the rebuilt table map
5167 /// rather than computed from the splice, because twig's blank line and
5168 /// quote prefix put the first bar at an offset only the reparse knows.
5169 ///
5170 /// The shape is the caller's: a menu offers a few, a dialog asks. Zero
5171 /// rows or columns is twig's refusal (a header with nothing under it is
5172 /// what its row delete refuses to leave), reported through `status`.
5173 pub fn insert_table(&mut self, rows: usize, cols: usize) {
5174 if self.read_only || self.refuse_unsupported("table", Gesture::InsertTable) {
5175 return;
5176 }
5177 self.caret = self.skip_trailing_close_delims(self.caret);
5178 if let Some((s, e)) = self.selection() {
5179 self.splice(s, e, "", EditKind::Other);
5180 }
5181 self.anchor = None;
5182 self.record_caret();
5183 let at = self.caret;
5184 if self.caret_parts_bare_paragraph() {
5185 let _ = self.editor.split_block(at);
5186 }
5187 match self.editor.insert_table(at, rows, cols) {
5188 Ok(change) => {
5189 self.last_edit_kind = None;
5190 self.refresh();
5191 self.anchor = None;
5192 self.caret = change.new.end;
5193 self.dirty = self.source != self.clean_source;
5194 self.status = None;
5195 self.clamp_caret();
5196 // Into the first header cell of the table just written: the
5197 // first table whose grid begins inside the splice.
5198 self.rebuild_map();
5199 let first_cell = self
5200 .vmap
5201 .tables
5202 .iter()
5203 .filter_map(|t| t.grid.first().and_then(|row| row.cells.first()))
5204 .find(|cell| cell.start >= change.new.start && cell.start < change.new.end)
5205 .map(|cell| (cell.start, cell.end));
5206 if let Some((start, end)) = first_cell {
5207 self.select_cell(start, end);
5208 }
5209 self.record_caret();
5210 }
5211 Err(e) => self.status = Some(format!("table: {e}")),
5212 }
5213 }
5214
5215 /// Whether the caret sits in a paragraph and nothing else — no list item, no
5216 /// quote, no fence, no table — with paragraph text still ahead of it. The
5217 /// one shape where parting the block around the caret is unambiguously what
5218 /// a rule button means; see
5219 /// [`insert_thematic_break`](Self::insert_thematic_break) for why every other
5220 /// container is left to take the rule after itself.
5221 ///
5222 /// The "text ahead" half is what keeps `split_block` from running at the
5223 /// one edge where its output composes badly. At a paragraph's end twig
5224 /// cannot mint the empty second half (no format spells an empty
5225 /// paragraph), so it writes only the separator — a blank line and the
5226 /// slot Enter leaves for the paragraph to come — and a block then aimed at
5227 /// the first half lands above a slot that nothing fills: `para\n` with the
5228 /// caret at 4 came out as `para\n\n* * *\n\n\n`. Trailing whitespace counts
5229 /// as nothing ahead, since the split would shed it as the second half's
5230 /// leading indent and leave the same slot. Which end of the newline a
5231 /// paragraph's span stops at differs between the formats (Markdown before
5232 /// it, djot after), which is why this reads the remaining bytes rather
5233 /// than comparing offsets. The paragraph's
5234 /// start is deliberately not the same case — see
5235 /// [`insert_thematic_break`](Self::insert_thematic_break) for why that
5236 /// split is kept.
5237 fn caret_parts_bare_paragraph(&mut self) -> bool {
5238 let caret = self.caret.min(self.source.len());
5239 let Ok(chain) = self.editor.ancestors_at(caret) else {
5240 return false;
5241 };
5242 let mut para_end = None;
5243 for m in chain {
5244 match m.kind {
5245 Kind::Para => para_end = Some(m.span.end.min(self.source.len())),
5246 Kind::ListItem
5247 | Kind::TaskListItem
5248 | Kind::BlockQuote
5249 | Kind::CodeBlock
5250 | Kind::Table => return false,
5251 _ => {}
5252 }
5253 }
5254 match para_end {
5255 Some(end) if end > caret => !self.source[caret..end].trim().is_empty(),
5256 _ => false,
5257 }
5258 }
5259
5260 /// The destination of the link under the caret — what a Link prompt shows so
5261 /// ⌘K on an existing link edits its URL instead of asking for it again.
5262 /// `None` when the caret stands in no link.
5263 ///
5264 /// An autolink carries no separate destination: its text *is* the URL, so
5265 /// that's what comes back for one.
5266 pub fn link_destination_at_caret(&mut self) -> Option<String> {
5267 self.link_destination_at(self.caret)
5268 }
5269
5270 /// The destination of the link at `off`.
5271 /// [`link_destination_at_caret`](Self::link_destination_at_caret) for a place
5272 /// the caret isn't.
5273 ///
5274 /// The offset form exists for the same reason
5275 /// [`footnote_at`](Self::footnote_at)'s does: a frontend drawing a *piece* of
5276 /// the document somewhere else — a footnote's text in a popover, say — has
5277 /// rows and runs but no caret in them, and still needs to know which of those
5278 /// runs a reader can follow.
5279 pub fn link_destination_at(&mut self, off: usize) -> Option<String> {
5280 self.nodes()
5281 .into_iter()
5282 .filter(|n| matches!(n.kind.as_str(), "link" | "url" | "email"))
5283 .filter(|n| n.span.start <= off && off < n.span.end)
5284 .max_by_key(|n| n.span.start)
5285 .and_then(|n| n.destination.or(n.text))
5286 }
5287
5288 /// Where the locator `id` lands in this document — the `#v2` half of a
5289 /// `chapter.dj#v2`, resolved to the block it names. `None` when nothing here
5290 /// answers to it.
5291 ///
5292 /// The other end of a link, and the reason this exists: without it a
5293 /// destination has only file granularity, so following a citation into a
5294 /// chapter drops the reader at the top of it to hunt for the verse. Which is
5295 /// also why it is a *document* query rather than a caret one — the document
5296 /// being asked is usually not the one the reader is in.
5297 ///
5298 /// Three readings, tried in order, because the same `#some-heading` is
5299 /// written three ways across the formats leaf opens:
5300 ///
5301 /// 1. **A declared id**, exactly as written: djot's `{#v1}` on a block, and
5302 /// the auto-ids djot mints for its headings. The only exact answer, so it
5303 /// goes first — a document that says `{#v1}` has settled the question.
5304 /// 2. **A declared id, slugged.** djot spells a heading's auto-id
5305 /// `Some-Heading-Here`; nearly every tool that *writes* a link to one
5306 /// spells it `#some-heading-here`. Comparing slugs is what lets a link
5307 /// authored anywhere land on a djot heading.
5308 /// 3. **A heading's text, slugged.** Markdown has no ids at all — twig mints
5309 /// none and `{#custom}` is literal text in a Markdown heading — so for
5310 /// the format most vaults are written in, the heading's own words are the
5311 /// only thing a fragment can name. This is the rule every Markdown
5312 /// renderer already follows, which is what makes `#a-heading` mean in
5313 /// diaryx what it means on the web.
5314 ///
5315 /// Ties go to the earliest match, then to the widest: a duplicated id is the
5316 /// document's mistake and the first one is the answer every anchor
5317 /// implementation gives, while preferring the wider span picks the section
5318 /// over the heading that opens it — more for a peek to show, same place to
5319 /// land.
5320 pub fn locate(&mut self, id: &str) -> Option<Landing> {
5321 let id = id.trim();
5322 if id.is_empty() {
5323 return None;
5324 }
5325 let nodes = self.nodes();
5326
5327 // Earliest wins, then widest. `Reverse` on the end because `min_by_key`
5328 // is picking, among nodes that start together, the one that ends last.
5329 let pick = |matches: &mut dyn Iterator<Item = &FlatNode>| {
5330 matches
5331 .min_by_key(|n| (n.span.start, std::cmp::Reverse(n.span.end)))
5332 .map(|n| Landing {
5333 start: n.span.start,
5334 end: n.span.end,
5335 })
5336 };
5337
5338 if let Some(landing) = pick(&mut nodes.iter().filter(|n| declared_id(n) == Some(id))) {
5339 return Some(landing);
5340 }
5341 let want = slug(id);
5342 if want.is_empty() {
5343 return None;
5344 }
5345 if let Some(landing) = pick(
5346 &mut nodes
5347 .iter()
5348 .filter(|n| declared_id(n).map(slug).as_deref() == Some(&*want)),
5349 ) {
5350 return Some(landing);
5351 }
5352
5353 // A heading by its words. Its span is one line, so the end comes from
5354 // where the *section* it opens gives out — the next heading that is not
5355 // under it, or the end of the document. A Markdown heading has no
5356 // section node to ask (twig only builds those for djot), and a peek that
5357 // showed the heading alone would answer "what does that say" with the
5358 // title of the thing it says.
5359 let heading = nodes
5360 .iter()
5361 .filter(|n| n.kind == Kind::Heading)
5362 .filter(|n| {
5363 n.content_span
5364 .clone()
5365 .and_then(|s| self.source.get(s))
5366 .is_some_and(|text| slug(text) == want)
5367 })
5368 .min_by_key(|n| n.span.start)?;
5369 let level = heading.level.unwrap_or(u32::MAX);
5370 let end = nodes
5371 .iter()
5372 .filter(|n| n.kind == Kind::Heading)
5373 .filter(|n| n.span.start > heading.span.start)
5374 .filter(|n| n.level.unwrap_or(u32::MAX) <= level)
5375 .map(|n| n.span.start)
5376 .min()
5377 .unwrap_or(self.source.len());
5378 Some(Landing {
5379 start: heading.span.start,
5380 end,
5381 })
5382 }
5383
5384 /// Write a footnote at the caret — the toolbar's Footnote button, and the
5385 /// one gesture in the footnote story that *authors* rather than follows.
5386 ///
5387 /// Both halves go in as one twig edit: the `[^1]` where the caret is, and
5388 /// the `[^1]:` definition at the end of the document. Half a footnote is not
5389 /// a footnote — a bare reference with nothing defining it renders as literal
5390 /// brackets — so a single button that wrote only the reference would leave
5391 /// the author to hand-spell the other half in a document that had just
5392 /// stopped showing them what the first half meant. One edit also means one
5393 /// undo takes both back.
5394 ///
5395 /// The definition's body is left empty and **the caret lands in it**, which
5396 /// is the whole point of pressing the button: nobody wants a reference to a
5397 /// note they have not written yet. Getting back to where they were writing
5398 /// is [`footnote_definition_at_caret`](Self::footnote_definition_at_caret) —
5399 /// the same return leg a reader following a reference already uses, so the
5400 /// author is left standing on the near end of a round trip that works.
5401 ///
5402 /// A selection collapses to its *end* rather than being replaced: a
5403 /// reference annotates the words before it, so "select the claim, add a
5404 /// footnote" should mark that claim, not consume it.
5405 pub fn insert_footnote(&mut self) {
5406 if self.read_only || self.refuse_unsupported("footnote", Gesture::InsertFootnote) {
5407 return;
5408 }
5409 let at = self.selection().map_or(self.caret, |(_, end)| end);
5410 self.anchor = None;
5411 self.caret = at;
5412 self.record_caret();
5413 let label = self.next_footnote_label();
5414 match self.editor.insert_footnote(at, &label) {
5415 Ok(change) => {
5416 self.last_edit_kind = None;
5417 self.refresh();
5418 self.anchor = None;
5419 // `change.new` runs from the reference to the end of the
5420 // document, so its start is the `[^1]` just written and
5421 // `footnote_at` resolves it to the note the same way a reader's
5422 // tap does — and to the note's *body*, which is already a caret
5423 // stop even when it is empty (the `[^1]:` marker draws as `[1] `
5424 // and has none), so this needs no snap on top. The fallback is
5425 // the reference's own offset: a format that spelled the pair some
5426 // way leaf can't read back should still leave the caret on the
5427 // edit rather than at the far end of a document it just grew.
5428 self.caret = self
5429 .footnote_at(change.new.start)
5430 .and_then(|note| note.offset)
5431 .unwrap_or(change.new.start);
5432 self.dirty = self.source != self.clean_source;
5433 self.status = None;
5434 self.clamp_caret();
5435 self.record_caret();
5436 }
5437 Err(e) => self.status = Some(format!("footnote: {e}")),
5438 }
5439 }
5440
5441 /// The label to give a footnote the author has not named: the lowest counting
5442 /// number no footnote in the document is already wearing.
5443 ///
5444 /// twig takes the label rather than minting one, because it holds no opinion
5445 /// about what a document's footnotes should be called — and it is right not
5446 /// to. Numbering them is what every author of a numbered note expects, and
5447 /// re-using a taken number would silently point the new reference at somebody
5448 /// else's note (twig reuses an existing definition rather than appending a
5449 /// second one, which is the right rule for citing a note twice on purpose and
5450 /// exactly the wrong accident to have by default).
5451 ///
5452 /// *References* are counted alongside definitions, not just definitions: a
5453 /// document carrying a dangling `[^2]` has a 2 that means something to
5454 /// whoever wrote it, and minting a definition for it here would answer a
5455 /// question nobody asked. Non-numeric labels (`[^why]`) are left out of the
5456 /// count entirely — they take no number, so they block none.
5457 fn next_footnote_label(&mut self) -> String {
5458 let mut taken: Vec<u32> = wysiwyg::footnote_definitions(&mut self.editor)
5459 .into_iter()
5460 .filter_map(|note| wysiwyg::footnote_label(&self.source, note.span.start))
5461 .filter_map(|label| label.parse().ok())
5462 .collect();
5463 taken.extend(
5464 self.nodes()
5465 .into_iter()
5466 .filter(|n| n.kind == Kind::FootnoteReference)
5467 .filter_map(|n| wysiwyg::footnote_reference_label(&self.source, n.span))
5468 .filter_map(|label| label.parse::<u32>().ok()),
5469 );
5470 (1..).find(|n| !taken.contains(n)).unwrap_or(1).to_string()
5471 }
5472
5473 /// The footnote reference under the caret, resolved to the note it names.
5474 /// [`footnote_at`](Self::footnote_at) at the caret's offset.
5475 pub fn footnote_at_caret(&mut self) -> Option<FootnoteRef> {
5476 self.footnote_at(self.caret)
5477 }
5478
5479 /// The footnote reference at `off`, resolved to the note it names — what a
5480 /// frontend shows when a reader activates a `[^1]`.
5481 ///
5482 /// A reference is not a link node, so
5483 /// [`link_destination_at_caret`](Self::link_destination_at_caret) does not
5484 /// (and should not) answer for one: a link names a destination to leave for,
5485 /// a reference names a note that is already in this document. Following one
5486 /// is a move within the page, which is why this hands back an `offset`
5487 /// rather than something to open.
5488 ///
5489 /// Offset-based rather than caret-only because the gesture that wants this
5490 /// most is the one that must not move the caret: a pointer hovering a `[1]`
5491 /// asks what note it names without disturbing where the reader was typing.
5492 /// The caret is just the offset a click already placed —
5493 /// [`footnote_at_caret`](Self::footnote_at_caret) passes it.
5494 ///
5495 /// `None` when `off` stands in no reference. A reference whose note the
5496 /// document never defines is *not* `None` — it answers with the label it
5497 /// looked for and no text, which is what lets a frontend say so instead of
5498 /// silently doing nothing.
5499 pub fn footnote_at(&mut self, off: usize) -> Option<FootnoteRef> {
5500 // Innermost-wins by latest start, the rule its link sibling uses.
5501 let span = self
5502 .nodes()
5503 .into_iter()
5504 .filter(|n| n.kind == Kind::FootnoteReference)
5505 .filter(|n| n.span.start <= off && off < n.span.end)
5506 .max_by_key(|n| n.span.start)?
5507 .span;
5508 let label = wysiwyg::footnote_reference_label(&self.source, span)?.to_string();
5509
5510 // The note itself. Definitions are roots beside `doc` rather than
5511 // children of it, so they're asked for directly — see
5512 // `wysiwyg::footnote_definitions`.
5513 let note = wysiwyg::footnote_definitions(&mut self.editor)
5514 .into_iter()
5515 .find(|m| wysiwyg::footnote_label(&self.source, m.span.start) == Some(&label));
5516 let Some(note) = note else {
5517 return Some(FootnoteRef {
5518 label,
5519 text: None,
5520 offset: None,
5521 end: None,
5522 });
5523 };
5524 let body = wysiwyg::footnote_body_span(&self.source, note.span.clone());
5525 Some(FootnoteRef {
5526 label,
5527 text: body
5528 .clone()
5529 .and_then(|b| self.source.get(b))
5530 .map(str::to_string),
5531 // The body's start, not the definition's — see `FootnoteRef::offset`.
5532 offset: body.clone().map(|b| b.start),
5533 end: body.map(|b| b.end),
5534 })
5535 }
5536
5537 /// The footnote *definition* the caret stands in, and where the reference
5538 /// that names it is. [`footnote_definition_at`](Self::footnote_definition_at)
5539 /// at the caret's offset.
5540 pub fn footnote_definition_at_caret(&mut self) -> Option<FootnoteDef> {
5541 self.footnote_definition_at(self.caret)
5542 }
5543
5544 /// The footnote definition spanning `off`, and where the reference that
5545 /// names it is — the return leg of [`footnote_at`](Self::footnote_at).
5546 ///
5547 /// The mirror image, deliberately: the same gesture that takes a reader from
5548 /// `[1]` down to the note takes them from the note back up to `[1]`, so
5549 /// following a footnote is a round trip rather than a fall. It needs no
5550 /// memory of how the reader arrived — the document says where the reference
5551 /// is — which is what makes it work for a reader who scrolled to the notes
5552 /// themselves, and what keeps it right after an edit moves either end.
5553 ///
5554 /// `None` when `off` stands in no definition. A definition nothing cites is
5555 /// *not* `None`, for [`FootnoteRef`]'s reason in reverse: it answers with
5556 /// its label and no offset, so a frontend can say "nothing refers to this"
5557 /// rather than offer a jump that goes nowhere.
5558 pub fn footnote_definition_at(&mut self, off: usize) -> Option<FootnoteDef> {
5559 // Definitions are roots beside `doc`, so `nodes()` — which walks the
5560 // document body — never reports one. They're asked for directly, the way
5561 // `footnote_at` asks for the note it resolves to.
5562 //
5563 // Closed at the end, unlike the half-open test its neighbours use. A
5564 // definition's span stops at its last content byte — the newline ending
5565 // the line is outside it — so `span.end` is the caret stop at the end of
5566 // the note's own row, not the first byte of anything after. Excluding it
5567 // meant the one caret an author is guaranteed to have, the one left
5568 // sitting at the end of the note they just typed, was in no definition at
5569 // all: writing a note and then asking to go back to its reference
5570 // answered nothing. Two definitions in a row still can't both match —
5571 // there is a blank line between them — and `max_by_key` decides anyway.
5572 let note = wysiwyg::footnote_definitions(&mut self.editor)
5573 .into_iter()
5574 .filter(|m| m.span.start <= off && off <= m.span.end)
5575 .max_by_key(|m| m.span.start)?;
5576 let label = wysiwyg::footnote_label(&self.source, note.span.start)?.to_string();
5577
5578 // The earliest reference carrying this label. `min` rather than a `find`,
5579 // because `nodes()` reports a flattened walk whose order is twig's
5580 // business, not document order. Bound first: the walk needs `&mut self`
5581 // and reading the labels back out needs `&self.source`.
5582 let nodes = self.nodes();
5583 let offset = nodes
5584 .into_iter()
5585 .filter(|n| n.kind == Kind::FootnoteReference)
5586 .filter(|n| {
5587 wysiwyg::footnote_reference_label(&self.source, n.span.clone()) == Some(&*label)
5588 })
5589 // Past the `[^`, onto the label — see `FootnoteDef::offset`.
5590 .map(|n| n.span.start + 2)
5591 .min();
5592 Some(FootnoteDef { label, offset })
5593 }
5594
5595 /// The destination of the image under the caret — what an image prompt shows
5596 /// so editing an existing image starts from its current URL instead of blank,
5597 /// the image analogue of [`link_destination_at_caret`](Self::link_destination_at_caret).
5598 /// `None` when the caret stands in no image. A caret resting just after a
5599 /// block image (its trailing stop) is still "in" it — the half-open span test
5600 /// excludes that offset, which is the intended precision: past the image is
5601 /// past it.
5602 pub fn image_destination_at_caret(&mut self) -> Option<String> {
5603 let off = self.caret;
5604 self.nodes()
5605 .into_iter()
5606 .filter(|n| n.kind == Kind::Image)
5607 .filter(|n| n.span.start <= off && off < n.span.end)
5608 .max_by_key(|n| n.span.start)
5609 .and_then(|n| n.destination)
5610 }
5611
5612 /// The language of the fenced code block the caret stands in — what a
5613 /// language prompt shows so editing it starts from the current value rather
5614 /// than blank. `None` when the caret is in no code block, or in one whose
5615 /// fence carries no language (or an indented block, which has no fence).
5616 pub fn code_language_at_caret(&mut self) -> Option<String> {
5617 let start = self.code_block_start_at_caret()?;
5618 wysiwyg::code_language(&self.source, start)
5619 }
5620
5621 /// Whether the caret stands in a fenced code block — the one a language
5622 /// prompt could edit. A frontend gates its "set language" affordance on this
5623 /// (an indented block, which can't carry a language, reports `false`).
5624 pub fn caret_in_fenced_code(&mut self) -> bool {
5625 self.code_block_start_at_caret()
5626 .is_some_and(|start| wysiwyg::code_info_span(&self.source, start).is_some())
5627 }
5628
5629 /// Set (or clear, with `""`) the language of the fenced code block the caret
5630 /// is in — the prompt's confirm. A no-op when the caret is in no fenced
5631 /// block, and a reported error for a language the format's fence cannot
5632 /// carry.
5633 ///
5634 /// twig rewrites the info string, so the fence's own width — measured
5635 /// against a body neither side touches — is kept, and a language holding a
5636 /// space, a line end or the fence character is refused rather than written
5637 /// out to reparse as something else. Leaf used to splice over the info span
5638 /// itself and `trim()` the input, which handled the one bad case it had
5639 /// thought of.
5640 pub fn set_code_language(&mut self, lang: &str) {
5641 // The read-only gate — this door reaches twig without the splice.
5642 if self.read_only {
5643 return;
5644 }
5645 if self.refuse_unsupported("code language", Gesture::SetCodeLanguage) {
5646 return;
5647 }
5648 if self.code_block_start_at_caret().is_none() {
5649 return;
5650 }
5651 let lang = lang.trim();
5652 // `None` clears the info string; `Some("")` asks for an empty one. Both
5653 // write a bare fence, and the prompt's empty value means "clear".
5654 let want = (!lang.is_empty()).then_some(lang);
5655 self.record_caret();
5656 match self.editor.set_code_language(self.caret, want) {
5657 Ok(_) => {
5658 self.last_edit_kind = None;
5659 self.refresh();
5660 self.anchor = None;
5661 self.dirty = self.source != self.clean_source;
5662 self.status = None;
5663 self.clamp_caret();
5664 self.record_caret();
5665 }
5666 Err(e) => self.status = Some(format!("code language: {e}")),
5667 }
5668 }
5669
5670 /// The `span.start` of the code block covering the caret — the anchor
5671 /// [`wysiwyg::code_info_span`] reads the fence from. `None` when the caret is
5672 /// in none.
5673 fn code_block_start_at_caret(&mut self) -> Option<usize> {
5674 let off = self.caret;
5675 self.nodes()
5676 .into_iter()
5677 .filter(|n| n.kind == Kind::CodeBlock && n.span.start <= off && off <= n.span.end)
5678 .max_by_key(|n| n.span.start)
5679 .map(|n| n.span.start)
5680 }
5681
5682 /// The source range of the text inside the link covering `off` — what sits
5683 /// between its `[` and `]`. `None` when twig reports no link there.
5684 fn link_text_span(&mut self, off: usize) -> Option<std::ops::Range<usize>> {
5685 self.nodes()
5686 .into_iter()
5687 // Two links can touch (`[a](x)[b](y)`), and then one's `span.end` is
5688 // the other's `span.start`; the link that starts latest at or before
5689 // `off` is the one `off` is actually in.
5690 .filter(|n| n.kind == Kind::Link && n.span.start <= off && off < n.span.end)
5691 .max_by_key(|n| n.span.start)
5692 .and_then(|n| n.content_span)
5693 }
5694
5695 // ── undo / redo ───────────────────────────────────────────────────────────
5696 // twig owns the history of *bytes* (it owns the buffer) and now carries the
5697 // caret through it too: `record_caret` stashes each state's caret in twig's
5698 // opaque per-step blob, and undo/redo hand it back with the source they
5699 // restore. So leaf keeps no history of its own — no parallel stacks to march
5700 // in lockstep and silently drift out of it.
5701
5702 /// Undo the last edit step (⌘Z / ^Z), putting the caret and selection back
5703 /// where they were when that step began.
5704 pub fn undo(&mut self) {
5705 if self.read_only {
5706 return;
5707 }
5708 let (undone, redoable) = (self.undo_steps, self.redo_steps);
5709 match self.editor.undo() {
5710 Ok(Some(change)) => {
5711 self.after_history(change);
5712 // `refresh` counted the restore as an edit; it was a step back.
5713 self.undo_steps = undone.saturating_sub(1);
5714 self.redo_steps = redoable + 1;
5715 }
5716 Ok(None) => {
5717 self.undo_steps = 0;
5718 self.status = Some("nothing to undo".into());
5719 }
5720 Err(e) => self.status = Some(format!("undo: {e}")),
5721 }
5722 }
5723
5724 /// Redo the last undone edit step (⇧⌘Z / ^Y), putting the caret and
5725 /// selection back where that step originally left them.
5726 pub fn redo(&mut self) {
5727 if self.read_only {
5728 return;
5729 }
5730 let (undone, redoable) = (self.undo_steps, self.redo_steps);
5731 match self.editor.redo() {
5732 Ok(Some(change)) => {
5733 self.after_history(change);
5734 // `refresh` counted the restore as an edit; it was a step forward.
5735 self.undo_steps = undone + 1;
5736 self.redo_steps = redoable.saturating_sub(1);
5737 }
5738 Ok(None) => {
5739 self.redo_steps = 0;
5740 self.status = Some("nothing to redo".into());
5741 }
5742 Err(e) => self.status = Some(format!("redo: {e}")),
5743 }
5744 }
5745
5746 /// Refresh the cached source and put the caret back where the step being
5747 /// undone/redone had it, clearing any active run.
5748 ///
5749 /// The caret comes from twig's blob for the restored state (what
5750 /// `record_caret` stored). `change` is only the fallback for a state with no
5751 /// blob — a caret at the end of the restored text, which is where this always
5752 /// landed before the blobs were kept. It is the edit site, not where the user
5753 /// was standing, so it's a floor and not the behaviour: undoing should hand
5754 /// back the document *and* the place you were working, which for an edit made
5755 /// anywhere but under the caret are two different places.
5756 fn after_history(&mut self, change: Change) {
5757 self.refresh();
5758 match self
5759 .editor
5760 .caret_blob()
5761 .ok()
5762 .and_then(|b| CaretState::from_blob(&b))
5763 {
5764 Some(state) => {
5765 self.caret = state.caret.min(self.source.len());
5766 self.anchor = state.anchor.map(|a| a.min(self.source.len()));
5767 }
5768 None => {
5769 self.caret = change.new.end.min(self.source.len());
5770 self.anchor = None;
5771 }
5772 }
5773 self.goal_col = None;
5774 self.last_edit_kind = None;
5775 self.dirty = self.source != self.clean_source;
5776 self.status = None;
5777 self.clamp_caret();
5778 }
5779
5780 // ── the file ──────────────────────────────────────────────────────────────
5781
5782 #[cfg(feature = "fs")]
5783 pub fn save(&mut self) {
5784 if self.is_untitled() {
5785 // No path to write and no name to invent: ⌘S on an untitled document
5786 // is a Save As, and only a frontend has a picker to ask with. Say so
5787 // rather than failing at the filesystem with an empty path.
5788 self.status = Some("untitled — save as…".into());
5789 return;
5790 }
5791 let path = self.path.clone();
5792 if self.write(&path) {
5793 self.mark_saved();
5794 }
5795 }
5796
5797 /// Save As: write the document to `path` and *move* it there — `self.path`
5798 /// becomes `path`, and every later [`Doc::save`] writes the new file. That's
5799 /// what Save As means; a copy would leave the user editing a document whose
5800 /// name is no longer where their keystrokes go.
5801 ///
5802 /// The move only happens if the bytes actually landed. A failed write leaves
5803 /// the path, `dirty`, and the disk watermark exactly as they were, with the
5804 /// same `save failed: …` status a failed [`Doc::save`] sets — the document
5805 /// must never come away believing it was saved.
5806 ///
5807 /// An existing `path` is overwritten, and the caller is the one that knows
5808 /// whether to ask first: a Save As picker has already run that prompt, and a
5809 /// second confirmation from down here would be the same question twice.
5810 ///
5811 /// `format` does **not** follow the new extension. The buffer is parsed as
5812 /// the format it was opened with, and re-reading it as another one is a
5813 /// conversion — a different, lossy operation that would throw away the undo
5814 /// history — not a rename. So `notes.md` saved as `notes.dj` holds Markdown
5815 /// in a `.dj` file, and `format_name()` keeps honestly saying `markdown`
5816 /// until it's reopened.
5817 #[cfg(feature = "fs")]
5818 pub fn save_as(&mut self, path: PathBuf) {
5819 if !self.write(&path) {
5820 return;
5821 }
5822 self.path = path;
5823 self.mark_saved();
5824 }
5825
5826 /// Put `source` on disk at `path`, reporting whether it got there. The one
5827 /// place leaf writes a document, so a save and a Save As can't disagree
5828 /// about what a failure looks like.
5829 #[cfg(feature = "fs")]
5830 fn write(&mut self, path: &Path) -> bool {
5831 match std::fs::write(path, self.source.as_bytes()) {
5832 Ok(()) => true,
5833 Err(e) => {
5834 self.status = Some(format!("save failed: {e}"));
5835 false
5836 }
5837 }
5838 }
5839
5840 /// Re-base the document's saved watermark to the current bytes: clears
5841 /// `dirty`, records `source` as the new clean state (so undoing back to here
5842 /// clears the flag again), and re-stamps the on-disk hash.
5843 ///
5844 /// [`Doc::save`]/[`Doc::save_as`] call this after a write lands. It is also
5845 /// the hook a **filesystem-free host** calls itself once it has persisted
5846 /// [`Doc::source`] its own way (a browser download, `localStorage`, a backend
5847 /// `PUT`) — which is why it is public and touches no filesystem: the bytes
5848 /// are already where that host wants them, and this just tells the model they
5849 /// are safe.
5850 pub fn mark_saved(&mut self) {
5851 self.clean_source = self.source.clone();
5852 self.dirty = false;
5853 // The bytes on disk are now ours, so this is the new watermark: without
5854 // re-stamping it, every save would report its own work as an external
5855 // change forever after.
5856 self.disk_hash = Some(hash_bytes(self.source.as_bytes()));
5857 self.status = Some(format!("saved {}", self.file_name()));
5858 }
5859
5860 /// What the file looks like now against the bytes leaf last read or wrote.
5861 ///
5862 /// Reads the file and hashes it (see `disk_hash` for why it isn't an mtime),
5863 /// so this is a filesystem round-trip, not a per-frame question — ask it
5864 /// when a window regains focus, on a timer, or before a save.
5865 ///
5866 /// This *only* reports the file. Whether the document also has unsaved edits
5867 /// is `dirty`, and the interesting case is the conjunction: `dirty` plus
5868 /// [`DiskState::Changed`] means a save overwrites someone's work and a
5869 /// [`Doc::reload`] discards the user's. leaf-core deliberately won't choose —
5870 /// it has no way to ask — so it hands a frontend both halves and lets it put
5871 /// the question to the person who can answer it.
5872 #[cfg(feature = "fs")]
5873 pub fn disk_state(&self) -> DiskState {
5874 let Some(want) = self.disk_hash else {
5875 return DiskState::Untitled;
5876 };
5877 match std::fs::read(&self.path) {
5878 Ok(bytes) if hash_bytes(&bytes) == want => DiskState::Unchanged,
5879 Ok(_) => DiskState::Changed,
5880 Err(e) if e.kind() == std::io::ErrorKind::NotFound => DiskState::Missing,
5881 Err(_) => DiskState::Unreadable,
5882 }
5883 }
5884
5885 /// Re-read the file and replace the document with what's there — the other
5886 /// answer to a [`DiskState::Changed`].
5887 ///
5888 /// **Discards unsaved changes, unconditionally.** It doesn't check `dirty`
5889 /// first: a frontend that wants to protect unsaved work asks (`dirty` +
5890 /// [`Doc::disk_state`]) *before* calling this, and one reloading a clean
5891 /// document shouldn't have to argue with a guard.
5892 ///
5893 /// **The undo history survives, and the reload is one step in it.** The
5894 /// whole buffer is spliced with the file's bytes through the same door every
5895 /// other edit goes through, as an [`EditKind::Other`] that coalesces with
5896 /// nothing on either side — so ^Z after a formatter or a `git checkout` has
5897 /// swapped the document out from under a reader gives them back what they
5898 /// were looking at, marked dirty, and ^Z again carries on into whatever they
5899 /// had done before it. This used to build a fresh parse and drop the stack,
5900 /// on the reasoning that twig's history belongs to the buffer and these are
5901 /// different bytes; that is true of *rebasing* a step onto them and not of
5902 /// recording the swap itself as one, which is all this is. A splice twig
5903 /// won't take falls back to the fresh parse, and only that path still costs
5904 /// the history.
5905 ///
5906 /// The caret keeps its byte offset, clamped to the new length; the selection
5907 /// is dropped. Anything cleverer would be a lie: leaf doesn't know how the
5908 /// file changed, so it can't know where the caret "still" is. Clamping keeps
5909 /// it where the user left it in the common case (a change further down the
5910 /// file, or none in the text they're sitting in), and never puts it
5911 /// somewhere invalid. A selection has two such offsets and no such excuse —
5912 /// silently reinterpreting one over changed bytes would arm the *next*
5913 /// keystroke to delete something the user never selected.
5914 ///
5915 /// Nothing is touched unless the whole reload succeeds; a failure leaves the
5916 /// document alone with a status.
5917 #[cfg(feature = "fs")]
5918 pub fn reload(&mut self) {
5919 if self.is_untitled() {
5920 self.status = Some("no file to reload".into());
5921 return;
5922 }
5923 let bytes = match std::fs::read(&self.path) {
5924 Ok(b) => b,
5925 Err(e) => {
5926 self.status = Some(format!("reload failed: {e}"));
5927 return;
5928 }
5929 };
5930 let Ok(source) = String::from_utf8(bytes) else {
5931 self.status = Some("reload failed: file is not UTF-8".into());
5932 return;
5933 };
5934 // Already these bytes — someone saved a file back unchanged, or leaf's
5935 // own write is being read back. Re-baseline against it and stop: a
5936 // splice of the text onto itself would put an undo step on the stack for
5937 // something nobody did.
5938 if source == self.source {
5939 self.disk_hash = Some(hash_bytes(source.as_bytes()));
5940 self.clean_source = source;
5941 self.dirty = false;
5942 self.status = Some(format!("reloaded {}", self.file_name()));
5943 return;
5944 }
5945 let caret = self.caret;
5946 // The pre-reload caret, so undoing the swap puts it back where the
5947 // reader was standing — the same bracketing `splice_exact` does.
5948 self.record_caret();
5949 if self
5950 .editor
5951 .edit_range(0, self.source.len(), &source)
5952 .is_ok()
5953 {
5954 self.refresh();
5955 } else {
5956 // twig wouldn't take the splice. Start over from the bytes, which is
5957 // what this always did, and is the one path that still costs the
5958 // history — `format` is the format this document *is*, not what the
5959 // (unchanged) name now says, see `save_as`.
5960 match new_editor(source.as_bytes(), self.format) {
5961 Ok(editor) => {
5962 self.editor = editor;
5963 self.source = source.clone();
5964 // Not going through `refresh`, so the revision has to move
5965 // here or every frontend keeps painting the old file from
5966 // cache.
5967 self.revision += 1;
5968 }
5969 Err(e) => {
5970 self.status = Some(format!("reload failed: {e}"));
5971 return;
5972 }
5973 }
5974 }
5975 self.disk_hash = Some(hash_bytes(source.as_bytes()));
5976 self.clean_source = self.source.clone();
5977 self.caret = caret.min(self.source.len());
5978 self.anchor = None;
5979 self.goal_col = None;
5980 self.last_edit_kind = None;
5981 self.dirty = false;
5982 self.status = Some(format!("reloaded {}", self.file_name()));
5983 self.clamp_caret();
5984 // And the post-reload caret, so a redo restores it.
5985 self.record_caret();
5986 }
5987
5988 /// Re-read the source from twig after it has changed the document. The one
5989 /// funnel every edit, undo, and redo comes through — so it's where the
5990 /// revision moves, and anything cached against the text dies here.
5991 fn refresh(&mut self) {
5992 if let Ok(s) = self.editor.source_str() {
5993 self.source = s;
5994 }
5995 self.revision += 1;
5996 // An edit is a step onto the history and the end of anything undone;
5997 // `undo`/`redo` come through here too and correct this after.
5998 self.undo_steps += 1;
5999 self.redo_steps = 0;
6000 self.clamp_caret();
6001 }
6002
6003 /// Whether [`undo`](Self::undo) has a step to take back — for a native
6004 /// Edit menu to enable its item by. See the note on `undo_steps` for what
6005 /// "has" means here.
6006 pub fn can_undo(&self) -> bool {
6007 !self.read_only && self.undo_steps > 0
6008 }
6009
6010 /// Whether [`redo`](Self::redo) has an undone step to restore.
6011 pub fn can_redo(&self) -> bool {
6012 !self.read_only && self.redo_steps > 0
6013 }
6014
6015 // ── caret movement ─────────────────────────────────────────────────────────
6016 // `extend` grows the selection (Shift+motion): it pins the anchor on the
6017 // first extended step and moves only the caret; an un-extended motion drops
6018 // the selection.
6019
6020 /// Place the caret at byte `offset` (clamped to a char boundary), extending
6021 /// the selection when `extend` is set. The public form of `move_to`, for a
6022 /// frontend that hit-tests pixels straight to a source offset.
6023 pub fn place_caret(&mut self, offset: usize, extend: bool) {
6024 self.goal_col = None;
6025 let before = self.caret;
6026 // A pixel hit-test can land between the visible caret stops — in the
6027 // blank gap a paragraph break is drawn with, or inside a hidden delimiter.
6028 // Snap to the nearest real stop so the caret can't come to rest where it
6029 // would draw in one place and type in another. The `(row, col)` click
6030 // path (`click`) already snaps this way through `offset_of_pos`; the
6031 // source view reaches every byte, so it snaps to nothing.
6032 let target = match self.view {
6033 View::Wysiwyg => self.vmap.snap_to_stop(offset.min(self.source.len())),
6034 // The source view reaches every byte, so there is no stop to snap
6035 // to — but "every byte" still means every *character* boundary. A
6036 // caret resting inside a multi-byte character draws nowhere real
6037 // and panics the next time anything slices there.
6038 View::Source => self.char_boundary_at_or_before(offset),
6039 };
6040 self.move_to(target, extend);
6041 self.clamp_caret();
6042 self.debug_assert_on_a_stop(before);
6043 }
6044
6045 /// Select the whole document (⌘A / Ctrl+A) — everything reachable in the
6046 /// active view, so in WYSIWYG it starts below hidden frontmatter (copy won't
6047 /// grab the metadata) while the source view still selects the literal whole.
6048 pub fn select_all(&mut self) {
6049 self.anchor = Some(self.caret_floor());
6050 self.caret = self.source.len();
6051 self.goal_col = None;
6052 self.last_edit_kind = None;
6053 self.status = None;
6054 }
6055
6056 /// Select the word (or whitespace / punctuation run) at `offset` — the
6057 /// double-click gesture. Anchors on the run's start with the caret at its
6058 /// end so a following Shift-motion extends from the far edge.
6059 pub fn select_word_at(&mut self, offset: usize) {
6060 let (s, e) = word_range_at(&self.source, offset.min(self.source.len()));
6061 self.anchor = Some(s);
6062 self.caret = e;
6063 self.goal_col = None;
6064 self.last_edit_kind = None;
6065 self.status = None;
6066 self.clamp_caret();
6067 }
6068
6069 /// Select the whole enclosing text block (paragraph, heading, list item's
6070 /// text…) at `offset` — the triple-click gesture. Reads the range straight
6071 /// from the AST (twig's `content_span`), so it selects the entire *logical*
6072 /// paragraph even when that paragraph soft-wraps across several visual rows —
6073 /// where a visual-row-based select breaks down, because one source offset at
6074 /// a wrap boundary belongs to two rows at once.
6075 pub fn select_block_at(&mut self, offset: usize) {
6076 let off = offset.min(self.source.len());
6077 let range = self
6078 .editor
6079 .ancestors_at(off)
6080 .ok()
6081 .and_then(|chain| {
6082 // Ancestors run root → deepest; the deepest node that is neither
6083 // an inline span nor a multi-block container is the text block
6084 // the caret sits in (a paragraph, a heading, a code block…).
6085 chain
6086 .into_iter()
6087 .rev()
6088 .find(|m| !wysiwyg::is_inline_kind(&m.kind) && !is_block_container(&m.kind))
6089 .map(|m| m.content_span.unwrap_or(m.span))
6090 })
6091 .unwrap_or_else(|| source_line_range(&self.source, off));
6092 self.anchor = Some(range.start.min(self.source.len()));
6093 self.caret = range.end.min(self.source.len());
6094 self.goal_col = None;
6095 self.last_edit_kind = None;
6096 self.status = None;
6097 self.clamp_caret();
6098 }
6099
6100 /// Select the exact source range `[start, end)` — anchor at `start`, caret
6101 /// at `end` — without snapping either end to a visible caret stop.
6102 ///
6103 /// The one caret verb that takes a range it was *handed* rather than one it
6104 /// worked out, for a host that already knows the bytes it means: a search
6105 /// hit, an annotation's footprint, a quote re-anchored through
6106 /// [`Doc::selection_quote`]. [`place_caret`](Self::place_caret) is the
6107 /// wrong tool for that, and not by a little — it snaps to the nearest
6108 /// *visible* stop, and where a range butts up against a hidden delimiter
6109 /// the nearest stop is the one before it, so selecting the "needle" of
6110 /// `**needle**` comes back with "needl" and an edit against it strands the
6111 /// "e".
6112 ///
6113 /// What `place_caret` does that is bookkeeping rather than snapping still
6114 /// happens here, because a host handing in a range is not asking to opt out
6115 /// of the invariants:
6116 ///
6117 /// - both ends are clamped into the document and up to
6118 /// [`caret_floor`](Self::caret_floor) — in WYSIWYG the leading
6119 /// frontmatter is hidden, and a caret parked in it draws nowhere and
6120 /// types into the metadata;
6121 /// - both land on character boundaries, so nothing slices a `é` in half;
6122 /// - the sticky vertical goal column is dropped, and any armed inline mark
6123 /// disarmed, since a range from outside inherits neither.
6124 ///
6125 /// An empty range is a caret rather than a selection —
6126 /// [`selection`](Self::selection) reports `None` for it, as it does for any
6127 /// anchor that has met the caret.
6128 pub fn select_range(&mut self, start: usize, end: usize) {
6129 let floor = self.caret_floor();
6130 let anchor = self.char_boundary_at_or_before(start.clamp(floor, self.source.len()));
6131 let caret = self.char_boundary_at_or_before(end.clamp(floor, self.source.len()));
6132 self.anchor = Some(anchor);
6133 self.caret = caret;
6134 self.goal_col = None;
6135 self.status = None;
6136 self.last_edit_kind = None;
6137 self.clear_pending();
6138 }
6139
6140 /// `offset` itself if it is a character boundary, else the boundary before
6141 /// it. An offset that isn't one draws nowhere real and panics the next time
6142 /// anything slices there.
6143 fn char_boundary_at_or_before(&self, offset: usize) -> usize {
6144 let mut o = offset.min(self.source.len());
6145 while o > 0 && !self.source.is_char_boundary(o) {
6146 o -= 1;
6147 }
6148 o
6149 }
6150
6151 /// The lowest source offset the caret may occupy in the active view. In
6152 /// WYSIWYG, leading frontmatter is hidden and unreachable, so the floor is
6153 /// the first rendered offset; the source view reaches everything, so it's 0.
6154 fn caret_floor(&self) -> usize {
6155 match self.view {
6156 View::Wysiwyg => self.vmap.content_start.min(self.source.len()),
6157 View::Source => 0,
6158 }
6159 }
6160
6161 /// Land in a table cell with its whole content selected — the anchor at the
6162 /// cell's start, the caret at its end — so a Tab/Return hop into a cell reads
6163 /// like tabbing into a form field: the text comes up selected, so typing
6164 /// replaces it and an arrow collapses to an edge. An empty cell (`start ==
6165 /// end`) collapses to a plain caret home (an empty selection is no selection).
6166 fn select_cell(&mut self, start: usize, end: usize) {
6167 self.select_range(start, end);
6168 }
6169
6170 fn move_to(&mut self, offset: usize, extend: bool) {
6171 if extend {
6172 if self.anchor.is_none() {
6173 self.anchor = Some(self.caret);
6174 }
6175 } else {
6176 self.anchor = None;
6177 }
6178 self.caret = offset.min(self.source.len()).max(self.caret_floor());
6179 self.status = None;
6180 // A caret move ends the current typing/deletion run, so the next edit
6181 // starts a fresh undo group rather than coalescing across the gap.
6182 self.last_edit_kind = None;
6183 // Moving away disarms any sticky mark — "start bold" applies only where
6184 // it was asked for, not wherever the caret next lands.
6185 self.clear_pending();
6186 }
6187
6188 // In the source view, motion walks source bytes / source lines. In the
6189 // WYSIWYG view it walks the rendered glyph grid (the visual map), which is
6190 // what steps the caret cleanly over hidden delimiters.
6191
6192 pub fn move_left(&mut self, extend: bool) {
6193 self.goal_col = None;
6194 if !extend && let Some((s, _e)) = self.selection() {
6195 self.move_to(s, false);
6196 return;
6197 }
6198 let target = match self.view {
6199 View::Source => {
6200 if self.caret > 0 {
6201 prev_boundary(&self.source, self.caret)
6202 } else {
6203 0
6204 }
6205 }
6206 // Walks caret *stops*, not columns: decoration (a table border, a
6207 // cell's padding) is stepped over in one press, and a hidden
6208 // delimiter never holds the caret up — though the end of a mark's
6209 // content is a stop of its own (`VisualMap::mark_ends`), so
6210 // leaving `**bold**` from past its `**` is a press onto the end of
6211 // the bold and another onto the `d`.
6212 View::Wysiwyg => self
6213 .vmap
6214 .caret_stop_before(self.caret)
6215 .unwrap_or(self.caret),
6216 };
6217 let before = self.caret;
6218 self.move_to(target, extend);
6219 self.debug_assert_on_a_stop(before);
6220 }
6221
6222 pub fn move_right(&mut self, extend: bool) {
6223 self.goal_col = None;
6224 if !extend && let Some((_s, e)) = self.selection() {
6225 self.move_to(e, false);
6226 return;
6227 }
6228 let target = match self.view {
6229 View::Source => {
6230 if self.caret < self.source.len() {
6231 next_boundary(&self.source, self.caret)
6232 } else {
6233 self.caret
6234 }
6235 }
6236 View::Wysiwyg => self.vmap.caret_stop_after(self.caret).unwrap_or(self.caret),
6237 };
6238 let before = self.caret;
6239 self.move_to(target, extend);
6240 self.debug_assert_on_a_stop(before);
6241 }
6242
6243 /// Move to the start of the previous word (⌥← / Ctrl+←).
6244 pub fn move_word_left(&mut self, extend: bool) {
6245 self.goal_col = None;
6246 let before = self.caret;
6247 let target = self.word_left_from(self.caret);
6248 self.move_to(target, extend);
6249 self.debug_assert_on_a_stop(before);
6250 }
6251
6252 /// Move to the end of the next word (⌥→ / Ctrl+→).
6253 pub fn move_word_right(&mut self, extend: bool) {
6254 self.goal_col = None;
6255 let before = self.caret;
6256 let target = self.word_right_from(self.caret);
6257 self.move_to(target, extend);
6258 self.debug_assert_on_a_stop(before);
6259 }
6260
6261 // Word boundaries are found in the space the *view* is in. The source view
6262 // walks the source, because there the source is what's rendered. WYSIWYG
6263 // walks the rendered text instead: `**` is invisible to the user, so it has
6264 // to be invisible to word motion too — a caret parked inside one draws in
6265 // the column after `bold` and types two bytes earlier, and a word-delete
6266 // that stops there shreds the markup into `a ** c`.
6267
6268 /// The word boundary to the left of `off` in the active view's space.
6269 fn word_left_from(&self, off: usize) -> usize {
6270 match self.view {
6271 View::Source => prev_word(&self.source, off),
6272 View::Wysiwyg => self.glyph_word_left(off),
6273 }
6274 }
6275
6276 /// The word boundary to the right of `off` in the active view's space.
6277 fn word_right_from(&self, off: usize) -> usize {
6278 match self.view {
6279 View::Source => next_word(&self.source, off),
6280 View::Wysiwyg => self.glyph_word_right(off),
6281 }
6282 }
6283
6284 /// The character class of the glyph drawn at stop `off`.
6285 ///
6286 /// Read from the source, because a stop points at the source byte its glyph
6287 /// came from — the source *is* where the rendered character is written. What
6288 /// makes the walk glyph space rather than source space is that it only ever
6289 /// visits stops, and the hidden bytes between them have none.
6290 fn class_at(&self, off: usize) -> Class {
6291 self.source
6292 .get(off..)
6293 .and_then(|s| s.chars().next())
6294 .map_or(Class::Space, classify)
6295 }
6296
6297 /// [`next_word`] in glyph space: skip any leading separators, then consume
6298 /// the following word run, with the stop table standing in for the source's
6299 /// characters.
6300 fn glyph_word_right(&self, from: usize) -> usize {
6301 let Some(mut off) = self.vmap.stop_at_or_after(from) else {
6302 return from;
6303 };
6304 let mut in_word = false;
6305 loop {
6306 match self.class_at(off) {
6307 Class::Word => in_word = true,
6308 _ if in_word => return off,
6309 _ => {}
6310 }
6311 match self.vmap.stop_after(off) {
6312 Some(next) => off = next,
6313 None => return off,
6314 }
6315 }
6316 }
6317
6318 /// [`prev_word`] in glyph space: skip separators walking left, then consume
6319 /// the preceding word run.
6320 fn glyph_word_left(&self, from: usize) -> usize {
6321 let Some(mut off) = self.vmap.stop_at_or_before(from) else {
6322 return from;
6323 };
6324 let mut in_word = false;
6325 while let Some(prev) = self.vmap.stop_before(off) {
6326 match self.class_at(prev) {
6327 Class::Word => in_word = true,
6328 _ if in_word => return off,
6329 _ => {}
6330 }
6331 off = prev;
6332 }
6333 off
6334 }
6335
6336 /// After a motion that walks the visual map, the caret must be *on* the map.
6337 /// A stop is the only offset where the caret draws and edits in the same
6338 /// place, and it's the invariant both a caret parked inside an emoji and one
6339 /// parked inside a `**` were quietly breaking.
6340 ///
6341 /// Only when the caret actually moved: a walk with nowhere to go leaves it
6342 /// where it was, which is wherever the floor or a frontend put it rather
6343 /// than somewhere this motion chose.
6344 fn debug_assert_on_a_stop(&self, before: usize) {
6345 debug_assert!(
6346 self.view != View::Wysiwyg
6347 || self.vmap.num_rows() == 0
6348 || self.caret == before
6349 || self.vmap.is_stop(self.caret),
6350 "motion left the caret at {}, which is not a caret stop: it would draw in \
6351 one place and type in another",
6352 self.caret
6353 );
6354 }
6355
6356 // Up and Down run off the ends of the document rather than stopping dead at
6357 // them: Up from the first row lands at the document's start, Down from the
6358 // last at its end. That's Cocoa's rule (`moveUp:`/`moveDown:` past the edge
6359 // are `moveToBeginningOfDocument:`/`moveToEndOfDocument:`), and holding ↓
6360 // reaching the end of the text is what a reader means by it.
6361 //
6362 // The views used to disagree here by accident rather than by decision: the
6363 // source view fell into the edge behaviour through `row_col_to_offset`
6364 // clamping an out-of-range row to the end of the string, while WYSIWYG had
6365 // no row below to walk to and did nothing at all. They share the rule now,
6366 // each in its own space — the source view reaches every byte, WYSIWYG only
6367 // the offsets it draws.
6368
6369 pub fn move_up(&mut self, extend: bool) {
6370 let (row, col) = self.caret_pos();
6371 let goal = self.goal_col.unwrap_or(col);
6372 let target = match self.view {
6373 View::Source => match row.checked_sub(1) {
6374 Some(r) => row_col_to_offset(&self.source, r, goal),
6375 None => self.reachable_start(),
6376 },
6377 // A table's border rules are drawn but hold no caret, so Up steps
6378 // over them to the row that does.
6379 View::Wysiwyg => match self.vmap.navigable_above(row) {
6380 Some(r) => self.row_target(r, goal),
6381 None => self.reachable_start(),
6382 },
6383 };
6384 self.step_vertical(target, goal, extend);
6385 }
6386
6387 pub fn move_down(&mut self, extend: bool) {
6388 let (row, col) = self.caret_pos();
6389 let goal = self.goal_col.unwrap_or(col);
6390 let target = match self.view {
6391 View::Source => match self.source_row_below(row) {
6392 Some(r) => row_col_to_offset(&self.source, r, goal),
6393 None => self.reachable_end(),
6394 },
6395 View::Wysiwyg => match self.vmap.navigable_below(row) {
6396 Some(r) => self.row_target(r, goal),
6397 None => self.reachable_end(),
6398 },
6399 };
6400 self.step_vertical(target, goal, extend);
6401 }
6402
6403 /// Land a vertical motion at `target`, latching the `goal` column it aimed
6404 /// with so the rest of the run keeps aiming there.
6405 ///
6406 /// A motion with nowhere to go changes *nothing*, the goal column included:
6407 /// the latch used to run before the early return at the top of the document,
6408 /// so an Up that did nothing still armed a column, and the next Down aimed
6409 /// at one the caret had never been in.
6410 fn step_vertical(&mut self, target: usize, goal: usize, extend: bool) {
6411 let before = self.caret;
6412 if target == before {
6413 return;
6414 }
6415 self.goal_col = Some(goal);
6416 self.move_to(target, extend);
6417 self.debug_assert_on_a_stop(before);
6418 }
6419
6420 /// The source line below `row`, or `None` when `row` is the last one. Lines
6421 /// are counted by newline, so a trailing one leaves a real, empty last line
6422 /// for the caret to sit on — the document ends below it, not on it.
6423 fn source_row_below(&self, row: usize) -> Option<usize> {
6424 let last = self.source.bytes().filter(|&b| b == b'\n').count();
6425 (row < last).then_some(row + 1)
6426 }
6427
6428 /// Where a vertical motion aiming at the `goal` column lands on visual row
6429 /// `r`: the column clamped to the row, mapped to its offset, then held
6430 /// inside the row's own [bounds](Self::row_bounds) — a wrapped row's last
6431 /// column belongs to the row below, and a gutter's column 0 points at the
6432 /// block rather than at this row.
6433 fn row_target(&self, r: usize, goal: usize) -> usize {
6434 let (start, end) = self.row_bounds(r);
6435 self.vmap
6436 .offset_of_pos(r, goal.min(self.vmap.row_width(r)))
6437 .clamp(start, end)
6438 }
6439
6440 /// The first and last offsets the caret can reach in the active view.
6441 ///
6442 /// Not the same span in both: the source view shows every byte, so it can
6443 /// reach every byte. WYSIWYG reaches only what it draws — hidden frontmatter
6444 /// sits below the first stop, and a document's trailing newline is drawn
6445 /// nowhere and so sits past the last.
6446 fn reachable_start(&self) -> usize {
6447 match self.view {
6448 View::Source => 0,
6449 View::Wysiwyg => self.vmap.stop_at_or_after(0).unwrap_or(self.caret),
6450 }
6451 }
6452
6453 fn reachable_end(&self) -> usize {
6454 match self.view {
6455 View::Source => self.source.len(),
6456 View::Wysiwyg => self
6457 .vmap
6458 .stop_at_or_before(self.source.len())
6459 .unwrap_or(self.caret),
6460 }
6461 }
6462
6463 /// The `[start, end]` offsets visual row `r` *draws* — everything on it,
6464 /// including the space a soft wrap ate off its end, which is drawn on this
6465 /// row however much the offset past it belongs to the next one.
6466 fn row_span(&self, r: usize) -> (usize, usize) {
6467 let start = self
6468 .vmap
6469 .row_start(r)
6470 .unwrap_or_else(|| self.vmap.offset_of_pos(r, 0));
6471 let end = self.vmap.offset_of_pos(r, self.vmap.row_width(r));
6472 (start.min(end), end)
6473 }
6474
6475 /// [`row_span`](Self::row_span) narrowed to where the caret can stand: a
6476 /// soft wrap's shared offset opens the row below (see `pos_of_offset`), so
6477 /// this row's last position is the one before it — the offset before the
6478 /// space the wrap ate, where the caret draws just past the row's last word
6479 /// and types there too.
6480 ///
6481 /// Aiming at the shared offset instead is what stalled End: it is the row's
6482 /// last *column*, so End pressed on the row reached it and then read back as
6483 /// the row below's start, where a second press ran on to that row's end and
6484 /// the next to the one after — End walking down the paragraph a row a press.
6485 fn row_bounds(&self, r: usize) -> (usize, usize) {
6486 let (start, end) = self.row_span(r);
6487 let wraps = self
6488 .vmap
6489 .navigable_below(r)
6490 .and_then(|b| self.vmap.row_start(b))
6491 .is_some_and(|off| off == end);
6492 match wraps {
6493 true => (start, self.vmap.stop_before(end).unwrap_or(end).max(start)),
6494 false => (start, end),
6495 }
6496 }
6497
6498 /// The `[start, end]` of the line Home and End aim at: the visual row in
6499 /// WYSIWYG, the logical line in the source view. Both ends are caret stops.
6500 ///
6501 /// A soft-wrapped row is a line here, because it is one to the eye and the
6502 /// eye is what these keys are aimed by — a reader pressing End means the end
6503 /// of the line they can see. (`select_block_at` wants the opposite and reads
6504 /// the AST for it: a triple-click grabs the whole paragraph, however many
6505 /// rows it folds into.)
6506 fn line_bounds(&self) -> (usize, usize) {
6507 let (row, _) = self.caret_pos();
6508 match self.view {
6509 View::Source => {
6510 let start = line_start(&self.source, row);
6511 (start, line_end_from(&self.source, start))
6512 }
6513 View::Wysiwyg => self.row_bounds(row),
6514 }
6515 }
6516
6517 /// The same line as [`line_bounds`](Self::line_bounds), as far as it is
6518 /// *drawn* — what a kill takes.
6519 ///
6520 /// The two part only at a soft wrap, over the space the wrap ate: the caret
6521 /// can't stand after it (that offset opens the row below, and End stopping
6522 /// there would walk), but it is on this row, and a kill that spared it would
6523 /// leave a double space behind where the row's text had been. Deleting it
6524 /// joins nothing — a wrap is drawn, not written.
