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::MarkColor;
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}
847
848impl Capabilities {
849 /// Resolve every flag for `format`, as leaf parses it. Pure and cheap —
850 /// twig computes each from a static table — but a frontend that wants to
851 /// hold them can.
852 ///
853 /// The extensions are not a parameter because they are not a choice a
854 /// caller makes: every leaf document is parsed with [`parse_extensions`],
855 /// so the format is the whole of what varies.
856 pub fn of(format: Format) -> Self {
857 let exts = parse_extensions();
858 let supports = |g| format.supports_with(exts, g);
859 let inline = |k| supports(Gesture::ToggleInline(k));
860 let container = |k| supports(Gesture::ToggleBlockContainer(k));
861 Self {
862 bold: inline(InlineKind::Strong),
863 italic: inline(InlineKind::Emph),
864 code: inline(InlineKind::Verbatim),
865 mark: inline(InlineKind::Mark),
866 underline: inline(InlineKind::Insert),
867 strike: inline(InlineKind::Delete),
868 mark_color: supports(Gesture::SetMarkColor),
869 superscript: inline(InlineKind::Superscript),
870 subscript: inline(InlineKind::Subscript),
871 heading: supports(Gesture::SetBlock),
872 blockquote: container(BlockContainerKind::BlockQuote),
873 bullet_list: container(BlockContainerKind::BulletList),
874 ordered_list: container(BlockContainerKind::OrderedList),
875 // Both halves of the checkbox story, and leaf offers no control that
876 // needs only one: the item gesture mints the box, the checked one
877 // ticks it, and a format spelling a `task_marker` spells both.
878 task: supports(Gesture::ToggleTaskItem) && supports(Gesture::ToggleTaskChecked),
879 link: supports(Gesture::InsertLink),
880 image: supports(Gesture::InsertImage),
881 thematic_break: supports(Gesture::InsertThematicBreak),
882 footnote: supports(Gesture::InsertFootnote),
883 code_language: supports(Gesture::SetCodeLanguage),
884 table: spells_pipe_tables(format),
885 cell_line_break: supports(Gesture::InsertLineBreak),
886 }
887 }
888}
889
890/// The source of [`Doc::identity`], one per document ever built.
891static NEXT_IDENTITY: std::sync::atomic::AtomicU64 = std::sync::atomic::AtomicU64::new(0);
892
893impl Doc {
894 #[cfg(feature = "fs")]
895 pub fn open(path: PathBuf) -> Result<Self> {
896 let bytes = std::fs::read(&path).with_context(|| format!("reading {}", path.display()))?;
897 Self::from_disk_bytes(path, bytes)
898 }
899
900 /// An empty document *named* `path`, for a file that isn't there yet — what
901 /// every other terminal editor gives you when you name a file that doesn't
902 /// exist. It is a real named document, not a [`Doc::blank`]: `is_untitled`
903 /// is false, so ⌘S writes straight to `path` with no Save As detour, and
904 /// the header shows the name the user asked for.
905 ///
906 /// The format comes from the extension, exactly as [`Doc::open`] reads it —
907 /// so `leaf notes.dj` starts a djot buffer rather than the Markdown
908 /// [`Doc::blank`] has to assume for want of a name. An extension leaf can't
909 /// parse is still an error: a mistyped flag or a stray argument should say
910 /// so, not open a buffer promising to save somewhere.
911 ///
912 /// The watermark is the hash of *no bytes*, not `None`, and that is the
913 /// whole trick: `None` means untitled, and would leave [`Doc::disk_state`]
914 /// answering [`DiskState::Untitled`] for a document that has a path and
915 /// intends to write to it. Hashing `""` instead makes the answers the true
916 /// ones — [`DiskState::Missing`] while the file still isn't there (a save
917 /// recreates it, which is exactly what this is for), and
918 /// [`DiskState::Changed`] if somebody creates it underneath us between
919 /// launch and save, so the frontend's overwrite prompt guards a new file as
920 /// it guards an opened one.
921 ///
922 /// Nothing is written here. A buffer that is never typed into never touches
923 /// the filesystem, and a `path` whose directory doesn't exist is allowed to
924 /// open — the write is where that fails, and it says so then.
925 #[cfg(feature = "fs")]
926 pub fn create(path: PathBuf) -> Result<Self> {
927 Self::from_disk_bytes(path, Vec::new())
928 }
929
930 /// [`Doc::open`] when the file is there, [`Doc::create`] when it isn't —
931 /// the call a CLI frontend wants for its path argument.
932 ///
933 /// The decision is made from the failed read itself rather than a `exists()`
934 /// check first, so there is no window between the two for the file to appear
935 /// or vanish in. Only `NotFound` opens a new buffer: a permissions error or
936 /// a directory in the way is still an error, because pretending those are
937 /// "no file yet" would offer to save over something leaf couldn't read.
938 #[cfg(feature = "fs")]
939 pub fn open_or_create(path: PathBuf) -> Result<Self> {
940 match std::fs::read(&path) {
941 Ok(bytes) => Self::from_disk_bytes(path, bytes),
942 Err(e) if e.kind() == std::io::ErrorKind::NotFound => Self::create(path),
943 Err(e) => Err(e).with_context(|| format!("reading {}", path.display())),
944 }
945 }
946
947 /// The shared body of [`Doc::open`] and [`Doc::create`]: bytes that are (or
948 /// stand in for) the file at `path`, parsed as the format its extension
949 /// names. Keeping the two on one path is what makes a new file's document
950 /// identical in every respect to an opened one but its contents.
951 #[cfg(feature = "fs")]
952 fn from_disk_bytes(path: PathBuf, bytes: Vec<u8>) -> Result<Self> {
953 let format = detect_format(&path)?;
954 let editor = new_editor(&bytes, format)?;
955 let source = String::from_utf8(bytes).map_err(|_| anyhow!("document is not UTF-8"))?;
956 let disk_hash = Some(hash_bytes(source.as_bytes()));
957 // Store the document's *absolute* path. A relative one (`leaf README.md`)
958 // has an empty parent, so a frontend can't resolve a relative image
959 // destination (``) against the document's directory and the
960 // picture silently falls back to its text placeholder. `absolute` is
961 // purely lexical — it prefixes the current directory and normalizes, but
962 // reads nothing and resolves no symlinks — so `file_name` and save are
963 // unchanged; it only gives `path.parent()` something to join against.
964 let path = std::path::absolute(&path).unwrap_or(path);
965 Ok(Doc::from_parts(editor, format, path, source, disk_hash))
966 }
967
968 /// Build a document from an in-memory string, the format named explicitly —
969 /// the portable, filesystem-free counterpart to [`Doc::open`] (which reads a
970 /// path and sniffs the format from its extension). A wasm or FFI host, which
971 /// has no path to read, uses this: it hands over bytes it fetched however it
972 /// could, and later persists [`Doc::source`] however it can (a browser
973 /// download, `localStorage`, a backend `PUT`) and calls [`Doc::mark_saved`].