6525 fn line_span(&self) -> (usize, usize) {
6526 let (row, _) = self.caret_pos();
6527 match self.view {
6528 View::Source => self.line_bounds(),
6529 View::Wysiwyg => self.row_span(row),
6530 }
6531 }
6532
6533 /// The first offset in `[start, end]` holding something other than
6534 /// whitespace, or `end` when the line holds nothing else — where Home aims.
6535 ///
6536 /// Walks the space the view is in, as word motion does: WYSIWYG steps stops,
6537 /// so a hidden delimiter is never taken for the line's first character (nor
6538 /// landed on), and the source view steps the source it is showing.
6539 fn first_non_space(&self, start: usize, end: usize) -> usize {
6540 let mut off = start;
6541 while off < end {
6542 if self.class_at(off) != Class::Space {
6543 return off;
6544 }
6545 off = match self.view {
6546 View::Source => next_boundary(&self.source, off),
6547 View::Wysiwyg => match self.vmap.stop_after(off) {
6548 Some(next) => next,
6549 None => return end,
6550 },
6551 };
6552 }
6553 end
6554 }
6555
6556 /// Home: to the first character on the line, or to column 0 when the caret
6557 /// is already on it — the two-press toggle every editor spells this way.
6558 /// The indentation is somewhere the caret has to be able to reach and almost
6559 /// never where a reader is headed, so it costs the second press.
6560 pub fn move_home(&mut self, extend: bool) {
6561 self.goal_col = None;
6562 let (start, end) = self.line_bounds();
6563 let text = self.first_non_space(start, end);
6564 let target = if self.caret == text { start } else { text };
6565 let before = self.caret;
6566 self.move_to(target, extend);
6567 self.debug_assert_on_a_stop(before);
6568 }
6569
6570 /// End: to the end of the line.
6571 pub fn move_end(&mut self, extend: bool) {
6572 self.goal_col = None;
6573 let (_, end) = self.line_bounds();
6574 let before = self.caret;
6575 self.move_to(end, extend);
6576 self.debug_assert_on_a_stop(before);
6577 }
6578
6579 /// Hop to the next (Tab) or previous (Shift+Tab) table cell, landing with the
6580 /// cell's whole content selected (see [`Self::select_cell`]). Returns `false`
6581 /// when the caret isn't in a table, or is already in the last/first cell — the
6582 /// frontend then does whatever Tab normally does (indent), so Tab keeps its
6583 /// meaning everywhere else.
6584 pub fn cell_hop(&mut self, forward: bool) -> bool {
6585 let Some((grid, r, c)) = self.table_grid_at(self.caret) else {
6586 return false;
6587 };
6588 // Flatten to document (row-major) order and step one cell either way.
6589 let i: usize = grid[..r].iter().map(Vec::len).sum::<usize>() + c;
6590 let flat: Vec<(usize, usize)> = grid.into_iter().flatten().collect();
6591 let next = if forward {
6592 i.checked_add(1)
6593 } else {
6594 i.checked_sub(1)
6595 };
6596 let Some(&(start, end)) = next.and_then(|j| flat.get(j)) else {
6597 return false; // at the table's edge; leave Tab to the frontend
6598 };
6599 self.select_cell(start, end);
6600 true
6601 }
6602
6603 /// Move the caret to the cell directly above (`down == false`) or below in
6604 /// the same column, landing with the cell's whole content selected (see
6605 /// [`Self::select_cell`]). Returns `false` at the grid's top/bottom edge (or
6606 /// when the caret isn't in a table), so the frontend can fall through — the
6607 /// vertical counterpart of [`Self::cell_hop`].
6608 ///
6609 /// A ragged row that is short a column clamps to its last cell, so Down never
6610 /// falls out of the table over a gap the row above happened to have.
6611 pub fn cell_move_vertical(&mut self, down: bool) -> bool {
6612 let Some((grid, r, c)) = self.table_grid_at(self.caret) else {
6613 return false;
6614 };
6615 let target = match down {
6616 true => r + 1,
6617 false if r == 0 => return false,
6618 false => r - 1,
6619 };
6620 let Some(row) = grid.get(target) else {
6621 return false;
6622 };
6623 let Some(&(start, end)) = row.get(c).or_else(|| row.last()) else {
6624 return false;
6625 };
6626 self.select_cell(start, end);
6627 true
6628 }
6629
6630 /// The table containing `off` as a row-major grid of `(start, end)` cell
6631 /// caret homes, plus the `(row, col)` the caret sits in — `None` when `off`
6632 /// isn't in a table. Read straight off the visual map's laid-out grid, so
6633 /// every cell (an empty one included, whose derived home twig gives no
6634 /// `content_span` for) is present and in the order Tab walks them.
6635 // Grid, row, column — three returns that only ever travel together, and a
6636 // named type for the pair of them would be read at one call site.
6637 #[allow(clippy::type_complexity)]
6638 fn table_grid_at(&self, off: usize) -> Option<(Vec<Vec<(usize, usize)>>, usize, usize)> {
6639 for t in &self.vmap.tables {
6640 let mut pos = None;
6641 let grid: Vec<Vec<(usize, usize)>> = t
6642 .grid
6643 .iter()
6644 .enumerate()
6645 .map(|(r, row)| {
6646 row.cells
6647 .iter()
6648 .enumerate()
6649 .map(|(c, cell)| {
6650 if pos.is_none() && off >= cell.start && off <= cell.end {
6651 pos = Some((r, c));
6652 }
6653 (cell.start, cell.end)
6654 })
6655 .collect()
6656 })
6657 .collect();
6658 if let Some((r, c)) = pos {
6659 return Some((grid, r, c));
6660 }
6661 }
6662 None
6663 }
6664
6665 // ── table key policy ──────────────────────────────────────────────────────
6666 // The three keys a table gives its own meaning — Tab, Return, Shift+Return —
6667 // as one policy every frontend shares, rather than each re-deriving it. Each
6668 // reports whether it acted *as a table key*; a `false` hands the key back to
6669 // the frontend's ordinary handling (indent, newline) so it keeps its meaning
6670 // everywhere else.
6671
6672 /// Tab / Shift+Tab inside a table. Tab steps to the next cell, appending a
6673 /// fresh row and entering it when it runs off the last one; Shift+Tab steps
6674 /// back and simply stays put at the very first cell. `false` when the caret
6675 /// isn't in a table.
6676 pub fn cell_tab(&mut self, forward: bool) -> bool {
6677 if !self.caret_in_table() {
6678 return false;
6679 }
6680 if self.cell_hop(forward) {
6681 return true;
6682 }
6683 // Off the last cell: grow the table by a row and step into its first
6684 // cell. (Shift+Tab at the first cell has nowhere to go and just holds.)
6685 if forward {
6686 self.append_row_and_enter(0);
6687 }
6688 true
6689 }
6690
6691 /// Return inside a table: drop to the cell below in the same column,
6692 /// appending a new row when the caret is already in the last one. `false`
6693 /// when the caret isn't in a table, so the frontend inserts a newline.
6694 pub fn cell_return(&mut self) -> bool {
6695 if !self.caret_in_table() {
6696 return false;
6697 }
6698 if self.cell_move_vertical(true) {
6699 return true;
6700 }
6701 // Already on the last row: grow one below and drop into the same column.
6702 let col = self.table_grid_at(self.caret).map_or(0, |(_, _, c)| c);
6703 self.append_row_and_enter(col);
6704 true
6705 }
6706
6707 /// Append a row below the caret's (last) row and land in `col` of it. The
6708 /// caret is in the last row, so twig's "insert below" makes the fresh row the
6709 /// table's new last — but twig re-spells the whole table, moving every byte,
6710 /// so the destination is read back from the rebuilt grid by the table's
6711 /// position (stable across a row insert), not from the pre-edit caret.
6712 fn append_row_and_enter(&mut self, col: usize) {
6713 let table = self.caret_table_index();
6714 self.table_insert_row(true);
6715 self.rebuild_map();
6716 let Some((start, end)) = table
6717 .and_then(|ti| self.vmap.tables.get(ti))
6718 .and_then(|t| t.grid.last())
6719 .and_then(|row| row.cells.get(col.min(row.cells.len().saturating_sub(1))))
6720 .map(|cell| (cell.start, cell.end))
6721 else {
6722 return;
6723 };
6724 self.select_cell(start, end);
6725 }
6726
6727 /// The index, among the document's tables, of the one the caret sits in —
6728 /// `None` when it's in none. Used to re-find a table after an edit re-spells
6729 /// it (a row insert leaves the table order unchanged).
6730 fn caret_table_index(&self) -> Option<usize> {
6731 let off = self.caret;
6732 self.vmap.tables.iter().position(|t| {
6733 t.grid
6734 .iter()
6735 .any(|row| row.cells.iter().any(|c| off >= c.start && off <= c.end))
6736 })
6737 }
6738
6739 /// Shift+Return inside a table: insert a hard line break *within* the current
6740 /// cell, via twig's `insert_line_break`. `false` when the caret isn't in a
6741 /// table, so the frontend inserts an ordinary line break.
6742 ///
6743 /// A table row is a single source line, so the newline-spelled hard break
6744 /// can't live in a cell. twig spells the in-cell break the format's way
6745 /// (`<br>` for Markdown) and reparses it as a *semantic* `hard_break`, so the
6746 /// break round-trips as structure the renderer reads back as a line — not the
6747 /// opaque raw HTML the old raw-splice left behind.
6748 ///
6749 /// Djot has no idiomatic in-cell break, so twig refuses it
6750 /// (`UnsupportedFormat`) rather than emit a `<br>` that any other djot reader
6751 /// would render as the literal text `<br>`. The gesture is still *consumed*
6752 /// there — returning `false` would let the frontend insert a real newline,
6753 /// which splits the one-line row — it just leaves the cell unchanged and says
6754 /// so on the status line. A rollback (`EditConflict`) is swallowed the same.
6755 ///
6756 /// Which formats refuse is [`Capabilities::cell_line_break`], and the two
6757 /// have to be read together: djot is not the only `false`, and naming it in
6758 /// the message was already a guess that HTML — which spells the break as its
6759 /// own `<br>` — would have made wrong.
6760 pub fn cell_line_break(&mut self) -> bool {
6761 if self.read_only || !self.caret_in_table() {
6762 return false;
6763 }
6764 self.record_caret();
6765 match self.editor.insert_line_break(self.caret) {
6766 Ok(change) => {
6767 self.last_edit_kind = None;
6768 self.refresh();
6769 self.caret = change.new.end;
6770 self.anchor = None;
6771 self.goal_col = None;
6772 self.clamp_caret();
6773 self.dirty = self.source != self.clean_source;
6774 self.status = None;
6775 self.record_caret();
6776 }
6777 Err(twig::Error::UnsupportedFormat) => {
6778 self.status = Some(format!(
6779 "in-cell line breaks aren't supported in {}",
6780 self.format_name()
6781 ));
6782 }
6783 Err(_) => {}
6784 }
6785 true
6786 }
6787
6788 /// Rebuild the visual map at the width the last build used. A structural edit
6789 /// bumps the revision and swaps the source in, but leaves the *map* stale;
6790 /// when a single gesture edits and then moves over the result (Tab appending
6791 /// a row, then stepping into it), the move needs the map to already show the
6792 /// edit rather than waiting for the frontend's next frame.
6793 fn rebuild_map(&mut self) {
6794 let wrap = self.vmap_key.as_ref().and_then(|(_, w, _)| *w);
6795 self.build_map(wrap);
6796 }
6797
6798 /// Move the caret to the very start of the document (⌘↑ on macOS,
6799 /// Ctrl+Home on Windows/Linux).
6800 pub fn move_doc_start(&mut self, extend: bool) {
6801 self.goal_col = None;
6802 self.move_to(0, extend);
6803 }
6804
6805 /// Move the caret to the very end of the document (⌘↓ on macOS,
6806 /// Ctrl+End on Windows/Linux).
6807 pub fn move_doc_end(&mut self, extend: bool) {
6808 self.goal_col = None;
6809 let end = self.source.len();
6810 self.move_to(end, extend);
6811 }
6812
6813 /// Point the caret at the body cell `(row, col)` the mouse landed on —
6814 /// `col` being a cell of the terminal grid, which is what a display column
6815 /// is. A click on the far cell of a wide character lands at that
6816 /// character's start; the mapping's own doc-comments carry the rule.
6817 pub fn click(&mut self, row: usize, col: usize, extend: bool) {
6818 self.goal_col = None;
6819 let target = match self.view {
6820 View::Source => row_col_to_offset(&self.source, row, col),
6821 View::Wysiwyg => self.vmap.offset_of_pos(row, col),
6822 };
6823 let before = self.caret;
6824 self.move_to(target, extend);
6825 self.debug_assert_on_a_stop(before);
6826 }
6827
6828 /// Settle `scroll` for a frame about to be drawn: follow the caret onto the
6829 /// screen if it has moved since the last frame, and never scroll past the
6830 /// last of `rows`.
6831 ///
6832 /// Only if it has *moved* — that's the whole point. Revealing the caret on
6833 /// every frame ties the viewport to it, and a scroll wheel that fights the
6834 /// caret for the viewport loses: the view snaps back the instant it tries to
6835 /// pass the caret's row, so the document can't be scrolled beyond what's
6836 /// already on screen. A caret move is the frontend's cue to follow; a scroll
6837 /// with the caret sitting still is the reader's cue to leave it alone.
6838 pub fn follow_caret(&mut self, caret_row: usize, height: usize, rows: usize) {
6839 if self.drawn_caret != Some(self.caret) {
6840 if caret_row < self.scroll {
6841 self.scroll = caret_row;
6842 } else if height > 0 && caret_row >= self.scroll + height {
6843 self.scroll = caret_row + 1 - height;
6844 }
6845 self.drawn_caret = Some(self.caret);
6846 }
6847 self.scroll = self.scroll.min(rows.saturating_sub(1));
6848 }
6849
6850 /// The caret's screen position `(row, col)` in the active view's grid, with
6851 /// `col` a display column: the cell to draw the caret in, which on a line of
6852 /// `你好` or emoji is not the count of characters before it.
6853 pub fn caret_pos(&self) -> (usize, usize) {
6854 match self.view {
6855 View::Source => offset_to_row_col(&self.source, self.caret),
6856 View::Wysiwyg => self.vmap.pos_of_offset(self.caret),
6857 }
6858 }
6859
6860 fn clamp_caret(&mut self) {
6861 if self.caret > self.source.len() {
6862 self.caret = self.source.len();
6863 }
6864 // In WYSIWYG the caret can't sit inside hidden frontmatter; lift it (and
6865 // any selection anchor) to the first rendered offset.
6866 let floor = self.caret_floor();
6867 if self.caret < floor {
6868 self.caret = floor;
6869 }
6870 if let Some(a) = self.anchor
6871 && a < floor
6872 {
6873 self.anchor = Some(floor);
6874 }
6875 while self.caret > 0 && !self.source.is_char_boundary(self.caret) {
6876 self.caret -= 1;
6877 }
6878 }
6879}
6880
6881// ── byte-offset ⇄ (row, col) helpers ─────────────────────────────────────────
6882
6883// Left/right motion and backspace/delete step by *grapheme cluster*, not
6884// codepoint, so an emoji (a ZWJ sequence) or a base letter plus its combining
6885// marks moves and deletes as the single character a user sees. Grapheme
6886// boundaries are a superset of char boundaries, so the caret stays valid for twig.
6887
6888/// How an insert of `text` groups for undo: a single typed character folds into
6889/// the run of typing around it, while a newline or a multi-character insert is a
6890/// step of its own.
6891fn typed_edit_kind(text: &str) -> EditKind {
6892 if text.chars().take(2).count() == 1 && text != "\n" {
6893 EditKind::Insert
6894 } else {
6895 EditKind::Other
6896 }
6897}
6898
6899fn prev_boundary(s: &str, i: usize) -> usize {
6900 let mut cursor = GraphemeCursor::new(i, s.len(), true);
6901 cursor.prev_boundary(s, 0).ok().flatten().unwrap_or(0)
6902}
6903
6904fn next_boundary(s: &str, i: usize) -> usize {
6905 let mut cursor = GraphemeCursor::new(i, s.len(), true);
6906 cursor.next_boundary(s, 0).ok().flatten().unwrap_or(s.len())
6907}
6908
6909// ── word boundaries ──────────────────────────────────────────────────────────
6910// The shared primitive behind word-wise motion, word deletion, and
6911// double-click-to-select-a-word. A "word" is a maximal run of one character
6912// class; whitespace and punctuation are their own classes, so motion skips
6913// cleanly between them the way native text fields do.
6914
6915#[derive(PartialEq, Eq, Clone, Copy)]
6916enum Class {
6917 Word,
6918 Space,
6919 Other,
6920}
6921
6922/// The source range of an inline node's own visible text — the part of it a
6923/// WYSIWYG caret can reach, as against the delimiters that only spell it.
6924/// `None` for a node with no interior to empty (a `str`, a break).
6925///
6926/// twig reports no `content_span` for `verbatim`/`inline_math`, whose text sits
6927/// one delimiter in from the span — the same place the renderer maps it to. A
6928/// longer fence (`` ``a`` ``) breaks that assumption, so the guess is checked
6929/// against the source rather than trusted: a range guessed wrong here is text
6930/// deleted wrong.
6931fn inline_content_span(n: &FlatNode, source: &str) -> Option<std::ops::Range<usize>> {
6932 if let Some(span) = n.content_span.clone() {
6933 return Some(span);
6934 }
6935 match n.kind.as_str() {
6936 "verbatim" | "inline_math" => {
6937 let text = n.text.as_ref()?;
6938 let start = n.span.start + 1;
6939 let range = start..start + text.len();
6940 (source.get(range.clone()) == Some(text.as_str())).then_some(range)
6941 }
6942 _ => None,
6943 }
6944}
6945
6946/// The `id` a node declares, or `None` for one that declares none — the
6947/// attribute djot writes for a `{#v1}` and mints for a heading.
6948///
6949/// A bare attribute (`{#v1 hidden}`'s `hidden`) has no value, and a bare `id`
6950/// names nothing, so it reads as absent rather than as the empty string.
6951fn declared_id(n: &FlatNode) -> Option<&str> {
6952 n.attrs.iter().find(|(k, _)| k == "id")?.1.as_deref()
6953}
6954
6955/// A heading's words reduced to the form a link fragment spells them in:
6956/// lowercase, runs of anything else collapsed to a single `-`, with none left
6957/// dangling at either end. `## Some Heading Here` → `some-heading-here`.
6958///
6959/// The rule every Markdown renderer follows, and applied to djot's own auto-ids
6960/// too so that `#some-heading-here` and `#Some-Heading-Here` are one question.
6961/// Unicode-aware (`is_alphanumeric`, not an ASCII test), because a heading in
6962/// any other language is still a heading someone will link to. Underscores
6963/// survive for the same reason they do on the web: they are word characters
6964/// wherever identifiers are written.
6965fn slug(text: &str) -> String {
6966 let mut out = String::new();
6967 let mut pending = false;
6968 for c in text.chars() {
6969 if c.is_alphanumeric() || c == '_' {
6970 if pending && !out.is_empty() {
6971 out.push('-');
6972 }
6973 pending = false;
6974 out.extend(c.to_lowercase());
6975 } else {
6976 pending = true;
6977 }
6978 }
6979 out
6980}
6981
6982fn is_block_container(kind: &Kind) -> bool {
6983 matches!(
6984 kind,
6985 Kind::Doc
6986 | Kind::Section
6987 | Kind::BlockQuote
6988 | Kind::BulletList
6989 | Kind::OrderedList
6990 | Kind::TaskList
6991 | Kind::ListItem
6992 | Kind::TaskListItem
6993 // Every `container` — a directive in any of its three forms, or a
6994 // promoted HTML element. A *text* directive is really inline, so
6995 // claiming it here is a small overreach, and the deliberate one this
6996 // function's kind-only peer `is_inline_kind` documents: the pair is
6997 // consulted together, and answering "block container" for something
6998 // inline is what keeps an ancestor walk from stopping short of the
6999 // paragraph that actually holds it.
7000 | Kind::Container
7001 )
7002}
7003
7004/// The `[start, end)` byte range of the source line containing `off` (newline
7005/// excluded) — the fallback when `off` sits outside any AST block (e.g. a blank
7006/// line between paragraphs).
7007fn source_line_range(s: &str, off: usize) -> std::ops::Range<usize> {
7008 let off = off.min(s.len());
7009 let start = s[..off].rfind('\n').map(|p| p + 1).unwrap_or(0);
7010 let end = s[off..].find('\n').map(|p| off + p).unwrap_or(s.len());
7011 start..end
7012}
7013
7014/// How many leading bytes an outdent takes off `line`: a whole indent level
7015/// where the line has one, and whatever it has where it has less.
7016///
7017/// A leading tab counts as a level on its own. It's indentation some other
7018/// editor wrote, and one tab is one level everywhere it came from — measuring it
7019/// in spaces it doesn't contain would leave it untouchable.
7020fn outdent_width(line: &str, unit: usize) -> usize {
7021 if line.starts_with('\t') {
7022 return 1;
7023 }
7024 line.bytes().take(unit).take_while(|b| *b == b' ').count()
7025}
7026
7027/// A list marker found at the head of a line, together with everything before it
7028/// that a sibling line has to repeat.
7029///
7030/// The three offsets differ only inside a block quote, where `> - b` opens with
7031/// a `> ` quote marker the line's own text doesn't own. Outside one they collapse:
7032/// `line_start == marker_start`, and `text` is the plain `" - "`.
7033#[derive(Clone, Debug)]
7034struct ListMarker {
7035 /// The line's first byte.
7036 line_start: usize,
7037 /// Where the marker proper begins, past any quote prefix. The offset to hand
7038 /// the AST: a quoted item's span opens at its bullet, not at the `>`.
7039 marker_start: usize,
7040 /// `line_start` through the marker's trailing space — quote prefix, indent
7041 /// and bullet together, which is what the next item's line opens with.
7042 text: String,
7043}
7044
7045impl ListMarker {
7046 /// Where the item's content starts — one past the marker's trailing space.
7047 fn content_start(&self) -> usize {
7048 self.line_start + self.text.len()
7049 }
7050}
7051
7052fn classify(c: char) -> Class {
7053 if c == '_' || c.is_alphanumeric() {
7054 Class::Word
7055 } else if c.is_whitespace() {
7056 Class::Space
7057 } else {
7058 Class::Other
7059 }
7060}
7061
7062/// The offset at the end of the next word to the right of `i` (⌥→ / Ctrl+→):
7063/// skip any leading separators, then consume the following word run.
7064fn next_word(s: &str, i: usize) -> usize {
7065 let mut off = i;
7066 let mut in_word = false;
7067 for c in s[i..].chars() {
7068 if classify(c) == Class::Word {
7069 in_word = true;
7070 } else if in_word {
7071 break;
7072 }
7073 off += c.len_utf8();
7074 }
7075 off
7076}
7077
7078/// The offset at the start of the word to the left of `i` (⌥← / Ctrl+←):
7079/// skip separators walking left, then consume the preceding word run.
7080fn prev_word(s: &str, i: usize) -> usize {
7081 let mut off = i;
7082 let mut in_word = false;
7083 for c in s[..i].chars().rev() {
7084 if classify(c) == Class::Word {
7085 in_word = true;
7086 } else if in_word {
7087 break;
7088 }
7089 off -= c.len_utf8();
7090 }
7091 off
7092}
7093
7094/// The `[start, end)` run of same-class characters surrounding `off` — the
7095/// word (or whitespace/punctuation run) a double-click selects. At end-of-text
7096/// the run ending there is used.
7097fn word_range_at(s: &str, off: usize) -> (usize, usize) {
7098 if s.is_empty() {
7099 return (0, 0);
7100 }
7101 let off = off.min(s.len());
7102 let reference = if off < s.len() {
7103 s[off..].chars().next()
7104 } else {
7105 s[..off].chars().next_back()
7106 };
7107 let Some(rc) = reference else {
7108 return (off, off);
7109 };
7110 let class = classify(rc);
7111
7112 let mut start = off;
7113 for c in s[..start].chars().rev() {
7114 if classify(c) == class {
7115 start -= c.len_utf8();
7116 } else {
7117 break;
7118 }
7119 }
7120 let mut end = off;
7121 for c in s[end..].chars() {
7122 if classify(c) == class {
7123 end += c.len_utf8();
7124 } else {
7125 break;
7126 }
7127 }
7128 (start, end)
7129}
7130
7131/// `(row, col)` of byte offset `off`, `col` counted in *display columns* from
7132/// the line's start — terminal cells, not characters, so the column names the
7133/// cell the caret is drawn in even on a line of `你好` or emoji.
7134fn offset_to_row_col(s: &str, off: usize) -> (usize, usize) {
7135 let off = off.min(s.len());
7136 let mut row = 0;
7137 let mut line_start = 0;
7138 for (i, &b) in s.as_bytes().iter().enumerate() {
7139 if i >= off {
7140 break;
7141 }
7142 if b == b'\n' {
7143 row += 1;
7144 line_start = i + 1;
7145 }
7146 }
7147 (row, wysiwyg::text_width(&s[line_start..off]))
7148}
7149
7150/// The byte offset at display column `col` of `row` (clamped to that line's
7151/// end) — the inverse of [`offset_to_row_col`], which it has to agree with.
7152///
7153/// A column landing *inside* a character — the second cell of `你`, or any cell
7154/// but the first of an emoji — resolves to that character's start, which is the
7155/// column the caret would have been drawn at to begin with. So both cells of a
7156/// wide character mean the character, and every offset survives the round trip
7157/// out to a column and back. The walk steps by grapheme cluster for the same
7158/// reason the caret does: a cluster is the character, and the cells belong to it
7159/// rather than to the codepoints spelling it.
7160fn row_col_to_offset(s: &str, row: usize, col: usize) -> usize {
7161 let start = line_start(s, row);
7162 let end = line_end_from(s, start);
7163 let mut off = start;
7164 let mut at = 0; // the display column `off` sits at
7165 while off < end {
7166 let next = next_boundary(s, off).min(end);
7167 let cells = wysiwyg::text_width(&s[off..next]);
7168 if at + cells > col {
7169 break; // `col` is one of this cluster's own cells
7170 }
7171 at += cells;
7172 off = next;
7173 }
7174 off
7175}
7176
7177fn line_start(s: &str, row: usize) -> usize {
7178 if row == 0 {
7179 return 0;
7180 }
7181 let mut r = 0;
7182 for (i, &b) in s.as_bytes().iter().enumerate() {
7183 if b == b'\n' {
7184 r += 1;
7185 if r == row {
7186 return i + 1;
7187 }
7188 }
7189 }
7190 s.len()
7191}
7192
7193fn line_end_from(s: &str, start: usize) -> usize {
7194 s[start..].find('\n').map(|p| start + p).unwrap_or(s.len())
7195}
7196
7197/// twig's node-kind name for an inline mark, back to the [`InlineKind`] a
7198/// frontend names when it calls [`Doc::toggle`] — the inverse of the mapping
7199/// twig applies writing the mark out, so the toolbar can light the same button
7200/// that made the node.
7201///
7202/// `None` for every other kind, including the inline nodes that aren't marks at
7203/// all (`str`, `link`, `image`, the math and break kinds): they're things a
7204/// caret stands in, not formatting a button toggles.
7205fn inline_kind(kind: &Kind) -> Option<InlineKind> {
7206 Some(match kind {
7207 Kind::Strong => InlineKind::Strong,
7208 Kind::Emph => InlineKind::Emph,
7209 Kind::Verbatim => InlineKind::Verbatim,
7210 Kind::Mark => InlineKind::Mark,
7211 Kind::Superscript => InlineKind::Superscript,
7212 Kind::Subscript => InlineKind::Subscript,
7213 Kind::Insert => InlineKind::Insert,
7214 Kind::Delete => InlineKind::Delete,
7215 _ => return None,
7216 })
7217}
7218
7219/// leaf's [`MarkColor`] as twig's — the palette twig writes as the emoji after
7220/// a highlight's opening `==`.
7221///
7222/// Two enums for one closed vocabulary, and the duplication is the boundary
7223/// working: core's is what a *frontend* names (`style::MarkColor`, beside the
7224/// [`Role`](crate::Role) that carries it into the glyph map) and twig's is what
7225/// the editor writes. Spelled as a match rather than routed through the two
7226/// crates' name strings so that a colour added on either side is a compile
7227/// error here, where the pairing is decided, rather than a runtime `None` that
7228/// would read as "clear the colour".
7229fn twig_mark_color(color: MarkColor) -> twig::MarkColor {
7230 match color {
7231 MarkColor::Red => twig::MarkColor::Red,
7232 MarkColor::Orange => twig::MarkColor::Orange,
7233 MarkColor::Yellow => twig::MarkColor::Yellow,
7234 MarkColor::Green => twig::MarkColor::Green,
7235 MarkColor::Blue => twig::MarkColor::Blue,
7236 MarkColor::Purple => twig::MarkColor::Purple,
7237 MarkColor::Brown => twig::MarkColor::Brown,
7238 }
7239}
7240
7241/// Where an offset lands after a splice it didn't make — twig's own rule, from
7242/// [`Change`]: shift anything at or past the replaced range's end by the length
7243/// the replacement gained or lost, and leave anything before it alone.
7244///
7245/// An offset *inside* the replaced range has no text of its own to ride any
7246/// more, and lands at the end of what replaced it: for
7247/// [`Doc::set_mark_color`] that is a caret standing on the colour prefix when
7248/// the prefix is cleared, which then sits where the highlighted text begins.
7249/// One node's attribute list, twig's own `(key, value)` pairs owned — what
7250/// every presentation gesture reads, edits one key of, and passes back whole.
7251type Attrs = Vec<(String, Option<String>)>;
7252
7253/// The name of the leaf directive a page break is — [`Doc::insert_page_break`]
7254/// writes it and the walker draws it, and a frontend that paginates matches a
7255/// [`DirectiveMark`](crate::wysiwyg::DirectiveMark) against it. One spelling,
7256/// stated once.
7257pub const PAGE_BREAK: &str = "page-break";
7258
7259/// `attrs` with `key` set to `value`, or removed when `value` is `None`, and
7260/// every other attribute kept in its place — the read-edit-write half of twig's
7261/// replace-not-merge contract for a `data-` key.
7262///
7263/// **A key that is already there is rewritten where it stands**, and only a key
7264/// the node did not have goes on the end. That is what makes the proposal's
7265/// worked example true: `class="lead center" id="intro"
7266/// data-line-height="1.5"`, right-aligned, is `class="lead right" id="intro"
7267/// data-line-height="1.5"` — the same document with one token changed, and a
7268/// one-line diff. Removing the key and pushing it back would reorder the
7269/// author's attributes on every press, so a document that passed through the
7270/// editor came out shuffled even where nothing about it had changed.
7271///
7272/// A duplicate key — which no format leaf opens can spell, but twig reports
7273/// verbatim — collapses onto the first of its copies, since twig is handed one
7274/// value for one key either way.
7275fn with_attr(attrs: &[(String, Option<String>)], key: &str, value: Option<&str>) -> Attrs {
7276 let mut out: Attrs = Vec::with_capacity(attrs.len() + 1);
7277 let mut written = false;
7278 for (k, v) in attrs {
7279 if k != key {
7280 out.push((k.clone(), v.clone()));
7281 continue;
7282 }
7283 if let Some(new) = value.filter(|_| !written) {
7284 out.push((k.clone(), Some(new.to_string())));
7285 written = true;
7286 }
7287 }
7288 if let Some(new) = value.filter(|_| !written) {
7289 out.push((key.to_string(), Some(new.to_string())));
7290 }
7291 out
7292}
7293
7294/// [`with_attr`] for a `class` token: every token `mine` claims is removed, and
7295/// `token` added, with the rest of the list kept in order.
7296///
7297/// `class` is a space-separated token list, and leaf owns three of the tokens in
7298/// it. A paragraph that arrives as `class="lead center"` and is right-aligned
7299/// goes out as `class="lead right"`; one whose last owned token goes and which
7300/// carried nothing else loses the key, so a block that has lost its whole
7301/// vocabulary is spelled bare again. `class` itself keeps its place among the
7302/// attributes, because [`with_attr`] does the writing.
7303fn with_class_token(
7304 attrs: &[(String, Option<String>)],
7305 mine: impl Fn(&str) -> bool,
7306 token: Option<&str>,
7307) -> Attrs {
7308 let kept: Vec<&str> = attrs
7309 .iter()
7310 .find(|(k, _)| k == "class")
7311 .and_then(|(_, v)| v.as_deref())
7312 .unwrap_or_default()
7313 .split_whitespace()
7314 .filter(|t| !mine(t))
7315 .collect();
7316 let class = kept.into_iter().chain(token).collect::<Vec<_>>().join(" ");
7317 with_attr(
7318 attrs,
7319 "class",
7320 (!class.is_empty()).then_some(class.as_str()),
7321 )
7322}
7323
7324/// An owned attribute list as the borrowed pairs twig's two attribute ops take.
7325///
7326/// A **bare** attribute — one twig reports with no value, such as HTML's `<p
7327/// hidden>` — is passed back as an empty one. Twig refuses a `None` outright
7328/// (djot has no bare attribute, so no format reads one back everywhere), and
7329/// `hidden=""` is the same document where `hidden` is; dropping it instead
7330/// would lose what the author wrote, which is the one thing these gestures
7331/// promise not to do.
7332fn attr_pairs(attrs: &[(String, Option<String>)]) -> Vec<(&str, Option<&str>)> {
7333 attrs
7334 .iter()
7335 .map(|(k, v)| (k.as_str(), Some(v.as_deref().unwrap_or_default())))
7336 .collect()
7337}
7338
7339fn reanchor(off: usize, change: &Change) -> usize {
7340 if off < change.old.start {
7341 return off;
7342 }
7343 if off < change.old.end {
7344 return change.new.end;
7345 }
7346 (off + change.new.end).saturating_sub(change.old.end)
7347}
7348
7349/// [`reanchor`] for an edit that respells the markup *around* a block and
7350/// leaves the block's own bytes alone — which is every attribute gesture.
7351///
7352/// `block` is that block's content span before and after the splice, so an
7353/// offset standing in the text keeps its distance from the text's start and how
7354/// many bytes twig wrote above it never enters the arithmetic. That is the whole
7355/// rule, and it is why nothing here knows how long a `<div …>` is: a second key
7356/// on the same div lengthens the attribute line, clearing the last one takes the
7357/// div away entirely, and both are the same sum. `None` where the splice named
7358/// no block at either end, which is every djot case — the `{…}` line is written
7359/// above the block, and the block itself only shifts past it.
7360///
7361/// Anywhere else it is `reanchor`'s own answer: untouched before the splice,
7362/// shifted by its delta after it, and at the splice's end for an offset that
7363/// stood in markup being rewritten — a caret inside djot's `{…}` line has no
7364/// text to keep.
7365fn reanchor_in_block(
7366 off: usize,
7367 change: &Change,
7368 block: Option<(&Range<usize>, &Range<usize>)>,
7369) -> usize {
7370 if let Some((was, now)) = block
7371 && was.start <= off
7372 && off <= was.end
7373 {
7374 return now.start + (off - was.start).min(now.end - now.start);
7375 }
7376 reanchor(off, change)
7377}
7378
7379/// A watermark for a file's contents (see `Doc::disk_hash`).
7380///
7381/// `DefaultHasher` is not stable across Rust releases, which doesn't matter: a
7382/// watermark is compared only against one taken by the same process moments
7383/// earlier, and never outlives it. 64 bits leaves a collision — an external edit
7384/// that hashes to exactly what leaf wrote — at odds no filesystem race gets near.
7385fn hash_bytes(bytes: &[u8]) -> u64 {
7386 use std::hash::{Hash, Hasher};
7387 let mut h = std::collections::hash_map::DefaultHasher::new();
7388 bytes.hash(&mut h);
7389 h.finish()
7390}
7391
7392#[cfg(feature = "fs")]
7393fn detect_format(path: &Path) -> Result<Format> {
7394 let ext = path
7395 .extension()
7396 .and_then(|e| e.to_str())
7397 .unwrap_or("")
7398 .to_ascii_lowercase();
7399 Ok(match ext.as_str() {
7400 "dj" | "djot" => Format::Djot,
7401 "md" | "markdown" => Format::Markdown,
7402 "xml" => Format::Xml,
7403 "html" | "htm" => Format::Html,
7404 other => return Err(anyhow!("unknown document extension: .{other}")),
7405 })
7406}
7407
7408#[cfg(test)]
7409mod tests {
7410 use super::*;
7411
7412 /// A document open in `view`. WYSIWYG motion reads the visual map, which the
7413 /// renderer stamps each frame, so the map is built here too — a WYSIWYG doc
7414 /// without one is a view no user is ever in.
7415 fn doc_in(view: View, name: &str, body: &str) -> Doc {
7416 // The fixture name doubles as the temp file's, so two tests picking the
7417 // same one raced under the parallel runner and read each other's body —
7418 // a green suite proving the wrong thing. The counter makes that
7419 // unreachable rather than asking every future caller to notice.
7420 static SEQ: std::sync::atomic::AtomicUsize = std::sync::atomic::AtomicUsize::new(0);
7421 let seq = SEQ.fetch_add(1, std::sync::atomic::Ordering::Relaxed);
7422 let mut p = std::env::temp_dir();
7423 p.push(format!("leaf_test_{name}_{seq}.md"));
7424 std::fs::write(&p, body).unwrap();
7425 let mut d = Doc::open(p).unwrap();
7426 d.view = view;
7427 if view == View::Wysiwyg {
7428 d.build_visual(80);
7429 }
7430 d
7431 }
7432
7433 // Source-view document for the source-behaviour tests. `Doc::open` now
7434 // defaults to WYSIWYG (leaf's default view), so pin the source view here;
7435 // `wysiwyg_doc` builds the rich-text variant on top of this.
7436 fn doc_with(name: &str, body: &str) -> Doc {
7437 doc_in(View::Source, name, body)
7438 }
7439
7440 /// Every visual row's drawn text — what the reader actually sees, which is
7441 /// the only thing the reveal preference is supposed to change.
7442 fn drawn_rows(d: &Doc) -> Vec<String> {
7443 d.vmap
7444 .rows
7445 .iter()
7446 .map(|r| r.glyphs.iter().map(|g| g.ch).collect())
7447 .collect()
7448 }
7449
7450 /// Put the caret at the first byte of `needle` and rebuild, so the row under
7451 /// it becomes the revealed line.
7452 fn caret_at(d: &mut Doc, needle: &str) {
7453 d.caret = d.source.find(needle).expect("needle in source");
7454 d.build_visual(80);
7455 }
7456
7457 #[test]
7458 fn blockquote_after_a_list_is_not_bulleted() {
7459 // twig nests a following top-level block quote under the `bullet_list`
7460 // (a direct child, not a `list_item`). The map must render it de-nested —
7461 // `│ quote`, never `• │ quote` — with a blank separator, like any block
7462 // that follows a list. Regression for the "combined list + blockquote" bug.
7463 let mut d = doc_in(View::Wysiwyg, "bq_after_list", "- item\n\n> quote\n");
7464 d.build_visual(80);
7465 let rows: Vec<String> = d
7466 .vmap
7467 .rows
7468 .iter()
7469 .map(|r| r.glyphs.iter().map(|g| g.ch).collect())
7470 .collect();
7471 assert!(
7472 rows.iter().any(|r| r == "│ quote"),
7473 "block quote should render on its own gutter, got rows: {rows:?}"
7474 );
7475 assert!(
7476 !rows.iter().any(|r| r.contains('•') && r.contains('│')),
7477 "no row should carry both a bullet and a quote gutter, got rows: {rows:?}"
7478 );
7479 }
7480
7481 // ── the map is built at most once per (revision, wrap) ───────────────────
7482 //
7483 // A frontend repaints for reasons that have nothing to do with the text — a
7484 // blinking caret, a scroll — and rebuilding the map is O(document). These
7485 // pin *that the cache fires*, which a passing suite can't tell you: a cache
7486 // that never hits is invisible to every other test in this file.
7487 //
7488 // The probe is to wreck the built map and ask for it again. A rebuild
7489 // repairs it; a cache hit hands the wreckage straight back. Nothing else
7490 // can distinguish the two from outside.
7491
7492 #[test]
7493 fn a_rebuild_with_nothing_changed_reuses_the_map() {
7494 let mut d = doc_in(View::Wysiwyg, "cache_hit", "# Title\n\nbody\n");
7495 d.build_visual(80);
7496 assert!(!d.vmap.rows.is_empty());
7497 d.vmap.rows.clear(); // wreck it
7498 d.build_visual(80);
7499 assert!(
7500 d.vmap.rows.is_empty(),
7501 "the map was rebuilt though nothing changed — the cache never fired"
7502 );
7503 }
7504
7505 #[test]
7506 fn an_edit_rebuilds_the_map() {
7507 let mut d = doc_in(View::Wysiwyg, "cache_edit", "# Title\n\nbody\n");
7508 d.build_visual(80);
7509 let before = d.revision();
7510 d.vmap.rows.clear();
7511 d.insert("x");
7512 d.build_visual(80);
7513 assert!(d.revision() > before, "an edit must move the revision");
7514 assert!(
7515 !d.vmap.rows.is_empty(),
7516 "an edited document must not paint from a stale map"
7517 );
7518 }
7519
7520 #[test]
7521 fn a_width_change_rebuilds_the_map() {
7522 // The map is a function of the wrap width too, so a resize is a miss
7523 // even though the text is untouched.
7524 let mut d = doc_in(
7525 View::Wysiwyg,
7526 "cache_width",
7527 "one two three four five six\n",
7528 );
7529 d.build_visual(80);
7530 d.vmap.rows.clear();
7531 d.build_visual(12);
7532 assert!(!d.vmap.rows.is_empty(), "a resize must rebuild the map");
7533 // And the unwrapped map is its own key, not the same as any width.
7534 d.vmap.rows.clear();
7535 d.build_visual_unwrapped();
7536 assert!(!d.vmap.rows.is_empty(), "unwrapped is a different map");
7537 }
7538
7539 #[test]
7540 fn a_motion_does_not_rebuild_the_map() {
7541 // The whole point: moving the caret changes nothing the map is built
7542 // from. If a motion bumped the revision, every arrow key would cost a
7543 // full rebuild and the cache would be worthless.
7544 let mut d = doc_in(View::Wysiwyg, "cache_motion", "# Title\n\nbody text\n");
7545 d.build_visual(80);
7546 let rev = d.revision();
7547 d.move_right(false);
7548 d.move_right(true);
7549 d.move_down(false);
7550 assert_eq!(d.revision(), rev, "a motion must not move the revision");
7551 d.vmap.rows.clear();
7552 d.build_visual(80);
7553 assert!(
7554 d.vmap.rows.is_empty(),
7555 "a motion should not rebuild the map"
7556 );
7557 }
7558
7559 #[test]
7560 fn saving_does_not_rebuild_the_map() {
7561 // Saving changes `dirty`, not the text.
7562 let mut d = doc_in(View::Wysiwyg, "cache_save", "# Title\n\nbody\n");
7563 d.insert("x");
7564 d.build_visual(80);
7565 let rev = d.revision();
7566 d.save();
7567 assert_eq!(d.revision(), rev, "a save must not move the revision");
7568 assert!(!d.dirty, "the save should have cleaned the document");
7569 }
7570
7571 #[test]
7572 fn a_reload_rebuilds_the_map() {
7573 // Reload replaces the text without going through `refresh`, so it has to
7574 // move the revision itself — else the editor paints the old file.
7575 let mut d = doc_in(View::Wysiwyg, "cache_reload", "# Title\n\nbody\n");
7576 d.build_visual(80);
7577 let rev = d.revision();
7578 std::fs::write(&d.path, "# Other\n\nwholly new\n").unwrap();
7579 d.reload();
7580 assert!(d.revision() > rev, "a reload must move the revision");
7581 d.build_visual(80);
7582 let text: String = d
7583 .vmap
7584 .rows
7585 .iter()
7586 .flat_map(|r| r.glyphs.iter().map(|g| g.ch))
7587 .collect();
7588 assert!(
7589 text.contains("wholly new"),
7590 "the reloaded text should be on screen, got {text:?}"
7591 );
7592 }
7593
7594 // ── golden-case harness ──────────────────────────────────────────────────
7595 // The pattern the whole parity suite can reuse: write a fixture with the
7596 // caret marked by `|`, run one action, and compare the rendered result —
7597 // also caret-marked — against the expected string. One readable line per
7598 // behavior, and it exercises the exact `Doc` ops both frontends call.
7599
7600 /// Split a `|`-marked fixture into `(source, caret_offset)`.
7601 fn parse_caret(marked: &str) -> (String, usize) {
7602 let caret = marked.find('|').expect("fixture needs a `|` caret marker");
7603 (marked.replacen('|', "", 1), caret)
7604 }
7605
7606 /// Render a doc's source with `|` at the caret (and `[`…`]` around any
7607 /// selection) so a result reads like the fixtures.
7608 fn render_caret(d: &Doc) -> String {
7609 // (offset, rank, char); rank keeps coincident markers ordered `[ | ]`
7610 // so the caret always renders inside its own selection.
7611 let mut marks: Vec<(usize, u8, char)> = vec![(d.caret, 1, '|')];
7612 if let Some((s, e)) = d.selection() {
7613 marks.push((s, 0, '['));
7614 marks.push((e, 2, ']'));
7615 }
7616 // Insert right-to-left: descending offset, then descending rank.
7617 marks.sort_by(|a, b| b.0.cmp(&a.0).then(b.1.cmp(&a.1)));
7618 let mut out = d.source.clone();
7619 for (at, _, ch) in marks {
7620 out.insert(at, ch);
7621 }
7622 out
7623 }
7624
7625 /// Load a `|`-marked fixture, run `action`, return the caret-marked result.
7626 fn golden(name: &str, marked: &str, action: impl FnOnce(&mut Doc)) -> String {
7627 golden_in(View::Source, name, marked, action)
7628 }
7629
7630 /// [`golden`] in a chosen view — the editing ops are the view's to share, so
7631 /// the same fixture has to read the same way in both.
7632 fn golden_in(view: View, name: &str, marked: &str, action: impl FnOnce(&mut Doc)) -> String {
7633 let (src, caret) = parse_caret(marked);
7634 let mut d = doc_in(view, name, &src);
7635 d.caret = caret;
7636 action(&mut d);
7637 render_caret(&d)
7638 }
7639
7640 #[test]
7641 fn word_motion_walks_word_by_word() {
7642 let g = |m, f: fn(&mut Doc)| golden("word_motion", m, f);
7643 assert_eq!(
7644 g("hello wor|ld", |d| d.move_word_left(false)),
7645 "hello |world"
7646 );
7647 assert_eq!(
7648 g("hello| world", |d| d.move_word_left(false)),
7649 "|hello world"
7650 );
7651 assert_eq!(
7652 g("hel|lo world", |d| d.move_word_right(false)),
7653 "hello| world"
7654 );
7655 assert_eq!(
7656 g("hello| world", |d| d.move_word_right(false)),
7657 "hello world|"
7658 );
7659 // Punctuation is its own class, so motion stops at the boundary.
7660 assert_eq!(g("|foo.bar", |d| d.move_word_right(false)), "foo|.bar");
7661 }
7662
7663 #[test]
7664 fn word_motion_extends_the_selection_when_asked() {
7665 assert_eq!(
7666 golden("word_sel", "hello |world", |d| d.move_word_right(true)),
7667 "hello [world|]"
7668 );
7669 }
7670
7671 #[test]
7672 fn delete_word_removes_a_whole_word() {
7673 let g = |m, f: fn(&mut Doc)| golden("del_word", m, f);
7674 assert_eq!(g("hello world|", |d| d.delete_word_back()), "hello |");
7675 assert_eq!(g("hello |world", |d| d.delete_word_forward()), "hello |");
7676 assert_eq!(g("foo |bar baz", |d| d.delete_word_back()), "|bar baz");
7677 }
7678
7679 // ── Home / End ───────────────────────────────────────────────────────────
7680
7681 #[test]
7682 fn home_toggles_between_the_line_s_text_and_its_margin() {
7683 // Source: the indentation is what the toggle is for. WYSIWYG resolves an
7684 // indent to the markup it spells everywhere it means one, so the fixture
7685 // with whitespace left to walk is a code block, which is verbatim.
7686 let g = |m, f: fn(&mut Doc)| golden("smart_home", m, f);
7687 assert_eq!(g(" inden|ted", |d| d.move_home(false)), " |indented");
7688 assert_eq!(g(" |indented", |d| d.move_home(false)), "| indented");
7689 assert_eq!(g("| indented", |d| d.move_home(false)), " |indented");
7690 // A line with no indentation has one place to go, so the toggle is a
7691 // no-op rather than a trip to nowhere.
7692 assert_eq!(g("hel|lo", |d| d.move_home(false)), "|hello");
7693 assert_eq!(g("|hello", |d| d.move_home(false)), "|hello");
7694
7695 let mut d = wysiwyg_doc("smart_home_wys", "```\n indented\n```\n");
7696 let indent = d.source.find(" indented").unwrap();
7697 d.caret = indent + 6; // inside "indented"
7698 d.move_home(false);
7699 assert_eq!(
7700 d.caret,
7701 indent + 4,
7702 "wysiwyg: Home aims at the code line's text"
7703 );
7704 d.move_home(false);
7705 assert_eq!(
7706 d.caret, indent,
7707 "wysiwyg: the second press takes the indent"
7708 );
7709 d.move_home(false);
7710 assert_eq!(d.caret, indent + 4, "wysiwyg: the toggle swaps back");
7711 }
7712
7713 #[test]
7714 fn end_takes_the_line_the_view_is_showing() {
7715 // The line differs by view for the same document, and that is the point:
7716 // a bare newline inside a paragraph is a soft break, which WYSIWYG draws
7717 // as a space on one row and the source view as two lines.
7718 let mut d = doc_with("end_src", "one two\nthree\n");
7719 d.caret = 1;
7720 d.move_end(false);
7721 assert_eq!(d.caret, 7, "source: the end of the source line");
7722
7723 let mut d = wysiwyg_doc("end_wys", "one two\nthree\n");
7724 d.caret = 1;
7725 d.move_end(false);
7726 assert_eq!(
7727 d.caret, 13,
7728 "wysiwyg: the end of the row, soft break and all"
7729 );
7730 }
7731
7732 #[test]
7733 fn home_and_end_extend_the_selection_when_asked() {
7734 for (view, tag) in VIEWS {
7735 let mut d = doc_in(view, &format!("home_end_ext_{tag}"), "hello world");
7736 d.caret = 6;
7737 d.move_end(true);
7738 assert_eq!(d.selection(), Some((6, 11)), "{tag}: End extends");
7739 let mut d = doc_in(view, &format!("home_ext_{tag}"), "hello world");
7740 d.caret = 6;
7741 d.move_home(true);
7742 assert_eq!(d.selection(), Some((0, 6)), "{tag}: Home extends");
7743 }
7744 }
7745
7746 // ── kill to the line's start / end ───────────────────────────────────────
7747
7748 #[test]
7749 fn kill_to_the_line_start_and_end_in_both_views() {
7750 for (view, tag) in VIEWS {
7751 // The gap that reads as a paragraph break in each view: the source
7752 // view's lines are the renderer's rows only where the source says so.
7753 let gap = if view == View::Source { "\n" } else { "\n\n" };
7754 let mut d = doc_in(
7755 view,
7756 &format!("kill_end_{tag}"),
7757 &format!("one two{gap}three\n"),
7758 );
7759 d.caret = 3;
7760 d.delete_to_line_end();
7761 assert_eq!(
7762 d.source,
7763 format!("one{gap}three\n"),
7764 "{tag}: ^K to the line's end"
7765 );
7766 assert_eq!(d.caret, 3, "{tag}: the caret stays where it kills from");
7767
7768 let mut d = doc_in(
7769 view,
7770 &format!("kill_start_{tag}"),
7771 &format!("one two{gap}three\n"),
7772 );
7773 d.caret = 7; // the end of the first line
7774 d.delete_to_line_start();
7775 assert_eq!(
7776 d.source,
7777 format!("{gap}three\n"),
7778 "{tag}: ⌘⌫ to the line's start"
7779 );
7780 assert_eq!(d.caret, 0, "{tag}");
7781 }
7782 }
7783
7784 #[test]
7785 fn a_kill_at_the_line_s_edge_leaves_the_lines_joined() {
7786 // The decision: at the boundary both kills do nothing, rather than
7787 // eating the line break. "Line" is the view's own — in WYSIWYG it ends
7788 // at a soft wrap as often as at a newline, where there is nothing
7789 // written to delete — and a source newline is only half of the blank
7790 // line between two paragraphs, so taking it leaves a soft break rather
7791 // than the join it looks like. Backspace and Delete are the keys for it.
7792 for (view, tag) in VIEWS {
7793 let gap = if view == View::Source { "\n" } else { "\n\n" };
7794 let src = format!("one{gap}three\n");
7795 let mut d = doc_in(view, &format!("kill_edge_end_{tag}"), &src);
7796 d.caret = 3; // the end of "one"
7797 d.delete_to_line_end();
7798 assert_eq!(
7799 d.source, src,
7800 "{tag}: ^K at the line's end joined it to the next"
7801 );
7802
7803 let mut d = doc_in(view, &format!("kill_edge_start_{tag}"), &src);
7804 d.caret = 3 + gap.len(); // the start of "three"
7805 d.delete_to_line_start();
7806 assert_eq!(
7807 d.source, src,
7808 "{tag}: ⌘⌫ at the line's start joined it to the last"
7809 );
7810 }
7811 }
7812
7813 #[test]
7814 fn a_kill_takes_the_selection_when_there_is_one() {
7815 // What every other delete here does with one, so these two as well.
7816 for (view, tag) in VIEWS {
7817 for (name, kill) in [
7818 (
7819 "end",
7820 (|d: &mut Doc| d.delete_to_line_end()) as fn(&mut Doc),
7821 ),
7822 ("start", |d: &mut Doc| d.delete_to_line_start()),
7823 ] {
7824 let mut d = doc_in(view, &format!("kill_sel_{name}_{tag}"), "one two three\n");
7825 d.anchor = Some(4);
7826 d.caret = 7; // "two"
7827 kill(&mut d);
7828 assert_eq!(
7829 d.source, "one three\n",
7830 "{tag}: {name} ignored the selection"
7831 );
7832 assert_eq!(d.selection(), None, "{tag}: {name}");
7833 }
7834 }
7835 }
7836
7837 #[test]
7838 fn a_kill_takes_the_markup_it_empties_with_it() {
7839 // The same hazard a word-delete has: a WYSIWYG range covers what the
7840 // user can see, which for `**bold**` is the word and never the
7841 // delimiters, so a kill that stopped at the text would leave `a ****` —
7842 // markup wrapped around nothing.
7843 let mut d = wysiwyg_doc("kill_widen", "a **bold**\n");
7844 d.caret = d.source.find("bold").unwrap();
7845 d.delete_to_line_end();
7846 assert_eq!(d.source, "a \n");
7847 }
7848
7849 #[test]
7850 fn a_kill_is_undone_in_one_step() {
7851 for (view, tag) in VIEWS {
7852 let mut d = doc_in(view, &format!("kill_undo_{tag}"), "one two three\n");
7853 d.caret = 3;
7854 d.delete_to_line_end();
7855 assert_eq!(d.source, "one\n", "{tag}");
7856 d.undo();
7857 assert_eq!(d.source, "one two three\n", "{tag}: a kill takes one undo");
7858 }
7859 }
7860
7861 #[test]
7862 fn select_block_grabs_the_whole_paragraph_from_any_wrapped_row() {
7863 // Regression: triple-click used move_home/move_end over visual rows, so
7864 // it only worked on a paragraph's first row (a wrap-boundary offset maps
7865 // to the earlier row). select_block_at reads the AST, so every offset in
7866 // the paragraph selects the whole thing.
7867 let body = "one two three four five six seven eight\n";
7868 let mut d = doc_with("sel_block", body);
7869 d.view = View::Wysiwyg;
7870 d.build_visual(12); // force the paragraph to wrap into several rows
7871 assert!(d.vmap.num_rows() > 1, "test needs a wrapped paragraph");
7872 let para = (0, "one two three four five six seven eight".len());
7873 for off in [0usize, 8, 19, 28, 38] {
7874 d.caret = 0;
7875 d.anchor = None;
7876 d.select_block_at(off);
7877 assert_eq!(
7878 d.selection(),
7879 Some(para),
7880 "offset {off} should select the paragraph"
7881 );
7882 }
7883 }
7884
7885 #[test]
7886 fn select_block_uses_content_span_for_a_heading() {
7887 let mut d = doc_with("sel_head", "# Title\n\nbody\n");
7888 d.select_block_at(4); // inside "Title"
7889 // content_span excludes the "# " marker.
7890 assert_eq!(d.selected_text(), Some("Title"));
7891 d.select_block_at(10); // inside "body"
7892 assert_eq!(d.selected_text(), Some("body"));
7893 }
7894
7895 #[test]
7896 fn select_all_spans_the_document() {
7897 let mut d = doc_with("sel_all", "abc\n\ndef\n");
7898 d.select_all();
7899 assert_eq!(d.selection(), Some((0, d.source.len())));
7900 }
7901
7902 #[test]
7903 fn select_word_at_picks_the_surrounding_word() {
7904 let mut d = doc_with("sel_word", "hello world\n");
7905 d.select_word_at(8); // inside "world"
7906 assert_eq!(d.selection(), Some((6, 11)));
7907 // Double-clicking at end-of-word still grabs the word to its left.
7908 d.select_word_at(5); // the space between the words
7909 assert_eq!(d.selection(), Some((5, 6)));
7910 }
7911
7912 #[test]
7913 fn word_helpers_respect_utf8_boundaries() {
7914 // "café" is 5 bytes ('é' is two); motion must land on char boundaries.
7915 assert_eq!(
7916 golden("utf8", "|café ok", |d| d.move_word_right(false)),
7917 "café| ok"
7918 );
7919 assert_eq!(golden("utf8b", "café |ok", |d| d.delete_word_back()), "|ok");
7920 }
7921
7922 #[test]
7923 fn typing_inserts_at_the_caret_and_advances_it() {
7924 let mut d = doc_with("type", "hello\n");
7925 d.insert("Hi ");
7926 assert_eq!(d.source, "Hi hello\n");
7927 assert_eq!(d.caret, 3);
7928 assert!(d.dirty);
7929 }
7930
7931 #[test]
7932 fn backspace_deletes_the_char_before_the_caret() {
7933 let mut d = doc_with("bs", "hello\n");
7934 d.caret = 3; // after "hel"
7935 d.backspace();
7936 assert_eq!(d.source, "helo\n");
7937 assert_eq!(d.caret, 2);
7938 }
7939
7940 #[test]
7941 fn typing_replaces_the_selection() {
7942 let mut d = doc_with("replace", "a word b\n");
7943 d.anchor = Some(2);
7944 d.caret = 6; // "word" selected
7945 d.insert("X");
7946 assert_eq!(d.source, "a X b\n");
7947 assert_eq!(d.caret, 3);
7948 assert_eq!(d.anchor, None);
7949 }
7950
7951 #[test]
7952 fn toggle_bold_wraps_then_unwraps_the_selection() {
7953 let mut d = doc_with("bold", "a word b\n");
7954 d.anchor = Some(2);
7955 d.caret = 6;
7956 d.toggle(InlineKind::Strong);
7957 assert_eq!(d.source, "a **word** b\n");
7958 // The toggled region stays selected, so a second toggle reverses it.
7959 d.toggle(InlineKind::Strong);
7960 assert_eq!(d.source, "a word b\n");
7961 d.toggle(InlineKind::Strong);
7962 assert_eq!(d.source, "a **word** b\n");
7963 }
7964
7965 #[test]
7966 fn toggle_code_wraps_then_unwraps_the_selection() {
7967 let mut d = doc_with("code_rt", "a word b\n");
7968 d.anchor = Some(2);
7969 d.caret = 6;
7970 d.toggle(InlineKind::Verbatim);
7971 assert_eq!(d.source, "a `word` b\n");
7972 d.toggle(InlineKind::Verbatim);
7973 assert_eq!(d.source, "a word b\n");
7974 }
7975
7976 #[test]
7977 fn sticky_bold_with_no_selection_wraps_the_next_typed_text() {
7978 // ⌘b at a bare caret, then type: the text comes out bold with no
7979 // selection ever made — the word-processor "start bold here" gesture.
7980 let mut d = doc_with("sticky_wrap", "xy\n");
7981 d.caret = 1; // between x and y
7982 d.toggle(InlineKind::Strong);
7983 assert_eq!(d.source, "xy\n", "arming a mark must not edit the document");
7984 d.insert("A");
7985 assert_eq!(d.source, "x**A**y\n");
7986 }
7987
7988 #[test]
7989 fn sticky_bold_lights_the_toolbar_before_any_typing() {
7990 // The button must light the instant ⌘b is pressed, or the mode is
7991 // invisible until the first character lands.
7992 let mut d = doc_with("sticky_light", "xy\n");
7993 d.caret = 1;
7994 assert!(!d.active_inline_marks().contains(InlineKind::Strong));
7995 d.toggle(InlineKind::Strong);
7996 assert!(d.active_inline_marks().contains(InlineKind::Strong));
7997 }
7998
7999 #[test]
8000 fn sticky_bold_toggled_off_types_normally_again() {
8001 // ⌘b, type, ⌘b, type: the first run is bold, the second is not — all
8002 // in the flow of typing, the exact sequence the user described.
8003 let mut d = doc_with("sticky_off", "\n");
8004 d.caret = 0;
8005 d.toggle(InlineKind::Strong);
8006 d.insert("a");
8007 d.insert("b"); // continues inside the run, no re-arming
8008 assert_eq!(d.source, "**ab**\n");
8009 d.toggle(InlineKind::Strong); // ⌘b again — shed bold
8010 d.insert("c");
8011 assert_eq!(d.source, "**ab**c\n");
8012 }
8013
8014 #[test]
8015 fn continued_typing_after_a_sticky_run_stays_in_the_run() {
8016 // Once a mark is realised the caret sits inside the run, so plain typing
8017 // extends it rather than starting a second, adjacent bold span.
8018 let mut d = doc_with("sticky_cont", "\n");
8019 d.caret = 0;
8020 d.toggle(InlineKind::Emph);
8021 d.insert("h");
8022 d.insert("i");
8023 assert_eq!(d.source, "*hi*\n");
8024 }
8025
8026 #[test]
8027 fn moving_the_caret_disarms_a_sticky_mark() {
8028 // Arming a mark and then moving away must not style text elsewhere.
8029 let mut d = doc_with("sticky_disarm", "xy\n");
8030 d.caret = 0;
8031 d.toggle(InlineKind::Strong);
8032 d.move_right(false); // caret 0 → 1, disarms
8033 assert!(!d.active_inline_marks().contains(InlineKind::Strong));
8034 d.insert("A");
8035 assert_eq!(d.source, "xAy\n", "the mark must not follow the caret");
8036 }
8037
8038 #[test]
8039 fn stacked_sticky_marks_apply_together() {
8040 // ⌘b then ⌘i before typing: the text comes out both bold and italic.
8041 let mut d = doc_with("sticky_stack", "\n");
8042 d.caret = 0;
8043 d.toggle(InlineKind::Strong);
8044 d.toggle(InlineKind::Emph);
8045 d.insert("x");
8046 // Land the caret on the styled character and confirm both marks are live.
8047 d.anchor = Some(d.source.find('x').unwrap());
8048 d.caret = d.anchor.unwrap() + 1;
8049 let marks = d.active_inline_marks();
8050 assert!(marks.contains(InlineKind::Strong), "bold: {}", d.source);
8051 assert!(marks.contains(InlineKind::Emph), "italic: {}", d.source);
8052 }
8053
8054 // ── the mark-edge rule (see `Doc::splice`) ───────────────────────────────
8055
8056 #[test]
8057 fn a_space_typed_in_a_bold_run_never_leaves_the_delimiters_showing() {
8058 // The reported bug, keystroke for keystroke: ⌘b, "bold", space, "hey".
8059 // The space inside the run made `**bold **`, which is *not* bold — four
8060 // literal asterisks — so the rich view drew them, correctly and
8061 // uselessly, until the next character happened to close the run again.
8062 let mut d = wysiwyg_doc("edge_typing", "a \n");
8063 d.caret = 2;
8064 d.toggle(InlineKind::Strong);
8065 for c in "bold".chars() {
8066 d.insert(&c.to_string());
8067 }
8068 assert_eq!(d.source, "a **bold**\n");
8069 d.insert(" ");
8070 assert_eq!(
8071 d.source, "a **bold** \n",
8072 "the space belongs outside the run"
8073 );
8074 assert!(
8075 d.active_inline_marks().contains(InlineKind::Strong),
8076 "bold is still what's being typed, so the button stays lit"
8077 );
8078 // What the writer is looking at while all this happens: their words.
8079 d.build_visual(80);
8080 let drawn: String = d.vmap.rows[0].glyphs.iter().map(|g| g.ch).collect();
8081 assert_eq!(drawn, "a bold ", "no delimiter ever surfaces: {}", d.source);
8082 for c in "hey".chars() {
8083 d.insert(&c.to_string());
8084 }
8085 assert_eq!(
8086 d.source, "a **bold hey**\n",
8087 "one bold phrase, not two runs"
8088 );
8089 }
8090
8091 #[test]
8092 fn typing_past_a_space_can_still_leave_the_bold_behind() {
8093 // The other half: the marks stay armed across the space, so ⌘b turns
8094 // them off again there and the next word is plain — the run isn't
8095 // rejoined by a caret that was told not to.
8096 let mut d = wysiwyg_doc("edge_shed", "\n");
8097 d.caret = 0;
8098 d.toggle(InlineKind::Strong);
8099 for c in "bold ".chars() {
8100 d.insert(&c.to_string());
8101 }
8102 assert_eq!(d.source, "**bold** \n");
8103 d.toggle(InlineKind::Strong);
8104 assert!(!d.active_inline_marks().contains(InlineKind::Strong));
8105 d.insert("x");
8106 assert_eq!(d.source, "**bold** x\n");
8107 }
8108
8109 #[test]
8110 fn a_space_typed_first_of_all_still_leaves_the_mark_armed() {
8111 // ⌘b and then a space before any word: the space is not marked (nothing
8112 // is), and the word after it is.
8113 let mut d = wysiwyg_doc("edge_space_first", "a\n");
8114 d.caret = 1;
8115 d.toggle(InlineKind::Strong);
8116 d.insert(" ");
8117 assert_eq!(d.source, "a \n");
8118 assert!(d.active_inline_marks().contains(InlineKind::Strong));
8119 d.insert("b");
8120 assert_eq!(d.source, "a **b**\n");
8121 }
8122
8123 #[test]
8124 fn a_space_typed_at_either_edge_of_an_existing_mark_steps_outside_it() {
8125 let mut d = wysiwyg_doc("edge_tail", "x **bold**\n");
8126 d.caret = 8; // the caret's home at the end of the run's text
8127 d.insert(" ");
8128 assert_eq!(
8129 d.source, "x **bold** \n",
8130 "the space lands past the delimiters"
8131 );
8132 assert_eq!(d.caret, 11, "and the caret stands past it, outside the run");
8133
8134 let mut d = wysiwyg_doc("edge_head", "x **bold** y\n");
8135 d.caret = 4; // in front of the "b"
8136 d.insert(" ");
8137 assert_eq!(d.source, "x **bold** y\n");
8138 assert_eq!(d.caret, 3, "in front of the run, where the space was typed");
8139 }
8140
8141 #[test]
8142 fn a_delete_that_backs_a_space_onto_a_delimiter_moves_the_delimiter() {
8143 // Backspace over the last letter of a bold phrase.
8144 let mut d = wysiwyg_doc("edge_bksp", "a **bold h**\n");
8145 d.caret = 10; // past the "h"
8146 d.backspace();
8147 assert_eq!(d.source, "a **bold** \n");
8148 assert_eq!(d.caret, 11, "the caret keeps the place on screen it had");
8149 assert!(d.active_inline_marks().contains(InlineKind::Strong));
8150 d.insert("x");
8151 assert_eq!(d.source, "a **bold x**\n", "and typing rejoins the run");
8152 }
8153
8154 #[test]
8155 fn deleting_the_last_of_a_run_takes_its_delimiters_with_it() {
8156 // `**b**` with the `b` gone is `****`: two delimiters with nothing to
8157 // mark, which is only text. The marks live on in the caret instead.
8158 let mut d = wysiwyg_doc("edge_empty", "a **b** c\n");
8159 d.caret = 5;
8160 d.backspace();
8161 assert_eq!(d.source, "a c\n");
8162 assert!(d.active_inline_marks().contains(InlineKind::Strong));
8163 d.insert("x");
8164 assert_eq!(d.source, "a **x** c\n");
8165 }
8166
8167 #[test]
8168 fn typing_over_a_whole_bold_word_keeps_it_bold() {
8169 let mut d = wysiwyg_doc("edge_replace", "a **bold** c\n");
8170 d.anchor = Some(4);
8171 d.caret = 8; // the word, not its delimiters
8172 d.insert("x");
8173 assert_eq!(d.source, "a **x** c\n");
8174 }
8175
8176 #[test]
8177 fn a_code_span_keeps_the_space_it_is_given() {
8178 // Backticks are not whitespace-sensitive the way `**` is: `` `code ` ``
8179 // is still verbatim, so nothing is re-spelt. The repair asks the parser
8180 // rather than a table of kinds, and this is the answer it gets.
8181 let mut d = wysiwyg_doc("edge_code", "a `code` c\n");
8182 d.caret = 7;
8183 d.insert(" ");
8184 assert_eq!(d.source, "a `code ` c\n");
8185 }
8186
8187 #[test]
8188 fn a_delete_from_a_runs_outer_edge_reaches_into_the_run() {
8189 // A run's closing delimiter has a caret home on each side of it, one
8190 // column apart on screen — and a plain ← off the space after a bold word
8191 // lands on the outer one. The character drawn behind the caret there is
8192 // still the last letter of the phrase, so that is what Backspace takes;
8193 // the byte behind it is a `*` nobody can see.
8194 let mut d = wysiwyg_doc("edge_outer_close", "**bold** x\n");
8195 d.caret = 9;
8196 d.move_left(false);
8197 assert_eq!(d.caret, 8, "← rests past the delimiters, not inside them");
8198 d.backspace();
8199 assert_eq!(
8200 d.source, "**bol** x\n",
8201 "a letter of the phrase, not its `*`"
8202 );
8203 assert_eq!(d.caret, 5);
8204
8205 // And the mirror in front of the opening delimiter, where Delete's
8206 // character is the first letter of the run.
8207 let mut d = wysiwyg_doc("edge_outer_open", "x**bold**\n");
8208 d.caret = 1;
8209 d.delete_forward();
8210 assert_eq!(d.source, "x**old**\n");
8211 assert_eq!(d.caret, 3, "inside the run, in front of what is left of it");
8212 }
8213
8214 #[test]
8215 fn a_delete_at_a_run_edge_never_eats_a_delimiter() {
8216 // The byte beside the caret at either edge of a bold word is a `*` the
8217 // rich view draws nothing for. Taking it is not the character delete the
8218 // key was pressed for — it unspells the run and puts a literal asterisk
8219 // on screen (`a *bold** c`). The visible character is the one that goes.
8220 let mut d = wysiwyg_doc("edge_open_bksp", "a **bold** c\n");
8221 d.caret = 4; // in front of the "b"
8222 d.backspace();
8223 assert_eq!(d.source, "a**bold** c\n", "the space goes, the run stands");
8224
8225 let mut d = wysiwyg_doc("edge_close_del", "a **bold** c\n");
8226 d.caret = 8; // past the "d"
8227 d.delete_forward();
8228 assert_eq!(d.source, "a **bold**c\n");
8229 assert_eq!(d.caret, 8, "and the caret stays inside the run");
8230 d.insert("x");
8231 assert_eq!(d.source, "a **boldx**c\n");
8232
8233 // A code span's backticks are hidden the same way, so they are covered
8234 // by the same rule and not by a list of kinds.
8235 let mut d = wysiwyg_doc("edge_open_code", "a `code` c\n");
8236 d.caret = 3;
8237 d.backspace();
8238 assert_eq!(d.source, "a`code` c\n");
8239 }
8240
8241 #[test]
8242 fn the_source_view_deletes_the_delimiter_byte_it_is_shown() {
8243 // The asterisks are on the screen there and the caret can stand between
8244 // them, so a delete takes exactly the byte it is aimed at.
8245 let mut d = doc_with("edge_open_src", "a **bold** c\n");
8246 d.caret = 4;
8247 d.backspace();
8248 assert_eq!(d.source, "a *bold** c\n");
8249
8250 let mut d = doc_with("edge_close_src", "a **bold** c\n");
8251 d.caret = 8;
8252 d.delete_forward();
8253 assert_eq!(d.source, "a **bold* c\n");
8254 }
8255
8256 #[test]
8257 fn backspacing_the_space_out_of_a_bold_phrase_leaves_the_caret_in_it() {
8258 // The reported bug, keystroke for keystroke: ⌘b, "bold", space, Backspace.
8259 // The space had stepped outside the run (the mark-edge rule), taking the
8260 // caret with it, so the delete put it back down on the far side of the
8261 // closing `**` — one place on screen, and the wrong side of it. Typing
8262 // came out plain and the toolbar went dark, with nothing to see.
8263 let mut d = wysiwyg_doc("edge_bksp_space", "\n");
8264 d.caret = 0;
8265 d.toggle(InlineKind::Strong);
8266 for c in "bold".chars() {
8267 d.insert(&c.to_string());
8268 }
8269 d.insert(" ");
8270 assert_eq!(d.source, "**bold** \n");
8271 d.backspace();
8272 assert_eq!(
8273 d.source, "**bold**\n",
8274 "the space goes, the delimiters stay"
8275 );
8276 assert_eq!(d.caret, 6, "and the caret comes back inside the run");
8277 assert!(
8278 d.active_inline_marks().contains(InlineKind::Strong),
8279 "so the button is still lit"
8280 );
8281 d.insert("x");
8282 assert_eq!(
8283 d.source, "**boldx**\n",
8284 "and the next character is still bold"
8285 );
8286 }
8287
8288 #[test]
8289 fn a_second_backspace_there_deletes_a_letter_of_the_phrase() {
8290 // What the stranded caret did next: the byte behind it was the closing
8291 // `*`, so a second press took that instead of a letter — `**bold*`, the
8292 // styling gone and an asterisk on the screen where the word had been.
8293 let mut d = wysiwyg_doc("edge_bksp_twice", "\n");
8294 d.caret = 0;
8295 d.toggle(InlineKind::Strong);
8296 for c in "bold ".chars() {
8297 d.insert(&c.to_string());
8298 }
8299 assert_eq!(d.source, "**bold** \n");
8300 d.backspace();
8301 d.backspace();
8302 assert_eq!(d.source, "**bol**\n", "the delete lands inside the run");
8303 assert_eq!(d.caret, 5);
8304 }
8305
8306 #[test]
8307 fn a_delete_that_ends_at_a_nested_run_settles_inside_every_delimiter() {
8308 // `***both***` closes two runs with one stack of asterisks: the caret has
8309 // to walk in through all of them, or it lands between the emph and the
8310 // strong and types half-marked.
8311 let mut d = wysiwyg_doc("edge_bksp_nested", "***both*** \n");
8312 d.caret = 11;
8313 d.backspace();
8314 assert_eq!(d.source, "***both***\n");
8315 assert_eq!(d.caret, 7, "past the last letter, inside both runs");
8316 d.insert("x");
8317 assert_eq!(d.source, "***bothx***\n");
8318 }
8319
8320 #[test]
8321 fn a_delete_that_ends_mid_run_leaves_the_caret_where_it_fell() {
8322 // The settle only moves a caret a run actually closed over. Ordinary
8323 // deletes — inside a run, or in plain prose — are untouched.
8324 let mut d = wysiwyg_doc("edge_bksp_mid", "a **bold** c\n");
8325 d.caret = 8;
8326 d.backspace();
8327 assert_eq!(d.source, "a **bol** c\n");
8328 assert_eq!(d.caret, 7);
8329
8330 let mut d = wysiwyg_doc("edge_bksp_plain", "plain\n");
8331 d.caret = 5;
8332 d.backspace();
8333 assert_eq!(d.source, "plai\n");
8334 assert_eq!(d.caret, 4);
8335 }
8336
8337 #[test]
8338 fn the_source_view_leaves_a_delete_where_it_landed() {
8339 // The delimiters are on the screen there, so the offset past them is a
8340 // place the caret can be seen to be — nothing to settle.
8341 let mut d = doc_with("edge_bksp_src", "**bold** \n");
8342 d.caret = 9;
8343 d.backspace();
8344 assert_eq!(d.source, "**bold**\n");
8345 assert_eq!(d.caret, 8);
8346 }
8347
8348 #[test]
8349 fn the_mark_edge_rule_clears_every_delimiter_of_a_nested_run() {
8350 // `***both***` closes two runs with one stack of asterisks; a space that
8351 // clears only the inner one lands against the outer's and breaks that
8352 // instead.
8353 let mut d = wysiwyg_doc("edge_nested", "a ***both***\n");
8354 d.caret = 9;
8355 d.insert(" ");
8356 assert_eq!(d.source, "a ***both*** \n");
8357 assert_eq!(d.caret, 13);
8358 d.insert("x");
8359 assert_eq!(d.source, "a ***both x***\n");
8360 }
8361
8362 #[test]
8363 fn the_mark_edge_repair_undoes_with_the_keystroke_that_caused_it() {
8364 // The delimiter shuffle is not an edit the writer made, so it is not a
8365 // step they have to undo past.
8366 let mut d = wysiwyg_doc("edge_undo", "a **bold**\n");
8367 d.caret = 8;
8368 d.insert(" ");
8369 assert_eq!(d.source, "a **bold** \n");
8370 d.undo();
8371 assert_eq!(d.source, "a **bold**\n");
8372 }
8373
8374 #[test]
8375 fn the_source_view_types_the_space_where_it_was_asked_to() {
8376 // The rule is a rich-view courtesy. In the source view the delimiters are
8377 // on the screen and the user is editing the bytes they can see.
8378 let mut d = doc_with("edge_src", "a **bold** c\n");
8379 d.caret = 8;
8380 d.insert(" ");
8381 assert_eq!(d.source, "a **bold ** c\n");
8382 }
8383
8384 #[test]
8385 fn toggling_a_mark_over_a_selection_leaves_its_edge_whitespace_out() {
8386 // Double-clicking a word takes the space after it; bolding that must not
8387 // spell `**word **`, which is not bold at all.
8388 let mut d = wysiwyg_doc("edge_sel", "a word b\n");
8389 d.anchor = Some(2);
8390 d.caret = 7; // "word "
8391 d.toggle(InlineKind::Strong);
8392 assert_eq!(d.source, "a **word** b\n");
8393 d.toggle(InlineKind::Strong);
8394 assert_eq!(d.source, "a word b\n");
8395 d.toggle(InlineKind::Strong);
8396 assert_eq!(
8397 d.source, "a **word** b\n",
8398 "reapplying the mark must not wrap stale delimiter offsets"
8399 );
8400 // And a selection of nothing but whitespace has no word to mark.
8401 let mut d = wysiwyg_doc("edge_sel_ws", "a word b\n");
8402 d.anchor = Some(6);
8403 d.caret = 7;
8404 d.toggle(InlineKind::Strong);
8405 assert_eq!(d.source, "a word b\n");
8406 assert!(d.status.is_some());
8407 }
8408
8409 #[test]
8410 fn set_block_turns_a_paragraph_into_a_heading_at_the_caret() {
8411 let mut d = doc_with("head_set", "hello\n");
8412 d.caret = 2; // caret inside the paragraph, no selection
8413 d.set_block(BlockKind::Heading(1));
8414 assert_eq!(d.source, "# hello\n");
8415 }
8416
8417 #[test]
8418 fn set_block_heading_works_in_wysiwyg_view() {
8419 // The app defaults to WYSIWYG; the caret is a source offset either way.
8420 let mut d = wysiwyg_doc("head_wys", "hello\n");
8421 d.caret = 2;
8422 d.set_block(BlockKind::Heading(1));
8423 assert_eq!(d.source, "# hello\n");
8424 }
8425
8426 #[test]
8427 fn toggle_heading_applies_switches_and_reverts() {
8428 let mut d = doc_with("head_toggle", "hello\n");
8429 d.caret = 2;
8430 d.toggle_heading(1);
8431 assert_eq!(d.source, "# hello\n"); // paragraph → H1
8432 d.toggle_heading(2);
8433 assert_eq!(d.source, "## hello\n"); // H1 → H2 (different level switches)
8434 d.toggle_heading(2);
8435 assert_eq!(d.source, "hello\n"); // same level reverts to paragraph
8436 }
8437
8438 #[test]
8439 fn preserve_enter_at_a_line_end_lands_the_caret_on_the_new_blank_line() {
8440 // Regression: Enter at the end of a soft-break line (mid-paragraph) opened
8441 // the blank line but the caret rendered on the *next* line, because the
8442 // separator was a non-navigable decoration row. In Preserve flow that
8443 // blank line is a real caret home — the caret must resolve onto it, and
8444 // typing there makes the soft break that continues the paragraph.
8445 let src = "line one:\nsecond line\n";
8446 let mut d = wysiwyg_doc("pre_enter_lineend", src);
8447 d.set_line_flow(LineFlow::Preserve);
8448 d.build_visual_unwrapped(); // the GUI path (pixel-wrapped)
8449 d.caret = 9; // the visual end of row 0, at the soft-break '\n'
8450 d.newline();
8451 d.build_visual_unwrapped();
8452 assert_eq!(d.source, "line one:\n\nsecond line\n");
8453 assert_eq!(
8454 d.caret, 10,
8455 "caret sits on the new blank line, not the next line"
8456 );
8457 // The blank line is row 1, and the caret resolves onto it — not row 2.
8458 assert_eq!(
8459 d.vmap.pos_of_offset(10),
8460 (1, 0),
8461 "caret renders on the blank row"
8462 );
8463 assert!(
8464 !d.vmap.rows[1].decoration,
8465 "the blank line is navigable in Preserve"
8466 );
8467 // Typing there makes a soft break: one paragraph, three lines.
8468 d.insert("new clause,");
8469 assert_eq!(d.source, "line one:\nnew clause,\nsecond line\n");
8470 }
8471
8472 #[test]
8473 fn preserve_enter_makes_a_soft_break_not_a_paragraph() {
8474 // Mid-paragraph: Enter splits the line with a single `\n`, a soft break
8475 // that keeps it one paragraph — where Fold would open a second paragraph.
8476 let mut d = wysiwyg_doc("pre_enter_mid", "abcdef\n");
8477 d.set_line_flow(LineFlow::Preserve);
8478 d.caret = 3;
8479 d.newline();
8480 assert_eq!(d.source, "abc\ndef\n", "mid-line Enter is a soft break");
8481
8482 // End-of-paragraph: Enter then typing continues the same paragraph on a
8483 // new line (a soft break), not a fresh paragraph.
8484 let mut d = wysiwyg_doc("pre_enter_end", "abc\n");
8485 d.set_line_flow(LineFlow::Preserve);
8486 d.caret = 3;
8487 d.newline();
8488 d.insert("def");
8489 assert_eq!(
8490 d.source, "abc\ndef\n",
8491 "end-of-line Enter + typing is a soft break"
8492 );
8493 }
8494
8495 #[test]
8496 fn preserve_double_enter_still_makes_a_paragraph() {
8497 // Two Enters in a row promote to a real paragraph break: the second lands
8498 // on the blank line the first opened and takes the empty-line branch.
8499 let mut d = wysiwyg_doc("pre_enter_dbl", "abc\n");
8500 d.set_line_flow(LineFlow::Preserve);
8501 d.caret = 3;
8502 d.newline();
8503 d.newline();
8504 d.insert("def");
8505 assert_eq!(
8506 d.source, "abc\n\ndef\n",
8507 "double Enter is a paragraph break"
8508 );
8509 }
8510
8511 #[test]
8512 fn preserve_backspace_joins_across_a_soft_break() {
8513 // Backspace is the symmetric undo of a Preserve Enter: over the `\n` of a
8514 // soft break it deletes the single newline and joins the two lines.
8515 let mut d = wysiwyg_doc("pre_bs", "abc\ndef\n");
8516 d.set_line_flow(LineFlow::Preserve);
8517 d.build_visual(80);
8518 d.caret = 4; // start of "def", just past the soft break
8519 d.backspace();
8520 assert_eq!(
8521 d.source, "abcdef\n",
8522 "Backspace joins across the soft break"
8523 );
8524 assert_eq!(d.caret, 3, "caret lands where the lines meet");
8525 }
8526
8527 #[test]
8528 fn fold_enter_still_starts_a_new_paragraph() {
8529 // The default flow is unchanged: a lone `\n` would render as an invisible
8530 // space, so Enter keeps opening the paragraph break that actually shows.
8531 let mut d = wysiwyg_doc("fold_enter", "abcdef\n");
8532 d.caret = 3;
8533 d.newline();
8534 assert_eq!(
8535 d.source, "abc\n\ndef\n",
8536 "Fold mid-line Enter is a paragraph break"
8537 );
8538 }
8539
8540 #[test]
8541 fn wysiwyg_one_enter_starts_a_new_paragraph() {
8542 // Regression: one Enter left the caret between the two newlines, so typing
8543 // made a soft break (one paragraph) and you needed a second Enter.
8544 let mut d = wysiwyg_doc("wys_enter", "abc\n");
8545 d.caret = 3;
8546 d.newline();
8547 d.insert("def");
8548 assert_eq!(d.source, "abc\n\ndef\n"); // two paragraphs, not "abc\ndef\n"
8549 }
8550
8551 #[test]
8552 fn enter_at_the_end_of_a_bold_run_keeps_its_closing_delimiter_attached() {
8553 // Regression: Enter at the caret's natural End-of-line resting place
8554 // after a bold run with nothing following it (on screen: right after
8555 // "bold", before the hidden closing "**") spliced the paragraph break
8556 // at that very byte offset — which sits *before* the closing "**" in
8557 // the source, since the delimiter is hidden and emits no glyph of its
8558 // own for `push_row`'s "end of row" fallback to count. That severed the
8559 // mark: "**bold**\n" became "**bold\n\n**\n", stranding the closing
8560 // "**" alone on the new line instead of leaving "**bold**" intact with
8561 // a fresh empty paragraph after it.
8562 let mut d = wysiwyg_doc("bold_eol_enter", "**bold**\n");
8563 d.move_end(false); // the WYSIWYG End key, from caret 0
8564 assert_eq!(
8565 d.caret, 6,
8566 "caret rests right after \"bold\", before the hidden \"**\""
8567 );
8568 d.newline();
8569 assert!(
8570 d.source.starts_with("**bold**"),
8571 "the closing ** must stay attached to \"bold\": got {:?}",
8572 d.source
8573 );
8574 assert_eq!(
8575 d.source, "**bold**\n\n\n",
8576 "a fresh empty paragraph follows the still-intact bold run"
8577 );
8578 }
8579
8580 #[test]
8581 fn source_view_enter_is_a_single_newline() {
8582 let mut d = doc_with("src_enter", "abc\n");
8583 d.caret = 3;
8584 d.newline();
8585 assert_eq!(d.source, "abc\n\n");
8586 }
8587
8588 #[test]
8589 fn heading_applies_at_the_end_of_a_paragraph() {
8590 // The caret at a line end sits at the doc level; set_block must still find
8591 // the block on that line.
8592 let mut d = doc_with("head_end", "abc\n");
8593 d.caret = 3; // end of "abc"
8594 d.toggle_heading(1);
8595 assert_eq!(d.source, "# abc\n");
8596 }
8597
8598 #[test]
8599 fn heading_on_an_empty_new_paragraph_creates_one() {
8600 let mut d = wysiwyg_doc("head_empty", "abc\n");
8601 d.caret = 3;
8602 d.newline(); // caret now on a fresh, empty paragraph
8603 d.toggle_heading(1);
8604 d.insert("Title");
8605 assert!(d.source.contains("# Title"), "got {:?}", d.source);
8606 }
8607
8608 #[test]
8609 fn a_heading_typed_on_a_blank_line_keeps_the_caret_on_its_own_row() {
8610 // The reported bug, end to end: click a blank line with another one under
8611 // it, press H1, type. The text landed in the heading and the caret's
8612 // offset was right (the source view drew it there), but the rich view
8613 // drew it two rows lower, on the trailing blank line — the empty `# `
8614 // heading had left every row below it short by the marker's two bytes,
8615 // and the blank line ended up claiming the heading's own end offset.
8616 let mut d = wysiwyg_doc("head_blank", "one\n\ntwo\n\n\n\n");
8617 d.build_visual_unwrapped();
8618 d.caret = d.vmap.offset_of_pos(4, 0); // the first of the two blank lines
8619 d.toggle_heading(1);
8620 for c in "title".chars() {
8621 d.insert(&c.to_string());
8622 d.build_visual_unwrapped(); // as a frontend does, one frame per key
8623 }
8624 assert_eq!(d.source, "one\n\ntwo\n\n# title\n\n");
8625 assert_eq!(
8626 d.caret_pos(),
8627 (4, 5),
8628 "the caret draws at the end of the heading"
8629 );
8630 }
8631
8632 #[test]
8633 fn clicking_an_empty_heading_types_after_its_marker() {
8634 // The same anchor from the other side: the empty heading's row is its own
8635 // caret home, so a click on it must land past the hidden `# `. Landing in
8636 // front of the hashes made the first keystroke un-heading the line.
8637 let mut d = wysiwyg_doc("head_click", "# \n");
8638 d.build_visual_unwrapped();
8639 d.caret = d.vmap.offset_of_pos(0, 0);
8640 d.insert("x");
8641 assert_eq!(d.source, "# x\n");
8642 }
8643
8644 #[test]
8645 fn wysiwyg_enter_after_a_heading_makes_a_paragraph() {
8646 let mut d = wysiwyg_doc("head_enter", "# Title\n");
8647 d.caret = 7; // end of the heading
8648 d.newline();
8649 d.insert("body");
8650 assert_eq!(d.source, "# Title\n\nbody\n");
8651 }
8652
8653 #[test]
8654 fn wysiwyg_enter_continues_a_bullet_list() {
8655 let mut d = wysiwyg_doc("wys_bullet", "- item\n");
8656 d.caret = 6; // end of "item"
8657 d.newline();
8658 d.insert("two");
8659 assert_eq!(d.source, "- item\n- two\n");
8660 }
8661
8662 #[test]
8663 fn wysiwyg_enter_increments_an_ordered_list() {
8664 let mut d = wysiwyg_doc("wys_ol", "1. one\n");
8665 d.caret = 6; // end of "one"
8666 d.newline();
8667 d.insert("two");
8668 assert_eq!(d.source, "1. one\n2. two\n");
8669 }
8670
8671 #[test]
8672 fn wysiwyg_backspace_after_leaving_a_list_collapses_the_gap_cleanly() {
8673 // Regression for the "extra newline" left between a list and the paragraph
8674 // below it. Enter, Enter leaves the list on a fresh empty paragraph
8675 // (`- item\n\n\n\nnext`, a navigable blank between the two blocks); one
8676 // Backspace should then take the caret cleanly back to the end of the list
8677 // item, `- item\n\nnext`, not delete a single newline and strand it on the
8678 // odd `- item\n\n\nnext` — a blank line the eye reads as one separator but
8679 // no caret can land on. The map is rebuilt between keystrokes exactly as a
8680 // frontend does, since Backspace reads the stop table to place the delete.
8681 let mut d = wysiwyg_doc("wys_exit_bksp", "- item\n\nnext\n");
8682 d.caret = 6; // end of "item"
8683 d.newline();
8684 d.build_visual(80);
8685 d.newline(); // leave the list onto a fresh empty paragraph
8686 d.build_visual(80);
8687 assert_eq!(
8688 d.source, "- item\n\n\n\nnext\n",
8689 "double-Enter opens the empty paragraph"
8690 );
8691 d.backspace();
8692 assert_eq!(
8693 d.source, "- item\n\nnext\n",
8694 "one Backspace collapses the whole gap"
8695 );
8696 assert_eq!(
8697 d.caret, 6,
8698 "and lands the caret back at the end of the list item"
8699 );
8700 }
8701
8702 #[test]
8703 fn wysiwyg_backspace_on_stacked_blank_lines_still_removes_just_one() {
8704 // The stop-wise delete must not over-reach when there is no block boundary
8705 // to cross: two blank lines in a row are one caret stop apart, so pressing
8706 // Enter on an empty line and then Backspace removes exactly the one newline
8707 // it added — the lone-Enter / lone-Backspace symmetry, preserved.
8708 let mut d = wysiwyg_doc("wys_stack", "abc\n\n\n");
8709 d.caret = 5; // the empty paragraph the first Enter already opened
8710 d.build_visual(80);
8711 d.newline();
8712 d.build_visual(80);
8713 assert_eq!(
8714 d.source, "abc\n\n\n\n",
8715 "Enter on the blank line adds one newline"
8716 );
8717 d.backspace();
8718 assert_eq!(
8719 d.source, "abc\n\n\n",
8720 "Backspace takes back exactly that one newline"
8721 );
8722 }
8723
8724 #[test]
8725 fn wysiwyg_enter_on_an_empty_list_item_exits_the_list() {
8726 let mut d = wysiwyg_doc("wys_exit", "- a\n- \n");
8727 d.caret = 6; // end of the empty "- " item
8728 d.newline();
8729 d.insert("p");
8730 assert_eq!(d.source, "- a\n\np\n");
8731 }
8732
8733 #[test]
8734 fn wysiwyg_enter_does_not_mistake_a_setext_underline_for_a_list() {
8735 // `text\n- \n` is a setext heading — the `- ` is its underline, not a
8736 // list item, though it reads as a `- ` marker byte-for-byte. Enter must
8737 // not take the list-exit path (which would splice the `- ` away as if
8738 // leaving an empty item); the AST guard sends it to a normal break and
8739 // leaves the underline intact.
8740 let mut d = wysiwyg_doc("wys_setext", "text\n- \n");
8741 assert!(
8742 d.nodes().iter().any(|n| n.kind == Kind::Heading),
8743 "precondition: twig parses this as a heading, not a list",
8744 );
8745 d.caret = 7; // on the `- ` underline line
8746 d.newline();
8747 assert!(
8748 d.source.contains("- "),
8749 "the setext underline survives, not spliced away as a list item: {:?}",
8750 d.source,
8751 );
8752 }
8753
8754 #[test]
8755 fn wysiwyg_enter_in_a_code_block_is_a_literal_newline() {
8756 let mut d = wysiwyg_doc("wys_code", "```\nabc\n```\n");
8757 d.caret = 7; // end of "abc" inside the fence
8758 d.newline();
8759 d.insert("def");
8760 assert_eq!(d.source, "```\nabc\ndef\n```\n");
8761 }
8762
8763 #[test]
8764 fn wysiwyg_enter_continues_a_block_quote() {
8765 // Enter opens a new *paragraph* inside the quote, not a second line of
8766 // the same one. `> quote\n> more` is a soft break, which under
8767 // `LineFlow::Fold` renders as a space — the keystroke would look like it
8768 // did nothing. The quoted blank line is what makes the break visible, and
8769 // it's the same thing Enter does in running prose.
8770 let mut d = wysiwyg_doc("wys_quote", "> quote\n");
8771 d.caret = 7; // end of "quote"
8772 d.newline();
8773 d.insert("more");
8774 assert_eq!(d.source, "> quote\n>\n> more\n");
8775 // Still one quote, now holding two paragraphs — not a quote and a stray
8776 // line that fell out of it.
8777 let quotes = d
8778 .nodes()
8779 .iter()
8780 .filter(|n| n.kind == Kind::BlockQuote)
8781 .count();
8782 assert_eq!(quotes, 1);
8783 }
8784
8785 #[test]
8786 fn set_block_makes_a_heading_at_the_caret() {
8787 let mut d = doc_with("head", "Title\n\nbody\n");
8788 d.caret = 0;
8789 d.set_block(BlockKind::Heading(2));
8790 assert_eq!(d.source, "## Title\n\nbody\n");
8791 d.set_block(BlockKind::Paragraph);
8792 assert_eq!(d.source, "Title\n\nbody\n");
8793 }
8794
8795 // ── block containers (quote / list) ──────────────────────────────────────
8796
8797 #[test]
8798 fn toggle_blockquote_wraps_the_block_at_the_caret_and_reverses() {
8799 let g = |m, f: fn(&mut Doc)| golden("quote", m, f);
8800 assert_eq!(g("hel|lo\n", |d| d.toggle_blockquote()), "> hel|lo\n");
8801 assert_eq!(g("> hel|lo\n", |d| d.toggle_blockquote()), "hel|lo\n");
8802 // A caret at a line end sits at the doc level; the block is still found.
8803 assert_eq!(g("hello|\n", |d| d.toggle_blockquote()), "> hello|\n");
8804 }
8805
8806 #[test]
8807 fn toggle_blockquote_keeps_the_caret_in_a_hard_wrapped_paragraph() {
8808 // Every source line of the paragraph gets its own `> `, so a caret left
8809 // on its old byte offset falls one prefix per line above it too far
8810 // back — inside the markup it just asked for rather than in its word.
8811 assert_eq!(
8812 golden("quote_wrap", "aaa\nb|bb\nccc\n", |d| d.toggle_blockquote()),
8813 "> aaa\n> b|bb\n> ccc\n"
8814 );
8815 }
8816
8817 #[test]
8818 fn toggle_blockquote_works_in_wysiwyg_view() {
8819 let g = |n, m, f: fn(&mut Doc)| golden_in(View::Wysiwyg, n, m, f);
8820 assert_eq!(
8821 g("q_wys", "hel|lo\n", |d| d.toggle_blockquote()),
8822 "> hel|lo\n"
8823 );
8824 assert_eq!(
8825 g("q_wys2", "> hel|lo\n", |d| d.toggle_blockquote()),
8826 "hel|lo\n"
8827 );
8828 }
8829
8830 #[test]
8831 fn toggle_list_makes_a_list_and_converts_between_the_kinds() {
8832 let g = |m, f: fn(&mut Doc)| golden("list", m, f);
8833 assert_eq!(g("hel|lo\n", |d| d.toggle_list(false)), "- hel|lo\n");
8834 assert_eq!(g("hel|lo\n", |d| d.toggle_list(true)), "1. hel|lo\n");
8835 // The *other* kind converts in place instead of nesting, which is what
8836 // makes the two buttons one three-state control.
8837 assert_eq!(g("- hel|lo\n", |d| d.toggle_list(true)), "1. hel|lo\n");
8838 assert_eq!(g("1. hel|lo\n", |d| d.toggle_list(false)), "- hel|lo\n");
8839 // Its own kind, over the only item the list holds, takes it off.
8840 assert_eq!(g("- hel|lo\n", |d| d.toggle_list(false)), "hel|lo\n");
8841 }
8842
8843 #[test]
8844 fn toggle_list_works_in_wysiwyg_view() {
8845 let g = |n, m, f: fn(&mut Doc)| golden_in(View::Wysiwyg, n, m, f);
8846 assert_eq!(
8847 g("l_wys", "hel|lo\n", |d| d.toggle_list(true)),
8848 "1. hel|lo\n"
8849 );
8850 assert_eq!(
8851 g("l_wys2", "1. hel|lo\n", |d| d.toggle_list(false)),
8852 "- hel|lo\n"
8853 );
8854 assert_eq!(
8855 g("l_wys3", "- hel|lo\n", |d| d.toggle_list(false)),
8856 "hel|lo\n"
8857 );
8858 }
8859
8860 #[test]
8861 fn a_list_over_a_selection_numbers_each_block_and_stays_selected() {
8862 // The selection has to grow with the markup: twig takes a container off
8863 // only a range covering every block it holds, so the second press can
8864 // reverse the first only if the result is what's selected.
8865 let mut d = doc_with("list_sel", "abc\n\ndef\n");
8866 d.select_all();
8867 d.toggle_list(true);
8868 assert_eq!(d.source, "1. abc\n\n2. def\n");
8869 assert_eq!(d.selection(), Some((0, d.source.len())));
8870 d.toggle_list(true);
8871 assert_eq!(d.source, "abc\n\ndef\n");
8872 }
8873
8874 #[test]
8875 fn toggle_blockquote_nests_a_partly_covered_quote() {
8876 // twig's rule: covering only some of a container's blocks nests, because
8877 // taking the quote off would drag its uncovered siblings out with it.
8878 let mut d = doc_with("quote_nest", "> a\n>\n> b\n");
8879 d.caret = 2; // in the first quoted paragraph only
8880 d.toggle_blockquote();
8881 assert_eq!(d.source, "> > a\n>\n> b\n");
8882 }
8883
8884 #[test]
8885 fn a_container_toggle_opens_an_empty_one_on_a_blank_line() {
8886 // A blank line used to be no block for twig to wrap —
8887 // `toggle_block_container` answered `NotFound` — so Quote and the list
8888 // buttons did nothing on the very line the H1 button works on, and leaf
8889 // lent twig a scratch paragraph to wrap and took it back out again.
8890 // twig 3.2.0 opens an empty container there itself, so what is left here
8891 // is where the caret lands: inside the marker that was just written.
8892 let mut d = doc_with("quote_blank", "\nabc\n");
8893 d.caret = 0;
8894 d.toggle_blockquote();
8895 assert_eq!(d.source, "> \nabc\n");
8896 assert_eq!(
8897 d.caret, 2,
8898 "the caret belongs inside the quote it just opened"
8899 );
8900 assert!(d.status.is_none(), "{:?}", d.status);
8901 assert!(d.dirty);
8902
8903 // And the paragraph below is still its own block: an empty container one
8904 // soft break from `abc` would take that paragraph into the quote with it.
8905 let mut d = wysiwyg_doc("quote_blank_rows", "\nabc\n");
8906 d.caret = 0;
8907 d.toggle_blockquote();
8908 d.build_visual(80);
8909 assert_eq!(drawn_rows(&d), ["│ ", "", "abc"]);
8910
8911 // The same from the other side: a blank line directly under a paragraph
8912 // earns the blank line an empty block needs, rather than being read as a
8913 // soft break inside that paragraph.
8914 let mut d = doc_with("list_blank_below", "abc\n");
8915 d.caret = 4;
8916 d.toggle_list(false);
8917 assert_eq!(d.source, "abc\n\n- ");
8918 assert_eq!(d.caret, 7);
8919 }
8920
8921 #[test]
8922 fn enter_at_the_end_of_a_quote_stays_in_the_quote() {
8923 // The gesture the rendering fix is for. `newline` inside a quote already
8924 // wrote the right source — `> a\n` becomes `> a\n>\n> \n`, twig's own
8925 // spelling — but the two marker lines it adds belonged to no node until
8926 // twig 3.2.0, so the gutter stopped at `a` and the line the writer had
8927 // just made drew as plain prose under the quote.
8928 let mut d = wysiwyg_doc("quote_enter", "> a\n");
8929 d.caret = 3; // past `a`, at the end of the quoted line
8930 d.newline();
8931 assert_eq!(d.source, "> a\n>\n> \n");
8932 d.build_visual(80);
8933 assert_eq!(drawn_rows(&d), ["│ a", "│ ", "│ "]);
8934 // And the caret is on the new line, not stranded on the old one.
8935 assert_eq!(d.caret, 8);
8936 }
8937
8938 #[test]
8939 fn opening_a_container_on_a_blank_line_is_one_undo_step() {
8940 // It was three edits — scratch, wrap, unscratch — coalesced into one, and
8941 // now it is twig's single edit. Either way one ⌘z has to put the blank
8942 // line back rather than undoing into a half-built document.
8943 for open in [
8944 &(|d: &mut Doc| d.toggle_blockquote()) as &dyn Fn(&mut Doc),
8945 &|d: &mut Doc| d.toggle_list(false),
8946 &|d: &mut Doc| d.toggle_list(true),
8947 ] {
8948 let mut d = doc_with("container_blank_undo", "a\n\n\n\nb\n");
8949 d.caret = 3;
8950 open(&mut d);
8951 assert_ne!(d.source, "a\n\n\n\nb\n");
8952 d.undo();
8953 assert_eq!(d.source, "a\n\n\n\nb\n");
8954 }
8955 }
8956
8957 #[test]
8958 fn a_container_toggle_is_one_undo_step() {
8959 let mut d = doc_with("quote_undo", "hello\n");
8960 d.caret = 3;
8961 d.insert("X"); // a typing run the structural edit must not fold into
8962 d.toggle_blockquote();
8963 assert_eq!(d.source, "> helXlo\n");
8964 d.undo();
8965 assert_eq!(d.source, "helXlo\n");
8966 }
8967
8968 // ── links ────────────────────────────────────────────────────────────────
8969
8970 #[test]
8971 fn insert_link_wraps_the_selection_and_leaves_its_text_selected() {
8972 let mut d = doc_with("link_sel", "word here\n");
8973 d.anchor = Some(0);
8974 d.caret = 4;
8975 d.insert_link("http://x.dev");
8976 assert_eq!(d.source, "[word](http://x.dev) here\n");
8977 // The text, not the destination — so a second press re-points the link
8978 // the first one made rather than nesting one inside it.
8979 assert_eq!(d.selected_text(), Some("word"));
8980 d.insert_link("http://y.dev");
8981 assert_eq!(d.source, "[word](http://y.dev) here\n");
8982 assert_eq!(d.selected_text(), Some("word"));
8983 }
8984
8985 #[test]
8986 fn insert_image_at_the_caret_spells_the_markup_and_lands_past_it() {
8987 let mut d = doc_with("img_caret", "before after\n");
8988 d.caret = 7; // between "before " and "after"
8989 d.insert_image("cat.png", "a cat");
8990 assert_eq!(d.source, "before after\n");
8991 // The caret sits just past the inserted image, nothing selected.
8992 assert_eq!(d.selection(), None);
8993 assert_eq!(d.caret, 7 + "".len());
8994 }
8995
8996 /// The bug a real vault hit: a filename with spaces in it. Markdown ends a
8997 /// destination at the first space, so the `format!` this used to be wrote
8998 /// something that was not an image at all — and the reader saw the markup as
8999 /// text. twig owns the spelling now, and moves it into the angle form.
9000 #[test]
9001 fn insert_image_spells_a_destination_with_spaces_so_it_stays_an_image() {
9002 let mut d = doc_with("img_space", "x\n");
9003 d.caret = 0;
9004 d.insert_image("Jesus Commands the Apostles to Rest.jpg", "");
9005 assert_eq!(
9006 d.source,
9007 "x\n"
9008 );
9009 // And it reads back as an image pointing at the unescaped path — the angle
9010 // brackets are spelling, not part of the destination.
9011 d.caret = 2;
9012 assert_eq!(
9013 d.image_destination_at_caret(),
9014 Some("Jesus Commands the Apostles to Rest.jpg".to_string())
9015 );
9016 }
9017
9018 /// A `)` in a caption or a filename must not close the image early.
9019 #[test]
9020 fn insert_image_escapes_a_paren_in_either_half() {
9021 let mut d = doc_with("img_paren", "x\n");
9022 d.caret = 0;
9023 d.insert_image("a)b.png", "");
9024 assert_eq!(d.source, "b.png)x\n");
9025 d.caret = 2;
9026 assert_eq!(d.image_destination_at_caret(), Some("a)b.png".to_string()));
9027 }
9028
9029 #[test]
9030 fn insert_image_uses_the_selection_as_alt_text() {
9031 let mut d = doc_with("img_sel", "caption here\n");
9032 d.anchor = Some(0);
9033 d.caret = 7; // "caption"
9034 d.insert_image("p.png", "ignored fallback");
9035 assert_eq!(d.source, " here\n");
9036 }
9037
9038 #[test]
9039 fn insert_image_with_no_alt_leaves_empty_brackets() {
9040 let mut d = doc_with("img_noalt", "\n");
9041 d.caret = 0;
9042 d.insert_image("logo.svg", "");
9043 assert_eq!(d.source, "\n");
9044 }
9045
9046 #[test]
9047 fn insert_media_spells_a_video_as_html_and_reads_it_back_as_a_block() {
9048 // The round trip is the point: it's no use writing markup the reader
9049 // can't pick up again. This is the pair that only holds from twig 2.5.1
9050 // on — before it, the one-line form went in fine and came back as a
9051 // paragraph of raw tags, publishing no media at all.
9052 let mut d = doc_with("vid_rt", "\n");
9053 d.caret = 0;
9054 d.insert_media(MediaKind::Video, "clip.mp4", "a clip");
9055 assert_eq!(
9056 d.source,
9057 "<video src=\"clip.mp4\" controls>a clip</video>\n"
9058 );
9059
9060 d.build_visual(80);
9061 assert_eq!(d.vmap.media.len(), 1, "reads back as one block media");
9062 assert_eq!(d.vmap.media[0].kind, MediaKind::Video);
9063 assert_eq!(d.vmap.media[0].destination, "clip.mp4");
9064 assert_eq!(d.vmap.media[0].alt, "a clip");
9065 }
9066
9067 #[test]
9068 fn insert_media_spells_audio_with_its_own_tag() {
9069 let mut d = doc_with("aud_rt", "\n");
9070 d.caret = 0;
9071 d.insert_media(MediaKind::Audio, "take.mp3", "");
9072 assert_eq!(d.source, "<audio src=\"take.mp3\" controls></audio>\n");
9073 d.build_visual(80);
9074 assert_eq!(d.vmap.media[0].kind, MediaKind::Audio);
9075 }
9076
9077 #[test]
9078 fn insert_media_uses_the_selection_as_fallback_text() {
9079 // The same courtesy `insert_image` does with alt: select a caption,
9080 // insert, and the caption labels the thing rather than being replaced.
9081 let mut d = doc_with("vid_sel", "the talk here\n");
9082 d.anchor = Some(0);
9083 d.caret = 8; // "the talk"
9084 d.insert_media(MediaKind::Video, "talk.mp4", "ignored fallback");
9085 assert_eq!(
9086 d.source,
9087 "<video src=\"talk.mp4\" controls>the talk</video> here\n"
9088 );
9089 }
9090
9091 #[test]
9092 fn insert_media_with_an_image_kind_is_just_insert_image() {
9093 let mut d = doc_with("img_via_media", "\n");
9094 d.caret = 0;
9095 d.insert_media(MediaKind::Image, "logo.svg", "x");
9096 assert_eq!(d.source, "\n");
9097 }
9098
9099 // ── thematic breaks ─────────────────────────────────────────────────────
9100
9101 /// The node the source parses as at `caret` — what confirms an inserted
9102 /// `---` actually reads back as a rule, not stray text or a setext heading.
9103 ///
9104 /// The *narrowest* node covering the offset. Every ancestor covers it too,
9105 /// and since twig 2.8 that includes the `doc` root, which now carries a real
9106 /// span (it reported none before, so taking the first match used to land on
9107 /// the block by luck and now always answers `"doc"`).
9108 fn kind_at(d: &mut Doc, caret: usize) -> Option<Kind> {
9109 d.nodes()
9110 .into_iter()
9111 .filter(|n| n.span.start <= caret && caret < n.span.end)
9112 .min_by_key(|n| n.span.end - n.span.start)
9113 .map(|n| n.kind)
9114 }
9115
9116 #[test]
9117 fn a_task_box_toggles_at_the_caret_and_reads_back() {
9118 let mut d = doc_with("task_toggle", "- [ ] todo\n- [x] done\n");
9119 d.caret = 8; // inside "todo"
9120 assert_eq!(d.task_checked_at_caret(), Some(false));
9121 d.toggle_task_checked();
9122 assert_eq!(d.source, "- [x] todo\n- [x] done\n");
9123 assert_eq!(d.task_checked_at_caret(), Some(true));
9124 d.toggle_task_checked();
9125 assert_eq!(d.source, "- [ ] todo\n- [x] done\n");
9126 }
9127
9128 #[test]
9129 fn a_click_toggles_a_box_without_taking_the_caret_with_it() {
9130 // The whole reason `toggle_task_at` exists apart from the caret form:
9131 // ticking a box elsewhere must not move the cursor out of what's being
9132 // typed.
9133 let mut d = doc_with("task_click", "- [ ] first\n- [ ] second\n");
9134 d.caret = 8; // inside "first"
9135 let second = d.source.find("second").unwrap();
9136 d.toggle_task_at(second);
9137 assert_eq!(d.source, "- [ ] first\n- [x] second\n");
9138 assert_eq!(d.caret, 8, "the caret stayed in the first item");
9139 }
9140
9141 #[test]
9142 fn a_plain_item_gains_and_loses_a_box() {
9143 let mut d = doc_with("task_mint", "- plain\n");
9144 d.caret = 4;
9145 assert_eq!(d.task_checked_at_caret(), None);
9146 d.toggle_task_item();
9147 assert_eq!(d.source, "- [ ] plain\n");
9148 assert_eq!(
9149 d.task_checked_at_caret(),
9150 Some(false),
9151 "a new box arrives unticked"
9152 );
9153 d.toggle_task_item();
9154 assert_eq!(d.source, "- plain\n");
9155 }
9156
9157 #[test]
9158 fn ticking_a_box_that_isnt_there_reports_rather_than_minting_one() {
9159 // `set checked` must not silently convert a bullet into a task — that is
9160 // `toggle_task_item`'s job, and twig refuses it here.
9161 let mut d = doc_with("task_none", "- plain\n");
9162 d.caret = 4;
9163 d.toggle_task_checked();
9164 assert_eq!(d.source, "- plain\n", "nothing written");
9165 assert!(
9166 d.status.is_some(),
9167 "the refusal should reach the status line"
9168 );
9169 }
9170
9171 #[test]
9172 fn a_task_item_in_a_quote_is_found_past_the_quote_marker() {
9173 let mut d = doc_with("task_quote", "> - [ ] nested\n");
9174 d.caret = d.source.find("nested").unwrap();
9175 assert_eq!(d.task_checked_at_caret(), Some(false));
9176 d.toggle_task_checked();
9177 assert_eq!(d.source, "> - [x] nested\n");
9178 }
9179
9180 #[test]
9181 fn insert_thematic_break_parts_the_paragraph_around_the_caret() {
9182 // A rule is a block, so twig's `insert_thematic_break` alone lands it
9183 // after the whole paragraph. `split_block` parts the paragraph first and
9184 // the rule is aimed at the *first* half, which is what a rule button is
9185 // understood to do — and what leaf spelled by hand until twig grew both
9186 // halves of the gesture.
9187 let mut d = doc_with("hr_mid", "before after\n");
9188 d.caret = 7; // between "before " and "after"
9189 d.insert_thematic_break();
9190 assert_eq!(d.source, "before \n\n---\n\nafter\n");
9191 assert_eq!(d.selection(), None);
9192 assert_eq!(
9193 kind_at(&mut d, "before \n\n".len()),
9194 Some(Kind::ThematicBreak)
9195 );
9196 }
9197
9198 #[test]
9199 fn insert_thematic_break_at_a_paragraph_s_end_splits_nothing() {
9200 // At the end there is nothing to part, and a split there writes the
9201 // separator anyway — a blank line and the empty slot the next paragraph
9202 // would fill — which the rule then landed above: `para\n\n* * *\n\n\n`,
9203 // two blank lines nothing fills. Now the rule lands after the paragraph,
9204 // where the split-and-aim was sending it regardless. Both formats, and
9205 // both shapes of a last line — terminated, and still being typed —
9206 // because the two reach the split through different doors: Markdown's
9207 // paragraph span stops before its newline, so `para\n` at 4 never split
9208 // there, but `para` at 4 did.
9209 for (fmt, rule) in [(Format::Markdown, "---"), (Format::Djot, "* * *")] {
9210 for src in ["para\n", "para"] {
9211 let mut d = Doc::from_source(src.into(), fmt).unwrap();
9212 d.caret = 4;
9213 d.insert_thematic_break();
9214 assert_eq!(d.source, format!("para\n\n{rule}\n"), "{fmt:?} {src:?}");
9215 assert_eq!(d.caret, d.source.len());
9216 }
9217 // Mid-document the slot sat between the rule and the next block.
9218 let mut d = Doc::from_source("para\n\nnext\n".into(), fmt).unwrap();
9219 d.caret = 4;
9220 d.insert_thematic_break();
9221 assert_eq!(d.source, format!("para\n\n{rule}\n\nnext\n"), "{fmt:?}");
9222 // Trailing whitespace is nothing to part either.
9223 let mut d = Doc::from_source("para \n".into(), fmt).unwrap();
9224 d.caret = 4;
9225 d.insert_thematic_break();
9226 assert_eq!(d.source, format!("para \n\n{rule}\n"), "{fmt:?}");
9227 }
9228 }
9229
9230 #[test]
9231 fn insert_thematic_break_at_a_paragraph_s_start_lands_before_it() {
9232 // The split at the start parts nothing, but it is kept on purpose:
9233 // `|para` becomes `\npara` with the caret on a blank line, and twig
9234 // (3.5.2) writes a rule aimed at a blank line ON that line — the only
9235 // way "before the paragraph" is reachable through a gesture that only
9236 // places after. Before 3.5.2 this came out as `\n\n---\n\npara`.
9237 for (fmt, rule) in [(Format::Markdown, "---"), (Format::Djot, "* * *")] {
9238 let mut d = Doc::from_source("para\n".into(), fmt).unwrap();
9239 d.caret = 0;
9240 d.insert_thematic_break();
9241 assert_eq!(d.source, format!("{rule}\n\npara\n"), "{fmt:?}");
9242 let mut d = Doc::from_source("prev\n\npara\n".into(), fmt).unwrap();
9243 d.caret = 6;
9244 d.insert_thematic_break();
9245 assert_eq!(d.source, format!("prev\n\n{rule}\n\npara\n"), "{fmt:?}");
9246 }
9247 }
9248
9249 #[test]
9250 fn insert_thematic_break_on_a_blank_line_takes_that_line() {
9251 // The gap between two blocks is where a click lands the caret; the
9252 // rule goes on the blank, one blank each side.
9253 let mut d = doc_with("hr_gap", "a\n\nb\n");
9254 d.caret = 2;
9255 d.insert_thematic_break();
9256 assert_eq!(d.source, "a\n\n---\n\nb\n");
9257 }
9258
9259 #[test]
9260 fn insert_table_at_a_paragraph_s_end_splits_nothing() {
9261 // The same door as the rule's, through the placement they share.
9262 let mut d = Doc::from_source("para\n".into(), Format::Djot).unwrap();
9263 d.caret = 4;
9264 d.insert_table(1, 1);
9265 assert_eq!(d.source, "para\n\n| |\n|---|\n| |\n");
9266 let mut d = doc_with("table_end_typed", "para");
9267 d.caret = 4;
9268 d.insert_table(1, 1);
9269 assert_eq!(d.source, "para\n\n| |\n| --- |\n| |\n");
9270 assert!(d.caret_in_table());
9271 }
9272
9273 #[test]
9274 fn insert_thematic_break_spells_the_rule_the_format_s_own_way() {
9275 // The whole point of delegating: `---` is Markdown's, `* * *` is djot's,
9276 // and leaf wrote the first into both until twig started spelling it.