974 ///
975 /// No file backs the result, so it starts untitled ([`Doc::is_untitled`] is
976 /// true) exactly like a [`Doc::blank`] that has been given content.
977 pub fn from_source(source: String, format: Format) -> Result<Self> {
978 let editor = new_editor(source.as_bytes(), format)?;
979 Ok(Doc::from_parts(
980 editor,
981 format,
982 PathBuf::new(),
983 source,
984 None,
985 ))
986 }
987
988 /// An untitled, empty document — the `+` button and a `leaf` launched with
989 /// no file argument. Nothing on disk backs it until a [`Doc::save_as`].
990 ///
991 /// It is Markdown, because a format has to be chosen before a name exists to
992 /// read one from: `detect_format` reads the extension and an untitled
993 /// document has neither. Markdown is what leaf's own files are, what its
994 /// block markers are already written for (`insert_block_prefix`), and the
995 /// extension a Save As will overwhelmingly pick — a wrong guess here would
996 /// mean typing djot into a buffer parsing it as Markdown. Note that Save As
997 /// *doesn't* revisit this: see [`Doc::save_as`].
998 pub fn blank() -> Result<Self> {
999 let format = Format::Markdown;
1000 let editor = new_editor(b"", format)?;
1001 // An empty `path` is the untitled marker (`path` is a public `PathBuf`
1002 // field two frontends already read; making it an `Option` to say this
1003 // would break both). `is_untitled` is the question to ask, not the
1004 // representation to copy.
1005 Ok(Doc::from_parts(
1006 editor,
1007 format,
1008 PathBuf::new(),
1009 String::new(),
1010 None,
1011 ))
1012 }
1013
1014 /// The fields every constructor agrees on, so `open` and `blank` can't drift
1015 /// apart in the ones neither of them has an opinion about.
1016 // `identity` is taken from a counter rather than from the `Doc`'s address,
1017 // which moves — a session that holds one is moved into and out of
1018 // containers freely, and an identity that changed with it would defeat the
1019 // one comparison it exists for.
1020 fn from_parts(
1021 editor: Editor,
1022 format: Format,
1023 path: PathBuf,
1024 source: String,
1025 disk_hash: Option<u64>,
1026 ) -> Self {
1027 Doc {
1028 editor,
1029 format,
1030 path,
1031 disk_hash,
1032 clean_source: source.clone(),
1033 source,
1034 caret: 0,
1035 anchor: None,
1036 dirty: false,
1037 status: None,
1038 read_only: false,
1039 highlights: Vec::new(),
1040 // leaf opens in the rich-text (WYSIWYG) view by default — the
1041 // markup-resolved surface is leaf's differentiator. Frontends can
1042 // still start in source view explicitly (e.g. a CLI flag), and ⌘e/⌥w
1043 // toggles at runtime.
1044 view: View::Wysiwyg,
1045 // `None` by default — the clean surface Diaryx ships, with typed
1046 // syntax kept literal; a markup-fluent frontend can climb the
1047 // ladder to `Shortcuts` or `Full`.
1048 markup_mode: MarkupMode::default(),
1049 // Fold by default — flowing prose that reflows to the viewport, the
1050 // behaviour every frontend had before this preference existed.
1051 line_flow: LineFlow::default(),
1052 last_edit_kind: None,
1053 pending_marks: InlineMarks::empty(),
1054 pending_at: None,
1055 goal_col: None,
1056 vmap: VisualMap::default(),
1057 smap: SourceMap::default(),
1058 // No map yet — the first `build_source` always builds.
1059 smap_key: None,
1060 revision: 0,
1061 undo_steps: 0,
1062 redo_steps: 0,
1063 // No map yet — the first `build_visual` always builds.
1064 vmap_key: None,
1065 identity: NEXT_IDENTITY.fetch_add(1, std::sync::atomic::Ordering::Relaxed),
1066 block_cache: wysiwyg::BlockCache::default(),
1067 media_rows: HashMap::new(),
1068 scroll: 0,
1069 body_origin: (0, 0),
1070 body_width: 0,
1071 body_height: 0,
1072 drawn_caret: None,
1073 }
1074 }
1075
1076 /// Whether this document has no file behind it yet — a [`Doc::blank`] that
1077 /// has never been saved. The question a ⌘S handler asks to know it should
1078 /// open a Save As picker instead ([`Doc::save`] won't guess a name), and the
1079 /// header asks to know the name it shows is a placeholder.
1080 pub fn is_untitled(&self) -> bool {
1081 self.path.as_os_str().is_empty()
1082 }
1083
1084 pub fn toggle_view(&mut self) {
1085 self.view = match self.view {
1086 View::Source => View::Wysiwyg,
1087 View::Wysiwyg => View::Source,
1088 };
1089 self.scroll = 0;
1090 self.status = None;
1091 // Entering WYSIWYG, the caret may be sitting in now-hidden frontmatter;
1092 // lift it to the first rendered offset.
1093 self.clamp_caret();
1094 }
1095
1096 /// The current markup-exposure preference (see [`MarkupMode`]).
1097 pub fn markup_mode(&self) -> MarkupMode {
1098 self.markup_mode
1099 }
1100
1101 /// Set the markup-exposure preference. Both of its axes take effect at
1102 /// once: the editing one on the next [`insert`](Self::insert), and the
1103 /// rendering one on the next build — which is why this drops the cached
1104 /// visual map and the per-block render cache, exactly as
1105 /// [`set_line_flow`](Self::set_line_flow) does.
1106 pub fn set_markup_mode(&mut self, mode: MarkupMode) {
1107 if self.markup_mode == mode {
1108 return;
1109 }
1110 self.markup_mode = mode;
1111 // Neither cache is keyed on the mode, and moving between `Full` and the
1112 // hidden modes changes every row the caret's line renders to — so
1113 // invalidate both explicitly.
1114 self.vmap_key = None;
1115 self.block_cache = wysiwyg::BlockCache::default();
1116 }
1117
1118 /// The source byte range of the line the caret sits on, when that line
1119 /// should render its raw delimiters — `None` in every mode and view that
1120 /// hides them, which is what the builder reads as "reveal nothing".
1121 ///
1122 /// A *source* line (newline to newline), not a visual row: a wrapped
1123 /// paragraph and a `LineFlow::Preserve` soft break both split one source
1124 /// line across several rows, and revealing half a delimiter pair because the
1125 /// other half wrapped would be worse than revealing neither. The range
1126 /// excludes the terminating newline and is empty-but-present on a blank
1127 /// line, which reveals nothing but still keys the caches correctly.
1128 ///
1129 /// Only in [`View::Wysiwyg`]: source view already shows every byte, so
1130 /// there is nothing there to reveal.