9277 let mut md = doc_with("hr_md", "para\n");
9278 md.caret = 2;
9279 md.insert_thematic_break();
9280 assert_eq!(md.source, "pa\n\n---\n\nra\n");
9281
9282 let mut dj = Doc::from_source("para\n".into(), Format::Djot).unwrap();
9283 dj.caret = 2;
9284 dj.insert_thematic_break();
9285 assert_eq!(dj.source, "pa\n\n* * *\n\nra\n");
9286 }
9287
9288 #[test]
9289 fn insert_table_parts_the_paragraph_and_lands_in_the_first_header_cell() {
9290 // The table goes *at* the caret the way the rule does: the paragraph is
9291 // parted first, and twig writes the grid after its first half. The
9292 // caret then sits in the first header cell — selected, as Tab would
9293 // leave it — so the next keystroke is the heading.
9294 let mut d = doc_with("table_mid", "before after\n");
9295 d.caret = 7;
9296 d.insert_table(2, 3);
9297 assert_eq!(
9298 d.source,
9299 "before \n\n| | | |\n| --- | --- | --- |\n| | | |\n| | | |\n\nafter\n"
9300 );
9301 assert!(d.caret_in_table());
9302 let first_bar = d.source.find('|').unwrap();
9303 assert!(
9304 d.caret > first_bar && d.caret < d.source.find("| ---").unwrap(),
9305 "caret {} is not in the header row",
9306 d.caret
9307 );
9308 d.insert("Name");
9309 assert!(d.source.starts_with("before \n\n| Name | | |\n"));
9310 // And the grid the table was written into is one the table keys walk
9311 // (over the map a frontend rebuilds after every edit).
9312 d.build_visual(80);
9313 assert!(d.cell_tab(true));
9314 d.insert("Qty");
9315 assert!(d.source.starts_with("before \n\n| Name | Qty | |\n"));
9316 }
9317
9318 #[test]
9319 fn insert_table_spells_the_grid_the_format_s_own_way() {
9320 // Djot's delimiter row is unpadded, and leaf never has to know that.
9321 let mut dj = Doc::from_source("para\n".into(), Format::Djot).unwrap();
9322 dj.caret = 2;
9323 dj.insert_table(1, 2);
9324 assert_eq!(dj.source, "pa\n\n| | |\n|---|---|\n| | |\n\nra\n");
9325 assert!(dj.caret_in_table());
9326 }
9327
9328 #[test]
9329 fn insert_table_refuses_where_the_format_spells_no_table() {
9330 let mut d = Doc::from_source("<p>ab</p>\n".into(), Format::Html).unwrap();
9331 d.caret = 4;
9332 d.insert_table(1, 1);
9333 assert_eq!(d.source, "<p>ab</p>\n");
9334 assert!(d.status.as_deref().unwrap_or("").contains("not supported"));
9335 assert!(!d.capabilities().table);
9336 }
9337
9338 #[test]
9339 fn insert_table_reports_a_zero_shape_and_writes_nothing() {
9340 let mut d = doc_with("table_zero", "para\n");
9341 d.caret = 2;
9342 d.insert_table(0, 2);
9343 assert_eq!(d.source, "para\n");
9344 assert!(d.status.as_deref().unwrap_or("").starts_with("table:"));
9345 }
9346
9347 #[test]
9348 fn clicking_below_a_final_thematic_break_can_type_after_it() {
9349 let mut d = wysiwyg_doc("hr_final_click", "---\n");
9350 d.build_visual(80);
9351 d.click(d.vmap.num_rows() + 2, 0, false);
9352 assert_eq!(d.caret, d.source.len(), "the caret belongs after the rule");
9353 d.insert("after");
9354 assert_eq!(d.source, "---\nafter");
9355 }
9356
9357 #[test]
9358 fn enter_in_a_nested_list_item_keeps_the_new_item_nested() {
9359 // The same bytes are two documents. In Markdown ` - b` is a nested item
9360 // and the next one belongs beside it, at its indent. In Djot a list
9361 // marker can't interrupt a paragraph, so those bytes are literal text in
9362 // item `a` and there is only one item — writing ` - ` under it would add
9363 // no item at all, just more text, and the new sibling has to go to
9364 // column zero. Both spellings come out of the *enclosing item's* line.
9365 let mut md = wysiwyg_doc("enter_nested_md", "- a\n - b\n");
9366 md.caret = "- a\n - b".len();
9367 md.newline();
9368 assert_eq!(md.source, "- a\n - b\n - \n");
9369 assert_eq!(list_items(&mut md), 3);
9370
9371 let mut dj = Doc::from_source("- a\n - b\n".into(), Format::Djot).unwrap();
9372 dj.view = View::Wysiwyg;
9373 dj.build_visual(80);
9374 dj.caret = "- a\n - b".len();
9375 dj.newline();
9376 assert_eq!(dj.source, "- a\n - b\n- \n");
9377 assert_eq!(list_items(&mut dj), 2);
9378
9379 // Where Djot's nesting is real — opened by a blank line — the indent is
9380 // reproduced there too, and the two formats agree again.
9381 let mut dj = Doc::from_source("- a\n\n - b\n".into(), Format::Djot).unwrap();
9382 dj.view = View::Wysiwyg;
9383 dj.build_visual(80);
9384 dj.caret = "- a\n\n - b".len();
9385 dj.newline();
9386 assert_eq!(dj.source, "- a\n\n - b\n - \n");
9387 assert_eq!(list_items(&mut dj), 3);
9388 }
9389
9390 #[test]
9391 fn tab_nests_an_item_at_the_column_its_own_marker_asks_for() {
9392 // Tab replaces the line's whole prefix with the one twig spells, so the
9393 // quote markers, the parent's indent and an ordered marker's extra
9394 // column are all its answer rather than leaf's arithmetic.
9395 for (name, body, caret, want) in [
9396 ("bullet", "- a\n- b\n", 6, "- a\n - b\n"),
9397 ("ordered", "1. a\n2. b\n", 8, "1. a\n 1. b\n"),
9398 ("quoted", "> - a\n> - b\n", 10, "> - a\n> - b\n"),
9399 // A checkbox is markup the item's own text wraps past, but a nested
9400 // list may only open at the *list* marker's column — four in from
9401 // there is a paragraph continuation, and `- [ ] a\n - [ ] b`
9402 // parses as one item, not two.
9403 ("task", "- [ ] a\n- [ ] b\n", 14, "- [ ] a\n - [ ] b\n"),
9404 (
9405 "quoted task",
9406 "> - [ ] a\n> - [ ] b\n",
9407 18,
9408 "> - [ ] a\n> - [ ] b\n",
9409 ),
9410 ] {
9411 let mut doc = wysiwyg_doc(name, body);
9412 doc.caret = caret;
9413 doc.indent();
9414 assert_eq!(doc.source, want, "{name}");
9415 // The nesting is real, not just indented text.
9416 assert_eq!(list_items(&mut doc), 2, "{name}");
9417 }
9418 }
9419
9420 #[test]
9421 fn backspace_only_outdents_where_the_format_says_there_is_an_item() {
9422 // The same bytes, the two formats disagreeing, and a gesture that used
9423 // to read the bytes. ` - b` is a nested item in Markdown, so Backspace
9424 // at its marker outdents. In Djot a marker can't interrupt a paragraph,
9425 // so those bytes are literal text inside item `a` — there is nothing to
9426 // outdent, and treating them as a marker turned one item into two, a
9427 // structural edit from a keystroke that should delete one character.
9428 //
9429 // twig's `line_prefix` is what tells them apart: it reports the marker
9430 // on the Markdown line and nothing on the Djot one, which is a
9431 // continuation. No byte scan can reach that answer.
9432 let src = "- a\n - b\n";
9433 let at = "- a\n - ".len();
9434
9435 let mut md = Doc::from_source(src.into(), Format::Markdown).unwrap();
9436 md.view = View::Wysiwyg;
9437 md.build_visual(80);
9438 md.caret = at;
9439 md.backspace();
9440 assert_eq!(md.source, "- a\n- b\n");
9441 assert_eq!(list_items(&mut md), 2);
9442
9443 let mut dj = Doc::from_source(src.into(), Format::Djot).unwrap();
9444 dj.view = View::Wysiwyg;
9445 dj.build_visual(80);
9446 dj.caret = at;
9447 dj.backspace();
9448 assert_eq!(dj.source, "- a\n -b\n"); // an ordinary character delete
9449 assert_eq!(list_items(&mut dj), 1); // and the structure is untouched
9450 }
9451
9452 #[test]
9453 fn enter_in_a_checklist_item_starts_another_unchecked_one() {
9454 // Leaf used to spell the next item from the marker bytes it scanned, and
9455 // its scanner stopped at the bullet — so Enter in a checklist wrote `- `
9456 // and dropped out of the checklist. twig reproduces the whole
9457 // continuation, and a fresh item is always unticked however the one above
9458 // it stands.
9459 for (name, body, want) in [
9460 ("unchecked", "- [ ] a\n", "- [ ] a\n- [ ] \n"),
9461 ("checked", "- [x] a\n", "- [x] a\n- [ ] \n"),
9462 ] {
9463 let mut doc = wysiwyg_doc(name, body);
9464 doc.caret = body.trim_end_matches('\n').len();
9465 doc.newline();
9466 assert_eq!(doc.source, want, "{name}");
9467 // Both items are checklist items — the new one is a box, not the
9468 // plain bullet the old marker scan left behind — and it is unticked
9469 // whichever way the one above it faces.
9470 let boxes: Vec<Option<bool>> = doc
9471 .nodes()
9472 .iter()
9473 .filter(|n| n.kind == Kind::TaskListItem)
9474 .map(|n| n.checked)
9475 .collect();
9476 assert_eq!(boxes.len(), 2, "{name}");
9477 assert_eq!(boxes[1], Some(false), "{name}");
9478 }
9479 }
9480
9481 #[test]
9482 fn a_split_takes_the_space_the_caret_was_in_front_of() {
9483 // Splicing a break at the caret strands the space the words were parted
9484 // at on the head of the second block, where it reads as an indent nobody
9485 // typed. twig's split consumes it.
9486 for (name, body, caret, want) in [
9487 ("para", "one two\n", 3, "one\n\ntwo\n"),
9488 ("item", "- one two\n", 5, "- one\n- two\n"),
9489 ("quote", "> one two\n", 5, "> one\n>\n> two\n"),
9490 // A heading takes leaf's own path, which has to match.
9491 ("heading", "# one two\n", 5, "# one\n\ntwo\n"),
9492 ] {
9493 let mut doc = wysiwyg_doc(name, body);
9494 doc.caret = caret;
9495 doc.newline();
9496 assert_eq!(doc.source, want, "{name}");
9497 }
9498 }
9499
9500 #[test]
9501 fn enter_at_the_end_of_a_heading_opens_a_paragraph() {
9502 // The one place leaf keeps its own break: `split_block` repeats the `#`,
9503 // and Enter after a title is how the body under it is asked for.
9504 let mut doc = wysiwyg_doc("head_enter", "# Title\n");
9505 doc.caret = "# Title".len();
9506 doc.newline();
9507 doc.insert("body");
9508 assert_eq!(doc.source, "# Title\n\nbody\n");
9509 assert_eq!(
9510 doc.nodes()
9511 .iter()
9512 .filter(|n| n.kind == Kind::Heading)
9513 .count(),
9514 1
9515 );
9516 }
9517
9518 #[test]
9519 fn enter_in_a_quoted_list_item_starts_the_next_quoted_item() {
9520 // A quoted item's marker doesn't open its line, so a scan that starts at
9521 // column zero finds a `>` where it wanted a bullet, calls the line "not a
9522 // list" and hands Enter to the plain-quote branch — which writes `> ` and
9523 // drops the list. The next item has to carry the whole prefix.
9524 for (name, body, want) in [
9525 ("flat", "> - a\n", "> - a\n> - \n"),
9526 ("sibling", "> - a\n> - b\n", "> - a\n> - b\n> - \n"),
9527 ("nested", "> - a\n> - b\n", "> - a\n> - b\n> - \n"),
9528 ("ordered", "> 1. a\n> 2. b\n", "> 1. a\n> 2. b\n> 3. \n"),
9529 ("twice quoted", "> > - a\n", "> > - a\n> > - \n"),
9530 ] {
9531 let mut doc = wysiwyg_doc(name, body);
9532 doc.caret = body.trim_end_matches('\n').len();
9533 doc.newline();
9534 assert_eq!(doc.source, want, "{name}");
9535 // The marker isn't just spelled right, it parses as an item.
9536 assert_eq!(list_items(&mut doc), body.lines().count() + 1, "{name}");
9537 }
9538 }
9539
9540 #[test]
9541 fn an_empty_quoted_item_leaves_the_list_and_stays_in_the_quote() {
9542 // Double-Enter exits the list. Unquoted that means a blank line, but a
9543 // *bare* blank line would end the quote too and drop the caret out of it,
9544 // so the separator keeps its `>` and the caret's line keeps its `> `.
9545 let mut doc = wysiwyg_doc("quoted_exit", "> - a\n> - \n");
9546 doc.caret = "> - a\n> - ".len();
9547 doc.newline();
9548 assert_eq!(doc.source, "> - a\n>\n> \n");
9549 assert_eq!(list_items(&mut doc), 1);
9550 // What "still in the quote" means for the next keystroke: the caret sits
9551 // behind the prefix, and what's typed there lands inside the quote as a
9552 // paragraph of its own — not as more of item `a`.
9553 doc.insert("x");
9554 assert_eq!(doc.source, "> - a\n>\n> x\n");
9555 assert!(
9556 doc.editor
9557 .ancestors_at(doc.caret - 1)
9558 .is_ok_and(|c| c.into_iter().any(|m| m.kind == Kind::BlockQuote))
9559 );
9560 }
9561
9562 #[test]
9563 fn backspace_at_a_quoted_marker_takes_the_marker_and_leaves_the_quote() {
9564 // The marker is hidden block markup, so Backspace over it is structural —
9565 // but only the marker is the list's. Splicing from the line start would
9566 // take the `>` with it and silently unquote the line.
9567 let mut doc = wysiwyg_doc("quoted_bksp", "> - a\n");
9568 doc.caret = "> - ".len();
9569 doc.backspace();
9570 assert_eq!(doc.source, "> a\n");
9571 assert_eq!(list_items(&mut doc), 0);
9572
9573 // A nested one outdents instead, moving the bullet within the quote
9574 // rather than moving the quote.
9575 let mut doc = wysiwyg_doc("quoted_outdent", "> - a\n> - b\n");
9576 doc.caret = "> - a\n> - ".len();
9577 doc.backspace();
9578 assert_eq!(doc.source, "> - a\n> - b\n");
9579 assert_eq!(list_items(&mut doc), 2);
9580 }
9581
9582 #[test]
9583 fn only_a_bare_paragraph_is_parted_around_the_caret() {
9584 // The split is deliberately narrow. Parting a fenced block would leave
9585 // two fences with a rule between them, and parting a list item would
9586 // mint an item nobody asked for on the way to a rule that lands after
9587 // the list either way — so both keep the whole block intact and take the
9588 // rule after it. A caret in a quote is likewise left alone.
9589 for (name, body, caret, want) in [
9590 (
9591 "code",
9592 "```\nfn x() {}\n```\n",
9593 8,
9594 "```\nfn x() {}\n```\n\n---\n",
9595 ),
9596 ("list", "- one two\n", 6, "- one two\n\n---\n"),
9597 ("quote", "> one two\n", 6, "> one two\n>\n> ---\n"),
9598 ] {
9599 let mut d = doc_with(&format!("hr_narrow_{name}"), body);
9600 d.caret = caret;
9601 d.insert_thematic_break();
9602 assert_eq!(d.source, want, "{name}: the block should stay whole");
9603 }
9604 }
9605
9606 #[test]
9607 fn insert_thematic_break_replaces_the_selection() {
9608 // Now that the rule lands *at* the caret again, replacing the selection
9609 // is coherent once more: the text goes, and the rule takes its place.
9610 // The space the deletion left leading the second half is consumed by the
9611 // split rather than opening the new paragraph with it.
9612 let mut d = doc_with("hr_sel", "one two three\n");
9613 d.anchor = Some(4);
9614 d.caret = 7; // "two"
9615 d.insert_thematic_break();
9616 assert_eq!(d.source, "one \n\n---\n\nthree\n");
9617 assert_eq!(d.selection(), None);
9618 }
9619
9620 #[test]
9621 fn insert_thematic_break_clears_a_code_block_and_a_table_rather_than_refusing() {
9622 // Both are blocks the rule lands *after*. Leaf used to refuse a fence,
9623 // because writing `---` into one is code, not a rule — twig now walks out
9624 // to the block that owns the caret's line, so there is nothing to refuse.
9625 let mut code = doc_with("hr_code", "```\nfn x() {}\n```\n");
9626 code.caret = 5; // inside the fenced code
9627 code.insert_thematic_break();
9628 assert_eq!(code.source, "```\nfn x() {}\n```\n\n---\n");
9629 assert_eq!(code.status, None, "no refusal to report any more");
9630
9631 let mut table = doc_with("hr_table", "| a | b |\n|---|---|\n| 1 | 2 |\n");
9632 table.caret = 3; // in the header row
9633 table.insert_thematic_break();
9634 assert_eq!(table.source, "| a | b |\n|---|---|\n| 1 | 2 |\n\n---\n");
9635 }
9636
9637 #[test]
9638 fn insert_thematic_break_in_a_list_item_ends_the_list() {
9639 // The un-indented rule cannot continue the list, so it closes the list
9640 // and lands at the top level rather than nested inside it.
9641 let mut d = doc_with("hr_list", "- one\n- two\n");
9642 d.caret = "- one\n- tw".len(); // mid "two"
9643 d.insert_thematic_break();
9644 d.build_visual(80);
9645 let rule_at = d.source.find("---").unwrap();
9646 assert_eq!(kind_at(&mut d, rule_at), Some(Kind::ThematicBreak));
9647 assert!(
9648 !d.nodes().iter().any(|n| n.kind == Kind::BulletList
9649 && n.span.start <= rule_at
9650 && rule_at < n.span.end),
9651 "the rule must not be nested inside the list"
9652 );
9653 }
9654
9655 #[test]
9656 fn insert_thematic_break_in_a_blockquote_stays_in_the_quote() {
9657 // Leaf used to end the quote. twig gives the rule the quote's own prefix,
9658 // which is the document the gesture was actually asked for.
9659 let mut d = doc_with("hr_quote", "> hello\n");
9660 d.caret = 4; // inside the quoted text
9661 d.insert_thematic_break();
9662 assert_eq!(d.source, "> hello\n>\n> ---\n");
9663 d.build_visual(80);
9664 let rule_at = d.source.find("---").unwrap();
9665 assert_eq!(kind_at(&mut d, rule_at), Some(Kind::ThematicBreak));
9666 assert!(
9667 d.nodes().iter().any(|n| n.kind == Kind::BlockQuote
9668 && n.span.start <= rule_at
9669 && rule_at < n.span.end),
9670 "the rule belongs to the quote it was asked for"
9671 );
9672 }
9673
9674 // ── typing against a block picture ────────────────────────────────────────
9675
9676 /// A rendered-view document with the caret parked on one of the picture's two
9677 /// stops, and the map already built — the state a frontend is in between
9678 /// drawing a frame and the next keystroke.
9679 fn doc_at_picture(name: &str, src: &str, side: MediaStop) -> Doc {
9680 let mut d = doc_in(View::Wysiwyg, name, src);
9681 d.build_visual_unwrapped();
9682 let start = src.find("".len(),
9686 };
9687 d
9688 }
9689
9690 /// The block media the map publishes, after rebuilding it — "is this still a
9691 /// picture, or has it become a line of text with an image in it?"
9692 fn media_count(d: &mut Doc) -> usize {
9693 d.build_visual_unwrapped();
9694 d.vmap.media.len()
9695 }
9696
9697 #[test]
9698 fn typing_past_a_block_picture_opens_a_paragraph_under_it() {
9699 // The accident this prevents: tap the blank page under a photo (which
9700 // lands on the picture's trailing stop), type, and `xy` is a
9701 // paragraph with an *inline* image — the photo stops being drawn.
9702 let mut d = doc_at_picture("pic_after", "hi\n\n\n", MediaStop::After);
9703 d.insert("xy");
9704 assert_eq!(d.source, "hi\n\n\n\nxy\n");
9705 assert_eq!(media_count(&mut d), 1, "still a picture");
9706 }
9707
9708 #[test]
9709 fn typing_in_front_of_a_block_picture_opens_a_paragraph_above_it() {
9710 let mut d = doc_at_picture("pic_before", "hi\n\n\n", MediaStop::Before);
9711 d.insert("xy");
9712 assert_eq!(d.source, "hi\n\nxy\n\n\n");
9713 assert_eq!(media_count(&mut d), 1);
9714 }
9715
9716 #[test]
9717 fn a_picture_that_opens_the_document_still_takes_a_paragraph_above_it() {
9718 let mut d = doc_at_picture("pic_first", "\n", MediaStop::Before);
9719 d.insert("x");
9720 assert_eq!(d.source, "x\n\n\n");
9721 assert_eq!(media_count(&mut d), 1);
9722 }
9723
9724 #[test]
9725 fn one_undo_puts_the_picture_back_the_way_it_was_found() {
9726 // The opened paragraph is part of the keystroke, not an edit the writer
9727 // made — so it undoes with the character, not a step later.
9728 let mut d = doc_at_picture("pic_undo", "hi\n\n\n", MediaStop::After);
9729 d.insert("x");
9730 assert_eq!(d.source, "hi\n\n\n\nx\n");
9731 d.undo();
9732 assert_eq!(d.source, "hi\n\n\n");
9733 }
9734
9735 #[test]
9736 fn pasting_against_a_block_picture_opens_a_paragraph_too() {
9737 // ⌘V dissolves the picture exactly as a keystroke does.
9738 let mut d = doc_at_picture("pic_paste", "hi\n\n\n", MediaStop::After);
9739 d.paste("pasted");
9740 assert_eq!(d.source, "hi\n\n\n\npasted\n");
9741 assert_eq!(media_count(&mut d), 1);
9742 }
9743
9744 #[test]
9745 fn typing_beside_an_inline_image_is_ordinary_editing() {
9746 // An inline image has no placeholder row and no stops of its own. Opening
9747 // a paragraph mid-sentence would be the bug, not the fix.
9748 let mut d = doc_in(View::Wysiwyg, "pic_inline", "see  here\n");
9749 d.build_visual_unwrapped();
9750 d.caret = "see ".len();
9751 d.insert("!");
9752 assert_eq!(d.source, "see ! here\n");
9753 }
9754
9755 #[test]
9756 fn source_view_types_raw_markup_against_an_image_untouched() {
9757 // Source view is for writing the markup itself; a break inserted behind
9758 // the writer's back there would be the editor arguing with them.
9759 let mut d = doc_in(View::Source, "pic_src", "\n");
9760 d.caret = "".len();
9761 d.insert("x");
9762 assert_eq!(d.source, "x\n");
9763 }
9764
9765 #[test]
9766 fn typing_over_a_selection_that_starts_at_a_picture_stop_replaces_it() {
9767 // A selection is replaced, not joined into, so there is nothing to
9768 // protect: the range takes the picture with it.
9769 let mut d = doc_at_picture("pic_sel", "hi\n\n\n", MediaStop::Before);
9770 d.anchor = Some(d.caret);
9771 d.caret = d.source.find("".len();
9772 d.insert("x");
9773 assert_eq!(d.source, "hi\n\nx\n");
9774 }
9775
9776 #[test]
9777 fn backspace_past_a_block_picture_deletes_the_picture_not_its_last_byte() {
9778 // What this actually cost: a real vault's photo, to one stray Backspace.
9779 // The caret past `` was deleting the closing paren — invisible
9780 // in the rendered view — and the photo became the text `\n", MediaStop::After);
9782 d.backspace();
9783 assert_eq!(d.source, "hi\n");
9784 assert_eq!(media_count(&mut d), 0, "the picture went, in one piece");
9785 d.undo();
9786 assert_eq!(
9787 d.source, "hi\n\n\n",
9788 "and comes back in one piece"
9789 );
9790 }
9791
9792 #[test]
9793 fn backspace_in_front_of_a_block_picture_steps_out_instead_of_merging_it() {
9794 // Deleting the break here would join the picture to the paragraph above,
9795 // where it is an *inline* image and stops being drawn. Step over the
9796 // boundary; the next press deletes in the paragraph the caret reached.
9797 let mut d = doc_at_picture("pic_bs_before", "hi\n\n\n", MediaStop::Before);
9798 d.backspace();
9799 assert_eq!(d.source, "hi\n\n\n", "nothing deleted");
9800 assert_eq!(d.caret, 2, "the caret stepped up to the end of `hi`");
9801 d.backspace();
9802 assert_eq!(d.source, "h\n\n\n", "and now it deletes there");
9803 assert_eq!(media_count(&mut d), 1, "the picture was never at risk");
9804 }
9805
9806 #[test]
9807 fn forward_delete_in_front_of_a_block_picture_deletes_the_picture() {
9808 // The mirror. A byte-step here eats the `!` and leaves a link.
9809 let mut d = doc_at_picture("pic_del", "hi\n\n\n\nbye\n", MediaStop::Before);
9810 d.delete_forward();
9811 assert_eq!(d.source, "hi\n\nbye\n");
9812 assert_eq!(media_count(&mut d), 0);
9813 }
9814
9815 #[test]
9816 fn forward_delete_past_a_block_picture_steps_over_the_boundary() {
9817 let mut d = doc_at_picture(
9818 "pic_del_after",
9819 "hi\n\n\n\nbye\n",
9820 MediaStop::After,
9821 );
9822 d.delete_forward();
9823 assert_eq!(d.source, "hi\n\n\n\nbye\n", "nothing deleted");
9824 assert_eq!(
9825 d.caret,
9826 d.source.find("bye").unwrap(),
9827 "the caret stepped down to `bye`"
9828 );
9829 }
9830
9831 #[test]
9832 fn a_picture_that_is_the_whole_document_still_deletes_cleanly() {
9833 let mut d = doc_at_picture("pic_only", "\n", MediaStop::After);
9834 d.backspace();
9835 assert_eq!(d.source, "\n");
9836 assert_eq!(media_count(&mut d), 0);
9837 }
9838
9839 #[test]
9840 fn a_word_delete_takes_the_picture_whole_or_steps_out_of_it() {
9841 // ⌥⌫ past a picture would otherwise eat a "word" of its markup.
9842 let mut d = doc_at_picture("pic_wordbs", "hi there\n\n\n", MediaStop::After);
9843 d.delete_word_back();
9844 assert_eq!(d.source, "hi there\n");
9845
9846 // And in front of one it runs *through* the paragraph break into the
9847 // prose above, which merges the picture inline — so it steps out first,
9848 // and the second press deletes the word it was aimed at.
9849 let mut d = doc_at_picture("pic_wordbs2", "hi there\n\n\n", MediaStop::Before);
9850 d.delete_word_back();
9851 assert_eq!(d.source, "hi there\n\n\n");
9852 d.delete_word_back();
9853 assert_eq!(
9854 d.source, "hi \n\n\n",
9855 "the word above went, the picture stayed"
9856 );
9857 assert_eq!(media_count(&mut d), 1);
9858 }
9859
9860 #[test]
9861 fn source_view_deletes_raw_markup_against_an_image_untouched() {
9862 let mut d = doc_in(View::Source, "pic_src_del", "\n");
9863 d.caret = "".len();
9864 d.backspace();
9865 assert_eq!(d.source, ";
9866 }
9867
9868 #[test]
9869 fn image_destination_at_caret_reads_the_image_under_the_caret() {
9870 let mut d = doc_with("img_read", "\n");
9871 d.caret = 3; // inside the image markup
9872 assert_eq!(d.image_destination_at_caret(), Some("cat.png".to_string()));
9873 // Past the image, the caret is in no image.
9874 d.caret = "".len();
9875 assert_eq!(d.image_destination_at_caret(), None);
9876 }
9877
9878 #[test]
9879 fn set_media_rows_reserves_blank_filler_rows_the_frontend_paints_over() {
9880 // The image is one placeholder row by default, and `set_media_rows` grows
9881 // it to the height the frontend measured: the label row plus blank
9882 // `decoration` fillers that hold the vertical space a raster is drawn into.
9883 let mut d = wysiwyg_doc("img_rows", "intro\n\n\n\nend\n");
9884 assert_eq!(d.vmap.media.len(), 1);
9885 let img_row = d.vmap.media[0].rows_span.start;
9886 assert_eq!(
9887 d.vmap.media[0].rows_span,
9888 img_row..img_row + 1,
9889 "default is one row"
9890 );
9891
9892 d.set_media_rows(HashMap::from([("cat.png".to_string(), 4)]));
9893 d.build_visual(80);
9894 assert_eq!(d.vmap.media.len(), 1, "still one image, now taller");
9895 let span = d.vmap.media[0].rows_span.clone();
9896 assert_eq!(span.end - span.start, 4, "reserves the four rows asked for");
9897 // The label row carries the mark and its glyphs; the three below are blank
9898 // decoration — drawn, but no caret and no text.
9899 assert!(
9900 d.vmap.rows[span.start].media.is_some(),
9901 "mark rides the first row"
9902 );
9903 for r in (span.start + 1)..span.end {
9904 assert!(d.vmap.rows[r].decoration, "filler row {r} is decoration");
9905 assert!(d.vmap.rows[r].glyphs.is_empty(), "filler row {r} is blank");
9906 assert!(
9907 d.vmap.rows[r].media.is_none(),
9908 "only the first row is marked"
9909 );
9910 }
9911 }
9912
9913 #[test]
9914 fn a_taller_image_adds_no_caret_stops_and_motion_steps_over_its_fillers() {
9915 // The extra rows are pure spacers: the caret's only homes stay the stop in
9916 // front of the image and the one just past it, so walking the document top
9917 // to bottom visits the same offsets whether the image is 1 row or 5.
9918 let body = "ab\n\n\n\ncd\n";
9919 let stops_at = |rows: usize| -> Vec<usize> {
9920 let mut d = wysiwyg_doc("img_stops", body);
9921 if rows > 1 {
9922 d.set_media_rows(HashMap::from([("p.png".to_string(), rows)]));
9923 d.build_visual(80);
9924 }
9925 d.caret = 0;
9926 let mut seen = vec![d.caret];
9927 loop {
9928 d.move_right(false);
9929 if *seen.last().unwrap() == d.caret {
9930 break;
9931 }
9932 seen.push(d.caret);
9933 }
9934 seen
9935 };
9936 assert_eq!(
9937 stops_at(1),
9938 stops_at(5),
9939 "reserving rows must not add stops"
9940 );
9941 }
9942
9943 #[test]
9944 fn insert_link_repoints_the_link_at_a_bare_caret() {
9945 let mut d = doc_with("link_repoint", "[word](http://x.dev)\n");
9946 d.caret = 3; // in the link's text, nothing selected
9947 d.insert_link("http://y.dev");
9948 assert_eq!(d.source, "[word](http://y.dev)\n");
9949 assert_eq!(d.selected_text(), Some("word"));
9950 }
9951
9952 #[test]
9953 fn insert_link_on_an_empty_range_autolinks_a_url() {
9954 // A link with no text of its own is an autolink, and twig spells it —
9955 // `<…>` is the canonical form and needs no text typed into it, so the
9956 // caret lands after it rather than selecting a finished link.
9957 let mut d = doc_with("link_empty", "\n");
9958 d.caret = 0;
9959 d.insert_link("http://x.dev");
9960 assert_eq!(d.source, "<http://x.dev>\n");
9961 assert_eq!(d.selection(), None);
9962 assert_eq!(d.caret, 14);
9963 }
9964
9965 #[test]
9966 fn insert_link_on_an_empty_range_falls_back_for_a_non_url() {
9967 // `<./notes.md>` is literal text in both formats and `<foo>` is raw HTML
9968 // in Markdown, so a destination that can't autolink doubles as the text
9969 // instead — which is then selected, ready to be typed over.
9970 let mut d = doc_with("link_rel", "\n");
9971 d.caret = 0;
9972 d.insert_link("./notes.md");
9973 assert_eq!(d.source, "[./notes.md](./notes.md)\n");
9974 assert_eq!(d.selection(), Some((1, 11)));
9975 d.insert("Notes");
9976 assert_eq!(d.source, "[Notes](./notes.md)\n");
9977 }
9978
9979 #[test]
9980 fn insert_link_repoints_the_autolink_the_caret_stands_in() {
9981 // The autolink's text is its URL, so re-pointing replaces the whole
9982 // node — the caret must not splice a second link inside the first.
9983 let mut d = doc_with("link_repoint_auto", "see <https://x.dev> ok\n");
9984 d.caret = 10;
9985 d.insert_link("https://y.dev");
9986 assert_eq!(d.source, "see <https://y.dev> ok\n");
9987 }
9988
9989 #[test]
9990 fn code_language_reads_and_edits_through_the_fence() {
9991 let mut d = doc_with("code_lang", "```rust\nlet x = 1;\n```\n");
9992 d.caret = 10; // inside the code body
9993 assert_eq!(d.code_language_at_caret().as_deref(), Some("rust"));
9994 assert!(d.caret_in_fenced_code());
9995
9996 d.set_code_language("python");
9997 assert!(
9998 d.source.starts_with("```python\n"),
9999 "source: {:?}",
10000 d.source
10001 );
10002 assert_eq!(d.code_language_at_caret().as_deref(), Some("python"));
10003
10004 // Clearing it leaves a bare fence and no label.
10005 d.set_code_language("");
10006 assert!(d.source.starts_with("```\n"), "source: {:?}", d.source);
10007 assert_eq!(d.code_language_at_caret(), None);
10008
10009 // A caret outside any code block edits nothing.
10010 let mut p = doc_with("code_lang_none", "just prose\n");
10011 assert!(!p.caret_in_fenced_code());
10012 p.set_code_language("rust");
10013 assert_eq!(p.source, "just prose\n");
10014 }
10015
10016 #[test]
10017 fn a_language_the_fence_cannot_carry_is_refused_not_written() {
10018 // Markdown's info string ends at whitespace, so `two words` would write
10019 // a fence that reads back with a different language than the one asked
10020 // for. twig refuses it; leaf reports that and leaves the source alone.
10021 // The old splice trimmed the ends and wrote whatever was left.
10022 let mut d = doc_with("code_lang_bad", "```rust\nx\n```\n");
10023 d.caret = 10;
10024 d.set_code_language("two words");
10025 assert_eq!(d.source, "```rust\nx\n```\n", "source should be untouched");
10026 assert!(d.status.is_some(), "the refusal should be reported");
10027 assert_eq!(d.code_language_at_caret().as_deref(), Some("rust"));
10028 }
10029
10030 #[test]
10031 fn link_destination_at_caret_reads_both_spellings() {
10032 let mut d = doc_with("link_dest", "see [t](https://x.dev) ok\n");
10033 d.caret = 5;
10034 assert_eq!(
10035 d.link_destination_at_caret().as_deref(),
10036 Some("https://x.dev")
10037 );
10038 d.caret = 0;
10039 assert_eq!(d.link_destination_at_caret(), None);
10040
10041 // An autolink has no `destination`; its text is the URL.
10042 let mut a = doc_with("link_dest_auto", "see <https://x.dev> ok\n");
10043 a.caret = 10;
10044 assert_eq!(
10045 a.link_destination_at_caret().as_deref(),
10046 Some("https://x.dev")
10047 );
10048 a.caret = 21;
10049 assert_eq!(a.link_destination_at_caret(), None);
10050 }
10051
10052 #[test]
10053 fn locate_finds_the_block_a_declared_id_names() {
10054 // The Book of Mormon shape: one document per chapter, one `{#v…}` per
10055 // verse. The locator has to land on the *verse*, which is the whole
10056 // reason a link carries one.
10057 let src = "{#v1}\nI, Nephi, having been born of goodly parents.\n\n\
10058 {#v2}\nYea, I make a record in the language of my father.\n";
10059 let mut d = Doc::from_source(src.to_string(), Format::Djot).unwrap();
10060 let v2 = d.locate("v2").expect("the document declares `{#v2}`");
10061 assert_eq!(
10062 d.source[v2.start..v2.end].trim_end(),
10063 "Yea, I make a record in the language of my father."
10064 );
10065 // The attribute line is not part of it: `start` is a place to put a
10066 // caret, and `{#v2}` is markup the caret has no business landing in.
10067 assert!(d.source[..v2.start].ends_with("{#v2}\n"));
10068 assert_eq!(d.locate("v99"), None);
10069 }
10070
10071 #[test]
10072 fn locate_reads_a_heading_by_its_words_when_the_format_mints_no_ids() {
10073 // Markdown has no ids at all — twig mints none, and `{#custom}` in a
10074 // Markdown heading is literal text. So `#the-second-part` can only be
10075 // the heading's own words, which is the rule every Markdown renderer
10076 // already follows and therefore the one a link was authored against.
10077 let src = "# Title\n\nintro\n\n## The Second Part\n\nbody\n\n## Third\n\nmore\n";
10078 let mut d = doc_with("locate_md", src);
10079 let hit = d.locate("the-second-part").expect("the heading's slug");
10080 assert!(d.source[hit.start..].starts_with("## The Second Part"));
10081 // Bounded by the next heading that isn't under it, so a peek shows the
10082 // section rather than only its title.
10083 assert_eq!(
10084 &d.source[hit.start..hit.end],
10085 "## The Second Part\n\nbody\n\n"
10086 );
10087
10088 // A subsection does not end its parent: `# Title` runs to `## Third`'s
10089 // sibling only because there is no other `#`, so it covers the lot.
10090 let title = d.locate("title").expect("the top heading");
10091 assert_eq!(title.end, d.source.len());
10092 }
10093
10094 #[test]
10095 fn locate_reads_a_djot_auto_id_however_the_link_spelled_it() {
10096 // djot mints `Some-Heading-Here`; a link to it is written
10097 // `#some-heading-here` by nearly everything that writes links. Both
10098 // spellings are one question.
10099 let src = "## Some Heading Here\n\nbody\n";
10100 let mut d = Doc::from_source(src.to_string(), Format::Djot).unwrap();
10101 let exact = d.locate("Some-Heading-Here").expect("djot's own spelling");
10102 let slugged = d.locate("some-heading-here").expect("the link's spelling");
10103 assert_eq!(exact, slugged);
10104 // The section, not the heading line — there is more to show than a title.
10105 assert_eq!(&d.source[exact.start..exact.end], src);
10106 }
10107
10108 #[test]
10109 fn locate_ignores_an_empty_locator_and_one_that_slugs_to_nothing() {
10110 let mut d = doc_with("locate_empty", "# Title\n\nbody\n");
10111 assert_eq!(d.locate(""), None);
10112 assert_eq!(d.locate(" "), None);
10113 // All punctuation: it names nothing, and must not be read as "match the
10114 // first heading whose slug is also empty".
10115 assert_eq!(d.locate("!!!"), None);
10116 }
10117
10118 #[test]
10119 fn locate_gives_a_duplicated_id_to_the_first_block_that_claims_it() {
10120 // The document's mistake, and the answer every other anchor
10121 // implementation gives — the alternative is for a link to mean whichever
10122 // of the two a walk happened to reach first.
10123 let src = "{#dup}\nfirst.\n\n{#dup}\nsecond.\n";
10124 let mut d = Doc::from_source(src.to_string(), Format::Djot).unwrap();
10125 let hit = d.locate("dup").expect("the first `{#dup}`");
10126 assert_eq!(d.source[hit.start..hit.end].trim_end(), "first.");
10127 }
10128
10129 #[test]
10130 fn insert_footnote_writes_both_halves_and_lands_the_caret_in_the_note() {
10131 // The button's whole job: a reference where the caret was, a definition
10132 // to give it meaning, and the caret waiting in the empty note so the
10133 // next keystroke is the note's first word.
10134 let mut d = doc_with("fn_insert", "A claim and more.\n");
10135 d.caret = 7; // just past "A claim"
10136 d.insert_footnote();
10137 assert!(
10138 d.source.starts_with("A claim[^1] and more."),
10139 "{:?}",
10140 d.source
10141 );
10142 assert!(
10143 d.source.contains("[^1]:"),
10144 "the definition too: {:?}",
10145 d.source
10146 );
10147 assert_eq!(d.status, None);
10148
10149 let reference = d.source.find("[^1]").unwrap();
10150 let note = d
10151 .footnote_at(reference + 2)
10152 .expect("the reference just written");
10153 assert_eq!(note.label, "1");
10154 assert_eq!(note.text.as_deref(), Some(""), "the note starts empty");
10155 assert_eq!(Some(d.caret), note.offset, "the caret waits in the note");
10156 // …and typing there is typing into the note, not near it.
10157 d.insert("the note");
10158 assert_eq!(
10159 d.footnote_at(reference + 2).and_then(|f| f.text),
10160 Some("the note".to_string())
10161 );
10162 }
10163
10164 #[test]
10165 fn insert_footnote_numbers_past_the_notes_already_written() {
10166 // A second press must not hand back a label somebody else is using: twig
10167 // reuses a defined label rather than appending a rival definition, so a
10168 // repeat of `1` would quietly point the new reference at the old note.
10169 let mut d = doc_with("fn_insert_number", "One[^1] two.\n\n[^1]: first\n");
10170 d.caret = 7; // past `[^1]`, before " two."
10171 d.insert_footnote();
10172 assert!(d.source.starts_with("One[^1][^2] two."), "{:?}", d.source);
10173 assert_eq!(d.source.matches("[^2]:").count(), 1);
10174 }
10175
10176 #[test]
10177 fn insert_footnote_counts_a_dangling_reference_and_ignores_a_named_one() {
10178 // `[^2]` with no definition is still a 2 that means something to whoever
10179 // wrote it — stepping over it would mint a note for their reference. A
10180 // word label takes no number, so it blocks none.
10181 let mut d = doc_with("fn_insert_dangling", "a[^2] b[^why] c\n\n[^why]: named\n");
10182 d.caret = d.source.find(" c").unwrap();
10183 d.insert_footnote();
10184 assert!(d.source.contains("[^1]:"), "1 is free: {:?}", d.source);
10185 assert!(
10186 d.source.starts_with("a[^2] b[^why][^1] c"),
10187 "{:?}",
10188 d.source
10189 );
10190 }
10191
10192 #[test]
10193 fn insert_footnote_marks_the_selection_rather_than_replacing_it() {
10194 // A reference annotates the words before it. Consuming the selection —
10195 // which is what an insert normally does — would delete the very claim
10196 // the author selected in order to footnote.
10197 let mut d = doc_with("fn_insert_sel", "A claim and more.\n");
10198 d.anchor = Some(2);
10199 d.caret = 7; // "claim" selected
10200 d.insert_footnote();
10201 assert!(
10202 d.source.starts_with("A claim[^1] and more."),
10203 "{:?}",
10204 d.source
10205 );
10206 }
10207
10208 #[test]
10209 fn a_note_just_written_still_knows_where_its_reference_is() {
10210 // The authoring loop in one test: press the button, type the note, ask to
10211 // go back. The caret ends at the note's last byte — which is the *end* of
10212 // the definition's span, the one offset the query used to exclude — so
10213 // this is where the round trip either works or doesn't.
10214 let mut d = doc_with("fn_insert_return", "A claim and more.\n");
10215 d.caret = 7;
10216 d.insert_footnote();
10217 d.insert("the note");
10218 assert_eq!(d.source, "A claim[^1] and more.\n\n[^1]: the note\n");
10219 let back = d
10220 .footnote_definition_at_caret()
10221 .expect("still in the note we just typed");
10222 assert_eq!(back.label, "1");
10223 // …and following it lands on the reference's label, where a reader's
10224 // return leg lands.
10225 assert_eq!(back.offset, Some(9));
10226 assert_eq!(&d.source[9..10], "1");
10227 }
10228
10229 #[test]
10230 fn insert_footnote_takes_one_undo_for_both_halves() {
10231 // twig writes the pair as a single edit; the point of that is here.
10232 let before = "A claim and more.\n";
10233 let mut d = doc_with("fn_insert_undo", before);
10234 d.caret = 7;
10235 d.insert_footnote();
10236 assert_ne!(d.source, before);
10237 d.undo();
10238 assert_eq!(d.source, before, "one undo takes back both halves");
10239 }
10240
10241 #[test]
10242 fn insert_footnote_refuses_a_format_that_cannot_spell_one() {
10243 // HTML is authorable — it spells the inline marks — and has no footnote.
10244 // The refusal says so rather than writing brackets that would render as
10245 // brackets.
10246 let src = "<p>A claim.</p>\n";
10247 let mut d = Doc::from_source(src.to_string(), Format::Html).unwrap();
10248 assert!(!Capabilities::of(Format::Html).footnote);
10249 d.caret = 5;
10250 d.insert_footnote();
10251 assert_eq!(d.source, src, "nothing written");
10252 assert!(d.status.is_some_and(|s| s.starts_with("footnote:")));
10253 }
10254
10255 #[test]
10256 fn insert_footnote_leaves_the_caret_on_a_real_stop_in_the_rich_view() {
10257 // The empty body is the one place this could go wrong: the definition
10258 // renders as a `[1] ` marker the caret cannot occupy, so a caret aimed a
10259 // byte early would draw up in the paragraph above the note it belongs to.
10260 let mut d = doc_in(View::Wysiwyg, "fn_insert_stop", "A claim and more.\n");
10261 d.place_caret(7, false);
10262 d.insert_footnote();
10263 d.build_visual(80); // the frame a frontend draws after the edit
10264 assert_eq!(
10265 d.vmap.snap_to_stop(d.caret),
10266 d.caret,
10267 "the caret sits on a stop"
10268 );
10269 let (row, _) = d.caret_pos();
10270 assert!(
10271 drawn_rows(&d)[row].contains("[1]"),
10272 "the caret is on the note's row, not above it: {:?}",
10273 drawn_rows(&d)
10274 );
10275 }
10276
10277 #[test]
10278 fn footnote_at_caret_resolves_a_reference_to_its_note() {
10279 // `[^1]` spans 7..11; its label byte is at 9. The definition follows a
10280 // blank line, as one has to.
10281 let mut d = doc_with("fn_at_caret", "A claim[^1] and more.\n\n[^1]: the note\n");
10282 d.caret = 9;
10283 let f = d
10284 .footnote_at_caret()
10285 .expect("the caret stands in a reference");
10286 assert_eq!(f.label, "1");
10287 assert_eq!(f.text.as_deref(), Some("the note"));
10288 // The offset points at the note's first word, not at the definition's
10289 // `[` — the marker is decoration with no caret stop on it.
10290 assert_eq!(f.offset, Some(29));
10291 assert_eq!(&d.source[29..37], "the note");
10292 // …and `end` closes the range, so a frontend can ask which rendered rows
10293 // the note occupies rather than re-deriving them from the text.
10294 assert_eq!(f.end, Some(37));
10295 assert_eq!(&d.source[f.offset.unwrap()..f.end.unwrap()], "the note");
10296 }
10297
10298 /// Two definitions in a row: each is its own note, and neither reaches into
10299 /// the other.
10300 ///
10301 /// A djot definition's span used to run past the blank line into the first
10302 /// byte of whatever followed, so this answered `"first note.\n\n["` — and the
10303 /// offsets named the *next* note's rows too, showing a reader two footnotes
10304 /// when they had asked about one. twig 3.1 ends the span after the block's
10305 /// own last line; the test outlives the workaround leaf carried for it.
10306 #[test]
10307 fn footnote_at_stops_a_note_at_the_definition_after_it() {
10308 let src = "Claim[^2a] and [^2b].\n\n[^2a]: first note.\n\n[^2b]: second note.\n";
10309 for format in [Format::Markdown, Format::Djot] {
10310 let mut d = Doc::from_source(src.to_string(), format).unwrap();
10311 d.caret = 7;
10312 let f = d.footnote_at_caret().expect("a reference");
10313 assert_eq!(f.text.as_deref(), Some("first note."), "in {format:?}");
10314 assert_eq!(
10315 &src[f.offset.unwrap()..f.end.unwrap()],
10316 "first note.",
10317 "in {format:?}"
10318 );
10319 }
10320 }
10321
10322 /// The other side of that boundary: a blank line *inside* a definition is
10323 /// interior to it, and the note keeps its second paragraph.
10324 ///
10325 /// This is what the old body scan cost. It stopped at the first line not
10326 /// indented under the note — a blank line is not — so a two-paragraph note
10327 /// came back as its first paragraph, and "go to note" framed half of it.
10328 /// Reading the span twig gives is both simpler and right.
10329 #[test]
10330 fn footnote_at_keeps_a_notes_second_paragraph() {
10331 let src = "Claim[^1].\n\n[^1]: first para.\n\n second para.\n\nAfter.\n";
10332 let mut d = Doc::from_source(src.to_string(), Format::Djot).unwrap();
10333 d.caret = 7;
10334 let f = d.footnote_at_caret().expect("a reference");
10335 assert_eq!(f.text.as_deref(), Some("first para.\n\n second para."));
10336 // And it stops there — `After.` is the next block, not more note.
10337 assert_eq!(
10338 &src[f.offset.unwrap()..f.end.unwrap()],
10339 f.text.as_deref().unwrap()
10340 );
10341 assert!(!f.text.as_deref().unwrap().contains("After"));
10342 }
10343
10344 #[test]
10345 fn footnote_at_bounds_a_note_whose_body_is_empty() {
10346 // `[^1]:` with nothing after it. The range is empty rather than
10347 // inverted, and still points inside the definition — which is what keeps
10348 // a frontend's row lookup from walking off into the block above.
10349 let src = "A claim[^1].\n\n[^1]:\n";
10350 let mut d = doc_with("fn_empty_body", src);
10351 d.caret = 9;
10352 let f = d.footnote_at_caret().expect("a reference");
10353 assert_eq!(f.text.as_deref(), Some(""));
10354 assert_eq!(f.offset, f.end, "an empty note is an empty range");
10355 assert!(f.offset.unwrap() >= src.find("[^1]:").unwrap());
10356 }
10357
10358 #[test]
10359 fn footnote_at_caret_ignores_a_caret_that_stands_in_no_reference() {
10360 let mut d = doc_with(
10361 "fn_at_caret_none",
10362 "A claim[^1] and more.\n\n[^1]: the note\n",
10363 );
10364 d.caret = 2; // in the prose
10365 assert_eq!(d.footnote_at_caret(), None);
10366 }
10367
10368 #[test]
10369 fn footnote_at_caret_is_not_a_link_query_and_vice_versa() {
10370 // The two are deliberately separate: a reference names a note in this
10371 // document, a link names somewhere to leave for, and answering one with
10372 // the other is what made a reference click do nothing at all.
10373 let mut d = doc_with("fn_vs_link", "a[^1] b [t](https://x.dev)\n\n[^1]: note\n");
10374 d.caret = 3; // the `1` of `[^1]`
10375 assert!(d.footnote_at_caret().is_some());
10376 assert_eq!(
10377 d.link_destination_at_caret(),
10378 None,
10379 "a reference is not a link"
10380 );
10381
10382 d.caret = 10; // inside the link's label
10383 assert_eq!(d.footnote_at_caret(), None, "a link is not a reference");
10384 assert_eq!(
10385 d.link_destination_at_caret().as_deref(),
10386 Some("https://x.dev")
10387 );
10388 }
10389
10390 #[test]
10391 fn footnote_at_caret_reports_an_undefined_reference_rather_than_nothing() {
10392 // A `[^99]` the document never defines is a real state — a note deleted
10393 // out from under its reference — and the label is what lets a frontend
10394 // say so. `None` here would be indistinguishable from "not on a
10395 // reference", which is the wrong thing to tell a reader.
10396 let mut d = doc_with("fn_undefined", "A claim[^99] and more.\n");
10397 d.caret = 9;
10398 let f = d
10399 .footnote_at_caret()
10400 .expect("the reference is still a reference");
10401 assert_eq!(f.label, "99");
10402 assert_eq!(f.text, None);
10403 assert_eq!(f.offset, None);
10404 }
10405
10406 #[test]
10407 fn footnote_at_caret_reads_a_word_label_and_a_multiline_note() {
10408 // Labels are not always numbers, and a note's body runs past its first
10409 // line — the indented continuation belongs to the note, so it comes back
10410 // with it (source bytes, verbatim, as documented).
10411 let src = "see[^note] here\n\n[^note]: first line\n second line\n";
10412 let mut d = doc_with("fn_word_label", src);
10413 d.caret = 6;
10414 let f = d
10415 .footnote_at_caret()
10416 .expect("the caret stands in a reference");
10417 assert_eq!(f.label, "note");
10418 assert_eq!(f.text.as_deref(), Some("first line\n second line"));
10419 }
10420
10421 #[test]
10422 fn footnote_at_answers_for_an_offset_the_caret_is_nowhere_near() {
10423 // The point of the offset form: a pointer hovering a reference asks what
10424 // note it names, and must not drag the caret along to ask.
10425 let mut d = doc_with("fn_at_off", "A claim[^1] and more.\n\n[^1]: the note\n");
10426 d.caret = 0;
10427 let f = d.footnote_at(9).expect("offset 9 stands in the reference");
10428 assert_eq!(f.label, "1");
10429 assert_eq!(f.text.as_deref(), Some("the note"));
10430 assert_eq!(d.caret, 0, "asking must not move the caret");
10431 assert_eq!(d.footnote_at(2), None, "offset 2 is prose");
10432 }
10433
10434 #[test]
10435 fn footnote_definition_at_caret_points_back_at_the_reference() {
10436 // The return leg. `[^1]` spans 7..11, so its label — the only byte of it
10437 // the caret can rest on — is at 9.
10438 let mut d = doc_with("fn_def", "A claim[^1] and more.\n\n[^1]: the note\n");
10439 d.caret = 30; // inside the note's body
10440 let f = d
10441 .footnote_definition_at_caret()
10442 .expect("the caret stands in a definition");
10443 assert_eq!(f.label, "1");
10444 assert_eq!(f.offset, Some(9));
10445 assert_eq!(&d.source[7..11], "[^1]");
10446 }
10447
10448 #[test]
10449 fn footnote_definition_at_covers_where_a_go_to_note_actually_lands() {
10450 // The two legs have to meet: wherever `footnote_at` sends the caret, the
10451 // definition query must answer for — otherwise arriving at a note leaves
10452 // the reader somewhere the way back isn't offered.
10453 let src = "A claim[^1] and more.\n\n[^1]: the note\n";
10454 let mut d = doc_with("fn_def_marker", src);
10455 let landed = d.footnote_at(9).unwrap().offset.unwrap();
10456 assert_eq!(
10457 d.footnote_definition_at(landed).and_then(|f| f.offset),
10458 Some(9),
10459 "the note a reference sends you to offers the way back"
10460 );
10461 }
10462
10463 #[test]
10464 fn footnote_definition_at_caret_ignores_prose_and_the_reference_itself() {
10465 // The two queries answer for disjoint places, which is what lets one
10466 // gesture mean "down to the note" in one and "back up" in the other
10467 // without either having to remember which way the reader is going.
10468 let mut d = doc_with("fn_def_none", "A claim[^1] and more.\n\n[^1]: the note\n");
10469 d.caret = 2; // prose
10470 assert_eq!(d.footnote_definition_at_caret(), None);
10471 d.caret = 9; // the reference
10472 assert_eq!(d.footnote_definition_at_caret(), None);
10473 assert!(
10474 d.footnote_at_caret().is_some(),
10475 "which is the reference's own query"
10476 );
10477 }
10478
10479 #[test]
10480 fn footnote_definition_at_caret_reports_an_orphan_note_rather_than_nothing() {
10481 // Nothing cites `[^2]`. Answering `None` would say "you are not in a
10482 // note", which is false and leaves a frontend unable to explain why the
10483 // way back is missing.
10484 let src = "A claim[^1].\n\n[^1]: cited\n\n[^2]: orphan\n";
10485 let mut d = doc_with("fn_def_orphan", src);
10486 d.caret = src.find("orphan").unwrap();
10487 let f = d
10488 .footnote_definition_at_caret()
10489 .expect("an orphan is still a definition");
10490 assert_eq!(f.label, "2");
10491 assert_eq!(f.offset, None);
10492 }
10493
10494 #[test]
10495 fn footnote_definition_at_caret_returns_to_the_first_of_repeated_references() {
10496 // One label, cited twice. The first is where the reader most likely came
10497 // from, and the only answer that doesn't depend on how they got here.
10498 let src = "One[^a] and two[^a].\n\n[^a]: the note\n";
10499 let mut d = doc_with("fn_def_repeat", src);
10500 d.caret = src.find("the note").unwrap();
10501 let f = d.footnote_definition_at_caret().expect("a definition");
10502 assert_eq!(
10503 f.offset,
10504 Some(5),
10505 "the first `[^a]`'s label, not the second's"
10506 );
10507 assert_eq!(&src[3..7], "[^a]");
10508 }
10509
10510 #[test]
10511 fn footnote_navigation_is_a_round_trip_through_placed_carets() {
10512 // Down and back up, each leg found from the document rather than from a
10513 // memory of the other — so it still works for a reader who scrolled to
10514 // the notes instead of jumping there.
10515 //
10516 // `place_caret` rather than assigning `caret`, because that is what a
10517 // frontend calls: it snaps to a real caret stop, and a jump that lands
10518 // on a byte the caret can't rest on would arrive somewhere the return
10519 // leg no longer answers for. `build_map` first, since snapping is a
10520 // no-op until the map exists — which is exactly how this went unnoticed
10521 // when the offsets pointed at the `[^` markers.
10522 let mut d = doc_with("fn_round", "A claim[^1] and more.\n\n[^1]: the note\n");
10523 d.build_map(None);
10524 d.place_caret(9, false);
10525 let down = d
10526 .footnote_at_caret()
10527 .expect("a reference")
10528 .offset
10529 .expect("a note");
10530 d.place_caret(down, false);
10531 let up = d
10532 .footnote_definition_at_caret()
10533 .expect("a definition")
10534 .offset
10535 .expect("a reference");
10536 d.place_caret(up, false);
10537 assert_eq!(d.caret, up, "the way back is a stop the caret can occupy");
10538 assert_eq!(
10539 d.footnote_at_caret().expect("back on the reference").label,
10540 "1"
10541 );
10542 }
10543
10544 #[test]
10545 fn insert_link_hands_the_destination_to_twig_raw() {
10546 // Escaping is twig's, and format-specific: Markdown ends a destination
10547 // at the first space and needs the `<…>` form, where djot would read
10548 // those angle brackets as part of the URL.
10549 let mut d = doc_with("link_space", "word\n");
10550 d.anchor = Some(0);
10551 d.caret = 4;
10552 d.insert_link("a b");
10553 assert_eq!(d.source, "[word](<a b>)\n");
10554 }
10555
10556 #[test]
10557 fn insert_link_reports_a_destination_no_format_can_carry() {
10558 let mut d = doc_with("link_bad", "word\n");
10559 d.anchor = Some(0);
10560 d.caret = 4;
10561 d.insert_link("a\nb");
10562 assert_eq!(d.source, "word\n"); // untouched, not quietly rewritten
10563 assert!(
10564 d.status.is_some(),
10565 "InvalidArgument should reach the status line"
10566 );
10567 assert!(!d.dirty);
10568 }
10569
10570 #[test]
10571 fn insert_link_works_in_wysiwyg_view() {
10572 let mut d = wysiwyg_doc("link_wys", "word here\n");
10573 d.anchor = Some(0);
10574 d.caret = 4;
10575 d.insert_link("http://x.dev");
10576 assert_eq!(d.source, "[word](http://x.dev) here\n");
10577 assert_eq!(d.selected_text(), Some("word"));
10578 // The map the caret has to keep riding is rebuilt each frame; motion
10579 // over the fresh one must still land on a real stop (the debug_assert).