1131 pub(crate) fn reveal_line(&self) -> Option<Range<usize>> {
1132 if !self.markup_mode.reveals_caret_line() || self.view != View::Wysiwyg {
1133 return None;
1134 }
1135 Some(source_line_range(&self.source, self.caret))
1136 }
1137
1138 /// The current soft-break flow preference (see [`LineFlow`]).
1139 pub fn line_flow(&self) -> LineFlow {
1140 self.line_flow
1141 }
1142
1143 /// Set the soft-break flow preference. The mode changes how every block lays
1144 /// out, so a change drops the cached visual map and the per-block render
1145 /// cache, forcing the next [`build_visual`] to rebuild under the new flow.
1146 ///
1147 /// [`build_visual`]: Self::build_visual
1148 pub fn set_line_flow(&mut self, mode: LineFlow) {
1149 if self.line_flow == mode {
1150 return;
1151 }
1152 self.line_flow = mode;
1153 // Both caches are keyed on `(revision, wrap)`, neither of which moved —
1154 // so invalidate them explicitly, or the next build would reuse rows laid
1155 // out under the old flow.
1156 self.vmap_key = None;
1157 self.block_cache = wysiwyg::BlockCache::default();
1158 }
1159
1160 pub fn view_name(&self) -> &'static str {
1161 match self.view {
1162 View::Source => "source",
1163 View::Wysiwyg => "wysiwyg",
1164 }
1165 }
1166
1167 /// Rebuild the WYSIWYG visual map for the current tree at `width` columns
1168 /// (called by the renderer each frame it's in the WYSIWYG view).
1169 /// Build the WYSIWYG map, wrapped at `width` display columns.
1170 ///
1171 /// Cheap to call every frame, which is what both frontends do: the map is a
1172 /// pure function of the document and the wrap width, so a call that would
1173 /// rebuild the same map returns the one already built. Only an edit (or a
1174 /// resize) pays.
1175 ///
1176 /// That isn't a micro-optimisation. A frontend repaints for reasons that have
1177 /// nothing to do with the text — a blinking caret, a scroll, a focus change —
1178 /// and rebuilding here is O(document): 23 ms on a 1 MB file, of which 5 ms is
1179 /// marshalling twig's AST across the C ABI. Paid twice a second by the GUI's
1180 /// blink timer, that was 14% of a core spent redrawing an unchanged document.
1181 /// (`cargo run --release -p leaf-core --example bench` for the numbers.)
1182 pub fn build_visual(&mut self, width: usize) {
1183 self.build_map(Some(width));
1184 }
1185
1186 /// Build the WYSIWYG map with each block as a single unwrapped row — for a
1187 /// frontend (the GUI) that wraps at its own proportional pixel width rather
1188 /// than a fixed character column.
1189 pub fn build_visual_unwrapped(&mut self) {
1190 self.build_map(None);
1191 }
1192
1193 /// Build the source view's syntax map ([`Doc::smap`]) — the styling for
1194 /// [`View::Source`], the way [`build_visual`](Self::build_visual) is the
1195 /// styling for [`View::Wysiwyg`].
1196 ///
1197 /// A frontend calls this before painting raw source. One that doesn't gets
1198 /// an empty map and paints unstyled text, so this is additive: nothing
1199 /// breaks by not calling it.
1200 ///
1201 /// Built at most once per revision, and the revision is the whole key — the
1202 /// map has no width and no caret in it, so it survives every resize, every
1203 /// motion, and every selection change.
1204 ///
1205 /// The builds it does do cost a whole-arena marshal, which is precisely what
1206 /// the WYSIWYG path works to avoid, so this has no incremental path where
1207 /// that one has two. From `cargo run --release -p leaf-core --example
1208 /// bench`, per keystroke, against the WYSIWYG build the source view is
1209 /// *not* doing:
1210 ///
1211 /// | size | nodes | marshal | `source::build` | (`wysiwyg::build`) |
1212 /// |------:|-------:|--------:|----------------:|-------------------:|
1213 /// | 10 KB| 613 | 0.16 ms| 0.07 ms | 0.28 ms |
1214 /// | 100 KB| 6 097 | 0.84 ms| 0.38 ms | 2.43 ms |
1215 /// | 1 MB| 60 601 | 5.67 ms| 3.12 ms | 23.39 ms |
1216 ///
1217 /// Linear, two thirds of it the marshal, and the build itself five to seven
1218 /// times cheaper than the one it stands in for at every size. Comfortable
1219 /// well past any document a person edits in a terminal — a megabyte is where
1220 /// it would want [`Editor::dirty_range`] and the same splice treatment
1221 /// `build_spliced` gives the other map. The door is open; nothing has needed
1222 /// it yet.
1223 pub fn build_source(&mut self) {
1224 if self.smap_key == Some(self.revision) {
1225 return;
1226 }
1227 let nodes = self.nodes();
1228 self.smap = source::build(&nodes, &self.source);
1229 self.smap_key = Some(self.revision);
1230 }
1231
1232 /// Tell the model how many visual rows each block image should reserve, keyed
1233 /// by the image's destination. A terminal frontend calls this once it has
1234 /// decoded and measured its pictures — core does no image I/O, so this is the
1235 /// only way it learns a height — and the next [`Doc::build_visual`] lays each
1236 /// placeholder out that tall (the label row plus blank filler rows the
1237 /// frontend paints the raster over). A destination left out of the map falls
1238 /// back to the bare one-row placeholder, which is also what a frontend that
1239 /// can't draw pictures (or lays them out in its own units, like the GUI) gets
1240 /// by never calling this.
1241 ///
1242 /// Cheap to call every frame with the same map: only a *change* invalidates
1243 /// the built map (and the block-row cache, since a height isn't part of a
1244 /// block's bytes and so wouldn't otherwise re-render it). Steady state is a
1245 /// no-op, so a frontend can just hand over its current measurements each frame.
1246 pub fn set_media_rows(&mut self, rows: HashMap<String, usize>) {
1247 if self.media_rows == rows {
1248 return;
1249 }
1250 self.media_rows = rows;
1251 // A height lives outside the block's source bytes, so the content-keyed
1252 // block cache would hand back the old-height rows on a hit. Drop it (and
1253 // the splice layout it carries) so the next build re-renders every block
1254 // at the new heights, and force that build by clearing the map key.
1255 self.block_cache = wysiwyg::BlockCache::default();
1256 self.vmap_key = None;
1257 }
1258
1259 /// The revision the document's text is at — bumped by every edit, undo,
1260 /// redo, and reload, and by nothing else. A frontend caches against this to
1261 /// tell a repaint that needs new work from one that doesn't.
1262 ///
1263 /// It counts *edits*, not distinct texts: typing `x` and deleting it again
1264 /// lands on the same text two revisions later. Work is only ever rebuilt
1265 /// needlessly, never wrongly reused.