10580 d.build_visual(80);
10581 d.move_right(false);
10582 d.move_left(false);
10583 }
10584
10585 #[test]
10586 fn click_maps_a_row_col_to_a_byte_offset() {
10587 let mut d = doc_with("click", "ab\ncd\n");
10588 d.click(1, 1, false); // row 1 ("cd"), col 1 -> the 'd'
10589 assert_eq!(d.caret, 4);
10590 }
10591
10592 // A pixel-hit-test placement (the GUI's `place_caret`) must land on a caret
10593 // stop just as the `(row, col)` click path does, so the caret can never come
10594 // to rest in the blank gap between two paragraphs — where it would draw in one
10595 // place and type in another.
10596 #[test]
10597 fn place_caret_snaps_out_of_the_blank_gap_between_paragraphs() {
10598 // "A\n\nB": offset 2 is the gap the paragraph break is drawn with, not a
10599 // caret stop (stops are 0,1,3,4).
10600 let mut d = wysiwyg_doc("place_gap", "A\n\nB");
10601 assert!(!d.vmap.is_stop(2), "offset 2 should be an unreachable gap");
10602 d.place_caret(2, false);
10603 assert!(d.vmap.is_stop(d.caret), "caret {} is not a stop", d.caret);
10604 assert_eq!(d.caret, 1, "should snap to the end of the paragraph above");
10605 }
10606
10607 #[test]
10608 fn place_caret_dragging_through_the_gap_keeps_selection_on_stops() {
10609 let mut d = wysiwyg_doc("place_gap_drag", "A\n\nB");
10610 d.place_caret(0, false); // anchor at the start of "A"
10611 d.place_caret(2, true); // drag into the gap
10612 assert!(d.vmap.is_stop(d.caret), "caret {} is not a stop", d.caret);
10613 let (s, e) = d.selection().expect("a selection");
10614 assert!(
10615 d.vmap.is_stop(s) && d.vmap.is_stop(e),
10616 "selection {s}..{e} off a stop"
10617 );
10618 }
10619
10620 #[test]
10621 fn place_caret_on_a_real_stop_is_left_untouched() {
10622 let mut d = wysiwyg_doc("place_stop", "A\n\nB");
10623 d.place_caret(3, false); // the start of "B" — a genuine stop
10624 assert_eq!(d.caret, 3);
10625 }
10626
10627 // An *empty paragraph* (two blank lines, an intentional blank line the user
10628 // opened) is a real caret stop, unlike the gap — a click into it must stay.
10629 #[test]
10630 fn place_caret_rests_in_an_empty_paragraph() {
10631 let mut d = wysiwyg_doc("place_empty_para", "A\n\n\n\nB");
10632 let empty = 3; // the navigable empty row's offset (stops: 0,1,3,5,6)
10633 assert!(d.vmap.is_stop(empty));
10634 d.place_caret(empty, false);
10635 assert_eq!(d.caret, empty);
10636 }
10637
10638 // The content end of a hidden mark is a home too (`VisualMap::mark_ends`):
10639 // a drag over the word `bold` ends there, and a caret placed there stays.
10640 #[test]
10641 fn place_caret_rests_at_the_end_of_a_hidden_marks_content() {
10642 let src = "| A | B |\n| --- | --- |\n| **bold** | other |\n";
10643 let mut d = wysiwyg_doc("place_mark_end", src);
10644 let start = src.find("bold").unwrap();
10645 d.place_caret(start, false);
10646 d.place_caret(start + 4, true);
10647 assert_eq!(d.selection(), Some((start, start + 4)), "the whole word");
10648 d.toggle(InlineKind::Strong);
10649 assert_eq!(d.source, src.replace("**bold**", "bold"));
10650 }
10651
10652 #[test]
10653 fn right_steps_onto_the_end_of_a_mark_and_then_past_its_delimiter() {
10654 let mut d = wysiwyg_doc("right_mark_end", "a **bold** b");
10655 d.caret = 7; // before the `d`
10656 d.move_right(false);
10657 assert_eq!(d.caret, 8, "onto the end of the bold");
10658 assert!(d.active_inline_marks().contains(InlineKind::Strong));
10659 d.move_right(false);
10660 assert_eq!(d.caret, 10, "past the closing `**`");
10661 assert!(!d.active_inline_marks().contains(InlineKind::Strong));
10662 d.move_left(false);
10663 assert_eq!(d.caret, 8);
10664 d.move_left(false);
10665 assert_eq!(d.caret, 7);
10666 // Typing at the inner home extends the bold.
10667 d.caret = 8;
10668 d.insert("!");
10669 assert_eq!(d.source, "a **bold!** b");
10670 }
10671
10672 #[test]
10673 fn a_marks_end_home_follows_an_edit_through_the_incremental_map() {
10674 // The splice path shifts the home with the block it is in, and the
10675 // re-rendered block finds its own again.
10676 let mut d = wysiwyg_doc("mark_end_splice", "x\n\na **bold** b\n\ny\n");
10677 d.build_visual_unwrapped();
10678 d.edit(0, 0, "zz");
10679 d.build_visual_unwrapped();
10680 wysiwyg::assert_maps_eq(&d.vmap, &reference_map(&d.source), "after a shift");
10681 assert!(d.vmap.is_stop(d.source.find("bold").unwrap() + 4));
10682 let at = d.source.find("bold").unwrap();
10683 d.edit(at, at, "very ");
10684 d.build_visual_unwrapped();
10685 wysiwyg::assert_maps_eq(&d.vmap, &reference_map(&d.source), "after a re-render");
10686 assert!(d.vmap.is_stop(d.source.find("bold").unwrap() + 4));
10687 }
10688
10689 fn wysiwyg_doc(name: &str, body: &str) -> Doc {
10690 doc_in(View::Wysiwyg, name, body)
10691 }
10692
10693 /// How many list items the source actually parses into — the check that a
10694 /// marker Leaf wrote is a marker the format agrees is one.
10695 fn list_items(doc: &mut Doc) -> usize {
10696 doc.editor
10697 .nodes()
10698 .unwrap()
10699 .iter()
10700 .filter(|n| n.kind == Kind::ListItem || n.kind == Kind::TaskListItem)
10701 .count()
10702 }
10703
10704 /// A from-scratch, cache-free WYSIWYG map for `source` — the ground truth the
10705 /// incremental (`build_spliced` / `build_cached`) path must always match.
10706 fn reference_map(source: &str) -> crate::wysiwyg::VisualMap {
10707 reference_map_revealing(source, None)
10708 }
10709
10710 /// [`reference_map`] with a reveal line — the ground truth for the
10711 /// `MarkupMode::Full` builds, where the map is a function of the caret's
10712 /// line as well as the text.
10713 fn reference_map_revealing(
10714 source: &str,
10715 reveal: Option<Range<usize>>,
10716 ) -> crate::wysiwyg::VisualMap {
10717 // The same parse `Doc` uses. With twig's plain defaults instead, the two
10718 // sides disagree on what the *document* is before the renderer is even
10719 // reached — a bare `:word` is a text directive to one and prose to the
10720 // other — and the mismatch reads as a splice bug that isn't one.
10721 let mut ed =
10722 twig::Editor::new_ext(source.as_bytes(), Format::Markdown, parse_extensions()).unwrap();
10723 let nodes = ed.nodes().unwrap();
10724 crate::wysiwyg::build(
10725 &nodes,
10726 source,
10727 None,
10728 false,
10729 &std::collections::HashMap::new(),
10730 reveal,
10731 )
10732 }
10733
10734 fn maps_differ(a: &crate::wysiwyg::VisualMap, b: &crate::wysiwyg::VisualMap) -> bool {
10735 if a.rows.len() != b.rows.len() {
10736 return true;
10737 }
10738 for (ra, rb) in a.rows.iter().zip(&b.rows) {
10739 if ra.end_src != rb.end_src || ra.glyphs.len() != rb.glyphs.len() {
10740 return true;
10741 }
10742 for (ga, gb) in ra.glyphs.iter().zip(&rb.glyphs) {
10743 if ga.ch != gb.ch || ga.src != gb.src {
10744 return true;
10745 }
10746 }
10747 }
10748 false
10749 }
10750
10751 #[test]
10752 fn incremental_build_matches_a_fresh_build_across_edits() {
10753 // Every `Doc` edit rebuilds through `build_spliced` (the single-block
10754 // fast path, gated on twig's `dirty_range`) or falls back to
10755 // `build_cached`. After each edit the map must be byte-identical to a
10756 // from-scratch build — this is the correctness net under the splice.
10757 let docs = [
10758 "# Title\n\nThe quick brown fox jumps.\n\nAnother paragraph here.\n\n- a\n- b\n",
10759 "para one\n\n> quote **bold** text\n> continued line\n\ntail paragraph\n",
10760 "alpha\n\nbeta\n\ngamma\n\ndelta\n\nepsilon\n\nzeta\n",
10761 // A footnote definition is a root beside `doc`, merged back into the
10762 // top-level list by `wysiwyg::top_blocks`. The random edits below
10763 // make and unmake definitions as they go (a deleted `:` turns one
10764 // back into a paragraph, and vice versa), which is exactly the
10765 // structural churn the splice path has to notice and bail out of.
10766 "text[^1] here\n\n[^1]: the note\n\nmore text[^b]\n\n[^b]: second\n",
10767 // A comment is a top-level block that draws no rows — a layout entry
10768 // at zero rows either side of blocks that do. The edits below type
10769 // into the blocks around it (a splice past a hidden block), and
10770 // break the comment open into prose and back (a structural change).
10771 "intro\n\n<!-- exec -->\n```\ncode\n```\n\nafter the comment\n\n<!-- trail -->\n",
10772 // Link reference definitions: a hidden block that an edit can turn
10773 // into a paragraph (a deleted `:`) and back, and whose own bytes an
10774 // edit can land in.
10775 "see [a] and [b]\n\n[a]: /a\n\nmid text\n\n[b]: /b\n",
10776 ];
10777 // A deterministic mix: mostly single characters (which stay inside one
10778 // block → splice), plus edits that reshape structure (a paragraph break,
10779 // a heading marker, a code fence → fallback), so both paths are exercised.
10780 let inserts = ["x", "y", "\n\n", "#", "`", " ", "z"];
10781 for src in docs {
10782 let mut d = wysiwyg_doc("diff", src);
10783 d.build_visual_unwrapped();
10784 wysiwyg::assert_maps_eq(&d.vmap, &reference_map(&d.source), "initial");
10785
10786 for step in 0..60usize {
10787 let len = d.source.len();
10788 let raw = (step * 13 + 5) % (len + 1);
10789 let pos = (raw..=len).find(|&i| d.source.is_char_boundary(i)).unwrap();
10790 let pre = d.source.clone();
10791 let action;
10792 if step % 3 == 0 && pos < len {
10793 let end = (pos + 1..=len)
10794 .find(|&i| d.source.is_char_boundary(i))
10795 .unwrap();
10796 action = format!("delete [{pos},{end})");
10797 d.edit(pos, end, "");
10798 } else {
10799 let ins = inserts[step % inserts.len()];
10800 action = format!("insert {ins:?} @ {pos}");
10801 d.edit(pos, pos, ins);
10802 }
10803 d.build_visual_unwrapped();
10804 if maps_differ(&d.vmap, &reference_map(&d.source)) {
10805 panic!(
10806 "FIRST MISMATCH at step {step}: {action}\n pre = {pre:?}\n post = {:?}",
10807 d.source
10808 );
10809 }
10810 }
10811 }
10812 }
10813
10814 /// A frontend is handed [`Doc::vmap`] and may present it differently:
10815 /// leaf-ratatui splices blank filler rows under an oversized heading so the
10816 /// raster it paints there has somewhere to stand, and leaves them in the map
10817 /// because the caret and the mouse both read it between frames. The splice
10818 /// path addresses that map by *row index*, against the block layout the last
10819 /// build recorded — so handed a map with rows in it that no block owns, it
10820 /// laid the re-rendered block over one of the fillers and carried the rows
10821 /// the block really occupied into the suffix. One stranded copy of the
10822 /// edited line, and everything below it a row further down, per keystroke.
10823 ///
10824 /// A map that isn't the one the layout describes is a map this path can't
10825 /// patch, whoever changed it and for whatever reason. It rebuilds instead.
10826 #[test]
10827 fn an_edit_over_a_map_a_frontend_reshaped_rebuilds_it_whole() {
10828 let mut d = wysiwyg_doc("reshaped", "# Title\n\nThe quick brown fox jumps.\n");
10829 d.build_visual_unwrapped();
10830
10831 // Stand in for the heading filler rows: two blank rows past the heading
10832 // that no block accounts for. Cloning a real row keeps every field
10833 // plausible — it is the row *count* the splice can't survive.
10834 let filler = d.vmap.rows[0].clone();
10835 d.vmap.rows.insert(1, filler.clone());
10836 d.vmap.rows.insert(1, filler);
10837
10838 // An edit inside the last block: the single-block case the splice path
10839 // is for, and the one the frontend hits on every keystroke.
10840 let at = d.source.len() - 1;
10841 d.edit(at, at, "!");
10842 d.build_visual_unwrapped();
10843
10844 wysiwyg::assert_maps_eq(&d.vmap, &reference_map(&d.source), "after the edit");
10845 }
10846
10847 #[test]
10848 fn incremental_build_matches_a_fresh_build_under_full_reveal() {
10849 // The same correctness net as `incremental_build_matches_a_fresh_build_
10850 // across_edits`, under `MarkupMode::Full` — where the map depends on
10851 // the caret's *line* as well as the text, so the two caches have a new
10852 // way to be wrong. Both are exercised: the block cache can hand back
10853 // rows built for a line that is no longer the revealed one, and the
10854 // splice path can reuse a suffix that still has yesterday's line raw.
10855 //
10856 // Caret motion is interleaved with the edits deliberately, because a
10857 // caret that only ever moved with the edit would never cross a line
10858 // without also dirtying it — the case where a stale reveal survives.
10859 let docs = [
10860 "# Title\n\n*one* and **two**\n\n[lk](http://x) and `code`\n\n- a *b*\n",
10861 "para *em* one\n\n> quote **bold** text\n\ntail ~~del~~ paragraph\n",
10862 ];
10863 let inserts = ["x", "*", "\n\n", "#", "`", " ", "_"];
10864 for src in docs {
10865 let mut d = wysiwyg_doc("reveal_diff", src);
10866 d.set_markup_mode(MarkupMode::Full);
10867
10868 for step in 0..60usize {
10869 let len = d.source.len();
10870 let raw = (step * 13 + 5) % (len + 1);
10871 let pos = (raw..=len).find(|&i| d.source.is_char_boundary(i)).unwrap();
10872 let pre = d.source.clone();
10873 let action;
10874 if step % 3 == 0 && pos < len {
10875 let end = (pos + 1..=len)
10876 .find(|&i| d.source.is_char_boundary(i))
10877 .unwrap();
10878 action = format!("delete [{pos},{end})");
10879 d.edit(pos, end, "");
10880 } else {
10881 let ins = inserts[step % inserts.len()];
10882 action = format!("insert {ins:?} @ {pos}");
10883 d.edit(pos, pos, ins);
10884 }
10885 // Walk the caret somewhere else in the document, independently
10886 // of where the edit landed.
10887 let want = (step * 29 + 11) % (d.source.len() + 1);
10888 d.caret = (want..=d.source.len())
10889 .find(|&i| d.source.is_char_boundary(i))
10890 .unwrap();
10891 d.build_visual_unwrapped();
10892
10893 let want = reference_map_revealing(&d.source, d.reveal_line());
10894 if maps_differ(&d.vmap, &want) {
10895 panic!(
10896 "FIRST MISMATCH at step {step}: {action}, caret {}\n pre = {pre:?}\n post = {:?}",
10897 d.caret, d.source
10898 );
10899 }
10900 }
10901 }
10902 }
10903
10904 #[test]
10905 fn caret_motion_across_lines_rebuilds_only_under_full() {
10906 // The cache-key change has to earn its keep in both directions: `Full`
10907 // must rebuild when the caret changes line (or the reveal would never
10908 // move), and the hidden modes must *not* (or every arrow key would pay
10909 // for a feature they don't use). The existing `cache_motion` test pins
10910 // the second for the default mode; this pins the pair against a mode
10911 // change alone.
10912 let body = "*one* here\n\n*two* there\n";
10913
10914 let mut full = doc_in(View::Wysiwyg, "motion_full", body);
10915 full.set_markup_mode(MarkupMode::Full);
10916 caret_at(&mut full, "one");
10917 let before = full.revision();
10918 caret_at(&mut full, "two");
10919 assert_eq!(full.revision(), before, "motion is not an edit");
10920 assert!(
10921 drawn_rows(&full).iter().any(|r| r == "*two* there"),
10922 "the map followed the caret: {:?}",
10923 drawn_rows(&full)
10924 );
10925
10926 let mut hidden = doc_in(View::Wysiwyg, "motion_hidden", body);
10927 caret_at(&mut hidden, "one");
10928 let key = hidden.vmap_key.clone();
10929 caret_at(&mut hidden, "two");
10930 assert_eq!(
10931 hidden.vmap_key, key,
10932 "a hidden mode rebuilds nothing on motion"
10933 );
10934 }
10935
10936 #[test]
10937 fn wysiwyg_down_crosses_a_paragraph_boundary() {
10938 // Regression: the blank separator row used to share the previous
10939 // paragraph's end offset, so Down got pinned at the boundary (while Up
10940 // still crossed). Both directions must step through it symmetrically.
10941 //
10942 // It's now stepped *over* rather than onto: the blank line between two
10943 // paragraphs is the boundary being drawn, not a line of the document, so
10944 // one press of Down crosses it. The goal column survives the crossing —
10945 // col 3 at the end of "abc" is col 3 at the end of "def".
10946 let mut d = wysiwyg_doc("wys_down", "abc\n\ndef\n");
10947 d.caret = 3; // end of "abc" (row 0)
10948 d.move_down(false);
10949 assert_eq!(d.caret_pos().0, 2, "Down should reach the second paragraph");
10950 assert_eq!(d.caret, 8); // end of "def", col 3 kept
10951 d.move_up(false);
10952 assert_eq!(d.caret_pos().0, 0, "Up should come back symmetrically");
10953 assert_eq!(d.caret, 3);
10954 }
10955
10956 #[test]
10957 fn wysiwyg_up_and_down_are_inverse_across_paragraphs() {
10958 // The second Up and the second Down here run off the ends of the
10959 // document, which is no longer a place a press is swallowed: they carry
10960 // the caret to the start and the end of the text. The claim in the
10961 // middle — that a Down retraces the Up that crossed the paragraph gap —
10962 // is the one this test is for, and it is asserted where it is made.
10963 let mut d = wysiwyg_doc("wys_updown", "abc\n\ndef\n");
10964 d.caret = 5; // start of "def"
10965 let start = d.caret_pos();
10966 d.move_up(false);
10967 assert_eq!(d.caret_pos().0, 0, "Up reaches the first paragraph");
10968 d.move_up(false);
10969 assert_eq!(d.caret, 0, "a second Up runs on to the document's start");
10970 d.move_down(false);
10971 assert_eq!(d.caret_pos(), start, "Down retraces Up exactly");
10972 d.move_down(false);
10973 assert_eq!(d.caret, 8, "a second Down runs on to the document's end");
10974 }
10975
10976 #[test]
10977 fn wysiwyg_new_paragraph_shows_before_typing() {
10978 // Regression: two Enters at the end of a paragraph produced trailing
10979 // newlines with no AST node, so the caret appeared stuck on the old line
10980 // until a character was typed. It must ride down onto the new line now.
10981 let mut d = doc_with("wys_newpara", "abc\n");
10982 d.view = View::Wysiwyg;
10983 d.caret = 3;
10984 d.insert("\n");
10985 d.insert("\n"); // source is now "abc\n\n\n", caret at 5
10986 assert_eq!(d.source, "abc\n\n\n");
10987 d.build_visual(80);
10988 let (row, _) = d.caret_pos();
10989 assert!(
10990 row >= 2,
10991 "caret should have moved down to the new line, got row {row}"
10992 );
10993 assert!(
10994 d.vmap.num_rows() >= 3,
10995 "the blank lines should render as rows"
10996 );
10997 }
10998
10999 #[test]
11000 fn wysiwyg_enter_between_paragraphs_lands_on_an_empty_line() {
11001 // The reported bug: Enter at the end of a paragraph that has another
11002 // paragraph below put the caret at the *start of the next paragraph* —
11003 // the empty paragraph it opened had no row, so the caret snapped onto
11004 // "World". It must now sit on its own empty line, with a blank spacer
11005 // above it (the paragraph gap).
11006 let mut d = wysiwyg_doc("wys_gap_mid", "Hello\n\nWorld\n");
11007 d.caret = 5; // end of "Hello"
11008 d.newline();
11009 d.build_visual(80);
11010 let (row, col) = d.caret_pos();
11011 assert_eq!(col, 0, "caret should start an empty line, not sit in text");
11012 assert_eq!(
11013 d.vmap.row_width(row),
11014 0,
11015 "caret's row must be empty, not 'World'"
11016 );
11017 assert!(
11018 row >= 2,
11019 "a blank spacer row should sit above the caret, got row {row}"
11020 );
11021 // The row above the caret is a real (empty) gap, and "Hello" stays put.
11022 assert_eq!(
11023 d.vmap.row_width(row - 1),
11024 0,
11025 "the row above the caret is a gap"
11026 );
11027 let row0: String = d.vmap.rows[0].glyphs.iter().map(|g| g.ch).collect();
11028 assert_eq!(row0, "Hello", "the paragraph above the caret must not move");
11029 }
11030
11031 #[test]
11032 fn wysiwyg_enter_at_eof_shows_a_gap_before_typing() {
11033 // At the document end a single Enter must also show the paragraph gap —
11034 // a blank spacer row above the caret — so the layout already matches how
11035 // it will look once the new paragraph has text.
11036 let mut d = wysiwyg_doc("wys_gap_eof", "Hello");
11037 d.caret = 5; // end of "Hello", no trailing newline
11038 d.newline(); // source becomes "Hello\n\n"
11039 d.build_visual(80);
11040 let (row, col) = d.caret_pos();
11041 assert_eq!(col, 0);
11042 assert!(
11043 row >= 2,
11044 "caret should sit below a blank spacer, got row {row}"
11045 );
11046 assert_eq!(
11047 d.vmap.row_width(row - 1),
11048 0,
11049 "the row above the caret is a gap"
11050 );
11051 }
11052
11053 #[test]
11054 fn wysiwyg_typing_after_enter_does_not_shift_the_caret_row() {
11055 // The spacer is view-only: typing the new paragraph must not reflow the
11056 // caret onto a different row — the transient view already matched the
11057 // settled one.
11058 let mut d = wysiwyg_doc("wys_no_reflow", "Hello\n\nWorld\n");
11059 d.caret = 5;
11060 d.newline();
11061 d.build_visual(80);
11062 let before = d.caret_pos();
11063 d.insert("New");
11064 d.build_visual(80);
11065 let after = d.caret_pos();
11066 assert_eq!(
11067 after.0, before.0,
11068 "typing must not move the caret to another row ({before:?} -> {after:?})"
11069 );
11070 }
11071
11072 #[test]
11073 fn wysiwyg_return_on_the_last_code_line_keeps_the_caret_in_the_block() {
11074 // Return at the end of the block's last line writes an empty line the
11075 // map used to drop, so the caret landed on `after` and the next
11076 // keystroke went into the paragraph below instead of into the code.
11077 let mut d = wysiwyg_doc("code_return", "prose\n\n```\nalpha\nbeta\n```\n\nafter\n");
11078 d.caret = d.source.find("beta").unwrap() + "beta".len();
11079 d.build_visual(80);
11080 let before = d.caret_pos().0;
11081
11082 d.newline();
11083 d.build_visual(80);
11084 assert_eq!(d.source, "prose\n\n```\nalpha\nbeta\n\n```\n\nafter\n");
11085
11086 let (row, col) = d.caret_pos();
11087 assert_eq!(row, before + 1, "the caret moves down one row");
11088 assert_eq!(col, 0, "onto the head of the empty line");
11089 let span = d.vmap.code_blocks[0].rows_span.clone();
11090 assert!(
11091 span.contains(&row),
11092 "caret row {row} is outside the block's rows {span:?}"
11093 );
11094
11095 // The whole point: what is typed next is code.
11096 d.insert("gamma");
11097 assert_eq!(d.source, "prose\n\n```\nalpha\nbeta\ngamma\n```\n\nafter\n");
11098 }
11099
11100 #[test]
11101 fn wysiwyg_hides_frontmatter_from_the_caret_and_copy() {
11102 let fm = "---\ntitle: hi\n---\n";
11103 let body = format!("{fm}# leaf\n\nbody\n");
11104 let mut d = wysiwyg_doc("wys_fm", &body);
11105 // Opening lifts the caret out of the now-hidden frontmatter.
11106 assert_eq!(
11107 d.caret,
11108 fm.len(),
11109 "caret should start at the first real block"
11110 );
11111 // Left at the content start can't step back into frontmatter.
11112 d.move_left(false);
11113 assert_eq!(d.caret, fm.len(), "left must not enter frontmatter");
11114 // Doc-start lands on the content floor, not offset 0.
11115 d.move_doc_start(false);
11116 assert_eq!(d.caret, fm.len());
11117 // Select-all + copy never include the frontmatter bytes.
11118 d.select_all();
11119 let sel = d.selected_text().unwrap().to_string();
11120 assert!(!sel.contains("title"), "copy leaked frontmatter: {sel:?}");
11121 assert!(
11122 sel.starts_with("# leaf"),
11123 "selection should begin at content: {sel:?}"
11124 );
11125 }
11126
11127 #[test]
11128 fn typing_in_a_frontmatter_only_document_lands_after_the_frontmatter() {
11129 // A fresh note is frontmatter and nothing else. With no rendered block
11130 // to floor the caret it opened at offset 0 — before the opening `---` —
11131 // so the first keystroke wrote itself in front of the metadata and the
11132 // file came out as `This---\ntitle: …`.
11133 let fm = "---\ntitle: 2026-08-29\nid: f8s32cd\n---\n";
11134 let mut d = wysiwyg_doc("wys_fm_only", fm);
11135 assert_eq!(d.caret, fm.len(), "caret must open past the frontmatter");
11136 // Nothing is rendered, so the caret draws at the origin of an empty view
11137 // — the same place an empty document puts it.
11138 assert_eq!(d.caret_pos(), (0, 0));
11139 d.insert("This");
11140 assert_eq!(d.source, format!("{fm}This"));
11141 }
11142
11143 /// `select_range` is the verb for a range a host already knows the bytes of,
11144 /// so it must not snap — and must still hold every invariant `place_caret`
11145 /// holds, the frontmatter floor above all.
11146 #[test]
11147 fn select_range_takes_the_range_as_given_but_still_floors_it() {
11148 let fm = "---\ntitle: foo\n---\n\n";
11149 let body = format!("{fm}body foo here\n");
11150 let mut d = wysiwyg_doc("wys_select_range", &body);
11151
11152 // The `foo` in the body: taken exactly, not snapped to a caret stop.
11153 let at = body.rfind("foo").unwrap();
11154 d.select_range(at, at + 3);
11155 assert_eq!(d.selection(), Some((at, at + 3)));
11156 assert_eq!(d.selected_text(), Some("foo"));
11157
11158 // The `foo` in the hidden frontmatter: below the floor, so both ends
11159 // come up to it rather than parking the caret in the metadata, where a
11160 // later keystroke would rewrite `title:`.
11161 let hidden = body.find("foo").unwrap();
11162 assert!(hidden < d.vmap.content_start);
11163 d.select_range(hidden, hidden + 3);
11164 assert!(
11165 d.caret >= d.vmap.content_start && d.anchor.unwrap() >= d.vmap.content_start,
11166 "a range under the floor must not leave the caret in the frontmatter"
11167 );
11168
11169 // Past the end, and mid-character, are both brought back to something
11170 // sliceable rather than panicking the next reader of the range.
11171 let multi = wysiwyg_doc("wys_select_range_utf8", "héllo\n");
11172 let mut d = multi;
11173 d.select_range(2, 9_999);
11174 assert_eq!(d.caret, d.source.len());
11175 assert!(d.source.is_char_boundary(d.anchor.unwrap()));
11176 assert!(d.source.is_char_boundary(d.caret));
11177 }
11178
11179 /// The bug `select_range` exists for: a match butting up against a hidden
11180 /// delimiter. `place_caret` snaps to the nearest *visible* stop, which is
11181 /// the one before the `**`.
11182 #[test]
11183 fn select_range_does_not_snap_off_a_hidden_delimiter() {
11184 let mut d = wysiwyg_doc("wys_select_range_bold", "a **needle** in it\n");
11185 let at = d.source.find("needle").unwrap();
11186 d.select_range(at, at + 6);
11187 assert_eq!(d.selected_text(), Some("needle"), "not \"needl\"");
11188 }
11189
11190 #[test]
11191 fn wysiwyg_backspace_at_content_start_leaves_frontmatter_intact() {
11192 // Backspace deletes `prev_boundary..caret` directly; at the first real
11193 // block that boundary is inside the hidden frontmatter, so it must be a
11194 // no-op rather than eating the closing `---`.
11195 let fm = "---\ntitle: hi\n---\n";
11196 let body = format!("{fm}leaf\n");
11197 let mut d = wysiwyg_doc("wys_fm_bs", &body);
11198 assert_eq!(d.caret, fm.len());
11199 d.backspace();
11200 assert_eq!(d.source, body, "backspace must not touch frontmatter");
11201 d.delete_word_back();
11202 assert_eq!(
11203 d.source, body,
11204 "word-delete must not touch frontmatter either"
11205 );
11206 }
11207
11208 #[test]
11209 fn wysiwyg_edits_inside_a_vis_directive_block_without_disturbing_its_fences() {
11210 // diaryx's `:::vis{.audience}` visibility block — any `:::name{.class}`
11211 // fenced div, really, since core parses these on for every document
11212 // now (`parse_extensions`). The container is a `directive` node, an
11213 // `is_block_container` kind like `block_quote`, so the caret works
11214 // inside its child paragraph exactly as it would inside a quote: typing
11215 // edits the paragraph, and the `:::vis{...}` / `:::` fences round-trip
11216 // untouched.
11217 let body = ":::vis{.public .family}\nhello\n:::\nafter\n";
11218 let mut d = wysiwyg_doc("wys_vis", body);
11219 d.caret = body.find("hello").unwrap() + "hello".len();
11220 d.insert("!");
11221 assert_eq!(
11222 d.source, ":::vis{.public .family}\nhello!\n:::\nafter\n",
11223 "typing inside the block edits its content in place"
11224 );
11225 assert!(
11226 d.source.contains(":::vis{.public .family}"),
11227 "opening fence survives"
11228 );
11229 assert!(d.source.contains(":::\nafter"), "closing fence survives");
11230 }
11231
11232 #[test]
11233 fn source_view_still_reaches_frontmatter() {
11234 // The metadata is only *hidden*, never lost: the source view edits and
11235 // selects it in full, and it's always preserved on save.
11236 let fm = "---\ntitle: hi\n---\n";
11237 let body = format!("{fm}# leaf\n");
11238 let mut d = doc_with("src_fm", &body);
11239 d.select_all();
11240 let sel = d.selected_text().unwrap();
11241 assert!(
11242 sel.contains("title"),
11243 "source view should select everything"
11244 );
11245 d.move_doc_start(false);
11246 assert_eq!(d.caret, 0, "source view can reach offset 0");
11247 }
11248
11249 const TABLE: &str = "| Name | Qty |\n|:-----|----:|\n| Pear | 3 |\n| Fig | 12 |\n";
11250
11251 #[test]
11252 fn wysiwyg_right_crosses_a_cell_border_without_stalling() {
11253 // The border and padding between two cells all share one source offset,
11254 // so a column-stepping caret would sit on `│` and then stall there
11255 // forever. Right must step: end of "Name" -> start of "Qty".
11256 let mut d = wysiwyg_doc("tbl_right", TABLE);
11257 d.caret = TABLE.find("Name").unwrap() + 4; // just after "Name"
11258 d.move_right(false);
11259 assert_eq!(
11260 d.caret,
11261 TABLE.find("Qty").unwrap(),
11262 "should land in the next cell"
11263 );
11264 let (r, c) = d.caret_pos();
11265 assert_eq!(d.vmap.rows[r].glyphs[c].ch, 'Q');
11266 }
11267
11268 #[test]
11269 fn wysiwyg_left_crosses_back_to_the_previous_cell() {
11270 let mut d = wysiwyg_doc("tbl_left", TABLE);
11271 d.caret = TABLE.find("Qty").unwrap();
11272 d.move_left(false);
11273 assert_eq!(
11274 d.caret,
11275 TABLE.find("Name").unwrap() + 4,
11276 "end of the previous cell"
11277 );
11278 }
11279
11280 #[test]
11281 fn wysiwyg_down_steps_over_a_table_rule() {
11282 // Between the header and the first body row sits a `├───┼───┤` rule.
11283 // It's drawn but holds no caret, so one Down must reach "Pear".
11284 let mut d = wysiwyg_doc("tbl_down", TABLE);
11285 d.caret = TABLE.find("Name").unwrap();
11286 d.move_down(false);
11287 assert_eq!(
11288 d.caret,
11289 TABLE.find("Pear").unwrap(),
11290 "one Down reaches the body row"
11291 );
11292 d.move_down(false);
11293 assert_eq!(d.caret, TABLE.find("Fig").unwrap());
11294 }
11295
11296 #[test]
11297 fn wysiwyg_tab_walks_the_cells_and_shift_tab_walks_back() {
11298 let mut d = wysiwyg_doc("tbl_tab", TABLE);
11299 d.caret = TABLE.find("Name").unwrap();
11300 // A hop lands with the destination cell's whole content selected, the
11301 // caret at its end — so typing replaces the cell like a form field.
11302 assert!(d.cell_hop(true));
11303 assert_eq!(
11304 d.selected_text(),
11305 Some("Qty"),
11306 "the target cell comes up selected"
11307 );
11308 assert_eq!(d.caret, TABLE.find("Qty").unwrap() + "Qty".len());
11309 assert!(d.cell_hop(true), "Tab wraps onto the next row's first cell");
11310 assert_eq!(d.selected_text(), Some("Pear"));
11311 assert!(d.cell_hop(false));
11312 assert_eq!(d.selected_text(), Some("Qty"));
11313 }
11314
11315 #[test]
11316 fn tab_outside_a_table_is_not_a_cell_hop() {
11317 // `cell_hop` reports false so the frontend can indent as usual.
11318 let mut d = wysiwyg_doc("tbl_none", "just a paragraph\n");
11319 d.caret = 4;
11320 assert!(!d.cell_hop(true));
11321 assert_eq!(d.caret, 4, "a refused hop leaves the caret alone");
11322 }
11323
11324 #[test]
11325 fn tab_at_the_last_cell_declines_rather_than_leaving_the_table() {
11326 let mut d = wysiwyg_doc("tbl_edge", TABLE);
11327 d.caret = TABLE.rfind("12").unwrap(); // the final cell
11328 assert!(!d.cell_hop(true), "no cell after the last one");
11329 d.caret = TABLE.find("Name").unwrap();
11330 assert!(!d.cell_hop(false), "no cell before the first one");
11331 }
11332
11333 #[test]
11334 fn wysiwyg_vertical_cell_motion_holds_the_column() {
11335 // Down/Up step to the cell above/below in the *same column*, not back to
11336 // the top-left the way a naive row/col motion over the picture would.
11337 let mut d = wysiwyg_doc("tbl_vert", TABLE);
11338 d.caret = TABLE.find("Qty").unwrap();
11339 // Each vertical hop selects the destination cell, holding the column.
11340 assert!(d.cell_move_vertical(true));
11341 assert_eq!(d.selected_text(), Some("3"), "Down holds column 1");
11342 assert!(d.cell_move_vertical(true));
11343 assert_eq!(d.selected_text(), Some("12"), "Down again, still column 1");
11344 assert!(!d.cell_move_vertical(true), "no row below the last");
11345 assert!(d.cell_move_vertical(false));
11346 assert_eq!(d.selected_text(), Some("3"), "Up holds column 1");
11347 assert!(d.cell_move_vertical(false));
11348 assert_eq!(d.selected_text(), Some("Qty"), "Up onto the header");
11349 assert!(!d.cell_move_vertical(false), "no row above the header");
11350 }
11351
11352 #[test]
11353 fn tab_off_the_last_cell_grows_a_row_and_enters_it() {
11354 let mut d = wysiwyg_doc("tbl_grow", TABLE);
11355 d.caret = TABLE.rfind("12").unwrap();
11356 let rows_before = d.source.matches('\n').count();
11357 assert!(d.cell_tab(true), "acts as a table key");
11358 assert_eq!(
11359 d.source.matches('\n').count(),
11360 rows_before + 1,
11361 "a fresh row was appended"
11362 );
11363 assert!(d.caret_in_table(), "the caret entered the new row");
11364 // The caret sits in the new row's first cell — past the old last cell.
11365 assert!(d.caret > TABLE.rfind("12").unwrap());
11366 }
11367
11368 #[test]
11369 fn return_in_a_table_drops_a_cell_and_grows_a_row_at_the_bottom() {
11370 let mut d = wysiwyg_doc("tbl_ret", TABLE);
11371 d.caret = TABLE.find("Name").unwrap();
11372 assert!(d.cell_return(), "acts as a table key");
11373 assert_eq!(
11374 d.selected_text(),
11375 Some("Pear"),
11376 "Return drops one cell, selecting it"
11377 );
11378 // From the last row, Return appends a row and enters it.
11379 d.caret = TABLE.rfind("Fig").unwrap();
11380 let rows_before = d.source.matches('\n').count();
11381 assert!(d.cell_return());
11382 assert_eq!(d.source.matches('\n').count(), rows_before + 1);
11383 assert!(d.caret_in_table());
11384 }
11385
11386 #[test]
11387 fn return_and_tab_outside_a_table_decline() {
11388 let mut d = wysiwyg_doc("tbl_decline", "just a paragraph\n");
11389 d.caret = 4;
11390 assert!(!d.cell_return(), "no table: the frontend inserts a newline");
11391 assert!(!d.cell_tab(true), "no table: the frontend indents");
11392 assert!(
11393 !d.cell_line_break(),
11394 "no table: the frontend breaks the line"
11395 );
11396 }
11397
11398 #[test]
11399 fn a_click_under_a_trailing_table_lands_past_it_and_enter_opens_a_line() {
11400 // A document that ends in a table used to end *inside* it: nothing
11401 // past the last cell was a caret stop, so a click in the blank space
11402 // under the grid snapped back into the table and there was no way to
11403 // write a line after it. The bottom border's end is that stop now.
11404 let mut d = wysiwyg_doc("tbl_trail", TABLE);
11405 let rows = d.vmap.num_rows();
11406 d.click(rows + 3, 0, false);
11407 let end = TABLE.trim_end_matches('\n').len();
11408 assert_eq!(d.caret, end, "the caret stands just past the table");
11409 assert!(!d.caret_in_table(), "past the table is outside it");
11410 assert!(!d.cell_return(), "Return there is the frontend's newline");
11411 d.newline();
11412 d.insert("after");
11413 assert_eq!(
11414 d.source,
11415 format!("{TABLE}\nafter\n"),
11416 "Enter opens a paragraph under the table"
11417 );
11418 }
11419
11420 #[test]
11421 fn typing_at_a_table_s_trailing_stop_opens_a_paragraph_first() {
11422 // The stop sits at the end of the table's last source line, and a
11423 // line glued under a table is a row of it — `| Fig | 12 |x` would be a
11424 // three-cell row. So the text gets a paragraph of its own, as it does
11425 // beside a block picture.
11426 let mut d = wysiwyg_doc("tbl_type", TABLE);
11427 d.caret = TABLE.trim_end_matches('\n').len();
11428 d.insert("x");
11429 assert_eq!(d.source, format!("{TABLE}\nx\n"));
11430 assert_eq!(d.caret, TABLE.len() + 2, "the caret follows the text");
11431 // And a paste, which joins the block exactly as typing would.
11432 let mut d = wysiwyg_doc("tbl_paste", TABLE);
11433 d.caret = TABLE.trim_end_matches('\n').len();
11434 d.paste("pasted");
11435 assert_eq!(d.source, format!("{TABLE}\npasted\n"));
11436 }
11437
11438 #[test]
11439 fn right_leaves_a_table_by_its_trailing_stop_and_backspace_steps_back_in() {
11440 let mut d = wysiwyg_doc("tbl_edge", TABLE);
11441 let last_cell_end = TABLE.rfind("12").unwrap() + 2;
11442 let end = TABLE.trim_end_matches('\n').len();
11443 d.caret = last_cell_end;
11444 d.move_right(false);
11445 assert_eq!(d.caret, end, "Right from the last cell leaves the table");
11446 // Backspace there takes no byte: the one behind the caret is the row's
11447 // closing `|`, which the rich view never drew. It steps back instead.
11448 d.backspace();
11449 assert_eq!(d.source, TABLE, "nothing deleted");
11450 assert_eq!(d.caret, last_cell_end, "back into the last cell");
11451 // Down from the last row lands on the same stop, and Up returns.
11452 d.move_down(false);
11453 assert_eq!(d.caret, end, "Down from the last row leaves the table");
11454 d.move_up(false);
11455 assert_eq!(d.caret, last_cell_end);
11456 }
11457
11458 #[test]
11459 fn a_table_s_trailing_stop_sits_between_it_and_the_text_below() {
11460 // With prose under the table, the stop is one hop between the last
11461 // cell and the paragraph — the shape a block picture's second stop has.
11462 let src = format!("{TABLE}\nafter\n");
11463 let mut d = wysiwyg_doc("tbl_mid", &src);
11464 d.caret = TABLE.rfind("12").unwrap() + 2;
11465 d.move_right(false);
11466 assert_eq!(d.caret, TABLE.trim_end_matches('\n').len());
11467 d.move_right(false);
11468 assert_eq!(d.caret, src.find("after").unwrap());
11469 // Typing at the stop still opens a paragraph, and the text below keeps
11470 // its own.
11471 d.move_left(false);
11472 d.insert("x");
11473 assert_eq!(d.source, format!("{TABLE}\nx\n\nafter\n"));
11474 }
11475
11476 #[test]
11477 fn shift_return_inserts_an_in_cell_break_the_renderer_reads_as_a_line() {
11478 let mut d = wysiwyg_doc("tbl_break", TABLE);
11479 d.caret = TABLE.find("Pear").unwrap() + 4; // just after "Pear"
11480 assert!(d.cell_line_break(), "acts as a table key");
11481 assert!(
11482 d.source.contains("Pear<br>"),
11483 "spelled as an inline <br>: {}",
11484 d.source
11485 );
11486 assert!(d.caret_in_table(), "still in the cell, past the break");
11487 // The break renders as a real line: the "Pear" cell now draws two lines,
11488 // so the table's picture is one row taller than a single-line table.
11489 d.build_visual(80);
11490 let table = &d.vmap.tables[0];
11491 let cell = &table.grid[1].cells[0]; // first body row, first column
11492 assert!(
11493 cell.glyphs.iter().any(|g| g.ch == '\n'),
11494 "the cell carries the break as a newline glyph for the frontend to split"
11495 );
11496 }
11497
11498 #[test]
11499 fn shift_return_in_a_markdown_cell_leaves_a_semantic_hard_break_not_raw_html() {
11500 // twig promotes the in-cell `<br>` to a `hard_break`, so the break reads
11501 // back as structure — the whole point of routing through insert_line_break
11502 // instead of splicing raw `<br>` bytes.
11503 let mut d = wysiwyg_doc("tbl_break_semantic", TABLE);
11504 d.caret = TABLE.find("Pear").unwrap() + 4;
11505 assert!(d.cell_line_break());
11506 let kinds: Vec<Kind> = d
11507 .editor
11508 .nodes()
11509 .unwrap()
11510 .iter()
11511 .map(|n| n.kind.clone())
11512 .collect();
11513 assert!(kinds.contains(&Kind::HardBreak), "got {kinds:?}");
11514 assert!(
11515 !kinds.contains(&Kind::RawInline),
11516 "still raw HTML: {kinds:?}"
11517 );
11518 }
11519
11520 #[test]
11521 fn backspace_over_an_in_cell_break_deletes_the_whole_br_not_a_byte() {
11522 // The `<br>` draws as one newline glyph, so Backspace over it must take
11523 // all four bytes — a one-byte delete would strand a visible `<br` in the
11524 // cell (the reported bug).
11525 let mut d = wysiwyg_doc("tbl_break_bs", TABLE);
11526 d.caret = TABLE.find("Pear").unwrap() + 4;
11527 assert!(d.cell_line_break());
11528 assert!(d.source.contains("Pear<br>"), "precondition: {}", d.source);
11529 d.backspace(); // caret sits just past the break
11530 assert!(
11531 !d.source.contains("<br"),
11532 "no half-deleted <br left: {}",
11533 d.source
11534 );
11535 assert!(
11536 d.source.contains("| Pear |"),
11537 "the cell is back to one line: {}",
11538 d.source
11539 );
11540 }
11541
11542 #[test]
11543 fn delete_forward_over_an_in_cell_break_deletes_the_whole_br() {
11544 let mut d = wysiwyg_doc("tbl_break_del", TABLE);
11545 d.caret = TABLE.find("Pear").unwrap() + 4;
11546 assert!(d.cell_line_break());
11547 d.caret = TABLE.find("Pear").unwrap() + 4; // back onto the break's start
11548 d.delete_forward();
11549 assert!(
11550 !d.source.contains("<br"),
11551 "no half-deleted <br: {}",
11552 d.source
11553 );
11554 assert!(
11555 d.source.contains("| Pear |"),
11556 "cell back to one line: {}",
11557 d.source
11558 );
11559 }
11560
11561 #[test]
11562 fn shift_return_in_a_djot_cell_is_swallowed_and_leaves_the_row_intact() {
11563 // Djot has no idiomatic in-cell break, so twig refuses it. The gesture is
11564 // still consumed (a real newline would split the one-line row), but the
11565 // cell must be left exactly as it was — no non-idiomatic `<br>` spliced in.
11566 let src = "| Name | Qty |\n|:-----|----:|\n| Pear | 3 |\n";
11567 let mut d = Doc::from_source(src.to_string(), Format::Djot).unwrap();
11568 d.caret = src.find("Pear").unwrap() + 4;
11569 assert!(d.caret_in_table(), "caret should be inside the djot table");
11570 assert!(
11571 d.cell_line_break(),
11572 "the key is consumed, not passed to the frontend"
11573 );
11574 assert_eq!(d.source, src, "the djot cell is left untouched");
11575 assert!(
11576 !d.source.contains("<br>"),
11577 "no non-idiomatic <br> spliced into djot"
11578 );
11579 assert!(
11580 d.status.is_some(),
11581 "the refusal is surfaced on the status line"
11582 );
11583 }
11584
11585 #[test]
11586 fn typing_in_a_cell_edits_that_cell() {
11587 // Editing comes free once offsets map correctly: the caret is a source
11588 // offset, so a normal splice lands inside the pipe table.
11589 let mut d = wysiwyg_doc("tbl_type", TABLE);
11590 d.caret = TABLE.find("Pear").unwrap() + 4;
11591 d.insert("s");
11592 assert!(d.source.contains("| Pears | 3 |"), "got {:?}", d.source);
11593 }
11594
11595 #[test]
11596 fn motion_and_delete_treat_an_emoji_as_one_character() {
11597 // 👨👩👧 is a single grapheme built from three emoji joined by ZWJ — 18
11598 // bytes, several codepoints. Right-arrow must clear it in one step, and
11599 // backspace must remove the whole cluster, not a stray joiner.
11600 let family = "👨👩👧";
11601 let mut d = doc_with("emoji", &format!("a{family}b\n"));
11602 d.caret = 1; // just after 'a', before the emoji
11603 d.move_right(false);
11604 assert_eq!(
11605 d.caret,
11606 1 + family.len(),
11607 "one step clears the whole cluster"
11608 );
11609 assert_eq!(&d.source[d.caret..d.caret + 1], "b");
11610
11611 d.backspace(); // delete the emoji as a unit
11612 assert_eq!(d.source, "ab\n");
11613 assert_eq!(d.caret, 1);
11614 }
11615
11616 #[test]
11617 fn motion_handles_a_combining_accent_as_one_character() {
11618 // "e" + U+0301 (combining acute) renders as one é.
11619 let mut d = doc_with("combining", "e\u{0301}x\n");
11620 d.caret = 0;
11621 d.move_right(false);
11622 assert_eq!(
11623 d.caret,
11624 "e\u{0301}".len(),
11625 "steps past base + combining mark"
11626 );
11627 }
11628
11629 #[test]
11630 fn undo_then_redo_round_trips_an_edit() {
11631 let mut d = doc_with("undo", "hello\n");
11632 d.caret = 5;
11633 d.insert("!");
11634 assert_eq!(d.source, "hello!\n");
11635 d.undo();
11636 assert_eq!(d.source, "hello\n");
11637 assert_eq!(d.caret, 5, "undo restores the caret");
11638 d.redo();
11639 assert_eq!(d.source, "hello!\n");
11640 }
11641
11642 #[test]
11643 fn a_run_of_typing_undoes_as_one_step() {
11644 let mut d = doc_with("coalesce", "\n");
11645 d.caret = 0;
11646 d.insert("a");
11647 d.insert("b");
11648 d.insert("c");
11649 assert_eq!(d.source, "abc\n");
11650 d.undo(); // the whole typed run, not just "c"
11651 assert_eq!(d.source, "\n");
11652 d.undo(); // nothing left — the run was one step
11653 assert_eq!(d.source, "\n");
11654 assert_eq!(d.status.as_deref(), Some("nothing to undo"));
11655 }
11656
11657 // ── IME composition ──────────────────────────────────────────────────────
11658
11659 #[test]
11660 fn a_composition_run_undoes_as_one_step() {
11661 let mut d = doc_with("compose", "\n");
11662 d.caret = 0;
11663 // What an IME does: each step replaces the last one's provisional bytes.
11664 d.edit_composing(0, 0, "k");
11665 d.edit_composing(0, 1, "か");
11666 d.edit_composing(0, 3, "かん");
11667 d.edit_composing(0, 6, "感"); // the commit
11668 d.end_composition();
11669 assert_eq!(d.source, "感\n");
11670 d.undo(); // the whole composition, not its last keystroke
11671 assert_eq!(d.source, "\n");
11672 assert_eq!(d.status.as_deref(), None, "the run was a single step");
11673 }
11674
11675 #[test]
11676 fn two_compositions_are_two_undo_steps() {
11677 let mut d = doc_with("compose_two", "\n");
11678 d.caret = 0;
11679 d.edit_composing(0, 0, "か");
11680 d.edit_composing(0, 3, "蚊");
11681 d.end_composition();
11682 d.edit_composing(3, 3, "き");
11683 d.edit_composing(3, 6, "木");
11684 d.end_composition();
11685 assert_eq!(d.source, "蚊木\n");
11686 d.undo();
11687 assert_eq!(d.source, "蚊\n", "only the second composition");
11688 d.undo();
11689 assert_eq!(d.source, "\n");
11690 }
11691
11692 #[test]
11693 fn a_composition_does_not_fold_into_the_typing_around_it() {
11694 let mut d = doc_with("compose_typing", "\n");
11695 d.caret = 0;
11696 d.insert("a");
11697 d.insert("b");
11698 d.edit_composing(2, 2, "か");
11699 d.edit_composing(2, 5, "蚊");
11700 d.end_composition();
11701 d.insert("c");
11702 assert_eq!(d.source, "ab蚊c\n");
11703 d.undo();
11704 assert_eq!(d.source, "ab蚊\n");
11705 d.undo();
11706 assert_eq!(d.source, "ab\n");
11707 d.undo();
11708 assert_eq!(d.source, "\n");
11709 }
11710
11711 #[test]
11712 fn ending_a_composition_that_never_began_leaves_a_typing_run_alone() {
11713 let mut d = doc_with("compose_spurious", "\n");
11714 d.caret = 0;
11715 d.insert("a");
11716 d.end_composition(); // an IME unmarking unprompted
11717 d.insert("b");
11718 assert_eq!(d.source, "ab\n");
11719 d.undo();
11720 assert_eq!(d.source, "\n", "still one typed run");
11721 }
11722
11723 // ── the clipboard's rich flavor ──────────────────────────────────────────
11724
11725 #[test]
11726 fn an_inline_selection_publishes_html_without_a_paragraph_wrapper() {
11727 let mut d = doc_with("sel_inline", "a **bold** c\n");
11728 d.anchor = Some(2);
11729 d.caret = 10; // `**bold**`, inside the paragraph
11730 assert_eq!(d.selection_html().as_deref(), Some("<strong>bold</strong>"));
11731 }
11732
11733 #[test]
11734 fn a_whole_block_selection_keeps_its_paragraph() {
11735 let mut d = doc_with("sel_block", "a **bold** c\n");
11736 d.anchor = Some(0);
11737 d.caret = 12; // the entire paragraph
11738 assert_eq!(
11739 d.selection_html().as_deref(),
11740 Some("<p>a <strong>bold</strong> c</p>")
11741 );
11742 }
11743
11744 #[test]
11745 fn a_multi_block_selection_keeps_its_structure() {
11746 let mut d = doc_with("sel_multi", "para\n\n- one\n- two\n");
11747 d.select_all();
11748 let html = d.selection_html().expect("renders");
11749 assert!(html.contains("<p>para</p>"), "{html:?}");
11750 assert!(html.contains("<li>one</li>"), "{html:?}");
11751 }
11752
11753 #[test]
11754 fn a_word_inside_a_heading_publishes_as_text_not_a_heading() {
11755 // The fragment `Head` is a paragraph standalone; the *document* says it
11756 // sits inside one block, so the wrapper is an artifact either way.
11757 let mut d = doc_with("sel_heading", "# Head line\n");
11758 d.anchor = Some(2);
11759 d.caret = 6;
11760 assert_eq!(d.selection_html().as_deref(), Some("Head"));
11761 }
11762
11763 #[test]
11764 fn no_selection_publishes_no_html() {
11765 let mut d = doc_with("sel_none", "a b\n");
11766 d.caret = 1;
11767 assert_eq!(d.selection_html(), None);
11768 }
11769
11770 #[test]
11771 fn pasting_html_converts_it_and_is_one_undo_step() {
11772 let mut d = doc_with("paste_html", "x\n");
11773 d.caret = 1;
11774 assert!(d.paste_html("<p>a <strong>b</strong> c</p>"));
11775 assert_eq!(d.source, "xa **b** c\n");
11776 d.undo();
11777 assert_eq!(d.source, "x\n", "the whole paste, in one step");
11778 }
11779
11780 #[test]
11781 fn pasting_html_replaces_the_selection() {
11782 let mut d = doc_with("paste_html_sel", "keep drop\n");
11783 d.anchor = Some(5);
11784 d.caret = 9;
11785 assert!(d.paste_html("<em>new</em>"));
11786 assert_eq!(d.source, "keep *new*\n");
11787 }
11788
11789 #[test]
11790 fn html_that_would_paste_garbage_declines_so_the_caller_falls_back() {
11791 let mut d = doc_with("paste_html_bad", "x\n");
11792 d.caret = 1;
11793 // twig builds no table from HTML; raw `<table>` in prose is worse than
11794 // the plain flavor the caller still holds.
11795 assert!(!d.paste_html("<table><tr><td>a</td></tr></table>"));
11796 assert_eq!(d.source, "x\n", "declined edits nothing");
11797 }
11798
11799 #[test]
11800 fn copy_then_paste_round_trips_through_the_html_flavor() {
11801 let mut d = doc_with("clip_round", "a **b** and [l](https://x.dev)\n");
11802 d.select_all();
11803 let html = d.selection_html().expect("renders");
11804 let mut into = doc_with("clip_round_dst", "\n");
11805 into.caret = 0;
11806 assert!(into.paste_html(&html));
11807 assert_eq!(into.source, "a **b** and [l](https://x.dev)\n");
11808 }
11809
11810 #[test]
11811 fn moving_the_caret_starts_a_new_undo_group() {
11812 let mut d = doc_with("break", "\n");
11813 d.caret = 0;
11814 d.insert("a");
11815 d.insert("b"); // "ab\n", caret at 2
11816 d.move_left(false); // breaks the run
11817 d.insert("X"); // "aXb\n"
11818 assert_eq!(d.source, "aXb\n");
11819 d.undo();
11820 assert_eq!(
11821 d.source, "ab\n",
11822 "first undo removes only the post-move insert"
11823 );
11824 d.undo();
11825 assert_eq!(d.source, "\n", "second undo removes the earlier run");
11826 }
11827
11828 #[test]
11829 fn undo_reverses_a_format_toggle() {
11830 let mut d = doc_with("fmt_undo", "a word b\n");
11831 d.anchor = Some(2);
11832 d.caret = 6;
11833 d.toggle(InlineKind::Strong);
11834 assert_eq!(d.source, "a **word** b\n");
11835 d.undo();
11836 assert_eq!(d.source, "a word b\n");
11837 }
11838
11839 #[test]
11840 fn undo_back_to_the_saved_state_clears_dirty() {
11841 let mut d = doc_with("dirty_undo", "hello\n");
11842 assert!(!d.dirty);
11843 d.caret = 5;
11844 d.insert("!");
11845 assert!(d.dirty);
11846 d.undo();
11847 assert!(
11848 !d.dirty,
11849 "undoing to the saved source is not a modification"
11850 );
11851 }
11852
11853 #[test]
11854 fn a_new_edit_invalidates_redo() {
11855 let mut d = doc_with("redo_inv", "\n");
11856 d.caret = 0;
11857 d.insert("a");
11858 d.undo();
11859 d.insert("b"); // diverges — the redo of "a" is now gone
11860 d.redo();
11861 assert_eq!(d.source, "b\n");
11862 }
11863
11864 #[test]
11865 fn can_undo_and_can_redo_follow_the_history_a_menu_would_enable_by() {
11866 let mut d = doc_with("can_undo", "hello\n");
11867 assert!(
11868 !d.can_undo() && !d.can_redo(),
11869 "a fresh document has no history"
11870 );
11871 d.caret = 5;
11872 d.insert("!");
11873 assert!(
11874 d.can_undo() && !d.can_redo(),
11875 "an edit is a step to take back"
11876 );
11877 d.undo();
11878 assert!(!d.can_undo() && d.can_redo(), "undone: only redo remains");
11879 d.redo();
11880 assert!(d.can_undo() && !d.can_redo(), "redone: back to undoable");
11881 d.undo();
11882 d.insert("?");
11883 assert!(
11884 d.can_undo() && !d.can_redo(),
11885 "a fresh edit ends the redo chain"
11886 );
11887 // A coalesced run over-counts steps — the bound is what a menu needs,
11888 // and it reconciles the moment twig reports the history empty.
11889 d.insert("a");
11890 d.insert("b");
11891 while d.can_undo() {
11892 d.undo();
11893 }
11894 assert_eq!(d.source, "hello\n");
11895 assert!(!d.can_undo());
11896 // A reading surface has nothing to undo, whatever the history holds.
11897 d.redo();
11898 d.set_read_only(true);
11899 assert!(!d.can_undo() && !d.can_redo());
11900 }
11901
11902 #[test]
11903 fn undo_on_empty_history_is_a_no_op() {
11904 let mut d = doc_with("undo_empty", "hi\n");
11905 d.undo();
11906 assert_eq!(d.source, "hi\n");
11907 assert_eq!(d.status.as_deref(), Some("nothing to undo"));
11908 }
11909
11910 #[test]
11911 fn a_one_character_paste_is_its_own_undo_step() {
11912 for view in [View::Source, View::Wysiwyg] {
11913 let mut d = doc_in(view, "paste_step", "ab\n");
11914 d.caret = 0;
11915 d.insert("x");
11916 d.insert("y"); // a run of typing
11917 d.paste("z"); // one character, but pasted — not part of that run
11918 assert_eq!(d.source, "xyzab\n");
11919 d.undo();
11920 assert_eq!(d.source, "xyab\n", "the paste undoes on its own");
11921 assert_eq!(d.caret, 2, "and hands back the caret it found");
11922 d.undo();
11923 assert_eq!(d.source, "ab\n", "the typed run is still one step under it");
11924 }
11925 }
11926
11927 #[test]
11928 fn the_same_character_typed_still_joins_the_run() {
11929 // The other half of the pair: `z` is a keystroke here and a paste above,
11930 // and the two undo differently. Nothing about the *string* says which —
11931 // which is why provenance has to come from the door the caller uses.
11932 for view in [View::Source, View::Wysiwyg] {
11933 let mut d = doc_in(view, "typed_run", "ab\n");
11934 d.caret = 0;
11935 d.insert("x");
11936 d.insert("y");
11937 d.insert("z");
11938 d.undo();
11939 assert_eq!(d.source, "ab\n", "one run, one step");
11940 }
11941 }
11942
11943 #[test]
11944 fn undo_restores_the_caret_to_where_it_was_not_to_the_edit_site() {
11945 for view in [View::Source, View::Wysiwyg] {
11946 let mut d = doc_in(view, "undo_caret", "hello world\n");
11947 d.caret = 11; // standing at the end of "world", away from the edit
11948 d.edit(0, 5, "goodbye");
11949 assert_eq!(d.source, "goodbye world\n");
11950 d.undo();
11951 assert_eq!(d.source, "hello world\n");
11952 // The undone edit ends at offset 5; the user was at 11.
11953 assert_eq!(d.caret, 11, "the caret comes back with the bytes");
11954 }
11955 }
11956
11957 #[test]
11958 fn undo_restores_the_selection_the_edit_replaced() {
11959 for view in [View::Source, View::Wysiwyg] {
11960 let mut d = doc_in(view, "undo_sel", "a word b\n");
11961 d.anchor = Some(2);
11962 d.caret = 6; // "word" selected
11963 d.insert("X");
11964 assert_eq!(d.source, "a X b\n");
11965 d.undo();
11966 assert_eq!(d.source, "a word b\n");
11967 assert_eq!(d.selection(), Some((2, 6)), "the selection comes back too");
11968 }
11969 }
11970
11971 #[test]
11972 fn redo_restores_the_caret_the_edit_left_behind() {
11973 for view in [View::Source, View::Wysiwyg] {
11974 let mut d = doc_in(view, "redo_caret", "hello world\n");
11975 d.caret = 11;
11976 d.edit(0, 5, "goodbye");
11977 assert_eq!(d.caret, 7, "the edit left the caret after its new text");
11978 d.undo();
11979 d.redo();
11980 assert_eq!(d.source, "goodbye world\n");
11981 assert_eq!(d.caret, 7, "redo puts it back where the edit had it");
11982 }
11983 }
11984
11985 #[test]
11986 fn undoing_a_typed_run_restores_the_caret_from_before_the_whole_run() {
11987 for view in [View::Source, View::Wysiwyg] {
11988 let mut d = doc_in(view, "run_caret", "hi\n");
11989 d.caret = 2;
11990 d.insert("a");
11991 d.insert("b");
11992 d.insert("c");
11993 assert_eq!(d.source, "hiabc\n");
11994 d.undo();
11995 assert_eq!(d.source, "hi\n");
11996 assert_eq!(d.caret, 2, "before the run, not before its last keystroke");
11997 d.redo();
11998 assert_eq!(d.caret, 5, "and redo restores the end of the whole run");
11999 }
12000 }
12001
12002 #[test]
12003 fn undo_restores_the_caret_across_a_format_toggle() {
12004 // A toggle reaches twig without going through `splice`, so it has to
12005 // record its own step — miss it and every stack depth below it is off by
12006 // one, and undo starts handing back another edit's caret.
12007 for view in [View::Source, View::Wysiwyg] {
12008 let mut d = doc_in(view, "fmt_caret", "a word b\n");
12009 d.caret = 8;
12010 d.anchor = Some(2);
12011 d.caret = 6;
12012 d.toggle(InlineKind::Strong);
12013 assert_eq!(d.source, "a **word** b\n");
12014 d.undo();
12015 assert_eq!(d.source, "a word b\n");
12016 assert_eq!(
12017 d.selection(),
12018 Some((2, 6)),
12019 "the toggled selection comes back"
12020 );
12021 }
12022 }
12023
12024 #[test]
12025 fn an_edit_after_an_undo_truncates_the_caret_history_with_twigs() {
12026 // The drift that would never announce itself: twig drops its redo stack
12027 // on any fresh edit, so a leaf redo entry that outlives it would restore
12028 // a caret from the timeline that edit abandoned.
12029 for view in [View::Source, View::Wysiwyg] {
12030 let mut d = doc_in(view, "redo_trunc", "hello world\n");
12031 d.caret = 11;
12032 d.edit(0, 5, "goodbye"); // step A, caret 11 → 7
12033 d.undo();
12034 assert_eq!(d.caret, 11);
12035 d.caret = 0;
12036 d.insert("X"); // diverges: A's redo is gone from twig
12037 assert_eq!(d.source, "Xhello world\n");
12038
12039 d.redo();
12040 assert_eq!(d.source, "Xhello world\n", "nothing to redo onto");
12041 assert_eq!(d.status.as_deref(), Some("nothing to redo"));
12042 d.undo();
12043 assert_eq!(d.source, "hello world\n");
12044 assert_eq!(
12045 d.caret, 0,
12046 "the surviving step's caret, not the dropped one"
12047 );
12048 }
12049 }
12050
12051 #[test]
12052 fn indent_and_outdent_move_the_caret_line_with_its_text() {
12053 for view in [View::Source, View::Wysiwyg] {
12054 let g = |m, f: fn(&mut Doc)| golden_in(view, "indent_line", m, f);
12055 assert_eq!(g("he|llo\n", |d| d.indent()), " he|llo\n");
12056 assert_eq!(g(" he|llo\n", |d| d.outdent()), "he|llo\n");
12057 // Indentation the caret is standing *in* collapses to the line start
12058 // rather than dragging the caret into the text.
12059 assert_eq!(g("| hello\n", |d| d.outdent()), "|hello\n");
12060 // A line with none to give back is left exactly as it was.
12061 assert_eq!(g("he|llo\n", |d| d.outdent()), "he|llo\n");
12062 // Less than a full level gives back what it has.
12063 assert_eq!(g(" he|llo\n", |d| d.outdent()), "he|llo\n");
12064 // A tab is one level however many spaces it isn't.
12065 assert_eq!(g("\the|llo\n", |d| d.outdent()), "he|llo\n");
12066 }
12067 }
12068
12069 #[test]
12070 fn one_indent_level_leaves_a_paragraph_a_paragraph() {
12071 // Why the level is two spaces and not the four both frontends type
12072 // today. Four is markdown's indented-code-block marker, so a Tab on a
12073 // paragraph would silently restyle it as code — a width that changes
12074 // what the document *means* isn't an indent. Pinned because the number
12075 // is the kind of thing a later list-aware pass would reach for.
12076 let mut d = doc_with("indent_kind", "hello\n");
12077 d.caret = 2;
12078 d.indent();
12079 assert_eq!(d.source, " hello\n");
12080 assert!(
12081 d.nodes().iter().any(|n| n.kind == Kind::Para),
12082 "still prose after a Tab"
12083 );
12084 assert!(!d.nodes().iter().any(|n| n.kind == Kind::CodeBlock));
12085
12086 // The four-space level this replaces, for contrast: same text, and twig
12087 // reparses the paragraph into a code block.
12088 let mut wide = doc_with("indent_kind_4", " hello\n");
12089 wide.build_visual(80);
12090 assert!(
12091 wide.nodes().iter().any(|n| n.kind == Kind::CodeBlock),
12092 "four spaces is a code block, not an indented paragraph"
12093 );
12094 }
12095
12096 #[test]
12097 fn indent_nests_a_list_item_under_its_parent() {
12098 // Tab indents a list item by its own marker width, landing its marker at
12099 // the parent's content column so twig reparses it as a nested list.
12100 for view in [View::Source, View::Wysiwyg] {
12101 let mut d = doc_in(view, "indent_nest", "- a\n- b\n");
12102 d.caret = 6; // on the second item
12103 d.indent();
12104 assert_eq!(d.source, "- a\n - b\n");
12105 let lists = d
12106 .nodes()
12107 .iter()
12108 .filter(|n| n.kind == Kind::BulletList)
12109 .count();
12110 assert_eq!(lists, 2, "the indented item is a nested list");
12111 }
12112 }
12113
12114 #[test]
12115 fn indent_nests_an_ordered_item_at_its_marker_width() {
12116 // An ordered marker `1. ` is three columns wide, so a two-space step
12117 // (which nests a bullet) leaves it flat. Regression: Tab must use the
12118 // marker width, three, so the item actually nests — and the source
12119 // renumbers so the sub-list restarts at 1 and the outer list resumes.
12120 for view in [View::Source, View::Wysiwyg] {
12121 let mut d = doc_in(view, "indent_ord", "1. a\n2. b\n3. c\n");
12122 d.caret = d.source.find('b').unwrap();
12123 d.indent();
12124 assert_eq!(d.source, "1. a\n 1. b\n2. c\n");
12125 let lists = d
12126 .nodes()
12127 .iter()
12128 .filter(|n| n.kind == Kind::OrderedList)
12129 .count();
12130 assert_eq!(lists, 2, "the indented item is a nested ordered list");
12131 }
12132 }
12133
12134 #[test]
12135 fn indent_leaves_a_lists_first_item_put() {
12136 // The first item of a list has no sibling above it to nest under, so Tab
12137 // is a no-op there — the marker stays at column zero rather than being
12138 // shoved into indentation twig can't read as a sub-list.
12139 for view in [View::Source, View::Wysiwyg] {
12140 let mut d = doc_in(view, "indent_first", "- a\n- b\n");
12141 d.caret = 1; // on the FIRST item
12142 d.indent();
12143 assert_eq!(d.source, "- a\n- b\n", "the first item doesn't nest");
12144 // The sibling below still nests, proving the guard is per-item.
12145 d.caret = d.source.find('b').unwrap();
12146 d.indent();
12147 assert_eq!(d.source, "- a\n - b\n");
12148 }
12149 }
12150
12151 #[test]
12152 fn hidden_mode_keeps_typed_markup_literal() {
12153 // The Diaryx default: typing `*hi*` gives the characters, not emphasis —
12154 // twig escapes what would open markup, so the source is `\*hi\*` and the
12155 // AST is a plain string. Formatting is the commands' job in this mode.
12156 let mut d = doc_in(View::Wysiwyg, "hidden_literal", "");
12157 d.insert("*hi*");
12158 assert_eq!(d.source, "\\*hi\\*");
12159 assert!(
12160 d.nodes()
12161 .iter()
12162 .all(|n| n.kind != Kind::Emph && n.kind != Kind::Strong)
12163 );
12164 }
12165
12166 #[test]
12167 fn hidden_mode_escapes_a_line_start_block_marker() {
12168 // A `#`/`-`/`>` at a line start would open a block, so Hidden mode keeps
12169 // it literal too — a Diaryx user's "# 1 idea" stays prose, not a heading.
12170 let mut d = doc_in(View::Wysiwyg, "hidden_block", "");
12171 d.insert("# hi");
12172 assert_eq!(d.source, "\\# hi");
12173 assert!(d.nodes().iter().all(|n| n.kind != Kind::Heading));
12174 }
12175
12176 #[test]
12177 fn authoring_modes_keep_typed_markup_live() {
12178 // Both authoring rungs of the ladder: typing `*hi*` really is emphasis
12179 // (no escape), the same as source view — escaping is `None`'s alone, and
12180 // it's the axis, not the reveal, that decides.
12181 for (view, mode) in [
12182 (View::Wysiwyg, MarkupMode::Shortcuts),
12183 (View::Wysiwyg, MarkupMode::Full),
12184 (View::Source, MarkupMode::None),
12185 ] {
12186 let mut d = doc_in(view, "live_markup", "");
12187 d.set_markup_mode(mode);
12188 d.insert("*hi*");
12189 assert_eq!(d.source, "*hi*", "{mode:?} in {view:?} types raw markup");
12190 }
12191 }
12192
12193 #[test]
12194 fn hidden_mode_overwrite_undoes_in_one_step() {
12195 // Typing over a selection escapes the replacement *and* stays a single
12196 // undo — the selection-delete and the literal insert fold together, so
12197 // one undo brings the whole selection back, like a plain overwrite.
12198 let mut d = doc_in(View::Wysiwyg, "hidden_overwrite", "a word b\n");
12199 d.anchor = Some(2);
12200 d.caret = 6; // "word"
12201 d.insert("*");
12202 assert_eq!(d.source, "a \\* b\n", "the replacement is escaped");
12203 d.undo();
12204 assert_eq!(d.source, "a word b\n");
12205 assert_eq!(d.selection(), Some((2, 6)), "one undo, selection restored");
12206 }
12207
12208 #[test]
12209 fn backspace_over_an_escaped_char_takes_the_hidden_backslash_too() {
12210 // Type `*` in Hidden mode → `\*` (drawn as one `*`); one Backspace clears
12211 // the whole visual character, never stranding the hidden `\`.
12212 let mut d = doc_in(View::Wysiwyg, "bsp_escape", "");
12213 d.insert("*");
12214 assert_eq!(d.source, "\\*");
12215 d.backspace();
12216 assert_eq!(d.source, "", "the escape backslash went with the *");
12217 // A *literal* backslash (source view, no escape) is an ordinary char.
12218 let mut s = doc_in(View::Source, "bsp_lit", "a\\b\n");
12219 s.caret = 3; // after `b`
12220 s.backspace();
12221 assert_eq!(s.source, "a\\\n", "only the b is deleted, the \\ stays");
12222 }
12223
12224 #[test]
12225 fn hidden_mode_leaves_structural_markup_alone() {
12226 // Enter continues a bullet list by writing a real `- ` marker (an
12227 // `insert_raw`, not the typing path), so Hidden mode's escaping never
12228 // touches it — the list keeps working.
12229 let mut d = doc_in(View::Wysiwyg, "hidden_struct", "- item\n");
12230 d.caret = 6;
12231 d.newline();
12232 d.insert("two");
12233 assert_eq!(d.source, "- item\n- two\n");
12234 }
12235
12236 #[test]
12237 fn markup_mode_defaults_to_none_and_round_trips() {
12238 // Diaryx's default is the clean `None` surface; a markup-fluent
12239 // frontend can climb the ladder, and the choice sticks.
12240 let mut d = doc_in(View::Wysiwyg, "markup_mode", "hi\n");
12241 assert_eq!(d.markup_mode(), MarkupMode::None, "None by default");
12242 for mode in [MarkupMode::Shortcuts, MarkupMode::Full, MarkupMode::None] {
12243 d.set_markup_mode(mode);
12244 assert_eq!(d.markup_mode(), mode);
12245 }
12246 }
12247
12248 #[test]
12249 fn full_mode_reveals_only_the_caret_line() {
12250 // The mode's whole claim: the caret's line shows its raw delimiters and
12251 // every other line stays resolved. Two paragraphs with identical markup
12252 // so the only difference between the rows is where the caret is.
12253 let mut d = doc_in(
12254 View::Wysiwyg,
12255 "reveal_caret_line",
12256 "*one* here\n\n*two* there\n",
12257 );
12258 d.set_markup_mode(MarkupMode::Full);
12259
12260 caret_at(&mut d, "one");
12261 let rows = drawn_rows(&d);
12262 assert!(
12263 rows.iter().any(|r| r == "*one* here"),
12264 "caret's line raw: {rows:?}"
12265 );
12266 assert!(
12267 rows.iter().any(|r| r == "two there"),
12268 "other line resolved: {rows:?}"
12269 );
12270
12271 // Move to the other paragraph: the reveal follows, and the line just
12272 // left goes back to being resolved.
12273 caret_at(&mut d, "two");
12274 let rows = drawn_rows(&d);
12275 assert!(
12276 rows.iter().any(|r| r == "*two* there"),
12277 "caret's line raw: {rows:?}"
12278 );
12279 assert!(
12280 rows.iter().any(|r| r == "one here"),
12281 "left line resolved: {rows:?}"
12282 );
12283 }
12284
12285 #[test]
12286 fn revealing_a_coloured_highlight_shows_the_emoji_that_spelled_it() {
12287 // The emoji is a delimiter, not content — so `MarkupMode::Full` owes it
12288 // the same treatment as an emphasis's `*`: hidden while the caret is
12289 // elsewhere, shown in full where the caret lands. That falls out of
12290 // `delims` reading the bytes between the mark's span and its content
12291 // span, which is exactly `==🔴 ` and `==`, rather than from a table
12292 // of spellings — so the no-space form `==🟢green==` reveals right too.
12293 let mut d = doc_in(
12294 View::Wysiwyg,
12295 "reveal_coloured_mark",
12296 "a ==🔴 red== one\n\nb ==plain== two\n",
12297 );
12298 d.set_markup_mode(MarkupMode::Full);
12299
12300 caret_at(&mut d, "red");
12301 let rows = drawn_rows(&d);
12302 assert!(
12303 rows.iter().any(|r| r == "a ==🔴 red== one"),
12304 "the caret's line shows the colour it was written with: {rows:?}"
12305 );
12306 assert!(
12307 rows.iter().any(|r| r == "b plain two"),
12308 "and every other line stays resolved: {rows:?}"
12309 );
12310
12311 // Away from it, the emoji goes back to being markup — the reader sees
12312 // the words and the wash.
12313 caret_at(&mut d, "two");
12314 let rows = drawn_rows(&d);
12315 assert!(
12316 rows.iter().any(|r| r == "a red one"),
12317 "resolved again: {rows:?}"
12318 );
12319 }
12320
12321 #[test]
12322 fn hidden_modes_never_reveal_wherever_the_caret_is() {
12323 // The two rungs below `Full` share a rendering: delimiters stay hidden
12324 // even under the caret. `Shortcuts` differing from `None` only in what
12325 // typing does is exactly the point of splitting the axes.
12326 for mode in [MarkupMode::None, MarkupMode::Shortcuts] {
12327 let mut d = doc_in(View::Wysiwyg, "reveal_hidden", "*one* here\n");
12328 d.set_markup_mode(mode);
12329 caret_at(&mut d, "one");
12330 let rows = drawn_rows(&d);
12331 assert!(
12332 rows.iter().any(|r| r == "one here"),
12333 "{mode:?} hides: {rows:?}"
12334 );
12335 assert!(
12336 !rows.iter().any(|r| r.contains('*')),
12337 "{mode:?} shows no `*`: {rows:?}"
12338 );
12339 }
12340 }
12341
12342 #[test]
12343 fn revealed_delimiters_are_the_authors_own_spelling() {
12344 // Delimiters are re-read from the source rather than synthesized per
12345 // kind, so a line comes back spelled the way it was written: `_em_` does
12346 // not turn into `*em*`, and a two-backtick fence keeps both backticks.
12347 let body = "_em_ and __st__ and ``lit ` tick`` and [lk](http://x) and ~~del~~\n";
12348 let mut d = doc_in(View::Wysiwyg, "reveal_spelling", body);
12349 d.set_markup_mode(MarkupMode::Full);
12350 caret_at(&mut d, "em");
12351 let rows = drawn_rows(&d);
12352 assert!(
12353 rows.iter().any(|r| r == body.trim_end()),
12354 "the revealed line is its own source: {rows:?}"
12355 );
12356 }
12357
12358 #[test]
12359 fn revealed_heading_shows_its_hashes() {
12360 // The `# ` marker is a block-level prefix, not an inline delimiter, so
12361 // it takes its own path — but it reveals on the same rule.
12362 let mut d = doc_in(View::Wysiwyg, "reveal_heading", "# Title\n\nbody\n");
12363 d.set_markup_mode(MarkupMode::Full);
12364
12365 caret_at(&mut d, "Title");
12366 assert!(
12367 drawn_rows(&d).iter().any(|r| r == "# Title"),
12368 "{:?}",
12369 drawn_rows(&d)
12370 );
12371
12372 caret_at(&mut d, "body");
12373 let rows = drawn_rows(&d);
12374 assert!(
12375 rows.iter().any(|r| r == "Title"),
12376 "hashes hidden again: {rows:?}"
12377 );
12378 }
12379
12380 #[test]
12381 fn revealed_delimiters_are_caret_stops() {
12382 // A delimiter that is drawn but can't be reached is worse than one
12383 // that's hidden: the mode exists so the markup can be *edited*. Every
12384 // revealed byte must be somewhere the caret can stand.
12385 let mut d = doc_in(View::Wysiwyg, "reveal_stops", "*em* x\n");
12386 d.set_markup_mode(MarkupMode::Full);
12387 caret_at(&mut d, "em");
12388 let opener = d.source.find('*').unwrap();
12389 assert!(d.vmap.is_stop(opener), "the opening `*` is a caret stop");
12390 assert!(
12391 d.vmap.is_stop(opener + 3),
12392 "the closing `*` is a caret stop"
12393 );
12394 }
12395
12396 #[test]
12397 fn setext_heading_reveals_nothing_across_its_newline() {
12398 // A setext heading's underline is on another line, so it is not the
12399 // caret line's to reveal — and emitting it would inject a `\n` glyph
12400 // that splits the row where the author wrote no break.
12401 let mut d = doc_in(View::Wysiwyg, "reveal_setext", "Title\n=====\n\nbody\n");
12402 d.set_markup_mode(MarkupMode::Full);
12403 caret_at(&mut d, "Title");
12404 let rows = drawn_rows(&d);
12405 assert!(
12406 rows.iter().any(|r| r == "Title"),
12407 "title renders alone: {rows:?}"
12408 );
12409 assert!(
12410 !rows.iter().any(|r| r.contains('=')),
12411 "no underline leaks in: {rows:?}"
12412 );
12413 }
12414
12415 #[test]
12416 fn markup_mode_axes_split_the_ladder() {
12417 // The two behaviours the ladder spells: `Shortcuts` is the middle rung
12418 // that authors markup but still hides it, and it's the only rung where
12419 // the two axes disagree.
12420 assert!(!MarkupMode::None.authors());
12421 assert!(!MarkupMode::None.reveals_caret_line());
12422 assert!(MarkupMode::Shortcuts.authors());
12423 assert!(!MarkupMode::Shortcuts.reveals_caret_line());
12424 assert!(MarkupMode::Full.authors());
12425 assert!(MarkupMode::Full.reveals_caret_line());
12426 }
12427
12428 #[test]
12429 fn indenting_an_empty_dash_item_under_text_dodges_the_setext_collapse() {
12430 // Tabbing an empty `- ` under a text line would spell `- hello\n - `,
12431 // which twig (correctly, per CommonMark — pandoc agrees) reparses as a
12432 // setext H2. leaf swaps the dash for a `*` so the item stays an empty
12433 // nested bullet and `hello` stays prose: the file round-trips instead of
12434 // hiding a heading the user never asked for.
12435 for view in [View::Source, View::Wysiwyg] {
12436 let mut d = doc_in(view, "setext_guard", "- hello\n- \n");
12437 d.caret = d.source.find("- \n").unwrap() + 2; // after the empty marker
12438 d.indent();
12439 assert_eq!(d.source, "- hello\n * \n");
12440 assert!(
12441 d.nodes().iter().all(|n| n.kind != Kind::Heading),
12442 "no heading"
12443 );
12444 // And it's genuinely a nested list, not a flat one.
12445 assert_eq!(
12446 d.nodes()
12447 .iter()
12448 .filter(|n| n.kind == Kind::BulletList)
12449 .count(),
12450 2
12451 );
12452 }
12453 }
12454
12455 #[test]
12456 fn indenting_a_dash_item_with_content_keeps_its_dash() {
12457 // With content, `- x` can't be a setext underline, so there's nothing to
12458 // dodge: the marker stays a dash and nests as an ordinary sub-bullet.
12459 let mut d = doc_in(View::Wysiwyg, "setext_ok", "- hello\n- x\n");
12460 d.caret = d.source.find('x').unwrap();
12461 d.indent();
12462 assert_eq!(d.source, "- hello\n - x\n");
12463 }
12464
12465 #[test]
12466 fn the_setext_swap_undoes_as_one_step_with_the_indent() {
12467 // The dash→`*` repair coalesces into the Tab, so a single undo restores
12468 // the whole pre-Tab state rather than stranding a half-collapsed doc.
12469 let mut d = doc_in(View::Wysiwyg, "setext_undo", "- hello\n- \n");
12470 d.caret = d.source.find("- \n").unwrap() + 2;
12471 d.indent();
12472 assert_eq!(d.source, "- hello\n * \n");
12473 d.undo();
12474 assert_eq!(d.source, "- hello\n- \n", "one undo, not two");
12475 }
12476
12477 #[test]
12478 fn indent_leaves_a_nested_lists_first_item_put_too() {
12479 // The guard is about siblings, not depth: the first item of an *inner*
12480 // list (already nested under `a`) still has nothing before it at its own
12481 // level, so Tab can't take it deeper.
12482 let mut d = doc_in(View::Wysiwyg, "indent_first_nested", "- a\n - b\n - c\n");
12483 d.caret = d.source.find('b').unwrap();
12484 d.indent();
12485 assert_eq!(d.source, "- a\n - b\n - c\n", "inner first item holds");
12486 // But `c` (a sibling of `b`) nests under `b`.
12487 d.caret = d.source.find('c').unwrap();
12488 d.indent();
12489 assert_eq!(d.source, "- a\n - b\n - c\n");
12490 }
12491
12492 #[test]
12493 fn backspace_at_a_nested_item_start_outdents_it() {
12494 // Backspace with the caret right after a nested item's marker gives back
12495 // one level of nesting, the mirror of Tab — and renumbers the flattened
12496 // ordered list back to a clean run.
12497 let mut d = doc_in(View::Wysiwyg, "bsp_outdent", "1. a\n 1. b\n2. c\n");
12498 d.caret = d.source.find('b').unwrap(); // start of the nested item's content
12499 d.backspace();
12500 assert_eq!(d.source, "1. a\n2. b\n3. c\n");
12501 }
12502
12503 #[test]
12504 fn backspace_at_a_top_level_item_start_strips_the_marker() {
12505 // At the outermost level there's no nesting left to give back, so the same
12506 // keystroke drops the bullet and leaves a plain paragraph.
12507 let mut d = doc_in(View::Wysiwyg, "bsp_strip", "- a\n- b\n");
12508 d.caret = d.source.find('b').unwrap(); // right after `- `
12509 d.backspace();
12510 assert_eq!(d.source, "- a\nb\n", "the marker is gone, the text stays");
12511 }
12512
12513 #[test]
12514 fn backspace_mid_item_still_deletes_a_character() {
12515 // The list behaviour is armed only at the item's content start; anywhere
12516 // else Backspace is the ordinary character delete.
12517 let mut d = doc_in(View::Wysiwyg, "bsp_mid", "- ab\n");
12518 d.caret = d.source.find('b').unwrap(); // between `a` and `b`
12519 d.backspace();
12520 assert_eq!(d.source, "- b\n");
12521 }
12522
12523 #[test]
12524 fn backspace_at_a_heading_start_strips_the_marker() {
12525 // The `# ` is markup the rich view hides, so Backspace over it takes the
12526 // whole marker and leaves a paragraph. Deleting a byte of it instead left
12527 // `#Title` — no longer a heading, with the hash now literal text the user
12528 // never typed and has to delete again.
12529 let mut d = doc_in(View::Wysiwyg, "bsp_head", "## Title\n");
12530 d.caret = d.source.find('T').unwrap(); // right after `## `
12531 d.backspace();
12532 assert_eq!(d.source, "Title\n");
12533 assert_eq!(
12534 d.caret, 0,
12535 "the caret stays with the text it was in front of"
12536 );
12537 }
12538
12539 #[test]
12540 fn backspace_at_a_heading_start_keeps_the_block_around_it() {
12541 // Only the heading's own marker goes — the quote (or list) it sits in is
12542 // untouched, exactly as un-heading it should be.
12543 let mut d = doc_in(View::Wysiwyg, "bsp_head_quote", "> # Title\n");
12544 d.caret = d.source.find('T').unwrap();
12545 d.backspace();
12546 assert_eq!(d.source, "> Title\n");
12547 }
12548
12549 #[test]
12550 fn backspace_at_a_heading_start_takes_its_closing_sequence_too() {
12551 // `# Title #`'s trailing hashes are hidden at the other end; leaving them
12552 // behind would surface the same stray hash the marker delete just avoided.
12553 let mut d = doc_in(View::Wysiwyg, "bsp_head_closed", "# Title #\n");
12554 d.caret = d.source.find('T').unwrap();
12555 d.backspace();
12556 assert_eq!(d.source, "Title\n");
12557 // And it's one edit: a single undo puts the whole heading back.
12558 d.undo();
12559 assert_eq!(d.source, "# Title #\n");
12560 }
12561
12562 #[test]
12563 fn backspace_mid_heading_still_deletes_a_character() {
12564 // The heading behaviour is armed only at the content's start; anywhere
12565 // else Backspace is the ordinary character delete.
12566 let mut d = doc_in(View::Wysiwyg, "bsp_head_mid", "# ab\n");
12567 d.caret = d.source.find('b').unwrap();
12568 d.backspace();
12569 assert_eq!(d.source, "# b\n");
12570 }
12571
12572 #[test]
12573 fn source_view_backspace_still_edits_the_heading_marker_literally() {
12574 // In source view the `# ` is text on the screen the user is deleting a
12575 // byte of, so it keeps its literal meaning — the same split the list
12576 // ladder and Enter draw between the two views.
12577 let mut d = doc_with("bsp_head_src", "# Title\n");
12578 d.caret = d.source.find('T').unwrap();
12579 d.backspace();
12580 assert_eq!(d.source, "#Title\n");
12581 }
12582
12583 #[test]
12584 fn outdent_unnests_an_ordered_item_in_one_press() {
12585 // Shift+Tab gives back exactly the marker width the indent added, so a
12586 // nested ordered item unnests in a single press, and the flattened list
12587 // renumbers back to a clean 1, 2, 3.
12588 let mut d = doc_with("outdent_ord", "1. a\n 2. b\n3. c\n");
12589 d.caret = d.source.find('b').unwrap();
12590 d.outdent();
12591 assert_eq!(d.source, "1. a\n2. b\n3. c\n");
12592 let lists = d
12593 .nodes()
12594 .iter()
12595 .filter(|n| n.kind == Kind::OrderedList)
12596 .count();
12597 assert_eq!(lists, 1, "back to one flat list");
12598 }
12599
12600 #[test]
12601 fn table_insert_row_adds_a_row_below_the_caret() {
12602 let mut d = doc_with("tbl_ins_row", "| a | b |\n| --- | --- |\n| 1 | 2 |\n");
12603 d.caret = d.source.find('1').unwrap(); // in the body row
12604 d.table_insert_row(true);
12605 assert_eq!(d.source, "| a | b |\n| --- | --- |\n| 1 | 2 |\n| | |\n");
12606 }
12607
12608 #[test]
12609 fn table_insert_and_delete_column_at_the_caret() {
12610 let mut d = doc_with("tbl_col", "| a | b |\n| --- | --- |\n| 1 | 2 |\n");
12611 d.caret = d.source.find('a').unwrap(); // column 0
12612 d.table_insert_column(true); // add a column to the right of `a`
12613 assert_eq!(
12614 d.source,
12615 "| a | | b |\n| --- | --- | --- |\n| 1 | | 2 |\n"
12616 );
12617 d.caret = d.source.find('b').unwrap(); // now the third column
12618 d.table_delete_column();
12619 assert_eq!(d.source, "| a | |\n| --- | --- |\n| 1 | |\n");
12620 }
12621
12622 // ── ragged formats ───────────────────────────────────────────────────────
12623 // No format spells every gesture. HTML writes the inline marks as a tag pair
12624 // and no heading, list, quote or link; Markdown spells five of the eight
12625 // marks — the highlight only because leaf parses with `highlight`, which is
12626 // why the question is asked with the extensions; djot spells all eight and
12627 // no in-cell break. leaf asks twig per
12628 // gesture (`Doc::supports`) and refuses at the door, rather than letting each
12629 // op discover the fact on its own — one of them didn't.
12630
12631 /// An HTML document in the rich view, ready for a gesture.
12632 fn html_doc(body: &str) -> Doc {
12633 let mut d = Doc::from_source(body.to_string(), Format::Html).unwrap();
12634 d.view = View::Wysiwyg;
12635 d.build_visual(80);
12636 d
12637 }
12638
12639 #[test]
12640 fn a_table_gesture_leaves_an_html_table_alone() {
12641 // The regression this guard exists for. twig's table editor consults no
12642 // `Syntax` table — it spells a grid, not a delimiter — so it rebuilt an
12643 // HTML `<table>` as a *pipe table* and reported success: the whole
12644 // element replaced by `| a | b |`, silently, on one press of a toolbar
12645 // button. Every grid op went the same way.
12646 let src = "<table><tr><td>a</td><td>b</td></tr><tr><td>c</td><td>d</td></tr></table>\n";
12647 // A table of named operations, which is what it looks like.
12648 #[allow(clippy::type_complexity)]
12649 let ops: [(&str, &dyn Fn(&mut Doc)); 7] = [
12650 ("insert row", &|d: &mut Doc| d.table_insert_row(true)),
12651 ("delete row", &|d: &mut Doc| d.table_delete_row()),
12652 ("insert column", &|d: &mut Doc| d.table_insert_column(true)),
12653 ("delete column", &|d: &mut Doc| d.table_delete_column()),
12654 ("align", &|d: &mut Doc| {
12655 d.table_set_alignment(Alignment::Right)
12656 }),
12657 ("move row", &|d: &mut Doc| d.table_move_row(true)),
12658 ("move column", &|d: &mut Doc| d.table_move_column(true)),
12659 ];
12660 for (name, op) in ops {
12661 let mut d = html_doc(src);
12662 d.caret = d.source.find('a').unwrap();
12663 assert!(d.caret_in_table(), "{name}: the caret really is in a table");
12664 op(&mut d);
12665 assert_eq!(d.source, src, "{name} rewrote an HTML table");
12666 assert!(
12667 !d.dirty,
12668 "{name} marked the document dirty without editing it"
12669 );
12670 assert!(d.status.is_some(), "{name} refused without saying why");
12671 }
12672 }
12673
12674 #[test]
12675 fn the_block_gestures_html_cannot_spell_are_refused_with_a_reason() {
12676 // A task box is a form control in HTML and a footnote has no native
12677 // spelling at all — the two gestures twig 3.5 still spells nothing
12678 // for, now that a quote, a list, a link and an image print through
12679 // its renderer (see the test below).
12680 let src = "<h1>Title</h1>\n<p>Hello world</p>\n<ul><li>one</li></ul>\n";
12681 // A table of named operations, which is what it looks like.
12682 #[allow(clippy::type_complexity)]
12683 let ops: [(&str, &dyn Fn(&mut Doc)); 3] = [
12684 ("task item", &|d: &mut Doc| d.toggle_task_item()),
12685 ("task tick", &|d: &mut Doc| d.toggle_task_checked()),
12686 ("footnote", &|d: &mut Doc| d.insert_footnote()),
12687 ];
12688 for (name, op) in ops {
12689 let mut d = html_doc(src);
12690 let at = d.source.find("Hello").unwrap();
12691 d.caret = at;
12692 d.anchor = Some(at + 5); // a selection, for the ops that want one
12693 op(&mut d);
12694 assert_eq!(d.source, src, "{name} edited an HTML document");
12695 assert!(
12696 !d.dirty,
12697 "{name} marked the document dirty without editing it"
12698 );
12699 let status = d.status.as_deref().unwrap_or("");
12700 assert!(
12701 status.contains("html"),
12702 "{name}: the refusal should name the format, got {status:?}"
12703 );
12704 }
12705 }
12706
12707 #[test]
12708 fn html_spells_a_quote_a_list_a_link_and_an_image_through_the_renderer() {
12709 // twig 3.5: where HTML has no marker alphabet it prints the fresh
12710 // node — a `<blockquote>` around the paragraph, a `<ul>`/`<ol>` with
12711 // the paragraph as its item, an `<a>` or `<img>` over the selection.
12712 // Until then every one of these was a refusal; now each is a real
12713 // edit, which is what the toolbar's capability flags say too.
12714 let src = "<h1>Title</h1>\n<p>Hello world</p>\n<ul><li>one</li></ul>\n";
12715 #[allow(clippy::type_complexity)]
12716 let ops: [(&str, &dyn Fn(&mut Doc), &str); 5] = [
12717 (
12718 "quote",
12719 &|d: &mut Doc| d.toggle_blockquote(),
12720 "<blockquote>",
12721 ),
12722 ("list", &|d: &mut Doc| d.toggle_list(false), "<ul>\n<li>"),
12723 (
12724 "ordered list",
12725 &|d: &mut Doc| d.toggle_list(true),
12726 "<ol>\n<li>",
12727 ),
12728 (
12729 "link",
12730 &|d: &mut Doc| d.insert_link("https://example.dev"),
12731 "<a href=\"https://example.dev\">Hello</a>",
12732 ),
12733 (
12734 "image",
12735 &|d: &mut Doc| d.insert_image("pic.png", "alt"),
12736 "<img alt=\"Hello\" src=\"pic.png\">",
12737 ),
12738 ];
12739 for (name, op, expect) in ops {
12740 let mut d = html_doc(src);
12741 let at = d.source.find("Hello").unwrap();
12742 d.caret = at;
12743 d.anchor = Some(at + 5);
12744 op(&mut d);
12745 assert!(d.source.contains(expect), "{name}: got {:?}", d.source);
12746 assert!(d.dirty, "{name}: a real edit");
12747 assert_eq!(
12748 d.status, None,
12749 "{name}: a supported gesture reports nothing"
12750 );
12751 }
12752 }
12753
12754 #[test]
12755 fn html_spells_a_heading_as_its_tag_pair() {
12756 // twig 3.4 rebuilds a heading or paragraph as its tag pair, attributes
12757 // along — the one block gesture whose HTML shape it can write. So ⌘2
12758 // in an HTML document is a real edit, and ⌘0 takes it back.
12759 let src = "<h1>Title</h1>\n<p>Hello world</p>\n";
12760 let mut d = html_doc(src);
12761 d.caret = d.source.find("Hello").unwrap();
12762 d.toggle_heading(2);
12763 assert_eq!(d.source, "<h1>Title</h1>\n<h2>Hello world</h2>\n");
12764 assert!(d.dirty);
12765 assert_eq!(d.status, None, "a supported gesture reports nothing");
12766 d.toggle_heading(2);
12767 assert_eq!(d.source, src, "the same level again is back to a paragraph");
12768 }
12769
12770 #[test]
12771 fn html_spells_the_inline_marks_and_the_rule() {
12772 // The other half, and why one per-document flag stopped being enough:
12773 // ⌘B in an HTML document writes `<strong>` — the tag the serializer
12774 // already emits and the parser reads straight back as the same mark —
12775 // and the rule button writes an `<hr>`. Refusing these on the old
12776 // "HTML is parse-only" reading would now be leaf's own limitation.
12777 let mut d = html_doc("<p>Hello world</p>\n");
12778 let at = d.source.find("world").unwrap();
12779 d.caret = at;
12780 d.anchor = Some(at + 5);
12781 d.toggle(InlineKind::Strong);
12782 assert_eq!(d.source, "<p>Hello <strong>world</strong></p>\n");
12783 assert!(d.dirty);
12784 assert_eq!(d.status, None, "a supported gesture reports nothing");
12785
12786 // And off again — the toggle reverses, which is the property that makes
12787 // authoring in HTML worth offering rather than a one-way trip.
12788 d.toggle(InlineKind::Strong);
12789 assert_eq!(d.source, "<p>Hello world</p>\n");
12790
12791 let mut d = html_doc("<p>Hello world</p>\n");
12792 d.caret = d.source.find("world").unwrap();
12793 d.insert_thematic_break();
12794 assert!(d.source.contains("<hr>"), "got {:?}", d.source);
12795 }
12796
12797 #[test]
12798 fn a_mark_the_format_cannot_spell_arms_nothing() {
12799 // `toggle` with a collapsed caret doesn't reach twig at all — it arms a
12800 // sticky mark for the next text typed. Guarding only the twig call
12801 // leaves that path live, promising a mark the gesture will not write and
12802 // then swallowing the error inside `insert`.
12803 //
12804 // Markdown carries this, on the superscript now rather than on the
12805 // highlight: `^x^` is text there in any configuration, whereas twig
12806 // 3.3.1 authors `==x==` for an editor holding the `highlight` extension,
12807 // which every leaf document does.
12808 let mut d = doc_with("mark", "Hello world\n");
12809 d.view = View::Wysiwyg;
12810 d.build_visual(80);
12811 d.caret = d.source.find("world").unwrap();
12812 d.toggle(InlineKind::Superscript);
12813 assert!(d.pending_marks.is_empty(), "no mark should be armed");
12814 assert!(d.status.as_deref().unwrap_or("").contains("markdown"));
12815 d.insert("X");
12816 assert_eq!(d.source, "Hello Xworld\n");
12817 }
12818
12819 #[test]
12820 fn markdown_authors_a_highlight_and_a_strikethrough() {
12821 // twig 3.3.1: the two marks Markdown reads and, until it, refused to
12822 // write. `==x==` is authorable because leaf's own `parse_extensions`
12823 // turns `highlight` on — twig will only mint bytes this editor's reparse
12824 // reads back — and `~~x~~` because GFM strikethrough is parsed by
12825 // default, so the refusal there was never right for any leaf document.
12826 for (kind, marked) in [
12827 (InlineKind::Mark, "a ==word== b\n"),
12828 (InlineKind::Delete, "a ~~word~~ b\n"),
12829 ] {
12830 let mut d = doc_with("author_mark", "a word b\n");
12831 d.anchor = Some(2);
12832 d.caret = 6;
12833 d.toggle(kind);
12834 assert_eq!(d.source, marked, "{kind:?}");
12835 assert_eq!(d.status, None, "{kind:?}: a supported gesture is silent");
12836 assert!(d.dirty, "{kind:?}");
12837 // The region stays selected, so the second press reverses it — the
12838 // property that separates authoring from a one-way trip.
12839 d.toggle(kind);
12840 assert_eq!(d.source, "a word b\n", "{kind:?}");
12841 }
12842 }
12843
12844 #[test]
12845 fn an_authored_highlight_reads_back_as_a_mark() {
12846 // The round trip the extension gate exists to protect: what the toggle
12847 // writes, the reparse must read back as a `mark` rather than as two
12848 // literal `=` pairs. A `Role::Mark` glyph is that answer, taken from the
12849 // rebuilt map rather than from the source text.
12850 let mut d = doc_with("mark_roundtrip", "a word b\n");
12851 d.view = View::Wysiwyg;
12852 d.build_visual(80);
12853 d.anchor = Some(2);
12854 d.caret = 6;
12855 d.toggle(InlineKind::Mark);
12856 assert_eq!(d.source, "a ==word== b\n");
12857 d.build_visual(80);
12858 let w = d
12859 .vmap
12860 .rows
12861 .iter()
12862 .flat_map(|r| r.glyphs.iter())
12863 .find(|g| g.ch == 'w')
12864 .expect("the highlighted word");
12865 assert_eq!(w.style.role, crate::Role::Mark(None));
12866 }
12867
12868 #[test]
12869 fn a_highlight_takes_a_colour_changes_it_and_gives_it_back() {
12870 // The three states of one gesture, in the order a palette is pressed:
12871 // an uncoloured highlight takes the prefix, a coloured one has it
12872 // replaced, and `None` takes it away with the space that was part of the
12873 // spelling.
12874 let mut d = doc_with("mark_colour", "a ==word== b\n");
12875 d.caret = d.source.find("word").unwrap();
12876 d.set_mark_color(Some(MarkColor::Red));
12877 assert_eq!(d.source, "a ==🔴 word== b\n");
12878 assert_eq!(d.status, None);
12879 assert!(d.dirty);
12880
12881 d.set_mark_color(Some(MarkColor::Blue));
12882 assert_eq!(d.source, "a ==🔵 word== b\n");
12883
12884 d.set_mark_color(None);
12885 assert_eq!(d.source, "a ==word== b\n");
12886 }
12887
12888 #[test]
12889 fn the_caret_keeps_its_place_in_the_text_across_a_colour() {
12890 // The prefix is written *before* the word, so an offset in the word has
12891 // to ride its width — a caret that stayed put would be a caret that
12892 // walked backwards through the text it was standing in.
12893 let mut d = doc_with("mark_colour_caret", "a ==word== b\n");
12894 let word = d.source.find("word").unwrap();
12895 d.caret = word + 2; // between `wo` and `rd`
12896 d.set_mark_color(Some(MarkColor::Red));
12897 assert_eq!(&d.source[d.caret..d.caret + 2], "rd", "still before `rd`");
12898
12899 // And back the other way when the prefix goes.
12900 d.set_mark_color(None);
12901 assert_eq!(&d.source[d.caret..d.caret + 2], "rd");
12902 }
12903
12904 #[test]
12905 fn the_colour_at_the_caret_is_what_the_palette_lights() {
12906 let mut d = doc_with("mark_colour_read", "a ==🔴 red== and ==plain== b\n");
12907 d.caret = d.source.find("red").unwrap();
12908 assert!(d.caret_in_mark());
12909 assert_eq!(d.mark_color_at_caret(), Some(MarkColor::Red));
12910
12911 d.caret = d.source.find("plain").unwrap();
12912 assert!(d.caret_in_mark(), "a highlight with no colour is still one");
12913 assert_eq!(d.mark_color_at_caret(), None);
12914
12915 d.caret = d.source.find(" and ").unwrap() + 2;
12916 assert!(!d.caret_in_mark());
12917 assert_eq!(d.mark_color_at_caret(), None);
12918 }
12919
12920 #[test]
12921 fn a_colour_without_a_highlight_says_so_and_writes_nothing() {
12922 // The gesture colours a highlight that exists; it does not make one.
12923 // Two presses is the price of a coloured highlight from bare text, and
12924 // the reason is undo — one press that spliced twice would take two
12925 // presses to take back.
12926 let mut d = doc_with("mark_colour_none", "a word b\n");
12927 d.caret = d.source.find("word").unwrap();
12928 d.set_mark_color(Some(MarkColor::Red));
12929 assert_eq!(d.source, "a word b\n");
12930 assert!(d.status.is_some(), "it should say why");
12931 assert!(!d.dirty);
12932
12933 // Clearing where there is nothing to clear is the same refusal, not a
12934 // quiet success — the caret is in no highlight either way.
12935 d.status = None;
12936 d.set_mark_color(None);
12937 assert_eq!(d.source, "a word b\n");
12938 assert!(d.status.is_some());
12939 }
12940
12941 #[test]
12942 fn clearing_an_uncoloured_highlight_is_a_quiet_no_op() {
12943 // twig answers this one *successfully* with a `Change` describing some
12944 // earlier edit, so a caller that trusted the change would jump the caret
12945 // to wherever that was. Core answers it before asking.
12946 let mut d = doc_with("mark_colour_noop", "a ==word== b\n");
12947 d.toggle(InlineKind::Strong); // an earlier edit for a stale change to name
12948 d.caret = d.source.find("word").unwrap();
12949 let (source, caret) = (d.source.clone(), d.caret);
12950 d.set_mark_color(None);
12951 assert_eq!(d.source, source);
12952 assert_eq!(
12953 d.caret, caret,
12954 "the caret must not ride a change that isn't one"
12955 );
12956 assert_eq!(d.status, None, "and it is not an error either");
12957 }
12958
12959 #[test]
12960 fn djot_spells_the_highlight_and_not_its_colour() {
12961 // The reason the palette is its own capability rather than the Highlight
12962 // button's: `{=word=}` is a highlight djot writes happily, and there is
12963 // no djot spelling for a colour on it.
12964 assert!(Capabilities::of(Format::Djot).mark);
12965 assert!(!Capabilities::of(Format::Djot).mark_color);
12966 assert!(Capabilities::of(Format::Markdown).mark_color);
12967
12968 let mut d = Doc::from_source("a {=word=} b\n".into(), Format::Djot).unwrap();
12969 d.caret = d.source.find("word").unwrap();
12970 assert!(
12971 d.caret_in_mark(),
12972 "the caret is in a highlight all the same"
12973 );
12974 d.set_mark_color(Some(MarkColor::Red));
12975 assert_eq!(d.source, "a {=word=} b\n");
12976 assert!(
12977 d.status.as_deref().unwrap_or("").contains("djot"),
12978 "and the refusal names the document's format: {:?}",
12979 d.status
12980 );
12981 }
12982
12983 #[test]
12984 fn a_coloured_highlight_is_one_undo_step_and_reads_back_as_its_colour() {
12985 // The round trip that matters for a palette: the bytes twig writes are
12986 // bytes its own reparse reads back as a colour, so the swatch that was
12987 // pressed is the swatch that lights afterwards.
12988 let mut d = doc_with("mark_colour_undo", "a word b\n");
12989 d.anchor = Some(2);
12990 d.caret = 6;
12991 d.toggle(InlineKind::Mark);
12992 d.caret = d.source.find("word").unwrap();
12993 d.set_mark_color(Some(MarkColor::Green));
12994 assert_eq!(d.source, "a ==🟢 word== b\n");
12995 assert_eq!(d.mark_color_at_caret(), Some(MarkColor::Green));
12996
12997 // One splice, one step: the colour comes off and the highlight stays.
12998 d.undo();
12999 assert_eq!(d.source, "a ==word== b\n");
13000 d.undo();
13001 assert_eq!(d.source, "a word b\n");
13002 }
13003
13004 #[test]
13005 fn every_colour_leaf_names_is_one_twig_writes() {
13006 // The two enums are one vocabulary, and this is what says so: each of
13007 // leaf's colours writes an emoji twig's reparse reads back as *that*
13008 // colour, so `twig_mark_color`'s table cannot quietly pair red with
13009 // orange.
13010 for color in MarkColor::ALL {
13011 let mut d = doc_with("mark_colour_all", "a ==word== b\n");
13012 d.caret = d.source.find("word").unwrap();
13013 d.set_mark_color(Some(color));
13014 assert_eq!(d.status, None, "{color:?}");
13015 assert_eq!(d.mark_color_at_caret(), Some(color), "{color:?}");
13016 }
13017 }
13018
13019 #[test]
13020 fn a_fresh_highlight_takes_a_colour_without_moving_the_caret_first() {
13021 // The two presses a coloured highlight is made of, in the state the
13022 // first one leaves: `toggle` selects the whole `==word==` and puts the
13023 // caret one past the closing `==`, which is *not* in the mark. Asking at
13024 // the caret alone would refuse to colour the highlight just written —
13025 // the selection's start is what answers.
13026 let mut d = doc_with("mark_colour_fresh", "a word b\n");
13027 d.anchor = Some(2);
13028 d.caret = 6;
13029 d.toggle(InlineKind::Mark);
13030 assert_eq!(d.source, "a ==word== b\n");
13031 assert_eq!(d.caret, 10, "the caret twig leaves, past the closing `==`");
13032
13033 assert!(d.caret_in_mark(), "the selected highlight is the one meant");
13034 d.set_mark_color(Some(MarkColor::Yellow));
13035 assert_eq!(d.source, "a ==🟡 word== b\n");
13036 assert_eq!(d.status, None);
13037 }
13038
13039 #[test]
13040 fn one_press_highlights_a_selection_and_colours_it() {
13041 // What a toolbar swatch means over a plain selection, and the undo it
13042 // has to have: one press, one step. Two steps would leave an uncoloured
13043 // highlight behind on the way back, which is a state the author never
13044 // asked for and never saw.
13045 let mut d = doc_with("highlight_one", "a word b\n");
13046 d.anchor = Some(2);
13047 d.caret = 6;
13048 d.highlight(Some(MarkColor::Purple));
13049 assert_eq!(d.source, "a ==\u{1F7E3} word== b\n");
13050 assert_eq!(d.status, None);
13051
13052 d.undo();
13053 assert_eq!(d.source, "a word b\n", "one press, one undo");
13054 }
13055
13056 #[test]
13057 fn one_press_on_an_existing_highlight_only_recolours_it() {
13058 // The other half: inside a highlight there is nothing to make, so the
13059 // compound is the plain gesture and the text is untouched.
13060 let mut d = doc_with("highlight_recolour", "a ==\u{1F534} word== b\n");
13061 d.caret = d.source.find("word").unwrap();
13062 d.highlight(Some(MarkColor::Blue));
13063 assert_eq!(d.source, "a ==\u{1F535} word== b\n");
13064 d.undo();
13065 assert_eq!(d.source, "a ==\u{1F534} word== b\n", "the highlight stays");
13066 }
13067
13068 #[test]
13069 fn one_press_with_no_colour_over_a_selection_just_highlights_it() {
13070 // `None` means "no colour", and over bare text that is the Highlight
13071 // button's own job. The fold must not happen here — there is no second
13072 // splice, and folding would take the *previous* edit into this one.
13073 let mut d = doc_with("highlight_none", "a word b and more\n");
13074 d.caret = d.source.find("more").unwrap() + 4; // after "more"
13075 d.insert("!"); // an earlier edit for a wrong fold to swallow
13076 d.anchor = Some(2);
13077 d.caret = 6;
13078 d.highlight(None);
13079 assert_eq!(d.source, "a ==word== b and more!\n");
13080
13081 d.undo();
13082 assert_eq!(
13083 d.source, "a word b and more!\n",
13084 "only the highlight came off"
13085 );
13086 d.undo();
13087 assert_eq!(
13088 d.source, "a word b and more\n",
13089 "and the edit before it survived"
13090 );
13091 }
13092
13093 #[test]
13094 fn one_press_at_a_bare_caret_in_no_highlight_writes_nothing() {
13095 // `toggle` at a collapsed caret arms a mark for text not yet typed, and
13096 // a colour cannot be armed with it — so the compound declines rather
13097 // than leaving half a promise.
13098 let mut d = doc_with("highlight_bare", "a word b\n");
13099 d.caret = 4;
13100 d.highlight(Some(MarkColor::Red));
13101 assert_eq!(d.source, "a word b\n");
13102 assert!(d.pending_marks.is_empty(), "and nothing armed");
13103 assert!(d.status.is_some());
13104 }
13105
13106 #[test]
13107 fn a_read_only_document_takes_no_colour() {
13108 let mut d = doc_with("mark_colour_ro", "a ==word== b\n");
13109 d.caret = d.source.find("word").unwrap();
13110 d.set_read_only(true);
13111 d.set_mark_color(Some(MarkColor::Red));
13112 assert_eq!(d.source, "a ==word== b\n");
13113 }
13114
13115 #[test]
13116 fn a_sticky_highlight_wraps_the_next_typed_text_in_markdown() {
13117 // The other door into `toggle`: no selection, so nothing reaches twig
13118 // until `insert` realises the armed mark. It is armed now — the guard
13119 // above asks `Doc::supports`, which asks with the extensions — and what
13120 // it writes is the same `==…==`.
13121 let mut d = doc_with("sticky_mark", "xy\n");
13122 d.caret = 1;
13123 d.toggle(InlineKind::Mark);
13124 assert!(d.pending_marks.contains(InlineKind::Mark));
13125 d.insert("Z");
13126 assert_eq!(d.source, "x==Z==y\n");
13127 }
13128
13129 #[test]
13130 fn html_documents_still_take_typed_text() {
13131 // The guard covers *markup* gestures and must not touch plain editing:
13132 // twig's splicer is language-neutral, and typing into an HTML document
13133 // is the thing that does work today.
13134 let mut d = html_doc("<p>Hello world</p>\n");
13135 d.caret = d.source.find("world").unwrap();
13136 d.insert("big ");
13137 assert_eq!(d.source, "<p>Hello big world</p>\n");
13138 assert!(d.dirty);
13139 d.backspace();
13140 assert_eq!(d.source, "<p>Hello bigworld</p>\n");
13141 d.undo();
13142 d.undo();
13143 assert_eq!(d.source, "<p>Hello world</p>\n");
13144 }
13145
13146 #[test]
13147 fn authorable_is_the_coarse_question_and_capabilities_the_useful_one() {
13148 // `authorable` only separates "there is a door in" from "there is not",
13149 // and HTML is on the near side of that line — which is exactly why a
13150 // toolbar must not be built from it.
13151 let html = Doc::from_source("<p>x</p>\n".into(), Format::Html).unwrap();
13152 assert!(html.authorable());
13153 assert!(
13154 !Doc::from_source("<r>x</r>".into(), Format::Xml)
13155 .unwrap()
13156 .authorable()
13157 );
13158
13159 let caps = html.capabilities();
13160 assert!(caps.bold && caps.italic && caps.code && caps.mark);
13161 assert!(caps.thematic_break && caps.cell_line_break);
13162 // A heading is a tag pair twig rebuilds (3.4), and since 3.5 so are a
13163 // quote, a list, a code block's language, a link and an image — each
13164 // printed as a fresh node where HTML has no marker to rewrite. A task
13165 // box is a form control and a footnote has no spelling, so those two
13166 // are what keeps the record ragged.
13167 assert!(caps.heading && caps.blockquote && caps.bullet_list);
13168 assert!(caps.link && caps.image && caps.code_language);
13169 assert!(!caps.task && !caps.footnote);
13170 // The one flag that isn't twig's answer: an HTML `<table>` is a grid
13171 // twig's table editor would happily re-emit as `| a | b |`.
13172 assert!(!caps.table);
13173
13174 // The two lightweight formats spell everything leaf offers — and still
13175 // differ from each other, which is the other half of why one boolean
13176 // can't serve.
13177 for fmt in [Format::Markdown, Format::Djot] {
13178 let caps = Capabilities::of(fmt);
13179 assert!(
13180 caps.heading && caps.blockquote && caps.ordered_list,
13181 "{fmt:?}"
13182 );
13183 assert!(
13184 caps.task && caps.link && caps.image && caps.table,
13185 "{fmt:?}"
13186 );
13187 }
13188 // Both spell the highlight and the strikethrough: djot natively, and
13189 // Markdown because `Capabilities` asks with `parse_extensions` rather
13190 // than with twig's defaults — `==x==` is text under those, and a mark
13191 // under the `highlight` leaf always parses with.
13192 for fmt in [Format::Markdown, Format::Djot] {
13193 let caps = Capabilities::of(fmt);
13194 assert!(caps.mark && caps.strike, "{fmt:?}");
13195 }
13196 // What still separates them, now that the highlight doesn't: djot has
13197 // no in-cell break, and Markdown spells neither of the scripts.
13198 assert!(Capabilities::of(Format::Djot).superscript);
13199 assert!(!Capabilities::of(Format::Markdown).superscript);
13200 assert!(Capabilities::of(Format::Markdown).cell_line_break);
13201 assert!(!Capabilities::of(Format::Djot).cell_line_break);
13202
13203 // A parse-only format answers no to every one of them, so the coarse
13204 // predicate and the record agree there.
13205 let caps = Capabilities::of(Format::Xml);
13206 assert!(!caps.bold && !caps.heading && !caps.table && !caps.thematic_break);
13207 }
13208
13209 #[test]
13210 fn a_refused_gesture_says_so_where_twig_would_have_said_it() {
13211 // The guard exists to name the *document's* format rather than twig's
13212 // internals, so the message has to survive being one leaf writes itself.