1266 pub fn revision(&self) -> u64 {
1267 self.revision
1268 }
1269
1270 /// The identity of the map presently in [`vmap`](Self::vmap) — what the last
1271 /// [`build_visual`](Self::build_visual) built it from, or the identity of an
1272 /// unbuilt map before the first one.
1273 ///
1274 /// This is *not* [`revision`](Self::revision). The revision says where the
1275 /// text is; this says where the map is, and the two part company the moment
1276 /// an edit lands, until something rebuilds. A frontend that keeps its own
1277 /// copy of the map — leaf-ratatui stashes core's before splicing filler rows
1278 /// under an oversized heading — compares this against the value it held when
1279 /// it took the copy, and learns whether `vmap` is still the map it stashed
1280 /// or one somebody else has since rebuilt. Restoring a copy over a newer
1281 /// map would paint a stale document; restoring nothing hands core's
1282 /// incremental rebuild a map it never built.
1283 ///
1284 /// "Somebody else" includes another document. The key names the `Doc`
1285 /// as well as the build, so a frontend that draws two documents through
1286 /// one stash — a host with several buffers, or one that opens the next
1287 /// document where the last one stood — never has the copy it took of one
1288 /// accepted by the other, however alike their builds are.
1289 pub fn visual_key(&self) -> VisualKey {
1290 VisualKey(self.identity, self.vmap_key.clone())
1291 }
1292
1293 /// The map, built at most once per `(revision, wrap)`. `clamp_caret` still
1294 /// runs on every call: the caret moves without the document changing, and
1295 /// keeping it on a legal stop is this function's job either way.
1296 fn build_map(&mut self, wrap: Option<usize>) {
1297 // Under `MarkupMode::Full` the map is a function of the caret's *line*
1298 // as well as the text, so the line joins the key: moving within a line
1299 // still reuses the map, and crossing into another one rebuilds it. In
1300 // every other mode `reveal_line` is `None` and the key is what it was,
1301 // so caret motion goes on costing nothing.
1302 let reveal = self.reveal_line();
1303 let key = (self.revision, wrap, reveal.clone());
1304 if self.vmap_key.as_ref() != Some(&key) {
1305 // Enumerate the top-level blocks cheaply — no whole-arena marshal.
1306 // A subtree is pulled only for the block(s) that actually changed, so
1307 // the FFI marshal shrinks from O(document) to O(edited block).
1308 let top = self.top_blocks();
1309
1310 // Fast path: when twig reports a dirty byte range, try to patch the
1311 // previous map in place — a single-block edit moves the prefix,
1312 // shifts the suffix, and re-renders only one block. `build_spliced`
1313 // returns `None` (and we fall back to the always-correct full rebuild)
1314 // whenever the edit reshaped the block structure, hit a table, or
1315 // there's no previous map to patch.
1316 // Preserve soft breaks as written when the flow preference asks for
1317 // it — the builder renders each as its own visual row instead of
1318 // folding it into the reflowed paragraph.
1319 let preserve_soft = self.line_flow == LineFlow::Preserve;
1320 let spliced = match self.editor.dirty_range() {
1321 Some(dirty) => {
1322 let prev = std::mem::take(&mut self.vmap);
1323 let source = &self.source;
1324 let cache = &mut self.block_cache;
1325 let media_rows = &self.media_rows;
1326 let editor = &mut self.editor;
1327 wysiwyg::build_spliced(
1328 prev,
1329 source,
1330 wrap,
1331 preserve_soft,
1332 &top,
1333 dirty,
1334 media_rows,
1335 reveal.clone(),
1336 cache,
1337 |id| editor.subtree(NodeId(id)).unwrap_or_default(),
1338 )
1339 }
1340 None => None,
1341 };
1342 self.vmap = spliced.unwrap_or_else(|| {
1343 let source = &self.source;
1344 let cache = &mut self.block_cache;
1345 let media_rows = &self.media_rows;
1346 let editor = &mut self.editor;
1347 wysiwyg::build_cached(
1348 &top,
1349 source,
1350 wrap,
1351 preserve_soft,
1352 media_rows,
1353 reveal,
1354 cache,
1355 |id| editor.subtree(NodeId(id)).unwrap_or_default(),
1356 )
1357 });
1358 // Acknowledge the dirty range so the next edit's range starts fresh.
1359 self.editor.clear_dirty();
1360 self.vmap_key = Some(key);
1361 }
1362 self.clamp_caret();
1363 }
1364
1365 fn nodes(&mut self) -> Vec<FlatNode> {
1366 self.editor.nodes().unwrap_or_default()
1367 }
1368
1369 /// The document's top-level blocks for the incremental render. See
1370 /// [`wysiwyg::top_blocks`] for why this isn't simply `child_spans(None)`.
1371 fn top_blocks(&mut self) -> Vec<QueryMatch> {
1372 wysiwyg::top_blocks(&mut self.editor)
1373 }
1374
1375 pub fn format_name(&self) -> &'static str {
1376 // `Format` is `#[non_exhaustive]` as of twig 3.0, so the wildcard is
1377 // required. It also covers `Asciidoc`, which twig parses but cannot
1378 // serialize — leaf never opens a document in it (see `Doc::open`).
1379 match self.format {
1380 Format::Djot => "djot",
1381 Format::Markdown => "markdown",
1382 Format::Xml => "xml",
1383 Format::Html => "html",
1384 _ => "unknown",
1385 }
1386 }
1387
1388 /// Whether this document's format offers *any* door in — `false` only for a
1389 /// wholly parse-only format (XML, AsciiDoc), where every gesture refuses and
1390 /// a frontend may as well open the file read-only.
1391 ///
1392 /// This is a much weaker claim than the name suggests, and driving per-button
1393 /// state from it is exactly the mistake to avoid: HTML answers `true` because
1394 /// it spells the inline marks with a tag pair (`<strong>`, `<em>`, `<code>`)
1395 /// while a heading, a quote, a list, a task box, a link and a code fence all
1396 /// remain unspellable there. Ask [`capabilities`](Self::capabilities) — or
1397 /// [`supports`](Self::supports) — per control.
1398 pub fn authorable(&self) -> bool {
1399 self.format.is_authorable()
1400 }
1401
1402 /// Whether this document can spell `gesture`, which is twig's own answer
1403 /// rather than a copy of it: `Format::supports_with` reads the same
1404 /// `Syntax` table the `Editor` method consults before refusing, chosen by
1405 /// the very [`parse_extensions`] this document's editor reparses with — so
1406 /// what the toolbar offers and what the splice will accept are one table.
1407 ///
1408 /// It is a fact about the *document*, not about the caret. `true` does not
1409 /// promise the gesture succeeds where it is standing — a link over a table
1410 /// border still fails — only that it will not fail with
1411 /// `UnsupportedFormat`. Gray out on `false`; don't read `true` as "this
1412 /// will work here".