13213 // Checked against a gesture twig also refuses, since that is the pair
13214 // most at risk of drifting apart — the task box, once the code
13215 // language stopped being one (twig 3.5).
13216 let mut d = html_doc("<p>Hello</p>\n");
13217 d.caret = d.source.find("Hello").unwrap();
13218 d.toggle_task_item();
13219 assert_eq!(d.status.as_deref(), Some("task: not supported in html"));
13220 assert!(!d.dirty);
13221 }
13222
13223 #[test]
13224 fn table_set_alignment_respells_the_delimiter() {
13225 let mut d = doc_with("tbl_align", "| a | b |\n| --- | --- |\n| 1 | 2 |\n");
13226 d.caret = d.source.find('b').unwrap();
13227 d.table_set_alignment(Alignment::Right);
13228 assert_eq!(d.source, "| a | b |\n| --- | ---: |\n| 1 | 2 |\n");
13229 }
13230
13231 #[test]
13232 fn each_empty_table_cell_has_its_own_editable_home() {
13233 // Regression: an empty cell has no twig content_span, so both cells of a
13234 // `| | |` row collapsed onto the row's start (before the first `│`).
13235 // Typing there inserted *before* the table (`hello| | |`); nav couldn't
13236 // tell the cells apart. Each empty cell must now have a distinct home
13237 // inside it.
13238 let mut d = wysiwyg_doc("tbl_empty", "| a | b |\n| --- | --- |\n| | |\n");
13239 let (c0, c1) = {
13240 let cells = &d.vmap.tables[0].grid[1].cells;
13241 (cells[0].start, cells[1].start)
13242 };
13243 assert!(
13244 c0 < c1,
13245 "the two empty cells have distinct homes: {c0} < {c1}"
13246 );
13247 d.caret = c0;
13248 d.insert("x");
13249 assert_eq!(
13250 d.source, "| a | b |\n| --- | --- |\n| x | |\n",
13251 "typed inside the cell"
13252 );
13253 }
13254
13255 #[test]
13256 fn arrows_step_into_each_empty_table_cell() {
13257 let mut d = wysiwyg_doc("tbl_empty_nav", "| a | b |\n| --- | --- |\n| | |\n");
13258 let (c0, c1) = {
13259 let cells = &d.vmap.tables[0].grid[1].cells;
13260 (cells[0].start, cells[1].start)
13261 };
13262 d.caret = d.source.find('b').unwrap(); // in the header's second cell
13263 let mut seen = std::collections::HashSet::new();
13264 for _ in 0..6 {
13265 d.move_right(false);
13266 seen.insert(d.caret);
13267 }
13268 assert!(
13269 seen.contains(&c0),
13270 "right arrow reaches the first empty cell"
13271 );
13272 assert!(
13273 seen.contains(&c1),
13274 "right arrow reaches the second empty cell"
13275 );
13276 }
13277
13278 #[test]
13279 fn table_op_off_a_table_is_a_no_op_with_a_status() {
13280 let mut d = doc_with("tbl_none", "just text\n");
13281 d.caret = 3;
13282 d.table_insert_row(true);
13283 assert_eq!(d.source, "just text\n", "nothing changed");
13284 assert!(d.status.is_some(), "a status explains why");
13285 assert!(!d.caret_in_table());
13286 }
13287
13288 #[test]
13289 fn enter_in_an_ordered_list_renumbers_the_following_items() {
13290 // Inserting an item mid-list left the source markers stale (`1. 2. 2. 3.`);
13291 // the renumber pass keeps them sequential, matching what the view draws.
13292 let mut d = wysiwyg_doc("enter_renumber", "1. a\n2. b\n3. c\n");
13293 d.caret = d.source.find('a').unwrap() + 1; // end of item a
13294 d.newline();
13295 d.insert("x");
13296 assert_eq!(d.source, "1. a\n2. x\n3. b\n4. c\n");
13297 }
13298
13299 #[test]
13300 fn outdent_with_nothing_to_give_back_records_no_undo_step() {
13301 for view in [View::Source, View::Wysiwyg] {
13302 let mut d = doc_in(view, "outdent_noop", "hello\n");
13303 d.caret = 2;
13304 d.outdent();
13305 assert_eq!(d.source, "hello\n");
13306 assert!(!d.dirty, "a no-op is not a modification");
13307 d.undo();
13308 assert_eq!(
13309 d.status.as_deref(),
13310 Some("nothing to undo"),
13311 "spends no undo step"
13312 );
13313 assert_eq!(d.source, "hello\n");
13314 }
13315 }
13316
13317 #[test]
13318 fn indent_shifts_every_selected_line_and_keeps_them_selected() {
13319 for view in [View::Source, View::Wysiwyg] {
13320 let mut d = doc_in(view, "indent_sel", "one\n\ntwo\n");
13321 d.anchor = Some(0);
13322 d.caret = 7; // through "two"
13323 d.indent();
13324 assert_eq!(
13325 d.source, " one\n\n two\n",
13326 "the blank line keeps no trailing pad"
13327 );
13328 // Selected, so a second Tab lands on the same lines rather than on
13329 // whatever the shifted offsets now cover.
13330 assert_eq!(d.selection(), Some((0, 12)));
13331 d.indent();
13332 assert_eq!(d.source, " one\n\n two\n");
13333 }
13334 }
13335
13336 #[test]
13337 fn outdent_takes_what_each_line_has_and_leaves_the_rest_alone() {
13338 for view in [View::Source, View::Wysiwyg] {
13339 let mut d = doc_in(view, "outdent_sel", " two\n one\nnone\n");
13340 d.anchor = Some(0);
13341 d.caret = 15;
13342 d.outdent();
13343 assert_eq!(d.source, "two\none\nnone\n");
13344 }
13345 }
13346
13347 #[test]
13348 fn a_tab_undoes_as_one_step_however_many_lines_it_moved() {
13349 for view in [View::Source, View::Wysiwyg] {
13350 let mut d = doc_in(view, "indent_undo", "one\n\ntwo\n");
13351 d.anchor = Some(0);
13352 d.caret = 7;
13353 d.indent();
13354 assert_eq!(d.source, " one\n\n two\n");
13355 d.undo();
13356 assert_eq!(d.source, "one\n\ntwo\n", "one step, not one per line");
13357 assert_eq!(
13358 d.selection(),
13359 Some((0, 7)),
13360 "with the selection it was aimed at"
13361 );
13362 d.redo();
13363 assert_eq!(d.source, " one\n\n two\n");
13364 assert_eq!(
13365 d.selection(),
13366 Some((0, 12)),
13367 "redo replays the caret the indent placed, not the one splice left"
13368 );
13369 }
13370 }
13371
13372 #[test]
13373 fn vertical_motion_keeps_the_column() {
13374 let mut d = doc_with("move", "abcd\nef\n");
13375 d.caret = 3; // "abc|d" on row 0, col 3
13376 d.move_down(false); // row 1 "ef" only has cols 0..2 -> clamps to end
13377 assert_eq!(d.caret, 7); // just after "ef"
13378 }
13379
13380 // ── goal column ──────────────────────────────────────────────────────────
13381
13382 #[test]
13383 fn vertical_motion_goal_column_survives_a_short_line() {
13384 // Regression: re-deriving the column from the clamped position on
13385 // every step permanently forgets it once a short line clamps it.
13386 // Down through "xy" (2 cols) and into "ghijkl" must return to col 4.
13387 let g = |m, f: fn(&mut Doc)| golden("goalcol", m, f);
13388 assert_eq!(
13389 g("abcd|ef\nxy\nghijkl\n", |d| {
13390 d.move_down(false); // clamps to end of "xy"
13391 d.move_down(false); // restores col 4 on the long line
13392 }),
13393 "abcdef\nxy\nghij|kl\n"
13394 );
13395 }
13396
13397 #[test]
13398 fn goal_column_state_is_set_by_vertical_motion_and_cleared_by_horizontal() {
13399 let mut d = doc_with("goalcol_state", "abcdef\nxy\nghijkl\n");
13400 assert_eq!(d.goal_col, None);
13401 d.caret = 4; // row 0, col 4
13402 d.move_down(false); // clamps into "xy"; goal stays the original col
13403 assert_eq!(d.goal_col, Some(4));
13404 assert_eq!(d.caret_pos(), (1, 2));
13405
13406 // A horizontal motion drops the goal column...
13407 d.move_left(false);
13408 assert_eq!(d.goal_col, None);
13409
13410 // ...so the next vertical motion picks up the *new* column (1), not
13411 // the stale one (4).
13412 d.move_down(false);
13413 assert_eq!(d.goal_col, Some(1));
13414 assert_eq!(d.caret_pos(), (2, 1));
13415 }
13416
13417 #[test]
13418 fn editing_clears_the_goal_column() {
13419 let mut d = doc_with("goalcol_edit", "abcdef\nxy\nghijkl\n");
13420 d.caret = 4;
13421 d.move_down(false);
13422 assert_eq!(d.goal_col, Some(4));
13423 d.insert("Z");
13424 assert_eq!(d.goal_col, None);
13425 }
13426
13427 #[test]
13428 fn vertical_motion_on_an_empty_document_is_a_no_op() {
13429 let mut d = doc_with("empty_vert", "");
13430 d.move_down(false);
13431 assert_eq!(d.caret, 0);
13432 d.move_up(false);
13433 assert_eq!(d.caret, 0);
13434 }
13435
13436 // ── the document's edges ─────────────────────────────────────────────────
13437
13438 #[test]
13439 fn vertical_motion_at_the_document_edges_runs_to_them_in_both_views() {
13440 // The reproduction, and the disagreement: Down on the last line ran to
13441 // the end of the document in the source view — by accident, an
13442 // out-of-range row clamping to the end of the string — and did nothing
13443 // whatever in the view leaf opens in. One rule now, in both.
13444 for (view, tag) in VIEWS {
13445 let mut d = doc_in(view, &format!("edge_{tag}"), "abc");
13446 d.caret = 1;
13447 d.move_down(false);
13448 assert_eq!(d.caret, 3, "{tag}: Down on the last line runs to the end");
13449 d.move_up(false);
13450 assert_eq!(d.caret, 0, "{tag}: Up on the first line runs to the start");
13451 }
13452 }
13453
13454 #[test]
13455 fn vertical_motion_at_the_edges_carries_the_column_across_the_lines_between() {
13456 // Down off the bottom is a motion like any other, so it latches a goal
13457 // column — and Up comes back to the column the caret left, not to the
13458 // one the document's end happened to be in.
13459 for (view, tag) in VIEWS {
13460 let gap = if view == View::Source { "\n" } else { "\n\n" };
13461 let src = format!("abcdef{gap}ghijkl");
13462 let mut d = doc_in(view, &format!("edge_goal_{tag}"), &src);
13463 d.caret = 2; // row 0, col 2
13464 d.move_down(false);
13465 assert_eq!(d.caret_pos().1, 2, "{tag}: Down keeps the column");
13466 d.move_down(false);
13467 assert_eq!(
13468 d.caret,
13469 src.len(),
13470 "{tag}: Down off the bottom reaches the end"
13471 );
13472 d.move_up(false);
13473 assert_eq!(
13474 d.caret_pos().1,
13475 2,
13476 "{tag}: Up returns to the column Down left"
13477 );
13478 }
13479 }
13480
13481 #[test]
13482 fn vertical_motion_with_nowhere_to_go_latches_no_goal_column() {
13483 // `goal_col.get_or_insert` ran *before* the early return at row 0, so an
13484 // Up that did nothing still armed a goal column, and the next Down aimed
13485 // at a column the caret had never been in.
13486 for (view, tag) in VIEWS {
13487 let mut d = doc_in(view, &format!("noop_goal_{tag}"), "abc\n\ndef");
13488 d.caret = 0;
13489 d.move_up(false);
13490 assert_eq!(d.caret, 0, "{tag}: already at the start");
13491 assert_eq!(d.goal_col, None, "{tag}: a no-op Up latched a goal column");
13492
13493 d.caret = d.source.len();
13494 d.move_down(false);
13495 assert_eq!(d.caret, d.source.len(), "{tag}: already at the end");
13496 assert_eq!(
13497 d.goal_col, None,
13498 "{tag}: a no-op Down latched a goal column"
13499 );
13500 }
13501 }
13502
13503 // ── soft wrap ────────────────────────────────────────────────────────────
13504 // Every other test here builds the map at 80 columns, where no fixture is
13505 // long enough to fold. A wrap is where one offset belongs to two rows at
13506 // once, and it broke everything that asks the caret what row it is on.
13507
13508 /// The wrapped fixture these cases share, folded at 12 columns into
13509 /// `one two ` / `three four ` / `five six ` / `seven eight`.
13510 fn wrapped_doc(name: &str) -> Doc {
13511 let mut d = wysiwyg_doc(name, "one two three four five six seven eight");
13512 d.build_visual(12);
13513 d
13514 }
13515
13516 #[test]
13517 fn home_and_end_work_from_a_wrapped_row() {
13518 // The reproduction: offset 19 is the `f` of "five", the first character
13519 // of the third row — and also the offset the second row ends at. It
13520 // resolved to the *second* row, so End aimed at a place the caret was
13521 // already in and did nothing, while Home walked backwards onto a row the
13522 // caret had left.
13523 let mut d = wrapped_doc("wrap_home_end");
13524 d.caret = 19;
13525 assert_eq!(
13526 d.caret_pos(),
13527 (2, 0),
13528 "the wrap boundary opens the third row"
13529 );
13530 d.move_end(false);
13531 assert_eq!(d.caret, 27, "End stalled at the wrap boundary");
13532 d.move_home(false);
13533 assert_eq!(d.caret, 19, "Home left the row the caret was on");
13534 }
13535
13536 #[test]
13537 fn end_of_a_wrapped_row_stays_put_when_pressed_again() {
13538 // The row's end is the last offset that is only ever its own: the offset
13539 // past it opens the row below, and aiming there would send a second
13540 // press on to *that* row's end, and a third to the next — End walking
13541 // down the paragraph rather than sitting where it landed.
13542 let mut d = wrapped_doc("wrap_end_twice");
13543 d.caret = 12; // inside "three", on the second row
13544 d.move_end(false);
13545 assert_eq!(
13546 d.caret, 18,
13547 "the end of `three four`, before the space the wrap ate"
13548 );
13549 assert_eq!(d.caret_pos(), (1, 10), "drawn on the row it is the end of");
13550 d.move_end(false);
13551 assert_eq!(d.caret, 18, "a second End moved the caret");
13552 d.move_home(false);
13553 assert_eq!(d.caret, 8, "Home takes the row's own start");
13554 }
13555
13556 #[test]
13557 fn vertical_motion_crosses_a_soft_wrap() {
13558 // Down aimed at the row below's column 0, an offset that resolved *up*
13559 // to the row above's end — so it landed on the offset it already had and
13560 // the caret could never leave a paragraph's first row.
13561 let mut d = wrapped_doc("wrap_down");
13562 d.caret = 0;
13563 for (want, row) in [(8, 1), (19, 2), (28, 3), (39, 3)] {
13564 d.move_down(false);
13565 assert_eq!(d.caret, want, "Down stalled");
13566 assert_eq!(d.caret_pos().0, row, "Down landed on the wrong row");
13567 }
13568 d.move_down(false);
13569 assert_eq!(d.caret, 39, "the last row's Down runs to the end and stops");
13570
13571 // ...and back up, one row per press. The goal column is the end of the
13572 // last row, past every other row's width, so each press clamps to the
13573 // row's own last offset rather than to the one that opens the next.
13574 let mut d = wrapped_doc("wrap_up");
13575 d.caret = 39;
13576 for (want, pos) in [(27, (2, 8)), (18, (1, 10)), (7, (0, 7)), (0, (0, 0))] {
13577 d.move_up(false);
13578 assert_eq!(d.caret, want, "Up stalled");
13579 assert_eq!(d.caret_pos(), pos, "Up landed on the wrong row");
13580 }
13581 }
13582
13583 #[test]
13584 fn a_kill_on_a_wrapped_row_stops_at_the_row() {
13585 // The kills take the same line Home and End do, so in WYSIWYG they take
13586 // the visual row — and a soft wrap has no newline in it to delete, so
13587 // nothing is joined by reaching the end of one.
13588 let mut d = wrapped_doc("wrap_kill");
13589 d.caret = 19; // the `f` of "five", opening the third row
13590 d.delete_to_line_end();
13591 // The space the wrap ate goes with the row it was drawn on: sparing it
13592 // would leave "four seven", two spaces where the row had been.
13593 assert_eq!(d.source, "one two three four seven eight");
13594
13595 // Backwards from the row's last caret position — which is *before* that
13596 // space, so this one survives, being on the far side of the caret.
13597 let mut d = wrapped_doc("wrap_kill_back");
13598 d.caret = 27;
13599 d.delete_to_line_start();
13600 assert_eq!(d.source, "one two three four seven eight");
13601 }
13602
13603 // ── document start / end ────────────────────────────────────────────────
13604
13605 #[test]
13606 fn move_doc_start_and_end_jump_to_the_edges() {
13607 let g = |m, f: fn(&mut Doc)| golden("doc_edges", m, f);
13608 assert_eq!(
13609 g("hello\nwor|ld\n", |d| d.move_doc_start(false)),
13610 "|hello\nworld\n"
13611 );
13612 assert_eq!(
13613 g("hel|lo\nworld\n", |d| d.move_doc_end(false)),
13614 "hello\nworld\n|"
13615 );
13616 // Already at the edge: a no-op.
13617 assert_eq!(g("|hello\n", |d| d.move_doc_start(false)), "|hello\n");
13618 assert_eq!(g("hello|\n", |d| d.move_doc_end(false)), "hello\n|");
13619 }
13620
13621 #[test]
13622 fn move_doc_start_and_end_extend_the_selection() {
13623 assert_eq!(
13624 golden("doc_edges_ext_end", "hello wor|ld\n", |d| d
13625 .move_doc_end(true)),
13626 "hello wor[ld\n|]"
13627 );
13628 assert_eq!(
13629 golden("doc_edges_ext_start", "hello wor|ld\n", |d| d
13630 .move_doc_start(true)),
13631 "[|hello wor]ld\n"
13632 );
13633 }
13634
13635 #[test]
13636 fn move_doc_start_and_end_on_an_empty_document_are_a_no_op() {
13637 let mut d = doc_with("empty_edges", "");
13638 d.move_doc_end(false);
13639 assert_eq!(d.caret, 0);
13640 d.move_doc_start(false);
13641 assert_eq!(d.caret, 0);
13642 }
13643
13644 // ── arrow collapses an active selection ─────────────────────────────────
13645
13646 #[test]
13647 fn arrow_collapses_selection_to_its_near_edge() {
13648 let mut d = doc_with("collapse", "hello world\n");
13649
13650 // Forward selection (anchor before caret): Right -> end, Left -> start.
13651 d.anchor = Some(2);
13652 d.caret = 7;
13653 d.move_right(false);
13654 assert_eq!((d.caret, d.anchor), (7, None));
13655
13656 d.anchor = Some(2);
13657 d.caret = 7;
13658 d.move_left(false);
13659 assert_eq!((d.caret, d.anchor), (2, None));
13660
13661 // Backward selection (anchor after caret): edges are the same
13662 // regardless of which end the caret started on.
13663 d.anchor = Some(7);
13664 d.caret = 2;
13665 d.move_right(false);
13666 assert_eq!((d.caret, d.anchor), (7, None));
13667
13668 d.anchor = Some(7);
13669 d.caret = 2;
13670 d.move_left(false);
13671 assert_eq!((d.caret, d.anchor), (2, None));
13672 }
13673
13674 #[test]
13675 fn arrow_with_extend_keeps_growing_the_selection() {
13676 let mut d = doc_with("collapse_extend", "hello world\n");
13677 d.anchor = Some(2);
13678 d.caret = 7;
13679 d.move_right(true); // extend: no collapse, caret steps one further
13680 assert_eq!((d.caret, d.anchor), (8, Some(2)));
13681 }
13682
13683 #[test]
13684 fn arrow_without_a_selection_moves_one_character_as_before() {
13685 let mut d = doc_with("no_collapse", "hello\n");
13686 d.caret = 2;
13687 d.move_right(false);
13688 assert_eq!(d.caret, 3);
13689 d.move_left(false);
13690 assert_eq!(d.caret, 2);
13691 }
13692
13693 /// Press Right until it stops, collecting the offsets walked through. Every
13694 /// caret bug in the WYSIWYG view shows up here as a walk that ends early:
13695 /// two stops sharing one source offset can't be moved between, so the caret
13696 /// stalls on the first of them and the walk never reaches the rest.
13697 fn walk_right(d: &mut Doc) -> Vec<usize> {
13698 let mut seen = vec![d.caret];
13699 for _ in 0..2000 {
13700 let before = d.caret;
13701 d.move_right(false);
13702 if d.caret == before {
13703 break;
13704 }
13705 seen.push(d.caret);
13706 }
13707 seen
13708 }
13709
13710 #[test]
13711 fn the_caret_crosses_a_soft_break() {
13712 // A newline inside a paragraph is a `soft_break`, which twig gives no
13713 // span of its own — the space it renders as used to borrow the offset of
13714 // the character before it, and a caret can't move without changing
13715 // offset. Right must walk clean off the end of the first line.
13716 let mut d = wysiwyg_doc("soft_break_walk", "one two\nthree four\n");
13717 d.caret = 0;
13718 let seen = walk_right(&mut d);
13719 assert_eq!(seen, (0..=18).collect::<Vec<_>>(), "walk stalled: {seen:?}");
13720 }
13721
13722 #[test]
13723 fn line_flow_preserve_resplits_the_map_and_defaults_to_fold() {
13724 // The paragraph holds one soft break. Folded (the default) it lays out as
13725 // a single reflowed row; Preserve re-lays it as a row per source line.
13726 // The setter must invalidate the cached map for the change to show, and
13727 // again on the way back — so a round trip returns to the folded layout.
13728 let mut d = wysiwyg_doc("line_flow", "one two\nthree four\n");
13729 assert_eq!(d.line_flow(), LineFlow::Fold, "fold is the default");
13730 d.build_visual(80);
13731 assert_eq!(d.vmap.num_rows(), 1, "fold: one flowing row");
13732
13733 d.set_line_flow(LineFlow::Preserve);
13734 d.build_visual(80);
13735 assert_eq!(d.vmap.num_rows(), 2, "preserve: a row per source line");
13736
13737 d.set_line_flow(LineFlow::Fold);
13738 d.build_visual(80);
13739 assert_eq!(d.vmap.num_rows(), 1, "fold again: back to one row");
13740 }
13741
13742 #[test]
13743 fn the_caret_still_crosses_a_preserved_soft_break() {
13744 // Preserve renders the soft break as a row boundary rather than a space,
13745 // but the caret must still reach every offset — the break's own offset is
13746 // the first row's end stop, so Right walks clean off the end of line one
13747 // onto line two, exactly as it does when the break is folded.
13748 let mut d = wysiwyg_doc("preserve_walk", "one two\nthree four\n");
13749 d.set_line_flow(LineFlow::Preserve);
13750 d.build_visual(80);
13751 d.caret = 0;
13752 let seen = walk_right(&mut d);
13753 assert_eq!(seen, (0..=18).collect::<Vec<_>>(), "walk stalled: {seen:?}");
13754 }
13755
13756 #[test]
13757 fn the_caret_walks_a_code_block() {
13758 // Every glyph of a code block used to map to the block's start, so the
13759 // whole block was a single offset and the caret couldn't move inside it.
13760 let src = "```rust\nlet x = 1;\nfn f() {}\n```\n";
13761 let mut d = wysiwyg_doc("code_walk", src);
13762 d.caret = 0;
13763 let seen = walk_right(&mut d);
13764 // The fences are markup: hidden, and no caret stop. The code between
13765 // them is reached a character at a time.
13766 let code = src.find("let").unwrap()..src.find("\n```").unwrap();
13767 for off in code.clone() {
13768 assert!(seen.contains(&off), "offset {off} unreachable: {seen:?}");
13769 }
13770 assert!(seen.contains(&code.end), "no stop after the last line");
13771 }
13772
13773 #[test]
13774 fn the_caret_walks_an_indented_code_block() {
13775 // An indented block's text has the four-space indent stripped, so it
13776 // isn't a verbatim slice and its lines have to be re-found. The caret
13777 // lands on the code, never in the indent.
13778 let src = " indented\n code\n";
13779 let mut d = wysiwyg_doc("indent_code_walk", src);
13780 d.caret = 0;
13781 let seen = walk_right(&mut d);
13782 assert!(seen.contains(&src.find("indented").unwrap()));
13783 assert!(seen.contains(&src.find("code").unwrap()));
13784 assert!(
13785 !seen.contains(&0) || seen[0] == 0,
13786 "the caret starts where it was put"
13787 );
13788 // Nothing in the stripped indent is a stop.
13789 for off in [1, 2, 3] {
13790 assert!(!seen.contains(&off), "landed in the indent at {off}");
13791 }
13792 }
13793
13794 #[test]
13795 fn the_caret_leaves_a_tight_heading() {
13796 // "# H" with text directly under it: the heading row's end and the
13797 // separator row's end are the same offset. Right used to find the
13798 // separator's copy, set the caret to where it already was, and stop.
13799 let mut d = wysiwyg_doc("tight_heading_walk", "# H\ntext\n");
13800 d.caret = 2; // the "H"
13801 let seen = walk_right(&mut d);
13802 assert!(
13803 seen.len() > 2,
13804 "Right stalled at the heading's end: {seen:?}"
13805 );
13806 assert!(
13807 seen.contains(&8),
13808 "never reached the end of \"text\": {seen:?}"
13809 );
13810 }
13811
13812 #[test]
13813 fn the_caret_skips_the_gap_between_two_paragraphs() {
13814 // The blank line between two paragraphs is the boundary itself. The
13815 // caret used to be able to sit on it, and typing there landed in the
13816 // previous paragraph — "A\n\nB" became "A\nx\nB", one paragraph with a
13817 // soft break, so the text visibly snapped back up.
13818 let mut d = wysiwyg_doc("gap_skip", "A\n\nB\n");
13819 d.caret = 1; // the end of "A"
13820 d.move_right(false);
13821 assert_eq!(d.caret, 3, "Right stopped in the gap");
13822 d.insert("x");
13823 assert_eq!(d.source, "A\n\nxB\n", "typing landed outside B");
13824 }
13825
13826 #[test]
13827 fn down_from_a_paragraph_lands_on_the_next_one() {
13828 let mut d = wysiwyg_doc("gap_down", "A\n\nB\n");
13829 d.caret = 0;
13830 d.move_down(false);
13831 assert_eq!(d.caret, 3, "Down stopped in the gap");
13832 }
13833
13834 #[test]
13835 fn clicking_the_gap_lands_on_real_text() {
13836 // A click can still *reach* the gap — it's drawn, so it's clickable.
13837 // It has to resolve to somewhere the caret can be.
13838 let mut d = wysiwyg_doc("gap_click", "A\n\nB\n");
13839 d.click(1, 0, false); // the gap row
13840 assert!(
13841 d.caret == 1 || d.caret == 3,
13842 "click left the caret in the gap at {}",
13843 d.caret
13844 );
13845 d.insert("x");
13846 // Either edge of the boundary is a fair place to land; inside it isn't.
13847 assert!(
13848 d.source == "Ax\n\nB\n" || d.source == "A\n\nxB\n",
13849 "click in the gap typed into the boundary: {:?}",
13850 d.source
13851 );
13852 }
13853
13854 #[test]
13855 fn enter_opens_an_empty_paragraph_the_caret_can_type_into() {
13856 // Enter inserts a paragraph break, which leaves a blank line spare on
13857 // either side of a new one. That middle line is a real empty paragraph:
13858 // the caret lands there, and typing makes a paragraph rather than
13859 // extending a neighbour.
13860 let mut d = wysiwyg_doc("gap_enter", "A\n\nB\n");
13861 d.caret = 1;
13862 d.newline();
13863 assert_eq!(d.source, "A\n\n\n\nB\n");
13864 d.build_visual(80);
13865 let (row, _) = d.caret_pos();
13866 assert!(
13867 d.vmap.row_is_navigable(row),
13868 "the caret landed on a gap row"
13869 );
13870 d.insert("x");
13871 assert_eq!(
13872 d.source, "A\n\nx\n\nB\n",
13873 "the new paragraph merged into a neighbour"
13874 );
13875 }
13876
13877 #[test]
13878 fn enter_at_the_end_of_the_document_opens_a_paragraph_too() {
13879 let mut d = wysiwyg_doc("gap_eof", "A\n");
13880 d.caret = 1;
13881 d.newline();
13882 d.build_visual(80);
13883 let (row, _) = d.caret_pos();
13884 assert!(
13885 d.vmap.row_is_navigable(row),
13886 "the caret landed on a gap row"
13887 );
13888 d.insert("x");
13889 assert!(
13890 d.source.starts_with("A\n\n") && d.source.contains('x'),
13891 "typing at the end merged into A: {:?}",
13892 d.source
13893 );
13894 }
13895
13896 #[test]
13897 fn triple_click_selects_a_paragraph_across_its_soft_breaks() {
13898 // A paragraph broken over two source lines is one paragraph. Selecting
13899 // it must not stop at the newline inside it — that newline is markup the
13900 // rich-text view exists to hide.
13901 let src = "one two\nthree four\n\nnext\n";
13902 let mut d = wysiwyg_doc("triple_para", src);
13903 d.select_block_at(2);
13904 assert_eq!(
13905 d.selected_text(),
13906 Some("one two\nthree four"),
13907 "stopped at the soft break"
13908 );
13909 }
13910
13911 #[test]
13912 fn the_wheel_can_scroll_away_from_a_caret_that_stays_put() {
13913 // The reader scrolls down past the caret's row. Nothing moved the
13914 // caret, so the view must stay where it was put — the old code revealed
13915 // the caret every frame, which dragged the view straight back and made
13916 // the document unscrollable past the caret.
13917 let mut d = wysiwyg_doc("scroll_free", "a\n\nb\n\nc\n\nd\n\ne\n");
13918 d.caret = 0;
13919 d.follow_caret(0, 3, 9); // first frame: the caret is at the top
13920 d.scroll = 4; // the wheel
13921 d.follow_caret(0, 3, 9);
13922 assert_eq!(
13923 d.scroll, 4,
13924 "the wheel was overruled by a caret that never moved"
13925 );
13926 }
13927
13928 #[test]
13929 fn moving_the_caret_brings_the_view_back_to_it() {
13930 let mut d = wysiwyg_doc("scroll_follow", "a\n\nb\n\nc\n\nd\n\ne\n");
13931 d.caret = 0;
13932 d.follow_caret(0, 3, 9);
13933 d.scroll = 6; // scrolled away
13934 d.move_right(false); // ...and now the caret moves
13935 let (row, _) = d.caret_pos();
13936 d.follow_caret(row, 3, 9);
13937 assert!(
13938 d.scroll <= row && row < d.scroll + 3,
13939 "caret row {row} off screen at scroll {}",
13940 d.scroll
13941 );
13942 }
13943
13944 #[test]
13945 fn scrolling_stops_at_the_last_row() {
13946 let mut d = wysiwyg_doc("scroll_clamp", "a\n\nb\n");
13947 d.caret = 0;
13948 d.follow_caret(0, 3, 3); // a first frame, so the caret isn't "new"
13949 d.scroll = 999; // the wheel, spun hard
13950 d.follow_caret(0, 3, 3);
13951 assert_eq!(d.scroll, 2, "scrolled into the void past the document");
13952 }
13953
13954 #[test]
13955 fn every_cell_of_a_wide_table_is_reachable() {
13956 // A table whose cells are far wider than the surface: the columns are
13957 // cut to fit and the text wraps inside them, so no cell hangs off the
13958 // right edge where the caret can never go.
13959 let src = "| Ingredient | Notes |\n|---|---|\n\
13960 | flour milled coarse | sift it twice before folding it in |\n";
13961 let mut d = wysiwyg_doc("wide_table_walk", src);
13962 d.build_visual(30);
13963 d.caret = 0;
13964 let seen = walk_right(&mut d);
13965 for word in ["Ingredient", "Notes", "coarse", "folding"] {
13966 let at = src.find(word).unwrap();
13967 assert!(seen.contains(&at), "{word:?} at {at} unreachable: {seen:?}");
13968 }
13969 }
13970
13971 // ── view parity ──────────────────────────────────────────────────────────
13972 // `doc_with` pins the source view, so everything above tests a view users
13973 // never start in — `Doc::open` opens in WYSIWYG. These run the motion and
13974 // deletion golden cases through *both*, plus the WYSIWYG cases the two
13975 // can't share: where the source carries markup the rendered text is a
13976 // different string, and the views agreeing would itself be the bug.
13977
13978 const VIEWS: [(View, &str); 2] = [(View::Source, "source"), (View::Wysiwyg, "wysiwyg")];
13979
13980 /// Run `action` in both views on one `|`-marked fixture and assert they
13981 /// agree. Plain prose only: with no markup to hide, WYSIWYG renders the
13982 /// source verbatim, so the two views are looking at the same text and any
13983 /// disagreement is one of them having lost the plot.
13984 fn both_views(name: &str, marked: &str, action: fn(&mut Doc)) -> String {
13985 let (src, caret) = parse_caret(marked);
13986 let run = |view: View, tag: &str| {
13987 let mut d = doc_in(view, &format!("{name}_{tag}"), &src);
13988 d.caret = caret;
13989 action(&mut d);
13990 render_caret(&d)
13991 };
13992 let source = run(VIEWS[0].0, VIEWS[0].1);
13993 let wysiwyg = run(VIEWS[1].0, VIEWS[1].1);
13994 assert_eq!(source, wysiwyg, "the views disagree on {marked:?}");
13995 source
13996 }
13997
13998 #[test]
13999 fn word_motion_agrees_across_the_views_on_plain_prose() {
14000 let g = both_views;
14001 assert_eq!(
14002 g("par_wl", "hello wor|ld", |d| d.move_word_left(false)),
14003 "hello |world"
14004 );
14005 assert_eq!(
14006 g("par_wl2", "hello| world", |d| d.move_word_left(false)),
14007 "|hello world"
14008 );
14009 assert_eq!(
14010 g("par_wr", "hel|lo world", |d| d.move_word_right(false)),
14011 "hello| world"
14012 );
14013 assert_eq!(
14014 g("par_wr2", "hello| world", |d| d.move_word_right(false)),
14015 "hello world|"
14016 );
14017 assert_eq!(
14018 g("par_punct", "|foo.bar", |d| d.move_word_right(false)),
14019 "foo|.bar"
14020 );
14021 assert_eq!(
14022 g("par_ext", "hello |world", |d| d.move_word_right(true)),
14023 "hello [world|]"
14024 );
14025 }
14026
14027 #[test]
14028 fn word_deletion_agrees_across_the_views_on_plain_prose() {
14029 let g = both_views;
14030 assert_eq!(
14031 g("par_db", "hello world|", |d| d.delete_word_back()),
14032 "hello |"
14033 );
14034 assert_eq!(
14035 g("par_df", "hello |world", |d| d.delete_word_forward()),
14036 "hello |"
14037 );
14038 assert_eq!(
14039 g("par_db2", "foo |bar baz", |d| d.delete_word_back()),
14040 "|bar baz"
14041 );
14042 assert_eq!(g("par_utf8", "café |ok", |d| d.delete_word_back()), "|ok");
14043 }
14044
14045 #[test]
14046 fn character_motion_and_deletion_agree_across_the_views_on_plain_prose() {
14047 let g = both_views;
14048 assert_eq!(g("par_r", "he|llo", |d| d.move_right(false)), "hel|lo");
14049 assert_eq!(g("par_l", "he|llo", |d| d.move_left(false)), "h|ello");
14050 assert_eq!(g("par_bs", "hel|lo", |d| d.backspace()), "he|lo");
14051 assert_eq!(g("par_del", "hel|lo", |d| d.delete_forward()), "hel|o");
14052 }
14053
14054 #[test]
14055 fn wysiwyg_motion_steps_a_grapheme_cluster_the_way_the_source_view_does() {
14056 // The reproduction: the stop table was built one stop per `char`, so
14057 // Right parked the caret 4 bytes into a ZWJ sequence — a place the
14058 // source view, which steps by grapheme, can't reach and backspace can't
14059 // survive. The two views must land on the same offset.
14060 let family = "👨👩👧"; // three emoji strung together with joiners: one cluster
14061 for (view, tag) in VIEWS {
14062 let mut d = doc_in(view, &format!("cluster_{tag}"), &format!("a{family}b\n"));
14063 d.caret = 1;
14064 d.move_right(false);
14065 assert_eq!(d.caret, 1 + family.len(), "{tag} parked inside the cluster");
14066
14067 // ...and the edit that used to sever a joiner off the front of it.
14068 d.backspace();
14069 assert_eq!(d.source, "ab\n", "{tag} split the cluster");
14070 assert_eq!(d.caret, 1);
14071 }
14072 }
14073
14074 #[test]
14075 fn wysiwyg_motion_treats_a_combining_accent_as_one_character() {
14076 for (view, tag) in VIEWS {
14077 let mut d = doc_in(view, &format!("combining_{tag}"), "e\u{0301}x\n");
14078 d.caret = 0;
14079 d.move_right(false);
14080 assert_eq!(
14081 d.caret,
14082 "e\u{0301}".len(),
14083 "{tag} stopped on the combining mark"
14084 );
14085 }
14086 }
14087
14088 #[test]
14089 fn no_wysiwyg_motion_can_park_the_caret_inside_a_cluster() {
14090 // The general form: whatever route the caret takes through a document
14091 // full of clusters, it never lands between the codepoints of one — so no
14092 // motion-then-backspace sequence can leave a dangling joiner behind.
14093 use unicode_segmentation::UnicodeSegmentation;
14094
14095 let src = "a👨👩👧b e\u{0301}mo👨👩👧ji\n\nnext 👩🚀 line\n";
14096 let mut d = wysiwyg_doc("cluster_walk", src);
14097 d.caret = 0;
14098 let boundaries: Vec<usize> = src
14099 .grapheme_indices(true)
14100 .map(|(i, _)| i)
14101 .chain(std::iter::once(src.len()))
14102 .collect();
14103 for off in walk_right(&mut d) {
14104 assert!(
14105 boundaries.contains(&off),
14106 "Right stopped at {off}, inside a grapheme cluster"
14107 );
14108 }
14109 }
14110
14111 #[test]
14112 fn wysiwyg_word_motion_stays_out_of_hidden_delimiters() {
14113 // The reproduction: ⌥→ from inside the opening `**` computed its
14114 // boundary over the raw source and landed on byte 8 — inside the
14115 // *closing* `**`, which `caret_pos` draws at column 6, immediately after
14116 // "bold". The caret drew past the bold word and sat inside it.
14117 let mut d = wysiwyg_doc("wys_word_delim", "a **bold** c\n");
14118 d.caret = 2;
14119 d.move_word_right(false);
14120 assert!(
14121 d.vmap.is_stop(d.caret),
14122 "landed at {}, not a caret stop",
14123 d.caret
14124 );
14125 assert_eq!(d.caret, 10, "should land on the space after \"bold\"");
14126 // The rendered row is "a bold c": column 6 is the space just past "bold",
14127 // and now the caret is really there rather than only drawn there.
14128 assert_eq!(d.caret_pos(), (0, 6));
14129
14130 // ...and back again: ⌥← returns to the "b", not into the opening `**`.
14131 d.move_word_left(false);
14132 assert_eq!(d.caret, 4);
14133 assert_eq!(d.caret_pos(), (0, 2));
14134 }
14135
14136 #[test]
14137 fn wysiwyg_word_delete_takes_the_markup_with_the_word() {
14138 // The reproduction: ⌥⌫ from after "bold" walked the raw source, stopped
14139 // inside the closing `**`, and left "a ** c\n" — delimiters with no
14140 // opener. Glyph space covers the word alone, which would leave
14141 // "a **** c": markup wrapped around nothing. The word and the styling
14142 // that was only ever the word's go together.
14143 let mut d = wysiwyg_doc("wys_word_del_back", "a **bold** c\n");
14144 d.caret = 10;
14145 d.delete_word_back();
14146 assert_eq!(d.source, "a c\n");
14147 assert_eq!(d.caret, 2);
14148
14149 let mut d = wysiwyg_doc("wys_word_del_fwd", "a **bold** c\n");
14150 d.caret = 4; // the "b"
14151 d.delete_word_forward();
14152 assert_eq!(d.source, "a c\n");
14153 }
14154
14155 #[test]
14156 fn wysiwyg_word_delete_empties_a_nested_mark_and_a_code_span_too() {
14157 let src = "a ***bold*** c\n";
14158 let mut d = wysiwyg_doc("wys_word_del_nest", src);
14159 d.caret = src.find(" c").unwrap();
14160 d.delete_word_back();
14161 assert_eq!(
14162 d.source, "a c\n",
14163 "the emph inside the strong empties it too"
14164 );
14165
14166 let src = "a `code` c\n";
14167 let mut d = wysiwyg_doc("wys_word_del_code", src);
14168 d.caret = src.find(" c").unwrap();
14169 d.delete_word_back();
14170 assert_eq!(d.source, "a c\n");
14171 }
14172
14173 #[test]
14174 fn wysiwyg_word_delete_keeps_a_mark_that_still_has_text() {
14175 // Only an *emptied* node goes. Take one word of two and the `**` still
14176 // has a job to do — over the word that's left, with the space the delete
14177 // pushed against the opening delimiter moved out in front of it, or the
14178 // run would be no run at all (`** words**` is literal asterisks — see
14179 // the mark-edge rule on `splice`).
14180 let src = "a **two words** c\n";
14181 let mut d = wysiwyg_doc("wys_word_del_partial", src);
14182 d.caret = src.find(" words").unwrap();
14183 d.delete_word_back();
14184 assert_eq!(d.source, "a **words** c\n");
14185 }
14186
14187 #[test]
14188 fn source_view_word_motion_still_walks_the_markup() {
14189 // The other half of the decision: in the source view the `**` are
14190 // characters like any other — they're on the screen, so word motion has
14191 // to stop at them and a word-delete has to leave them behind. Only
14192 // WYSIWYG hides them, so only WYSIWYG steps over them.
14193 let g = |n, m, f: fn(&mut Doc)| golden(n, m, f);
14194 assert_eq!(
14195 g("src_word_motion", "a |**bold** c\n", |d| d
14196 .move_word_right(false)),
14197 "a **bold|** c\n"
14198 );
14199 // The same caret as the WYSIWYG reproduction, and the opposite outcome:
14200 // here "a ** c\n" is right, because `bold**` is what's to the left of it.
14201 assert_eq!(
14202 g("src_word_del", "a **bold**| c\n", |d| d.delete_word_back()),
14203 "a **| c\n"
14204 );
14205 }
14206
14207 #[test]
14208 fn every_wysiwyg_motion_lands_on_a_caret_stop() {
14209 // The single invariant both bugs violated: the caret draws and edits at
14210 // the same place only when it's on a stop. `debug_assert_on_a_stop`
14211 // makes the same claim in-place; this pins it from the outside, over a
14212 // document with every kind of thing the map has to be careful about.
14213 // At two widths: the wide one every other test builds at, where no
14214 // fixture folds, and one narrow enough that they all do. A soft wrap is
14215 // where an offset stops being on exactly one row, and testing only the
14216 // width that never wraps is how the caret came to be pinned at the first
14217 // one Down reached.
14218 let src = "# Title\n\na **bold** e\u{0301}mo👨👩👧ji `x` c\n\n\
14219 - item one\n\n| A | B |\n|---|---|\n| x | y |\n";
14220 // A table of named operations, which is what it looks like.
14221 #[allow(clippy::type_complexity)]
14222 let motions: [(&str, fn(&mut Doc)); 8] = [
14223 ("right", |d| d.move_right(false)),
14224 ("left", |d| d.move_left(false)),
14225 ("word_right", |d| d.move_word_right(false)),
14226 ("word_left", |d| d.move_word_left(false)),
14227 ("down", |d| d.move_down(false)),
14228 ("up", |d| d.move_up(false)),
14229 ("home", |d| d.move_home(false)),
14230 ("end", |d| d.move_end(false)),
14231 ];
14232 for width in [80, 12] {
14233 let mut d = wysiwyg_doc("stop_invariant", src);
14234 d.build_visual(width);
14235 let stops: Vec<usize> = (0..=src.len()).filter(|&o| d.vmap.is_stop(o)).collect();
14236 assert!(stops.len() > 20, "fixture should have plenty of stops");
14237 for start in stops {
14238 for (name, motion) in &motions {
14239 d.caret = start;
14240 d.anchor = None;
14241 motion(&mut d);
14242 assert!(
14243 d.vmap.is_stop(d.caret),
14244 "{name} from {start} at width {width} landed at {} — not a caret stop",
14245 d.caret
14246 );
14247 }
14248 }
14249 }
14250 }
14251
14252 #[test]
14253 fn no_wysiwyg_motion_is_a_dead_end() {
14254 // Down held to the bottom of a document reaches the bottom, and Up held
14255 // to the top reaches the top — from anywhere, at a width that wraps. The
14256 // invariant above says a motion lands somewhere legal; this one says it
14257 // gets somewhere at all, which is what a caret pinned at a wrap boundary
14258 // was quietly failing to do while every assertion around it held.
14259 let src = "# Title\n\none two three four five six seven eight nine ten\n\n\
14260 - item one two three four five\n\nlast\n";
14261 for width in [80, 12] {
14262 let mut d = wysiwyg_doc("no_dead_end", src);
14263 d.build_visual(width);
14264 let stops: Vec<usize> = (0..=src.len()).filter(|&o| d.vmap.is_stop(o)).collect();
14265 let (first, last) = (stops[0], stops[stops.len() - 1]);
14266 for &start in &stops {
14267 for (name, motion, want) in [
14268 (
14269 "down",
14270 (|d: &mut Doc| d.move_down(false)) as fn(&mut Doc),
14271 last,
14272 ),
14273 ("up", |d: &mut Doc| d.move_up(false), first),
14274 ] {
14275 d.caret = start;
14276 d.anchor = None;
14277 d.goal_col = None;
14278 // Every row, plus the presses the edges take, plus slack.
14279 for _ in 0..d.vmap.num_rows() + 4 {
14280 motion(&mut d);
14281 }
14282 assert_eq!(
14283 d.caret, want,
14284 "{name} held from {start} at width {width} never arrived"
14285 );
14286 }
14287 }
14288 }
14289 }
14290 // ── display columns ──────────────────────────────────────────────────────
14291 // A `col` is a terminal cell, not a character. The two are the same number
14292 // for the ASCII the fixtures above are written in, which is how they came
14293 // apart in the first place: `你` is one character drawn in two cells, so a
14294 // column counted in characters names a cell the text isn't in — one earlier
14295 // for every wide character to its left.
14296
14297 #[test]
14298 fn a_wide_character_is_two_columns_wide() {
14299 // The reproduction: `你` is one char and two cells, so the caret just
14300 // past it drew at column 1 — inside the character it had already left.
14301 for (view, tag) in VIEWS {
14302 let mut d = doc_in(view, &format!("wide_col_{tag}"), "你好\n");
14303 d.caret = "你".len();
14304 assert_eq!(d.caret_pos(), (0, 2), "{tag}: caret drew inside 你");
14305 d.caret = "你好".len();
14306 assert_eq!(d.caret_pos(), (0, 4), "{tag}");
14307 }
14308 }
14309
14310 #[test]
14311 fn a_cluster_is_as_wide_as_it_is_drawn_not_as_its_codepoints_measure() {
14312 // `👨👩👧` is five codepoints — two-cell, joiner, two-cell, joiner,
14313 // two-cell — measuring six cells one at a time, but the character they
14314 // spell is drawn in two. Width belongs to the cluster, not the glyph,
14315 // and the frontends measure it the same way.
14316 let family = "👨👩👧";
14317 for (view, tag) in VIEWS {
14318 let src = format!("a{family}b\n");
14319 let mut d = doc_in(view, &format!("wide_cluster_{tag}"), &src);
14320 d.caret = 1 + family.len();
14321 assert_eq!(
14322 d.caret_pos(),
14323 (0, 3),
14324 "{tag}: 'a' is one cell, the family two"
14325 );
14326 }
14327 }
14328
14329 #[test]
14330 fn both_cells_of_a_wide_character_mean_the_character() {
14331 // Clicking the far half of `好` is still clicking `好`: half a character
14332 // is not a place the caret can be, so it comes to rest at the
14333 // character's start — the column it would have been drawn at anyway.
14334 for (view, tag) in VIEWS {
14335 let mut d = doc_in(view, &format!("wide_click_{tag}"), "你好\n");
14336 for col in [2, 3] {
14337 d.caret = 0;
14338 d.click(0, col, false);
14339 assert_eq!(d.caret, "你".len(), "{tag}: click at col {col}");
14340 assert_eq!(d.caret_pos(), (0, 2), "{tag}: click at col {col}");
14341 }
14342 // Past the last cell is the line's end, as it is for ASCII.
14343 d.click(0, 9, false);
14344 assert_eq!(d.caret, "你好".len(), "{tag}: click past the end");
14345 }
14346 }
14347
14348 #[test]
14349 fn every_offset_survives_the_trip_out_to_a_column_and_back() {
14350 // The mapping is only a mapping if it inverts: the cell the caret is
14351 // drawn in has to be the cell that brings it back to the same offset.
14352 // Over a fixture where a character may be one cell or two, and one
14353 // codepoint or five.
14354 use unicode_segmentation::UnicodeSegmentation;
14355
14356 let src = "ab 你好 c\n\n👨👩👧 e\u{0301}x 漢字\n\nplain ascii\n";
14357
14358 let mut d = doc_in(View::Source, "roundtrip_source", src);
14359 // Every offset the source view's caret can occupy: it steps by grapheme
14360 // cluster, so those are its boundaries.
14361 for (off, _) in src
14362 .grapheme_indices(true)
14363 .chain(std::iter::once((src.len(), "")))
14364 {
14365 d.caret = off;
14366 let (row, col) = d.caret_pos();
14367 d.click(row, col, false);
14368 assert_eq!(d.caret, off, "source: {off} → ({row}, {col}) → {}", d.caret);
14369 }
14370
14371 // And in WYSIWYG, where the offsets the caret can occupy are the map's
14372 // stops rather than every boundary.
14373 let mut d = doc_in(View::Wysiwyg, "roundtrip_wysiwyg", src);
14374 let stops: Vec<usize> = (0..=src.len()).filter(|&o| d.vmap.is_stop(o)).collect();
14375 assert!(stops.len() > 20, "fixture should have plenty of stops");
14376 for off in stops {
14377 d.caret = off;
14378 let (row, col) = d.caret_pos();
14379 d.click(row, col, false);
14380 assert_eq!(
14381 d.caret, off,
14382 "wysiwyg: {off} → ({row}, {col}) → {}",
14383 d.caret
14384 );
14385 }
14386 }
14387
14388 #[test]
14389 fn vertical_motion_aims_at_a_column_the_reader_can_see() {
14390 // Down from under `世` lands under the glyph in that cell, not two
14391 // characters further along the line. The goal is a column, so a line of
14392 // wide characters and a line of ASCII line up the way they're drawn.
14393 //
14394 // The gap differs by view: a bare newline inside a paragraph is a soft
14395 // break, which WYSIWYG draws as a space on a single row. The views share
14396 // a grid only where the source's lines are the renderer's rows too.
14397 for (view, tag) in VIEWS {
14398 let gap = if view == View::Source { "\n" } else { "\n\n" };
14399 let src = format!("你好世{gap}abcdef\n");
14400 let mut d = doc_in(view, &format!("goal_wide_{tag}"), &src);
14401 d.caret = "你好".len();
14402 assert_eq!(d.caret_pos().1, 4, "{tag}: `世` is drawn at column 4");
14403 d.move_down(false);
14404 assert_eq!(d.caret_pos().1, 4, "{tag}: goal column lost");
14405 assert!(
14406 d.source[d.caret..].starts_with('e'),
14407 "{tag}: landed on the wrong glyph"
14408 );
14409 }
14410 }
14411
14412 #[test]
14413 fn a_goal_column_landing_inside_a_wide_character_lands_on_it() {
14414 // Down from column 3 onto `你好`, whose characters start at columns 0
14415 // and 2: column 3 is the *second* cell of `好`. There is nowhere to be
14416 // between the cells of one character, so the caret rests on it — and on
14417 // its start, which is the only offset there that is a caret stop.
14418 for (view, tag) in VIEWS {
14419 let gap = if view == View::Source { "\n" } else { "\n\n" };
14420 let src = format!("abcdef{gap}你好\n");
14421 let mut d = doc_in(view, &format!("goal_inside_{tag}"), &src);
14422 let line = src.find('你').unwrap();
14423 d.caret = 3;
14424 d.move_down(false);
14425 assert_eq!(d.caret, line + "你".len(), "{tag}: landed off `好`'s start");
14426 assert_eq!(d.caret_pos().1, 2, "{tag}: drew between `好`'s cells");
14427 }
14428 }
14429
14430 #[test]
14431 fn a_caret_in_a_table_cell_of_wide_text_draws_where_the_text_is() {
14432 // The column the cell's text is laid out in is measured in cells, so the
14433 // caret walking that text has to be too — the two agreeing is the whole
14434 // point of the grid staying square.
14435 let mut d = wysiwyg_doc("table_wide", "| A | B |\n|---|---|\n| 你好 | y |\n");
14436 let at = d.source.find("你").unwrap();
14437 d.caret = at;
14438 let (row, col) = d.caret_pos();
14439 // `│ ` opens the row, so the cell's text starts at column 2; `好` is two
14440 // cells further along.
14441 assert_eq!(col, 2, "the cell's first character");
14442 d.move_right(false);
14443 assert_eq!(
14444 d.caret_pos(),
14445 (row, 4),
14446 "`好` is drawn past `你`'s two cells"
14447 );
14448 assert_eq!(d.caret, at + "你".len());
14449 }
14450
14451 // ── active inline marks ───────────────────────────────────────────────────
14452
14453 /// The marks at a `|`-marked fixture's caret, in `InlineMarks::iter` order.
14454 fn marks(view: View, name: &str, marked: &str) -> Vec<InlineKind> {
14455 let (src, caret) = parse_caret(marked);
14456 let mut d = doc_in(view, name, &src);
14457 d.caret = caret;
14458 d.active_inline_marks().iter().collect()
14459 }
14460
14461 /// The marks over the selection `[start, end)`.
14462 fn marks_over(view: View, name: &str, src: &str, start: usize, end: usize) -> Vec<InlineKind> {
14463 let mut d = doc_in(view, name, src);
14464 d.anchor = Some(start);
14465 d.caret = end;
14466 d.active_inline_marks().iter().collect()
14467 }
14468
14469 #[test]
14470 fn a_caret_in_a_mark_reports_it() {
14471 for (view, tag) in VIEWS {
14472 let m = |marked| marks(view, &format!("marks_in_{tag}"), marked);
14473 assert_eq!(m("a **bo|ld** b"), [InlineKind::Strong], "{tag}");
14474 assert_eq!(m("a *it|alic* b"), [InlineKind::Emph], "{tag}");
14475 assert_eq!(m("a `co|de` b"), [InlineKind::Verbatim], "{tag}");
14476 // Plain text under no mark lights nothing — the toolbar's resting state.
14477 assert_eq!(m("a| **bold** b"), [], "{tag}");
14478 assert!(m("plain t|ext").is_empty(), "{tag}");
14479 }
14480 }
14481
14482 #[test]
14483 fn nested_marks_all_report() {
14484 // Bold *and* italic: a toolbar lights both buttons, so the set has both —
14485 // the ancestor chain is a chain, and every mark on it is in force.
14486 for (view, tag) in VIEWS {
14487 assert_eq!(
14488 marks(
14489 view,
14490 &format!("marks_nested_{tag}"),
14491 "**bold and *bo|th*** end"
14492 ),
14493 [InlineKind::Strong, InlineKind::Emph],
14494 "{tag}"
14495 );
14496 }
14497 }
14498
14499 #[test]
14500 fn the_caret_at_a_marks_edge_reports_it_where_typing_would_extend_it() {
14501 // The offsets a WYSIWYG caret actually reaches at a bold run's edges are
14502 // the first byte of its text and the byte after its last — both inside
14503 // the mark's span, both places typing lands inside the bold. The offset
14504 // past the closing delimiter is the next text, and reports nothing.
14505 let src = "a **bold** b";
14506 let inner_start = src.find("bold").unwrap(); // 4
14507 let inner_end = inner_start + "bold".len(); // 8, on the closing `**`
14508 for (view, tag) in VIEWS {
14509 let mut d = doc_in(view, &format!("marks_edge_{tag}"), src);
14510 for off in [2, 3, inner_start, inner_end, 9] {
14511 d.caret = off;
14512 assert!(
14513 d.active_inline_marks().contains(InlineKind::Strong),
14514 "{tag}: offset {off} is inside the strong span"
14515 );
14516 }
14517 for off in [0, 1, 10, 11, 12] {
14518 d.caret = off;
14519 assert!(
14520 !d.active_inline_marks().contains(InlineKind::Strong),
14521 "{tag}: offset {off} is outside the strong run"
14522 );
14523 }
14524 }
14525 }
14526
14527 #[test]
14528 fn a_mark_ends_the_same_way_at_the_end_of_the_buffer_as_in_the_middle() {
14529 // Regression: twig resolves an offset that is one node's end and the
14530 // next one's start to the node that *starts* there, so `**bold**|\n`
14531 // isn't bold. With nothing following there's no tie to break and the
14532 // chain still ended at the mark, which made a trailing `\n` — not the
14533 // text — decide whether the caret after a bold word reported bold. It's
14534 // the offset past the mark either way, and typing there is plain either
14535 // way. A blank document typed into is exactly this shape.
14536 for (view, tag) in VIEWS {
14537 let m = |name: String, marked| marks(view, &name, marked);
14538 assert_eq!(
14539 m(format!("marks_eob_{tag}"), "**bold**|"),
14540 [],
14541 "{tag}: no trailing newline"
14542 );
14543 assert_eq!(
14544 m(format!("marks_eol_{tag}"), "**bold**|\n"),
14545 [],
14546 "{tag}: with one"
14547 );
14548 // And the last offset that *is* in the mark still is.
14549 assert_eq!(
14550 m(format!("marks_eob_in_{tag}"), "**bold*|*"),
14551 [InlineKind::Strong],
14552 "{tag}"
14553 );
14554 }
14555 }
14556
14557 #[test]
14558 fn a_selection_reports_a_mark_only_when_it_covers_the_whole_thing() {
14559 let src = "a **bold** b";
14560 let (b, d_) = (src.find("bold").unwrap(), src.find("bold").unwrap() + 4);
14561 for (view, tag) in VIEWS {
14562 let m = |s, e| marks_over(view, &format!("marks_sel_{tag}"), src, s, e);
14563 // The whole bold word, and a slice of it.
14564 assert_eq!(m(b, d_), [InlineKind::Strong], "{tag}: the whole word");
14565 assert_eq!(m(b + 1, d_ - 1), [InlineKind::Strong], "{tag}: a slice");
14566 // Ending exactly at the closing delimiter's start is still all-bold:
14567 // an exclusive end sits *past* the last selected character, so the
14568 // question is asked of the character, not the boundary.
14569 assert_eq!(
14570 m(b, d_ + 2),
14571 [InlineKind::Strong],
14572 "{tag}: through the close"
14573 );
14574 // Half in, half out: Bold lit here would claim a press turns it off.