1413 pub fn supports(&self, gesture: Gesture) -> bool {
1414 self.format.supports_with(parse_extensions(), gesture)
1415 }
1416
1417 /// Every control's enabled state in one read — what a toolbar builds itself
1418 /// from when a document opens or its format changes. See [`Capabilities`].
1419 pub fn capabilities(&self) -> Capabilities {
1420 Capabilities::of(self.format)
1421 }
1422
1423 /// Refuse a gesture this document's format cannot spell, saying so in the
1424 /// status line. `true` means the caller must return without calling twig.
1425 ///
1426 /// Most of these refusals duplicate one twig would make anyway, and they are
1427 /// made here regardless because a message naming the *document's* format
1428 /// reads better than one naming twig's internals. Two of them are not
1429 /// duplicates and are the reason this is a guard rather than an error
1430 /// translation:
1431 ///
1432 /// - The table family (see [`table_op`](Self::table_op)) consults no
1433 /// `Syntax` table, so twig does not refuse it at all.
1434 /// - [`toggle`](Self::toggle) at a collapsed caret never reaches twig — it
1435 /// arms a sticky mark for text not yet typed, which is a promise `insert`
1436 /// could not keep.
1437 fn refuse_unsupported(&mut self, what: &str, gesture: Gesture) -> bool {
1438 self.refuse_unless(what, self.supports(gesture))
1439 }
1440
1441 /// [`refuse_unsupported`](Self::refuse_unsupported) against a capability leaf
1442 /// answers itself — today only [`spells_pipe_tables`].
1443 fn refuse_unless(&mut self, what: &str, supported: bool) -> bool {
1444 if supported {
1445 return false;
1446 }
1447 self.status = Some(format!("{what}: not supported in {}", self.format_name()));
1448 true
1449 }
1450
1451 /// The name to show for this document. An untitled one has no file to name
1452 /// it, and both frontends put this straight on screen — an empty path
1453 /// renders as an empty header, so it says so instead.
1454 pub fn file_name(&self) -> String {
1455 if self.is_untitled() {
1456 return "untitled".into();
1457 }
1458 self.path
1459 .file_name()
1460 .map(|s| s.to_string_lossy().into_owned())
1461 .unwrap_or_else(|| self.path.display().to_string())
1462 }
1463
1464 /// The selection as an ordered `[start, end)` byte range, or `None` when the
1465 /// caret and anchor coincide (an empty selection is no selection).
1466 pub fn selection(&self) -> Option<(usize, usize)> {
1467 self.anchor
1468 .map(|a| (a.min(self.caret), a.max(self.caret)))
1469 .filter(|(s, e)| s != e)
1470 }
1471
1472 /// The selected text, or `None` when there's no selection — the source
1473 /// slice a copy/cut hands to the system clipboard.
1474 pub fn selected_text(&self) -> Option<&str> {
1475 self.selection().map(|(s, e)| &self.source[s..e])
1476 }
1477
1478 /// The selection as a quote with a little of what surrounds it — the shape
1479 /// a host that cites, annotates, or searches for a passage wants, cut from
1480 /// the **source** rather than from anything rendered, so the quote is
1481 /// findable in the document again by plain string search.
1482 ///
1483 /// `context` is a count of characters (not bytes) on each side, clipped at
1484 /// the document's edges; the slices land on char boundaries by
1485 /// construction. `None` when nothing is selected.
1486 pub fn selection_quote(&self, context: usize) -> Option<Quote> {
1487 let (start, end) = self.selection()?;
1488 let mut before = start;
1489 for _ in 0..context {
1490 match self.source[..before].chars().next_back() {
1491 Some(c) => before -= c.len_utf8(),
1492 None => break,
1493 }
1494 }
1495 let mut after = end;
1496 for _ in 0..context {
1497 match self.source[after..].chars().next() {
1498 Some(c) => after += c.len_utf8(),
1499 None => break,
1500 }
1501 }
1502 Some(Quote {
1503 exact: self.source[start..end].to_string(),
1504 prefix: self.source[before..start].to_string(),
1505 suffix: self.source[end..after].to_string(),
1506 start,
1507 end,
1508 })
1509 }
1510
1511 /// Whether the document refuses to change — see the field.
1512 pub fn read_only(&self) -> bool {
1513 self.read_only
1514 }
1515
1516 /// Turn the read-only gate on or off. A frontend preference like
1517 /// [`set_markup_mode`](Self::set_markup_mode): nothing about the document
1518 /// itself changes, only what may be done to it from here on.
1519 pub fn set_read_only(&mut self, on: bool) {
1520 self.read_only = on;
1521 }
1522
1523 /// The host-painted ranges, sorted by start — see [`Highlight`].
1524 pub fn highlights(&self) -> &[Highlight] {
1525 &self.highlights
1526 }
1527
1528 /// Replace the host-painted ranges wholesale. The whole set each time,
1529 /// rather than add/remove verbs: the host owns the list (it derives it
1530 /// from its own state — annotations, search hits), and a replace can
1531 /// never leave the two disagreeing about what should be on screen.
1532 pub fn set_highlights(&mut self, mut highlights: Vec<Highlight>) {
1533 highlights.retain(|h| h.start < h.end);
1534 highlights.sort_by_key(|h| (h.start, h.end));
1535 self.highlights = highlights;
1536 }
1537
1538 /// The highlight covering source `offset`, if one does — first by start
1539 /// when several overlap, which makes overlapping washes resolvable rather
1540 /// than undefined. What a frontend asks when the reader activates a spot.
1541 ///
1542 /// [`Highlight::covering`] is the whole of it: the frontends paint by
1543 /// asking the same question per glyph, against a slice they were handed
1544 /// rather than against a `Doc`, and one answer for both is what keeps a
1545 /// wash and an activation agreeing about which range a spot is in.
1546 pub fn highlight_at(&self, offset: usize) -> Option<&Highlight> {
1547 Highlight::covering(&self.highlights, offset)
1548 }
1549
1550 /// The AST breadcrumb at the caret (root → deepest), e.g.
1551 /// `doc › para › strong`. Read live from twig via `ancestors_at`.
1552 pub fn breadcrumb(&mut self) -> String {
1553 match self.editor.ancestors_at(self.caret) {
1554 Ok(chain) => chain
1555 .iter()
1556 .map(|m| m.kind.as_str())
1557 .collect::<Vec<_>>()
1558 .join(" › "),
1559 Err(_) => String::new(),
1560 }
1561 }
1562
1563 // ── editing ──────────────────────────────────────────────────────────────
1564
1565 /// Replace the byte range `[start, end)` with `text`, re-anchoring the caret
1566 /// after it. The public form of the internal splice — a pixel frontend that
1567 /// hit-tests to a byte offset (or an IME that hands back an explicit range)
1568 /// edits through this, the same twig `edit_range` the caret ops use.
1569 pub fn edit(&mut self, start: usize, end: usize, text: &str) {
1570 self.splice(start, end, text, EditKind::Other);
1571 }
1572
1573 /// Insert typed `text` at the caret, replacing the selection if there is one.
1574 /// A single typed character coalesces with the run of typing before it; a
1575 /// newline or a multi-character insert is its own undo step.
1576 ///
1577 /// Typed input only — clipboard text goes through [`paste`](Self::paste).
1578 pub fn insert(&mut self, text: &str) {
1579 // The read-only gate, up front: the paths below reach twig by several
1580 // verbs, not all of them through the splice — see the field.
1581 if self.read_only {
1582 return;
1583 }
1584 // Typing against a block picture would dissolve it — see
1585 // `open_paragraph_at_block_media`. Give the text a paragraph first, so
1586 // what the caret was standing beside stays a picture.
1587 self.open_paragraph_at_block_media(text);
1588 // Armed sticky marks (⌘b with no selection) turn the next typed text
1589 // bold/italic/… and then retire — see `insert_with_marks`. Whitespace is
1590 // the exception: it takes no mark of its own and keeps the delta armed
1591 // for the character behind it — see `insert_space_with_marks`.
1592 let pending = self.pending_here();
1593 if !pending.is_empty() && self.selection().is_none() && !text.is_empty() {
1594 if text.trim().is_empty() {
1595 self.insert_space_with_marks(self.caret, text, pending);
1596 } else {
1597 self.insert_with_marks(self.caret, text, pending);
1598 }
1599 return;
1600 }
1601 // `MarkupMode::None`: typed syntax stays literal — twig escapes
1602 // anything that would open markup, so a Diaryx user never mints
1603 // formatting by keyboard (it comes from commands instead). The other two
1604 // rungs of the ladder author markup from what you type, which is the
1605 // whole difference between them and this one. Only in the rendered view
1606 // (source view is for typing raw markup) and only where the format has a
1607 // literal spelling at all: escaping is a backslash before a byte from the
1608 // format's own alphabet, and a format with no such alphabet (HTML escapes
1609 // with entities, XML spells nothing) would have `\&` written into it,
1610 // which is two literal characters and not an escape. Marks (⌘b) still
1611 // format — that path returned above; and leaf's own structural inserts go
1612 // through `insert_raw`, never here, so a list marker or quote gutter is
1613 // written as the markup it is.
1614 if !self.markup_mode.authors()
1615 && self.view == View::Wysiwyg
1616 && !text.is_empty()
1617 && self.supports(Gesture::InsertLiteral)
1618 {
1619 self.insert_literal_typed(text);
1620 return;
1621 }
1622 self.insert_raw(text);
1623 }
1624
1625 /// Insert `text` verbatim at the caret (replacing any selection) — the plain
1626 /// path with no Hidden-mode literal escaping. leaf's own structural inserts
1627 /// (a list marker, a quote gutter, an in-cell `<br>`) call this: they ARE
1628 /// markup by design and must not be escaped.
1629 fn insert_raw(&mut self, text: &str) {
1630 let (s, e) = self.selection().unwrap_or((self.caret, self.caret));
1631 self.splice(s, e, text, typed_edit_kind(text));
1632 }
1633
1634 /// Open a paragraph for text about to be inserted at one of a block media's
1635 /// two caret stops, and leave the caret standing in it.
1636 ///
1637 /// A block image is a paragraph whose entire content is the picture, and the
1638 /// caret's only homes on it are in front of it and just past it (see
1639 /// [`VisualMap::block_media_stop`]). Text inserted at either offset joins
1640 /// *that* paragraph — and a paragraph holding anything besides the image is
1641 /// no longer a block image but a line of text with an inline one in it. The
1642 /// frontend that was painting a photo there paints a text run instead; the
1643 /// picture is still in the file, and nothing said a word. Those two offsets
1644 /// are also exactly where a click on the picture lands, so the whole accident
1645 /// is one tap and one keystroke.
1646 ///
1647 /// So the break goes in first and the text lands in the new empty paragraph —
1648 /// what pressing Return before typing would have done, which is a habit no
1649 /// one should have to learn from losing a photo. A no-op everywhere else, and
1650 /// over a selection (which is replaced, not joined into).
1651 ///
1652 /// A picture inside a quote or a list leaves its container, because `\n\n`
1653 /// ends the block. The alternative is worse: the `\n> ` / next-item
1654 /// continuation [`newline`](Self::newline) writes stays in the same
1655 /// *paragraph*, which is the thing being prevented.
1656 ///
1657 /// Only in the rendered view. Source view is for typing raw markup, where
1658 /// putting a character against an image is exactly what it looks like.
1659 fn open_paragraph_at_block_media(&mut self, text: &str) {
1660 if self.view != View::Wysiwyg || text.is_empty() || text == "\n" {
1661 return;
1662 }
1663 if self.selection().is_some() {
1664 return;
1665 }
1666 // The map may be a revision behind (nothing has drawn since the last
1667 // edit), and this asks it about offsets — a stale answer would splice a
1668 // break into the wrong place. Free when it is already current, which it
1669 // is whenever a frontend drew a frame between keystrokes.
1670 self.rebuild_map();
1671 let at = self.caret;
1672 let Some((side, _)) = self.vmap.block_media_stop(at) else {
1673 return;
1674 };
1675 if !self.splice(at, at, "\n\n", EditKind::Other) {
1676 return;
1677 }
1678 // The break is part of the keystroke, not an edit of its own: leave the
1679 // run marked as typing so the character about to arrive folds into it and
1680 // one undo puts the document back the way it was found. (A paste, or a
1681 // multi-character insert, is `EditKind::Other` and stays its own step —
1682 // as it would have been anywhere else in the document.)
1683 self.last_edit_kind = Some(EditKind::Insert);
1684 if side == MediaStop::Before {
1685 // The break went in above the picture and the caret rode to the end
1686 // of it — which is still hard against the picture. Step back onto the
1687 // blank line it opened, so the text lands above rather than in front.
1688 self.caret = at;
1689 }
1690 }
1691
1692 /// A delete key pressed at one of a block picture's two caret stops, handled
1693 /// as the picture being an *atom* rather than a run of bytes. Returns whether
1694 /// the key was consumed.
1695 ///
1696 /// The caret rests in front of a block image and just past it, never inside
1697 /// its markup — which the rendered view doesn't show. So the byte a delete
1698 /// key nominally takes there is one the writer cannot see, and taking it
1699 /// leaves the picture as broken markup rather than as anything anyone asked
1700 /// for: Backspace at the stop past `` removes the closing paren, and
1701 /// a photo becomes the literal text `
1704 /// prevents from the typing side, and it cost this repository's own test vault
1705 /// a photo before it was found.