14575 assert_eq!(m(0, d_), [], "{tag}: leading plain text");
14576 assert_eq!(m(b, src.len()), [], "{tag}: trailing plain text");
14577 }
14578 }
14579
14580 #[test]
14581 fn a_selection_across_two_runs_of_the_same_mark_reports_nothing() {
14582 // Both ends are bold, but the space between them isn't — two runs are two
14583 // nodes, which is exactly what the node id catches and a kind-only
14584 // comparison would not.
14585 let src = "**one** **two**";
14586 for (view, tag) in VIEWS {
14587 let m = marks_over(view, &format!("marks_runs_{tag}"), src, 2, 13);
14588 assert_eq!(m, [], "{tag}: `one** **two` is not all bold");
14589 }
14590 }
14591
14592 #[test]
14593 fn marks_read_the_document_as_it_is_edited() {
14594 // The point of asking twig every frame instead of caching: the answer has
14595 // to follow the toggle that changed it.
14596 let mut d = wysiwyg_doc("marks_live", "one two\n");
14597 d.anchor = Some(0);
14598 d.caret = 3;
14599 assert!(d.active_inline_marks().is_empty(), "plain to start");
14600 d.toggle(InlineKind::Strong);
14601 assert_eq!(d.source, "**one** two\n");
14602 // `toggle` leaves the bolded text selected, so the button it lit stays lit.
14603 assert!(d.active_inline_marks().contains(InlineKind::Strong));
14604 d.toggle(InlineKind::Strong);
14605 assert!(d.active_inline_marks().is_empty(), "and off again");
14606 }
14607
14608 #[test]
14609 fn a_link_is_not_an_inline_mark() {
14610 // `link`/`str` are inline nodes, but nothing on the inline toolbar
14611 // toggles them — a set with a "link mark" in it would have no button.
14612 for (view, tag) in VIEWS {
14613 assert_eq!(
14614 marks(view, &format!("marks_link_{tag}"), "a [te|xt](u) b"),
14615 [],
14616 "{tag}"
14617 );
14618 }
14619 }
14620
14621 // ── blank documents ───────────────────────────────────────────────────────
14622
14623 #[test]
14624 fn a_blank_document_is_untitled_empty_and_markdown() {
14625 let mut d = Doc::blank().unwrap();
14626 assert!(d.is_untitled());
14627 assert_eq!(d.path, PathBuf::new());
14628 assert_eq!(
14629 d.file_name(),
14630 "untitled",
14631 "the header has to show something"
14632 );
14633 assert_eq!(d.format_name(), "markdown");
14634 assert_eq!(d.source, "");
14635 assert!(!d.dirty, "nothing typed yet is nothing to lose");
14636 assert_eq!(d.disk_state(), DiskState::Untitled);
14637 // And it's a document you can be in: the default view renders it.
14638 d.build_visual(80);
14639 assert_eq!(d.caret, 0);
14640 }
14641
14642 #[test]
14643 fn saving_an_untitled_document_asks_for_a_name_instead_of_writing() {
14644 let mut d = Doc::blank().unwrap();
14645 d.insert("hello");
14646 assert!(d.dirty);
14647 d.save();
14648 assert_eq!(d.status.as_deref(), Some("untitled — save as…"));
14649 assert!(d.dirty, "it must not come away believing it saved");
14650 assert!(d.is_untitled(), "and it still has no file");
14651 }
14652
14653 #[test]
14654 fn a_blank_document_becomes_a_real_one_at_the_first_save_as() {
14655 let p = temp_path("blank_save_as");
14656 let mut d = Doc::blank().unwrap();
14657 // Plain text — a blank doc opens in Hidden mode, where a typed `#` would
14658 // be kept literal (`\#`); this test is about save-as, not escaping (which
14659 // has its own test), so it types nothing that escaping would touch.
14660 d.insert("hi");
14661 d.save_as(p.clone());
14662 assert_eq!(std::fs::read_to_string(&p).unwrap(), "hi");
14663 assert!(!d.is_untitled());
14664 assert!(!d.dirty);
14665 assert_eq!(d.file_name(), p.file_name().unwrap().to_string_lossy());
14666 assert_eq!(
14667 d.disk_state(),
14668 DiskState::Unchanged,
14669 "the watermark is stamped"
14670 );
14671 // And ⌘S is a plain save from here on.
14672 d.insert("!");
14673 d.save();
14674 assert_eq!(std::fs::read_to_string(&p).unwrap(), "hi!");
14675 let _ = std::fs::remove_file(&p);
14676 }
14677
14678 // ── a file that isn't there yet ───────────────────────────────────────────
14679
14680 /// A unique path in the temp dir with the given extension, guaranteed not to
14681 /// exist — what `leaf notes.md` is handed when the file has never been made.
14682 fn missing_path(name: &str, ext: &str) -> PathBuf {
14683 static SEQ: std::sync::atomic::AtomicUsize = std::sync::atomic::AtomicUsize::new(0);
14684 let seq = SEQ.fetch_add(1, std::sync::atomic::Ordering::Relaxed);
14685 let mut p = std::env::temp_dir();
14686 p.push(format!("leaf_test_new_{name}_{seq}.{ext}"));
14687 let _ = std::fs::remove_file(&p);
14688 p
14689 }
14690
14691 #[test]
14692 fn a_file_that_doesnt_exist_opens_as_an_empty_named_document() {
14693 let p = missing_path("named", "md");
14694 let mut d = Doc::open_or_create(p.clone()).unwrap();
14695
14696 assert_eq!(d.source, "", "nothing was read, so there's nothing in it");
14697 assert!(!d.dirty, "an untouched new buffer has nothing to lose");
14698 assert!(
14699 !d.is_untitled(),
14700 "it has the name the user asked for — ^S must not detour to Save As"
14701 );
14702 assert_eq!(d.file_name(), p.file_name().unwrap().to_str().unwrap());
14703 assert!(d.path.is_absolute(), "the same absolute path `open` stores");
14704 assert!(!p.exists(), "and opening it wrote nothing");
14705 // And it's a document you can be in.
14706 d.build_visual(80);
14707 assert_eq!(d.caret, 0);
14708 }
14709
14710 #[test]
14711 fn a_new_file_is_created_by_its_first_save() {
14712 let p = missing_path("first_save", "md");
14713 let mut d = Doc::open_or_create(p.clone()).unwrap();
14714 d.insert("hello\n");
14715 assert!(d.dirty);
14716 d.save();
14717
14718 assert_eq!(
14719 std::fs::read_to_string(&p).unwrap(),
14720 "hello\n",
14721 "a plain ^S wrote it — no Save As, no name to invent"
14722 );
14723 assert!(!d.dirty);
14724 assert_eq!(d.disk_state(), DiskState::Unchanged);
14725 let _ = std::fs::remove_file(&p);
14726 }
14727
14728 #[test]
14729 fn a_new_file_takes_its_format_from_the_extension() {
14730 // The one thing `blank` can't do: with no name it has to assume Markdown,
14731 // and typing djot into a Markdown parse is the wrong buffer.
14732 let dj = missing_path("format", "dj");
14733 assert_eq!(Doc::open_or_create(dj).unwrap().format_name(), "djot");
14734 let md = missing_path("format", "md");
14735 assert_eq!(Doc::open_or_create(md).unwrap().format_name(), "markdown");
14736 }
14737
14738 #[test]
14739 fn a_new_file_reports_itself_missing_until_it_is_saved() {
14740 // Not `Untitled` — that's the answer for a document with no path, and it
14741 // would tell a frontend there is nothing a save could collide with. Here
14742 // there is a path, and the file simply isn't at it yet.
14743 let p = missing_path("disk_state", "md");
14744 let mut d = Doc::open_or_create(p.clone()).unwrap();
14745 assert_eq!(d.disk_state(), DiskState::Missing);
14746
14747 // Somebody else creates it while the buffer is open: that's an overwrite
14748 // the frontend has to be able to prompt about, exactly as for an opened
14749 // file. Their bytes, not ours, so `Changed`.
14750 std::fs::write(&p, "theirs\n").unwrap();
14751 assert_eq!(d.disk_state(), DiskState::Changed);
14752
14753 // Saving makes the file ours and re-stamps the watermark.
14754 d.insert("ours\n");
14755 d.save();
14756 assert_eq!(d.disk_state(), DiskState::Unchanged);
14757 assert_eq!(std::fs::read_to_string(&p).unwrap(), "ours\n");
14758 let _ = std::fs::remove_file(&p);
14759 }
14760
14761 #[test]
14762 fn open_or_create_still_opens_a_file_that_is_there() {
14763 let d = doc_with("open_or_create_existing", "body\n");
14764 let reopened = Doc::open_or_create(d.path.clone()).unwrap();
14765 assert_eq!(reopened.source, "body\n");
14766 assert_eq!(reopened.disk_state(), DiskState::Unchanged);
14767 }
14768
14769 #[test]
14770 fn a_missing_file_with_no_readable_extension_is_still_an_error() {
14771 // A mistyped flag or a stray argument must not become a buffer promising
14772 // to save somewhere — the same refusal `open` gives a real file.
14773 let mut p = std::env::temp_dir();
14774 p.push("leaf_test_new_bad_ext.wat");
14775 assert!(Doc::open_or_create(p).is_err());
14776 let mut none = std::env::temp_dir();
14777 none.push("leaf_test_new_no_ext");
14778 assert!(Doc::open_or_create(none).is_err());
14779 }
14780
14781 #[test]
14782 fn a_new_file_in_a_directory_that_doesnt_exist_opens_but_wont_save() {
14783 // Opening reads nothing, so there is nothing to fail on yet; the write is
14784 // where it fails, and it says so rather than claiming a save.
14785 let p = std::env::temp_dir().join("leaf_test_no_such_dir_c41/doc.md");
14786 let mut d = Doc::open_or_create(p).unwrap();
14787 d.insert("x");
14788 d.save();
14789 assert!(
14790 d.status.as_deref().unwrap().starts_with("save failed:"),
14791 "got {:?}",
14792 d.status
14793 );
14794 assert!(d.dirty, "it must not come away believing it saved");
14795 }
14796
14797 // ── save as ───────────────────────────────────────────────────────────────
14798
14799 /// A unique path in the temp dir that no fixture wrote — a Save As target.
14800 fn temp_path(name: &str) -> PathBuf {
14801 static SEQ: std::sync::atomic::AtomicUsize = std::sync::atomic::AtomicUsize::new(0);
14802 let seq = SEQ.fetch_add(1, std::sync::atomic::Ordering::Relaxed);
14803 let mut p = std::env::temp_dir();
14804 p.push(format!("leaf_test_target_{name}_{seq}.md"));
14805 let _ = std::fs::remove_file(&p);
14806 p
14807 }
14808
14809 #[test]
14810 fn save_as_moves_the_document_and_leaves_the_old_file_alone() {
14811 let mut d = doc_with("save_as_move", "original\n");
14812 let old = d.path.clone();
14813 let new = temp_path("save_as_move");
14814 d.insert("edited: ");
14815 d.save_as(new.clone());
14816
14817 assert_eq!(std::fs::read_to_string(&new).unwrap(), "edited: original\n");
14818 assert_eq!(
14819 std::fs::read_to_string(&old).unwrap(),
14820 "original\n",
14821 "Save As doesn't touch the file it came from"
14822 );
14823 assert_eq!(d.path, new, "the document moved");
14824 assert!(!d.dirty);
14825 assert_eq!(
14826 d.status.as_deref(),
14827 Some(&*format!("saved {}", d.file_name()))
14828 );
14829
14830 // Every later save follows it, which is the whole difference from a copy.
14831 d.caret = 0;
14832 d.insert("re-");
14833 d.save();
14834 assert_eq!(
14835 std::fs::read_to_string(&new).unwrap(),
14836 "re-edited: original\n"
14837 );
14838 assert_eq!(std::fs::read_to_string(&old).unwrap(), "original\n");
14839 let _ = std::fs::remove_file(&new);
14840 }
14841
14842 #[test]
14843 fn save_as_overwrites_an_existing_target() {
14844 // The picker already asked; asking again down here is the same question
14845 // twice, and the second one has no way to be answered.
14846 let new = temp_path("save_as_over");
14847 std::fs::write(&new, "theirs\n").unwrap();
14848 let mut d = doc_with("save_as_over", "ours\n");
14849 d.save_as(new.clone());
14850 assert_eq!(std::fs::read_to_string(&new).unwrap(), "ours\n");
14851 let _ = std::fs::remove_file(&new);
14852 }
14853
14854 #[test]
14855 fn a_save_as_that_fails_leaves_the_document_where_it_was() {
14856 let mut d = doc_with("save_as_fail", "body\n");
14857 let old = d.path.clone();
14858 d.insert("x");
14859 // A directory that doesn't exist: the write can't land.
14860 let bad = std::env::temp_dir().join("leaf_test_no_such_dir_9f2/doc.md");
14861 d.save_as(bad);
14862
14863 assert_eq!(
14864 d.path, old,
14865 "the document must not move to a file that isn't there"
14866 );
14867 assert!(d.dirty, "and must not believe it saved");
14868 assert!(
14869 d.status.as_deref().unwrap().starts_with("save failed:"),
14870 "the same failure a plain save reports, got {:?}",
14871 d.status
14872 );
14873 // The original is still the document's file, and still saveable.
14874 d.save();
14875 assert_eq!(std::fs::read_to_string(&old).unwrap(), "xbody\n");
14876 assert!(!d.dirty);
14877 }
14878
14879 #[test]
14880 fn save_as_renames_without_reparsing_the_format() {
14881 // `.dj` on the name doesn't make the buffer djot: it was parsed as
14882 // Markdown and still is, and saying otherwise would be a conversion the
14883 // user never asked for (and an undo history thrown away to do it).
14884 let mut d = doc_with("save_as_format", "**b**\n");
14885 let mut new = temp_path("save_as_format");
14886 new.set_extension("dj");
14887 d.save_as(new.clone());
14888 assert_eq!(d.format_name(), "markdown");
14889 let _ = std::fs::remove_file(&new);
14890 }
14891
14892 // ── external change / reload ──────────────────────────────────────────────
14893
14894 #[test]
14895 fn an_untouched_file_reports_unchanged() {
14896 let mut d = doc_with("disk_clean", "body\n");
14897 assert_eq!(d.disk_state(), DiskState::Unchanged);
14898 // Editing the buffer is not editing the file.
14899 d.insert("x");
14900 assert_eq!(d.disk_state(), DiskState::Unchanged);
14901 assert!(d.dirty);
14902 // Saving re-stamps the watermark rather than reporting our own bytes back.
14903 d.save();
14904 assert_eq!(d.disk_state(), DiskState::Unchanged);
14905 }
14906
14907 #[test]
14908 fn a_file_written_underneath_reports_changed() {
14909 let mut d = doc_with("disk_changed", "body\n");
14910 std::fs::write(&d.path, "someone else\n").unwrap();
14911 assert_eq!(d.disk_state(), DiskState::Changed);
14912 // Dirty *and* changed is the clobber: both halves are readable, and
14913 // leaf-core takes neither side.
14914 d.insert("x");
14915 assert!(d.dirty && d.disk_state() == DiskState::Changed);
14916 // Saving anyway is allowed — the frontend asked, or chose not to.
14917 d.save();
14918 assert_eq!(std::fs::read_to_string(&d.path).unwrap(), "xbody\n");
14919 assert_eq!(d.disk_state(), DiskState::Unchanged);
14920 }
14921
14922 #[test]
14923 fn a_file_rewritten_with_the_same_bytes_is_unchanged() {
14924 // The hash is what makes this honest: the file was written (a fresh
14925 // mtime), and nothing about the document is stale.
14926 let d = doc_with("disk_same_bytes", "body\n");
14927 std::fs::write(&d.path, "body\n").unwrap();
14928 assert_eq!(d.disk_state(), DiskState::Unchanged);
14929 }
14930
14931 #[test]
14932 fn a_deleted_file_reports_missing() {
14933 let mut d = doc_with("disk_missing", "body\n");
14934 std::fs::remove_file(&d.path).unwrap();
14935 assert_eq!(d.disk_state(), DiskState::Missing);
14936 // A save recreates it, and the document is whole again.
14937 d.save();
14938 assert_eq!(d.disk_state(), DiskState::Unchanged);
14939 assert_eq!(std::fs::read_to_string(&d.path).unwrap(), "body\n");
14940 }
14941
14942 #[test]
14943 fn reload_replaces_the_document_with_the_file() {
14944 for (view, tag) in VIEWS {
14945 let mut d = doc_in(view, &format!("reload_{tag}"), "one\n\ntwo\n");
14946 d.insert("edited ");
14947 assert!(d.dirty);
14948 std::fs::write(&d.path, "one\n\ntwo\n\nthree\n").unwrap();
14949 d.reload();
14950
14951 assert_eq!(d.source, "one\n\ntwo\n\nthree\n", "{tag}");
14952 assert!(!d.dirty, "{tag}: the file is what we have");
14953 assert_eq!(d.disk_state(), DiskState::Unchanged, "{tag}");
14954 assert_eq!(
14955 d.status.as_deref(),
14956 Some(&*format!("reloaded {}", d.file_name()))
14957 );
14958 // The reloaded tree is live, not the old parse.
14959 d.caret = d.source.find("three").unwrap();
14960 assert_eq!(d.breadcrumb(), "doc › para › str", "{tag}");
14961 }
14962 }
14963
14964 #[test]
14965 fn reload_clamps_the_caret_and_drops_the_selection() {
14966 let mut d = doc_with("reload_caret", "a long first line\n");
14967 d.caret = 12;
14968 d.anchor = Some(4);
14969 std::fs::write(&d.path, "short\n").unwrap();
14970 d.reload();
14971 assert_eq!(d.caret, d.source.len(), "clamped into the shorter file");
14972 assert_eq!(
14973 d.anchor, None,
14974 "a selection over bytes that changed is a lie"
14975 );
14976 assert!(d.selection().is_none());
14977
14978 // A caret the file still has room for stays put.
14979 let mut d = doc_with("reload_caret_keep", "one\n\ntwo\n");
14980 d.caret = 2;
14981 std::fs::write(&d.path, "one\n\ntwo\n\nthree\n").unwrap();
14982 d.reload();
14983 assert_eq!(d.caret, 2);
14984 }
14985
14986 /// A silent reload is something that happened *to* a reader — a formatter,
14987 /// a `git checkout` — so it has to be undoable like anything else that
14988 /// changes the document, and undoable as one step rather than as however
14989 /// many the file happens to differ by.
14990 #[test]
14991 fn reload_is_one_undo_step_and_keeps_the_history_under_it() {
14992 let mut d = doc_with("reload_undo", "body\n");
14993 d.insert("x");
14994 assert_eq!(d.source, "xbody\n");
14995 std::fs::write(&d.path, "replaced\n").unwrap();
14996 d.reload();
14997 assert_eq!(d.source, "replaced\n");
14998 assert!(!d.dirty, "a reload lands clean");
14999
15000 // One ^Z takes the whole swap off, and hands back the unsaved work it
15001 // replaced — which is unsaved again, because the file no longer says it.
15002 d.undo();
15003 assert_eq!(d.source, "xbody\n", "the reload comes off in one step");
15004 assert!(d.dirty, "and what it comes back to is unsaved");
15005 // …and the history under it is still there.
15006 d.undo();
15007 assert_eq!(
15008 d.source, "body\n",
15009 "the typing before the reload undoes too"
15010 );
15011 // Redo walks back up through the reload.
15012 d.redo();
15013 d.redo();
15014 assert_eq!(d.source, "replaced\n");
15015 }
15016
15017 /// A file rewritten with the bytes it already had is not an edit, so it
15018 /// must not leave an undo step behind for something nobody did.
15019 #[test]
15020 fn reloading_identical_bytes_pushes_no_undo_step() {
15021 let mut d = doc_with("reload_same", "body\n");
15022 d.insert("x");
15023 std::fs::write(&d.path, "xbody\n").unwrap();
15024 d.reload();
15025 assert_eq!(d.source, "xbody\n");
15026 assert!(!d.dirty, "the file now says what the buffer does");
15027 d.undo();
15028 assert_eq!(
15029 d.source, "body\n",
15030 "one step back is the typing, not a no-op"
15031 );
15032 }
15033
15034 #[test]
15035 fn a_reload_that_cant_read_leaves_the_document_alone() {
15036 let mut d = doc_with("reload_gone", "body\n");
15037 d.insert("x");
15038 std::fs::remove_file(&d.path).unwrap();
15039 d.reload();
15040 assert_eq!(d.source, "xbody\n", "the unsaved work is still here");
15041 assert!(d.dirty);
15042 assert!(
15043 d.status.as_deref().unwrap().starts_with("reload failed:"),
15044 "{:?}",
15045 d.status
15046 );
15047
15048 // And an untitled document has nothing to reload from.
15049 let mut d = Doc::blank().unwrap();
15050 d.insert("typed");
15051 d.reload();
15052 assert_eq!(d.source, "typed");
15053 assert_eq!(d.status.as_deref(), Some("no file to reload"));
15054 }
15055
15056 #[test]
15057 fn a_read_only_document_refuses_every_door() {
15058 let mut d = doc_with("readonly", "one two three\n");
15059 d.insert("x");
15060 assert!(d.dirty, "writable first, so the undo step exists");
15061 d.set_read_only(true);
15062 let before = d.source.clone();
15063 d.insert("y");
15064 d.backspace();
15065 d.undo();
15066 d.redo();
15067 assert_eq!(d.source, before, "no door moved a byte");
15068 d.set_read_only(false);
15069 d.undo();
15070 assert_ne!(d.source, before, "off again, the same doors work");
15071 }
15072
15073 /// The doors that go to twig's own verbs rather than through the splice.
15074 /// Typed text in the rendered view under the default markup mode is the
15075 /// everyday one — it is what a keystroke in leaf-web or the Apple views
15076 /// becomes — and it walked straight past the gate.
15077 #[test]
15078 fn a_read_only_document_refuses_the_doors_around_the_splice() {
15079 let mut d = wysiwyg_doc(
15080 "readonly-doors",
15081 "one two three\n\n| a | b |\n|---|---|\n| c | d |\n",
15082 );
15083 d.set_markup_mode(MarkupMode::None);
15084 d.set_read_only(true);
15085 let before = d.source.clone();
15086 d.place_caret(3, false);
15087 d.insert("y");
15088 d.insert_link("https://example.com");
15089 d.insert_image("a.png", "alt");
15090 d.insert_thematic_break();
15091 d.insert_footnote();
15092 d.place_caret(0, false);
15093 d.place_caret(3, true);
15094 d.toggle(InlineKind::Strong);
15095 d.toggle_heading(2);
15096 d.set_block(BlockKind::Paragraph);
15097 d.toggle_list(false);
15098 d.toggle_blockquote();
15099 d.toggle_task_item();
15100 d.newline();
15101 d.indent();
15102 d.set_code_language("rust");
15103 let in_cell = d.source.find("| c").unwrap() + 2;
15104 d.place_caret(in_cell, false);
15105 assert!(d.caret_in_table(), "the caret is in the grid");
15106 assert!(!d.cell_line_break(), "the cell break reports the refusal");
15107 assert_eq!(d.source, before, "no door moved a byte");
15108 assert!(!d.dirty, "nothing to save");
15109 d.set_read_only(false);
15110 d.place_caret(3, false);
15111 d.insert("y");
15112 assert_ne!(d.source, before, "off again, the same doors work");
15113 }
15114
15115 #[test]
15116 fn a_selection_quote_carries_its_context_on_char_boundaries() {
15117 let mut d = doc_with("quote", "before 你好 exact 世界 after\n");
15118 let start = d.source.find("exact").unwrap();
15119 d.place_caret(start, false);
15120 d.place_caret(start + "exact".len(), true);
15121 let q = d.selection_quote(3).unwrap();
15122 assert_eq!(q.exact, "exact");
15123 assert_eq!(
15124 q.prefix, "你好 ",
15125 "chars, not bytes — the multibyte pair counts as two"
15126 );
15127 assert_eq!(q.suffix, " 世界");
15128 assert_eq!(&d.source[q.start..q.end], "exact");
15129 // At the edges the context clips rather than erring.
15130 d.place_caret(0, false);
15131 d.place_caret(6, true);
15132 let q = d.selection_quote(40).unwrap();
15133 assert_eq!(q.prefix, "");
15134 assert_eq!(q.exact, "before");
15135 // No selection is no quote.
15136 d.place_caret(0, false);
15137 assert!(d.selection_quote(3).is_none());
15138 }
15139
15140 #[test]
15141 fn highlights_are_kept_sorted_and_answer_point_queries() {
15142 let mut d = doc_with("hl", "one two three\n");
15143 d.set_highlights(vec![
15144 Highlight {
15145 start: 8,
15146 end: 13,
15147 id: "b".into(),
15148 color: None,
15149 marker: None,
15150 },
15151 Highlight {
15152 start: 0,
15153 end: 3,
15154 id: "a".into(),
15155 color: Some("#ffe066".into()),
15156 marker: None,
15157 },
15158 Highlight {
15159 start: 5,
15160 end: 5,
15161 id: "empty".into(),
15162 color: None,
15163 marker: None,
15164 },
15165 ]);
15166 assert_eq!(
15167 d.highlights()
15168 .iter()
15169 .map(|h| h.id.as_str())
15170 .collect::<Vec<_>>(),
15171 ["a", "b"],
15172 "sorted by start, the empty range dropped"
15173 );
15174 assert_eq!(d.highlight_at(1).map(|h| h.id.as_str()), Some("a"));
15175 assert_eq!(d.highlight_at(3), None, "end is exclusive");
15176 assert_eq!(d.highlight_at(8).map(|h| h.id.as_str()), Some("b"));
15177 d.set_highlights(Vec::new());
15178 assert!(d.highlights().is_empty(), "a replace is a replace");
15179 }
15180
15181 /// `Highlight::covering` and the cursor over it are what both painters ask
15182 /// per glyph, so they have to answer the same as the scan they replaced —
15183 /// including in the gaps, which is where most glyphs are.
15184 #[test]
15185 fn covering_answers_from_a_sorted_list_without_scanning_all_of_it() {
15186 let hl = |start: usize, end: usize, id: &str| Highlight {
15187 start,
15188 end,
15189 id: id.into(),
15190 color: None,
15191 marker: None,
15192 };
15193 // Disjoint, as search hits are: in a range, in a gap, and past the end.
15194 let hits: Vec<Highlight> = (0..20).map(|i| hl(i * 10, i * 10 + 3, "hit")).collect();
15195 assert_eq!(Highlight::covering(&hits, 0).map(|h| h.start), Some(0));
15196 assert_eq!(Highlight::covering(&hits, 102).map(|h| h.start), Some(100));
15197 assert_eq!(
15198 Highlight::covering(&hits, 105),
15199 None,
15200 "a gap covers nothing"
15201 );
15202 assert_eq!(Highlight::covering(&hits, 103), None, "end is exclusive");
15203 assert_eq!(Highlight::covering(&hits, 9_999), None);
15204 assert_eq!(Highlight::covering(&[], 0), None);
15205
15206 // Nested: first by start, so a hit inside an annotation still resolves
15207 // to the annotation — and the range that stops short doesn't mask it.
15208 let nested = vec![hl(0, 20, "outer"), hl(5, 10, "inner")];
15209 assert_eq!(
15210 Highlight::covering(&nested, 7).map(|h| h.id.as_str()),
15211 Some("outer")
15212 );
15213 assert_eq!(
15214 Highlight::covering(&nested, 15).map(|h| h.id.as_str()),
15215 Some("outer")
15216 );
15217 }
15218
15219 /// The cursor is an optimisation, so the only thing worth asserting is that
15220 /// it is not also a change of answer — at every offset, over a list with a
15221 /// nest in it, walked forwards and then backwards.
15222 #[test]
15223 fn the_highlight_cursor_answers_exactly_what_a_fresh_scan_would() {
15224 let hl = |start: usize, end: usize, id: &str| Highlight {
15225 start,
15226 end,
15227 id: id.into(),
15228 color: None,
15229 marker: None,
15230 };
15231 let mut list = vec![
15232 hl(0, 20, "outer"),
15233 hl(5, 10, "inner"),
15234 hl(30, 33, "hit"),
15235 hl(40, 43, "hit"),
15236 ];
15237 list.sort_by_key(|h| (h.start, h.end));
15238
15239 let mut cursor = HighlightCursor::new(&list);
15240 for offset in 0..50 {
15241 assert_eq!(
15242 cursor.at(offset).map(|h| h.id.as_str()),
15243 Highlight::covering(&list, offset).map(|h| h.id.as_str()),
15244 "cursor disagrees at {offset}"
15245 );
15246 }
15247 // Backwards: the cursor re-seats rather than answering from where it
15248 // had got to, so a painter that revisits a row is still told the truth.
15249 for offset in (0..50).rev() {
15250 assert_eq!(
15251 cursor.at(offset).map(|h| h.id.as_str()),
15252 Highlight::covering(&list, offset).map(|h| h.id.as_str()),
15253 "cursor disagrees walking back at {offset}"
15254 );
15255 }
15256 }
15257
15258 // ── the presentation vocabulary ─────────────────────────────────────────
15259
15260 /// A document in `format`, for the gesture tests that want more than the
15261 /// Markdown `doc_with` writes.
15262 fn fmt_doc(body: &str, format: Format) -> Doc {
15263 Doc::from_source(body.to_string(), format).unwrap()
15264 }
15265
15266 /// Alignment is a block property, so the gesture is `set_block_attrs` on
15267 /// the caret's block whatever is selected — and each format spells it its
15268 /// own way: djot's `{…}` line above the block, a `<div>` around it in
15269 /// Markdown (the format has nowhere else to put it), the tag in HTML.
15270 #[test]
15271 fn set_alignment_spells_the_class_the_format_s_own_way() {
15272 let mut dj = fmt_doc("hello\n", Format::Djot);
15273 dj.caret = 1;
15274 dj.set_alignment(Some(Align::Center));
15275 assert_eq!(dj.source, "{.center}\nhello\n");
15276 assert!(dj.dirty);
15277 assert_eq!(dj.status, None);
15278
15279 let mut md = fmt_doc("hello\n", Format::Markdown);
15280 md.caret = 1;
15281 md.set_alignment(Some(Align::Right));
15282 assert_eq!(md.source, "<div class=\"right\">\n\nhello\n\n</div>\n");
15283
15284 let mut html = fmt_doc("<p>hello</p>\n", Format::Html);
15285 html.caret = html.source.find("hello").unwrap();
15286 html.set_alignment(Some(Align::Justify));
15287 assert_eq!(html.source, "<p class=\"justify\">hello</p>\n");
15288 }
15289
15290 /// Each gesture edits **one key and keeps the rest** — twig's contract is
15291 /// replace-not-merge, so leaf reads the node's attributes, edits its own
15292 /// key out of them, and passes the list back whole. A document from
15293 /// elsewhere passes through the editor unharmed.
15294 #[test]
15295 fn a_presentation_gesture_keeps_every_attribute_it_did_not_write() {
15296 let mut d = fmt_doc(
15297 "{.lead .center #intro data-line-height=\"1.5\"}\nhello\n",
15298 Format::Djot,
15299 );
15300 d.caret = d.source.find("hello").unwrap();
15301 d.set_alignment(Some(Align::Right));
15302 // `center` goes, `lead` stays, and neither the id nor the spacing is
15303 // touched.
15304 // The serializer picks the order; what matters is which keys survive.
15305 assert!(d.source.contains(".lead"), "{:?}", d.source);
15306 assert!(d.source.contains(".right"), "{:?}", d.source);
15307 assert!(!d.source.contains(".center"), "{:?}", d.source);
15308 assert!(d.source.contains("#intro"), "{:?}", d.source);
15309 assert!(
15310 d.source.contains("data-line-height=\"1.5\""),
15311 "{:?}",
15312 d.source
15313 );
15314 assert_eq!(d.alignment_at_caret(), Some(Align::Right));
15315 assert_eq!(d.line_spacing_at_caret(), Some(LineSpacing::OneHalf));
15316
15317 // And the other way round: the spacing gesture leaves the classes be.
15318 d.set_line_spacing(Some(LineSpacing::Double));
15319 assert!(d.source.contains(".lead"), "{:?}", d.source);
15320 assert!(d.source.contains(".right"), "{:?}", d.source);
15321 assert_eq!(d.line_spacing_at_caret(), Some(LineSpacing::Double));
15322 }
15323
15324 /// Clearing is the same gesture with `None`: the key goes, the tokens leaf
15325 /// owns go out of `class`, and a block left with nothing at all is spelled
15326 /// bare again — in Markdown by unwrapping the div twig wrapped it in.
15327 #[test]
15328 fn none_clears_a_key_and_an_empty_set_unwraps_the_block() {
15329 let mut dj = fmt_doc("{.lead .center}\nhello\n", Format::Djot);
15330 dj.caret = dj.source.find("hello").unwrap();
15331 dj.set_alignment(None);
15332 assert_eq!(dj.source, "{.lead}\nhello\n", "the foreign class stays");
15333 assert_eq!(dj.alignment_at_caret(), None);
15334
15335 let mut bare = fmt_doc("{.center}\nhello\n", Format::Djot);
15336 bare.caret = bare.source.find("hello").unwrap();
15337 bare.set_alignment(None);
15338 assert_eq!(
15339 bare.source, "hello\n",
15340 "the last key takes the line with it"
15341 );
15342
15343 let mut md = fmt_doc("hello\n", Format::Markdown);
15344 md.caret = 1;
15345 md.set_alignment(Some(Align::Center));
15346 assert_eq!(md.source, "<div class=\"center\">\n\nhello\n\n</div>\n");
15347 md.caret = md.source.find("hello").unwrap();
15348 md.set_line_spacing(Some(LineSpacing::OneFifteen));
15349 assert_eq!(
15350 md.source, "<div class=\"center\" data-line-height=\"1.15\">\n\nhello\n\n</div>\n",
15351 "the second key rewrites the div rather than nesting a second"
15352 );
15353 md.caret = md.source.find("hello").unwrap();
15354 md.set_alignment(None);
15355 md.caret = md.source.find("hello").unwrap();
15356 md.set_line_spacing(None);
15357 assert_eq!(md.source, "hello\n", "an empty set unwraps the div");
15358 }
15359
15360 /// Size, face and colour are the run's over a selection and the block's
15361 /// with none — so "make this paragraph larger" is a click with the caret in
15362 /// it rather than a select-all first.
15363 #[test]
15364 fn a_run_gesture_wraps_a_selection_and_sets_the_block_without_one() {
15365 // With a selection: a span, in each format's own spelling.
15366 let mut dj = fmt_doc("a big b\n", Format::Djot);
15367 dj.anchor = Some(2);
15368 dj.caret = 5;
15369 dj.set_font_size(Some(SizeStep::Large));
15370 assert_eq!(dj.source, "a [big]{data-size=\"large\"} b\n");
15371 assert_eq!(dj.font_size_at_caret(), Some(SizeStep::Large));
15372
15373 let mut md = fmt_doc("a big b\n", Format::Markdown);
15374 md.anchor = Some(2);
15375 md.caret = 5;
15376 md.set_text_color(Some(MarkColor::Blue));
15377 assert_eq!(md.source, "a <span data-color=\"blue\">big</span> b\n");
15378 assert_eq!(md.text_color_at_caret(), Some(MarkColor::Blue));
15379
15380 // Without one: the caret's block, through the block gesture.
15381 let mut block = fmt_doc("a big b\n", Format::Djot);
15382 block.caret = 3;
15383 block.set_font_family(Some(FontFamily::Monospace));
15384 assert_eq!(block.source, "{data-font=\"monospace\"}\na big b\n");
15385 assert_eq!(block.font_family_at_caret(), Some(FontFamily::Monospace));
15386 }
15387
15388 /// twig re-styles the span a range already lies in rather than nesting a
15389 /// second, and an empty set unwraps it — so a second press of the menu
15390 /// fixes the size instead of building `[[big]{.a}]{.b}`, and the entry that
15391 /// means "the theme's own" takes the span away.
15392 #[test]
15393 fn a_second_run_gesture_re_styles_the_span_and_none_unwraps_it() {
15394 let mut d = fmt_doc("a big b\n", Format::Djot);
15395 d.anchor = Some(2);
15396 d.caret = 5;
15397 d.set_font_size(Some(SizeStep::Large));
15398 assert_eq!(d.source, "a [big]{data-size=\"large\"} b\n");
15399
15400 // The selection `wrap_range_attrs` left behind covers the whole span;
15401 // colouring it now keeps the size, because the gesture reads the span's
15402 // attributes before it edits its own key.
15403 d.set_text_color(Some(MarkColor::Red));
15404 assert_eq!(
15405 d.source, "a [big]{data-size=\"large\" data-color=\"red\"} b\n",
15406 "one span, both keys"
15407 );
15408 assert_eq!(d.font_size_at_caret(), Some(SizeStep::Large));
15409 assert_eq!(d.text_color_at_caret(), Some(MarkColor::Red));
15410
15411 d.set_text_color(None);
15412 assert_eq!(d.source, "a [big]{data-size=\"large\"} b\n");
15413 d.set_font_size(None);
15414 assert_eq!(d.source, "a big b\n", "the last key unwraps the span");
15415 assert_eq!(d.font_size_at_caret(), None);
15416 }
15417
15418 /// The queries read the nearest node that names the property: the span the
15419 /// caret is in, then its block, then the `div`s around it.
15420 #[test]
15421 fn a_presentation_query_reads_the_nearest_node_that_names_it() {
15422 let mut d = fmt_doc(
15423 "{.center data-size=\"small\" data-font=\"serif\"}\nx [y]{data-size=\"xx-large\"} z\n",
15424 Format::Djot,
15425 );
15426 // In the span: its own size, the block's face and alignment.
15427 d.caret = d.source.find('y').unwrap();
15428 assert_eq!(d.font_size_at_caret(), Some(SizeStep::XxLarge));
15429 assert_eq!(d.font_family_at_caret(), Some(FontFamily::Serif));
15430 assert_eq!(d.alignment_at_caret(), Some(Align::Center));
15431 assert_eq!(d.line_spacing_at_caret(), None);
15432 assert_eq!(d.text_color_at_caret(), None);
15433
15434 // Outside it: the block's size.
15435 d.caret = d.source.find('x').unwrap();
15436 assert_eq!(d.font_size_at_caret(), Some(SizeStep::Small));
15437
15438 // And through a Markdown div, which is where a Markdown block's
15439 // attributes live.
15440 let mut md = fmt_doc(
15441 "<div class=\"center\" data-size=\"large\">\n\nhello\n\n</div>\n",
15442 Format::Markdown,
15443 );
15444 md.caret = md.source.find("hello").unwrap();
15445 assert_eq!(md.alignment_at_caret(), Some(Align::Center));
15446 assert_eq!(md.font_size_at_caret(), Some(SizeStep::Large));
15447
15448 // A document that names none of it answers `None` everywhere, which is
15449 // "the theme's own" and what every toolbar draws unlit.
15450 let mut plain = doc_with("plain_presentation", "hello\n");
15451 plain.caret = 1;
15452 assert_eq!(plain.alignment_at_caret(), None);
15453 assert_eq!(plain.line_spacing_at_caret(), None);
15454 assert_eq!(plain.font_size_at_caret(), None);
15455 assert_eq!(plain.font_family_at_caret(), None);
15456 assert_eq!(plain.text_color_at_caret(), None);
15457 }
15458
15459 /// A djot fenced div is anonymous the way an attributed span is, and is a
15460 /// block all the same — the *form* is the whole of what tells them apart.
15461 /// Read as a span it poisoned both halves: the run gesture copied the div's
15462 /// entire attribute set onto the span it minted, duplicating the `id`, and
15463 /// the run and block queries answered off a node the walker draws nothing
15464 /// for.
15465 #[test]
15466 fn a_djot_fenced_div_is_not_an_attributed_span() {
15467 let src = "{.center data-size=\"small\" #box}\n:::\nhello world\n:::\n";
15468 let mut d = fmt_doc(src, Format::Djot);
15469 let at = d.source.find("world").unwrap();
15470 d.anchor = Some(at);
15471 d.caret = at + "world".len();
15472 d.set_text_color(Some(MarkColor::Red));
15473 assert_eq!(
15474 d.source,
15475 "{.center data-size=\"small\" #box}\n:::\nhello [world]{data-color=\"red\"}\n:::\n",
15476 "the span carries its own key and nothing of the div's"
15477 );
15478
15479 // And the queries stop at the block: a djot div is not a `<div>`, the
15480 // walker lends its keys to nothing inside it, and a query that said
15481 // otherwise would tick a menu entry no glyph on screen obeys.
15482 assert_eq!(d.text_color_at_caret(), Some(MarkColor::Red));
15483 assert_eq!(d.font_size_at_caret(), None);
15484 assert_eq!(d.alignment_at_caret(), None);
15485 }
15486
15487 /// Clearing a property the block does not name and a `div` around it does
15488 /// would write nothing and change nothing — twig's `set_block_attrs`
15489 /// reaches one node, and the div is not it. The gesture says so instead of
15490 /// leaving the author pressing an entry that never ticks.
15491 #[test]
15492 fn clearing_a_property_an_enclosing_div_names_says_so_and_writes_nothing() {
15493 // Markdown, two paragraphs in one div: not the sole-child shape twig
15494 // writes, so `block_attrs_at_caret` reads the paragraph and the
15495 // paragraph names none of it.
15496 let src = "<div class=\"center\" data-line-height=\"1.5\" data-size=\"large\">\n\nhello\n\nworld\n\n</div>\n";
15497 let mut md = fmt_doc(src, Format::Markdown);
15498 md.caret = md.source.find("hello").unwrap();
15499 assert_eq!(md.alignment_at_caret(), Some(Align::Center));
15500
15501 md.set_alignment(None);
15502 assert_eq!(md.source, src, "nothing written");
15503 assert!(!md.dirty);
15504 assert_eq!(
15505 md.status.as_deref(),
15506 Some("alignment: set on the div around the block")
15507 );
15508 assert_eq!(md.alignment_at_caret(), Some(Align::Center));
15509
15510 // The same for a `data-` key, at both levels — the block pair and the
15511 // run three, the run three at a bare caret being the block gesture.
15512 md.set_line_spacing(None);
15513 assert_eq!(md.source, src);
15514 assert_eq!(
15515 md.status.as_deref(),
15516 Some("line spacing: set on the div around the block")
15517 );
15518 md.set_font_size(None);
15519 assert_eq!(md.source, src);
15520 assert_eq!(
15521 md.status.as_deref(),
15522 Some("size: set on the div around the block")
15523 );
15524
15525 // HTML has no sole-child fold at all: a block's attributes go on the
15526 // block, so the div around one is always out of reach.
15527 let html_src = "<div class=\"center\"><p>hi</p></div>\n";
15528 let mut html = fmt_doc(html_src, Format::Html);
15529 html.caret = html.source.find("hi").unwrap();
15530 assert_eq!(html.alignment_at_caret(), Some(Align::Center));
15531 html.set_alignment(None);
15532 assert_eq!(html.source, html_src);
15533 assert!(!html.dirty);
15534 assert_eq!(
15535 html.status.as_deref(),
15536 Some("alignment: set on the div around the block")
15537 );
15538
15539 // And it is a refusal, not a rule against clearing: a block that names
15540 // the property itself still loses it, div or no div.
15541 let mut own = fmt_doc(
15542 "<div class=\"center\"><p class=\"right\">hi</p></div>\n",
15543 Format::Html,
15544 );
15545 own.caret = own.source.find("hi").unwrap();
15546 own.set_alignment(None);
15547 assert_eq!(own.source, "<div class=\"center\"><p>hi</p></div>\n");
15548 assert_eq!(own.status, None);
15549 }
15550
15551 /// An edited key is rewritten **where it stands**. The proposal's worked
15552 /// example is the test: a paragraph that came in as `id="intro"
15553 /// class="lead center" data-line-height="1.5"` and is right-aligned goes
15554 /// out as the same list with one token changed. Removing the key and
15555 /// pushing it back shuffled a document's attributes on every press.
15556 #[test]
15557 fn an_edited_key_keeps_its_place_among_the_attributes() {
15558 let mut html = fmt_doc(
15559 "<p id=\"intro\" class=\"lead center\" data-line-height=\"1.5\">hello</p>\n",
15560 Format::Html,
15561 );
15562 html.caret = html.source.find("hello").unwrap();
15563 html.set_alignment(Some(Align::Right));
15564 assert_eq!(
15565 html.source,
15566 "<p id=\"intro\" class=\"lead right\" data-line-height=\"1.5\">hello</p>\n"
15567 );
15568
15569 // A `data-` key the same way, and a key the block did not have still
15570 // goes on the end.
15571 html.caret = html.source.find("hello").unwrap();
15572 html.set_line_spacing(Some(LineSpacing::Double));
15573 assert_eq!(
15574 html.source,
15575 "<p id=\"intro\" class=\"lead right\" data-line-height=\"2\">hello</p>\n"
15576 );
15577 html.caret = html.source.find("hello").unwrap();
15578 html.set_font_size(Some(SizeStep::Large));
15579 assert_eq!(
15580 html.source,
15581 "<p id=\"intro\" class=\"lead right\" data-line-height=\"2\" data-size=\"large\">hello</p>\n"
15582 );
15583
15584 // Djot writes the same list in its own spelling, and the order is the
15585 // author's there too.
15586 let mut dj = fmt_doc(
15587 "{#intro .lead .center data-line-height=\"1.5\"}\nhello\n",
15588 Format::Djot,
15589 );
15590 dj.caret = dj.source.find("hello").unwrap();
15591 dj.set_alignment(Some(Align::Right));
15592 assert_eq!(
15593 dj.source,
15594 "{#intro .lead .right data-line-height=\"1.5\"}\nhello\n"
15595 );
15596 }
15597
15598 /// A page break is a block, so twig alone lands one after the caret's whole
15599 /// block; the paragraph is parted at the caret first, exactly as
15600 /// `insert_thematic_break` parts it, and each format spells the directive
15601 /// its own way.
15602 #[test]
15603 fn insert_page_break_parts_the_paragraph_and_spells_the_directive() {
15604 let mut md = doc_with("page_break_md", "hello world\n");
15605 md.caret = 5;
15606 md.insert_page_break();
15607 assert_eq!(md.source, "hello\n\n::page-break\n\nworld\n");
15608 assert!(md.dirty);
15609 assert_eq!(md.status, None);
15610
15611 let mut dj = fmt_doc("hello world\n", Format::Djot);
15612 dj.caret = 5;
15613 dj.insert_page_break();
15614 assert_eq!(dj.source, "hello\n\n::: page-break\n:::\n\nworld\n");
15615
15616 // At a block's end there is no second half to mint, so the break simply
15617 // follows the block — the rule the rule button already has.
15618 let mut end = doc_with("page_break_end", "hello\n");
15619 end.caret = 5;
15620 end.insert_page_break();
15621 assert_eq!(end.source, "hello\n\n::page-break\n");
15622
15623 // And it reaches the map as the placeholder row a frontend paginates on.
15624 end.view = View::Wysiwyg;
15625 end.build_visual(80);
15626 assert_eq!(
15627 end.vmap
15628 .rows
15629 .iter()
15630 .find_map(|r| r.leaf_directive.as_ref())
15631 .map(|m| m.name.as_str()),
15632 Some(PAGE_BREAK)
15633 );
15634 }
15635
15636 /// The vocabulary's capabilities, per format. The two block properties are
15637 /// `SetBlockAttrs` and the three run ones `WrapRangeAttrs`, which is why
15638 /// AsciiDoc can align a paragraph and not size a run: its `[#id.role]#text#`
15639 /// keeps an id and a role and has no slot for a `data-` key.
15640 #[test]
15641 fn the_presentation_capabilities_are_ragged_per_format() {
15642 for fmt in [Format::Markdown, Format::Djot, Format::Html] {
15643 let c = Capabilities::of(fmt);
15644 assert!(c.alignment, "{fmt:?} alignment");
15645 assert!(c.line_spacing, "{fmt:?} line spacing");
15646 assert!(c.font_size, "{fmt:?} size");
15647 assert!(c.font_family, "{fmt:?} face");
15648 assert!(c.text_color, "{fmt:?} colour");
15649 }
15650 // Markdown spells both only under the extensions leaf parses with — a
15651 // `<div>` and a `<span>` read back as containers under `html_elements`,
15652 // and `::page-break` as a directive under `directives`. Ask twig's own
15653 // defaults and the answer is no, which is why `Capabilities` is built
15654 // with `supports_with`.
15655 assert!(!Format::Markdown.supports(Gesture::SetBlockAttrs));
15656 assert!(!Format::Markdown.supports(Gesture::WrapRangeAttrs));
15657 assert!(!Format::Markdown.supports(Gesture::InsertDirective));
15658
15659 let adoc = Capabilities::of(Format::Asciidoc);
15660 assert!(adoc.alignment && adoc.line_spacing, "AsciiDoc's `[…]` line");
15661 assert!(
15662 !adoc.font_size && !adoc.font_family && !adoc.text_color,
15663 "AsciiDoc has no inline spelling that keeps a data- key"
15664 );
15665
15666 // XML spells none of it, and neither page break.
15667 let xml = Capabilities::of(Format::Xml);
15668 assert!(!xml.alignment && !xml.font_size && !xml.page_break);
15669 assert!(Capabilities::of(Format::Markdown).page_break);
15670 assert!(Capabilities::of(Format::Djot).page_break);
15671
15672 // And those two *only*, though twig spells the gesture in HTML and
15673 // AsciiDoc as well: it spells it differently there —
15674 // `<page-break></page-break>` and `<<<` — and the walker reads neither,
15675 // so the button would write a break that draws as nothing at all in
15676 // HTML and as an empty unlabelled row in AsciiDoc. The flag describes
15677 // what leaf can show, not what twig can write. See
15678 // `docs/tasks/page-break-in-html-and-asciidoc.md`.
15679 let exts = parse_extensions();
15680 assert!(Format::Html.supports_with(exts, Gesture::InsertDirective));
15681 assert!(Format::Asciidoc.supports_with(exts, Gesture::InsertDirective));
15682 assert!(!Capabilities::of(Format::Html).page_break);
15683 assert!(!Capabilities::of(Format::Asciidoc).page_break);
15684 }
15685
15686 /// A format that cannot spell a property refuses in its own words and
15687 /// writes nothing — the guard every other gesture has.
15688 #[test]
15689 fn a_presentation_gesture_a_format_cannot_spell_is_refused_with_a_reason() {
15690 let src = "<doc><p>hello</p></doc>\n";
15691 #[allow(clippy::type_complexity)]
15692 let ops: [(&str, &dyn Fn(&mut Doc)); 6] = [
15693 ("alignment", &|d: &mut Doc| {
15694 d.set_alignment(Some(Align::Center))
15695 }),
15696 ("line spacing", &|d: &mut Doc| {
15697 d.set_line_spacing(Some(LineSpacing::Double))
15698 }),
15699 ("size", &|d: &mut Doc| {
15700 d.set_font_size(Some(SizeStep::Large))
15701 }),
15702 ("face", &|d: &mut Doc| {
15703 d.set_font_family(Some(FontFamily::Serif))
15704 }),
15705 ("colour", &|d: &mut Doc| {
15706 d.set_text_color(Some(MarkColor::Red))
15707 }),
15708 ("page break", &|d: &mut Doc| d.insert_page_break()),
15709 ];
15710 for (name, op) in ops {
15711 let mut d = fmt_doc(src, Format::Xml);
15712 let at = d.source.find("hello").unwrap();
15713 d.caret = at;
15714 d.anchor = Some(at + 5);
15715 op(&mut d);
15716 assert_eq!(d.source, src, "{name} edited an XML document");
15717 assert!(!d.dirty, "{name} marked the document dirty");
15718 let status = d.status.as_deref().unwrap_or("");
15719 assert!(
15720 status.contains("xml"),
15721 "{name}: the refusal should name the format, got {status:?}"
15722 );
15723 }
15724
15725 // AsciiDoc is the ragged one: the block gesture works where the run
15726 // gesture does not, and a *selection* is what tells the two apart.
15727 let mut adoc = fmt_doc("hello world\n", Format::Asciidoc);
15728 adoc.anchor = Some(0);
15729 adoc.caret = 5;
15730 adoc.set_font_size(Some(SizeStep::Large));
15731 assert_eq!(adoc.source, "hello world\n", "no inline spelling");
15732 assert!(adoc.status.is_some());
15733 }
15734
15735 /// A read-only document takes none of it, and a caret on a blank line has
15736 /// no block to carry an attribute — both say so rather than writing.
15737 #[test]
15738 fn a_presentation_gesture_respects_read_only_and_a_blank_line() {
15739 let mut ro = fmt_doc("hello\n", Format::Djot);
15740 ro.read_only = true;
15741 ro.caret = 1;
15742 ro.set_alignment(Some(Align::Center));
15743 assert_eq!(ro.source, "hello\n");
15744
15745 let mut blank = fmt_doc("a\n\n\nb\n", Format::Djot);
15746 blank.caret = 2; // the empty line between the two paragraphs
15747 blank.set_alignment(Some(Align::Center));
15748 assert_eq!(blank.source, "a\n\n\nb\n");
15749 assert!(
15750 blank.status.as_deref().unwrap_or("").contains("no block"),
15751 "got {:?}",
15752 blank.status
15753 );
15754 }
15755
15756 /// A block attribute gesture keeps the caret on the **text** it was on, not
15757 /// on the byte offset it had. Markdown has nowhere to put a paragraph's
15758 /// attributes but a `<div>` around it, and twig splices the div and the
15759 /// block it wraps as one region — so a caret that kept its offset landed in
15760 /// the markup, and every press after the first answered "no block at the
15761 /// caret" with the toolbar's queries reading nothing.
15762 #[test]
15763 fn a_markdown_block_gesture_keeps_the_caret_on_its_text() {
15764 let word = |d: &Doc| d.caret - d.source.find("brown").unwrap();
15765 let mut md = fmt_doc("the quick brown fox\n", Format::Markdown);
15766 md.caret = md.source.find("brown").unwrap() + 2; // "br|own"
15767
15768 // Wrapping: the div and two blank lines open above the block.
15769 md.set_alignment(Some(Align::Center));
15770 assert_eq!(
15771 md.source,
15772 "<div class=\"center\">\n\nthe quick brown fox\n\n</div>\n"
15773 );
15774 assert_eq!(word(&md), 2, "the caret left its word: {}", md.caret);
15775 assert_eq!(md.alignment_at_caret(), Some(Align::Center));
15776
15777 // Re-styling: the attribute line changes length under the same caret,
15778 // and the second press reaches the same block rather than nothing.
15779 md.set_alignment(Some(Align::Right));
15780 assert_eq!(
15781 md.source, "<div class=\"right\">\n\nthe quick brown fox\n\n</div>\n",
15782 "a second press re-styles the div"
15783 );
15784 assert_eq!(md.status, None);
15785 assert_eq!(word(&md), 2);
15786
15787 // A second key on the same div — the line grows, the caret rides it.
15788 md.set_line_spacing(Some(LineSpacing::Double));
15789 assert_eq!(
15790 md.source,
15791 "<div class=\"right\" data-line-height=\"2\">\n\nthe quick brown fox\n\n</div>\n"
15792 );
15793 assert_eq!(word(&md), 2);
15794 assert_eq!(md.line_spacing_at_caret(), Some(LineSpacing::Double));
15795
15796 // Unwrapping: the line shrinks, and then the div goes altogether.
15797 md.set_alignment(None);
15798 assert_eq!(
15799 md.source,
15800 "<div data-line-height=\"2\">\n\nthe quick brown fox\n\n</div>\n"
15801 );
15802 assert_eq!(word(&md), 2);
15803 md.set_line_spacing(None);
15804 assert_eq!(md.source, "the quick brown fox\n", "the last key unwraps");
15805 assert_eq!(word(&md), 2, "the caret came back down with the block");
15806 assert_eq!(md.alignment_at_caret(), None);
15807 assert_eq!(md.status, None);
15808 }
15809
15810 /// The same rule in djot, where the spelling is a `{…}` line *above* the
15811 /// block rather than a wrapper around it: inserting it pushes the block
15812 /// down, re-styling it changes the line's length, and clearing the last key
15813 /// takes the line away again. The caret rides all three.
15814 #[test]
15815 fn a_djot_attribute_line_keeps_the_caret_on_its_text() {
15816 let word = |d: &Doc| d.caret - d.source.find("brown").unwrap();
15817 let mut dj = fmt_doc("the quick brown fox\n", Format::Djot);
15818 dj.caret = dj.source.find("brown").unwrap() + 2;
15819
15820 dj.set_alignment(Some(Align::Center));
15821 assert_eq!(dj.source, "{.center}\nthe quick brown fox\n");
15822 assert_eq!(word(&dj), 2);
15823 assert_eq!(dj.alignment_at_caret(), Some(Align::Center));
15824
15825 dj.set_line_spacing(Some(LineSpacing::Double));
15826 assert_eq!(
15827 dj.source, "{.center data-line-height=\"2\"}\nthe quick brown fox\n",
15828 "a second press edits the line the first wrote"
15829 );
15830 assert_eq!(word(&dj), 2);
15831
15832 dj.set_alignment(None);
15833 assert_eq!(dj.source, "{data-line-height=\"2\"}\nthe quick brown fox\n");
15834 assert_eq!(word(&dj), 2);
15835
15836 dj.set_line_spacing(None);
15837 assert_eq!(dj.source, "the quick brown fox\n");
15838 assert_eq!(word(&dj), 2);
15839 assert_eq!(dj.status, None);
15840 }
15841
15842 /// The run gestures with no selection are the block gesture, so they keep
15843 /// the caret the same way — and a heading keeps it inside the heading's own
15844 /// text, past the `# ` its content span starts after. A selection rides
15845 /// along whole: a block gesture is not a run gesture, and what was selected
15846 /// before the press is still selected after it.
15847 #[test]
15848 fn a_block_gesture_carries_a_selection_and_a_heading_caret_too() {
15849 // No selection: the run gesture goes through the block door.
15850 let mut md = fmt_doc("the quick brown fox\n", Format::Markdown);
15851 md.caret = md.source.find("brown").unwrap() + 2;
15852 md.set_font_size(Some(SizeStep::Large));
15853 assert_eq!(
15854 md.source,
15855 "<div data-size=\"large\">\n\nthe quick brown fox\n\n</div>\n"
15856 );
15857 assert_eq!(md.caret - md.source.find("brown").unwrap(), 2);
15858 assert_eq!(md.font_size_at_caret(), Some(SizeStep::Large));
15859 md.set_font_size(Some(SizeStep::Small));
15860 assert_eq!(
15861 md.font_size_at_caret(),
15862 Some(SizeStep::Small),
15863 "the second press reached the same block"
15864 );
15865
15866 // A selection: alignment is the block's whatever is selected, and the
15867 // words stay selected.
15868 let mut sel = fmt_doc("the quick brown fox\n", Format::Markdown);
15869 let at = sel.source.find("brown").unwrap();
15870 sel.anchor = Some(at);
15871 sel.caret = at + 5;
15872 sel.set_alignment(Some(Align::Center));
15873 let now = sel.source.find("brown").unwrap();
15874 assert_eq!(sel.selection(), Some((now, now + 5)), "the words moved out");
15875
15876 // A heading: the content span starts past the `# `.
15877 let mut h = fmt_doc("# hi there\n\nbody\n", Format::Markdown);
15878 h.caret = h.source.find("there").unwrap() + 1;
15879 h.set_alignment(Some(Align::Right));
15880 assert_eq!(
15881 h.source,
15882 "<div class=\"right\">\n\n# hi there\n\n</div>\n\nbody\n"
15883 );
15884 assert_eq!(h.caret, h.source.find("there").unwrap() + 1);
15885 assert_eq!(h.alignment_at_caret(), Some(Align::Right));
15886 }
15887}