1706 ///
1707 /// So the key aimed *at* the picture deletes the picture, whole — Backspace
1708 /// when it is behind the caret, Delete when it is in front — which is what
1709 /// every editor does with an embed, and one undo away. The key aimed *away*
1710 /// from it would otherwise delete the paragraph break and merge a neighbour
1711 /// into the picture's own paragraph, which dissolves it just as surely; it
1712 /// steps the caret over the boundary instead and leaves the
1713 /// next press to delete in the block it has reached — the same "first press
1714 /// steps out of the atom, second press deletes" every delete key here gets,
1715 /// word-deletes included (⌥⌫ in front of a picture is aimed at the prose
1716 /// above, and reaches it on the second press rather than taking the break and
1717 /// the picture with it on the first).
1718 fn delete_around_block_media(&mut self, forward: bool) -> bool {
1719 // The map answers about offsets, so it has to be this revision's — see
1720 // the same call in `open_paragraph_at_block_media`.
1721 self.rebuild_map();
1722 let Some((side, span)) = self.vmap.block_media_stop(self.caret) else {
1723 return false;
1724 };
1725 let aimed_at_it = side
1726 == if forward {
1727 MediaStop::Before
1728 } else {
1729 MediaStop::After
1730 };
1731 if !aimed_at_it {
1732 let over = if forward {
1733 self.vmap.stop_after(self.caret)
1734 } else {
1735 self.vmap.stop_before(self.caret)
1736 };
1737 if let Some(off) = over.filter(|&o| o >= self.caret_floor()) {
1738 self.caret = off;
1739 self.anchor = None;
1740 self.goal_col = None;
1741 }
1742 return true;
1743 }
1744 // Take the break that held the picture apart from its neighbour with it,
1745 // so the delete doesn't leave a blank paragraph standing where the
1746 // picture was. The last arm is a picture that is the whole document.
1747 let (from, to) = if self.source[..span.start].ends_with("\n\n") {
1748 (span.start - 2, span.end)
1749 } else if self.source[span.end..].starts_with("\n\n") {
1750 (span.start, span.end + 2)
1751 } else {
1752 (span.start, span.end)
1753 };
1754 self.splice(from.max(self.caret_floor()), to, "", EditKind::Other);
1755 true
1756 }
1757
1758 /// The Hidden-mode typing path: replace any selection, then insert `text`
1759 /// escaped so it stays literal. When it replaces a selection the two edits
1760 /// fold into one undo step, so an overwrite undoes atomically (and restores
1761 /// the selection) exactly as a plain one does.
1762 fn insert_literal_typed(&mut self, text: &str) {
1763 let kind = typed_edit_kind(text);
1764 match self.selection() {
1765 Some((s, e)) => {
1766 if !self.splice(s, e, "", EditKind::Other) {
1767 return;
1768 }
1769 // Typing over a whole marked run takes its delimiters with it
1770 // (the empty content couldn't hold them — see
1771 // `repair_mark_edges`) and leaves its marks armed at the caret.
1772 // The text taking the run's place inherits them, exactly as it
1773 // would have by landing inside a run that survived.
1774 let pending = self.pending_here();
1775 if !pending.is_empty() && !text.trim().is_empty() {
1776 self.insert_with_marks(self.caret, text, pending);
1777 return;
1778 }
1779 self.insert_literal_at(self.caret, text, kind, true);
1780 }
1781 None => {
1782 self.insert_literal_at(self.caret, text, kind, false);
1783 }
1784 }
1785 }
1786
1787 /// The sticky-mark delta that is live right now: the marks armed by [`toggle`]
1788 /// at a collapsed caret, but only while the caret still stands where they
1789 /// were armed and nothing is selected. Empty otherwise, so a stale delta
1790 /// never styles text it wasn't meant for.
1791 fn pending_here(&self) -> InlineMarks {
1792 if self.anchor.is_none() && self.pending_at == Some(self.caret) {
1793 self.pending_marks
1794 } else {
1795 InlineMarks::empty()
1796 }
1797 }
1798
1799 /// Drop the armed sticky marks — any caret motion, selection, or edit does
1800 /// this, so "start bold here" only ever applies at the exact spot it was
1801 /// asked for.
1802 fn clear_pending(&mut self) {
1803 self.pending_marks = InlineMarks::empty();
1804 self.pending_at = None;
1805 }
1806
1807 /// Insert `text` at `at` carrying the armed sticky `marks`: a mark not yet in
1808 /// force is wrapped around the freshly typed text; a mark the caret already
1809 /// stands inside is *shed* — the text is inserted past the run's end so it
1810 /// lands unmarked ("type normally again"). The caret comes to rest inside any
1811 /// added runs, so continued typing inherits the marks with no re-wrapping,
1812 /// and the delta is cleared: the marks now live in the document, not here.
1813 fn insert_with_marks(&mut self, at: usize, text: &str, marks: InlineMarks) {
1814 let base = self.mark_spans_at(at);
1815 let base_set: InlineMarks = base.iter().map(|(k, _)| *k).collect();
1816 // Nothing to shed, and a run of exactly these marks standing just behind
1817 // the caret: carry on writing *that* run rather than opening a second
1818 // one beside it.
1819 if base_set.is_empty() && self.rejoin_run(at, text, marks) {
1820 return;
1821 }
1822 // Shed the marks we're turning off: step the insertion point past the
1823 // end of each run the caret sits in, so the new text falls outside it.
1824 let mut ins_at = at;
1825 for (kind, span) in &base {
1826 if marks.contains(*kind) {
1827 ins_at = ins_at.max(span.end);
1828 }
1829 }
1830 if !self.splice_exact(ins_at, ins_at, text, EditKind::Other) {
1831 return;
1832 }
1833 // The plain splice inserted exactly `text` at `ins_at`; that byte range
1834 // is the content every added mark wraps.
1835 let (mut cs, mut ce) = (ins_at, ins_at + text.len());
1836 for kind in marks.iter() {
1837 if !base_set.contains(kind) {
1838 let (ncs, nce) = self.wrap_span(cs, ce, kind);
1839 cs = ncs;
1840 ce = nce;
1841 }
1842 }
1843 self.caret = ce.min(self.source.len());
1844 self.anchor = None;
1845 self.last_edit_kind = None;
1846 // Realised: the marks are in the document now, and the caret sits inside
1847 // them, so there is no delta left to carry. Arm nothing, but remember the
1848 // spot so a *further* toggle before typing starts a clean delta here.
1849 self.pending_marks = InlineMarks::empty();
1850 self.pending_at = Some(self.caret);
1851 self.clamp_caret();
1852 self.record_caret();
1853 }
1854
1855 /// Carry on the marked run just behind `at` — moving its closing delimiters
1856 /// out past the new text — instead of opening a second run of the same marks
1857 /// beside it. Returns whether it did.
1858 ///
1859 /// This is the far half of the mark-edge rule (see [`splice`](Self::splice)).
1860 /// A space typed after a bold word steps the caret out of the run, because
1861 /// `**bold **` is not bold; the next character has to step back *in*, or the
1862 /// writer who typed one bold phrase is left with `**bold** **and**` — two
1863 /// runs that read the same to a reader but spell the file in a way nobody
1864 /// wrote. Only whitespace may stand in the gap (a run doesn't reach across
1865 /// words it isn't marking), and the marks behind it must be exactly the ones
1866 /// armed — a run of *some* other kind is a neighbour, not this phrase.
1867 fn rejoin_run(&mut self, at: usize, text: &str, marks: InlineMarks) -> bool {
1868 if text.is_empty() || text.trim() != text {
1869 return false;
1870 }
1871 let gap_at = self.source[..at].trim_end_matches([' ', '\t']).len();
1872 // Walk in through the delimiters stacked at that point, innermost last:
1873 // `***both*** ` closes two runs with one `***`, and rejoining means
1874 // getting behind all of them.
1875 let (mut cut, mut kinds) = (gap_at, InlineMarks::empty());
1876 while let Some((kind, content_end)) = self
1877 .editor
1878 .ancestors_at(prev_boundary(&self.source, cut))
1879 .unwrap_or_default()
1880 .into_iter()
1881 .filter(|m| m.span.end == cut)
1882 .find_map(|m| Some((inline_kind(&m.kind)?, m.content_span.clone()?.end)))
1883 {
1884 if content_end >= cut {
1885 break; // a mark with no closing delimiter to step behind
1886 }
1887 kinds.insert(kind);
1888 cut = content_end;
1889 }
1890 if cut == gap_at || kinds != marks {
1891 return false;
1892 }
1893 // Re-spell the tail: the gap, then the new text, then the delimiters that
1894 // used to close in front of them — read out of the document rather than
1895 // written from a table, so whatever twig spells them with is what moves.
1896 let tail = format!(
1897 "{}{text}{}",
1898 &self.source[gap_at..at],
1899 &self.source[cut..gap_at]
1900 );
1901 if !self.splice_exact(cut, at, &tail, EditKind::Other) {
1902 return false;
1903 }
1904 self.caret = (cut + (at - gap_at) + text.len()).min(self.source.len());
1905 self.anchor = None;
1906 self.last_edit_kind = None;
1907 self.pending_marks = InlineMarks::empty();
1908 self.pending_at = Some(self.caret);
1909 self.clamp_caret();
1910 self.record_caret();
1911 true
1912 }
1913
1914 /// Insert typed whitespace at a caret with sticky marks armed. Whitespace is
1915 /// never itself wrapped: a mark around a space draws nothing a reader can
1916 /// see, and in Markdown and Djot it draws its own delimiters instead
1917 /// (`** **`). So the space goes in unmarked — outside any run the armed
1918 /// marks are shedding — and the marks stay armed for the character after it,
1919 /// which rejoins the run (see [`rejoin_run`](Self::rejoin_run)).
1920 fn insert_space_with_marks(&mut self, at: usize, text: &str, marks: InlineMarks) {
1921 let base = self.mark_spans_at(at);
1922 // What the *next* character carries: the armed delta resolved against the
1923 // marks in force here, which the space must not quietly drop.
1924 let want = base
1925 .iter()
1926 .map(|(k, _)| *k)
1927 .collect::<InlineMarks>()
1928 .xor(marks);
1929 let mut ins_at = at;
1930 for (kind, span) in &base {
1931 if marks.contains(*kind) {
1932 ins_at = ins_at.max(span.end);
1933 }
1934 }
1935 if !self.splice(ins_at, ins_at, text, typed_edit_kind(text)) {
1936 return;
1937 }
1938 self.rearm(want);
1939 self.record_caret();
1940 }
1941
1942 /// Wrap `[s, e)` in `kind` via twig and return the byte span the *content*
1943 /// (not the delimiters) occupies afterwards. Markdown/Djot inline delimiters
1944 /// are symmetric (`**`…`**`, `_`…`_`, `` ` ``…`` ` ``), so the bytes twig
1945 /// added split evenly around the content — half the growth on each side.
1946 fn wrap_span(&mut self, s: usize, e: usize, kind: InlineKind) -> (usize, usize) {
1947 // The read-only gate — this door reaches twig without the splice.
1948 if self.read_only {
1949 return (s, e);
1950 }
1951 match self.editor.toggle_inline(s, e, kind) {
1952 Ok(change) => {
1953 self.last_edit_kind = None;
1954 self.refresh();
1955 self.dirty = self.source != self.clean_source;
1956 let added = (change.new.end - change.new.start).saturating_sub(e - s);
1957 let half = added / 2;
1958 (change.new.start + half, change.new.end - half)
1959 }
1960 // Unsupported here (e.g. mark on Markdown): leave the text unwrapped
1961 // rather than lose the keystroke.
1962 Err(e2) => {
1963 self.status = Some(format!("{kind:?}: {e2}"));
1964 (s, e)
1965 }
1966 }
1967 }
1968
1969 /// The safe offset to splice a block-level break at, given a caret that may
1970 /// sit exactly between an inline mark's content and its own closing
1971 /// delimiter (`content_span.end == off < span.end` for some enclosing mark
1972 /// — the WYSIWYG caret's natural resting place at the end of `**bold**`
1973 /// with nothing following it on the line: the closing `**` renders no
1974 /// glyph of its own, so the caret's "end of line" offset lands right
1975 /// before it). Splicing a paragraph/list/quote break at `off` itself would
1976 /// sever the delimiter from its content, stranding it alone on the new
1977 /// line. Walks out to the *outermost* such mark's `span.end` instead, so
1978 /// nested marks closing at the same point (`**_x_**`) all clear together.
1979 /// A no-op everywhere else — mid-run, or past real trailing content, no
1980 /// mark's `content_span` ends exactly at `off`.
1981 fn skip_trailing_close_delims(&mut self, off: usize) -> usize {
1982 let off = off.min(self.source.len());
1983 self.editor
1984 .ancestors_at(off)
1985 .unwrap_or_default()
1986 .into_iter()
1987 .filter(|m| inline_kind(&m.kind).is_some())
1988 .filter(|m| off < m.span.end && m.content_span.as_ref().is_some_and(|c| c.end == off))
1989 .map(|m| m.span.end)
1990 .max()
1991 .unwrap_or(off)
1992 }
1993
1994 /// The offset a *delete* aimed at the character before `off` should stop at,
1995 /// when `off` is the start of a run's text and the bytes behind it are that
1996 /// run's opening delimiter. The rich view draws no glyph for a `**`, so the
1997 /// byte behind the caret at the start of a bold word is not a character the
1998 /// writer can see, let alone one they aimed Backspace at: taking it leaves
1999 /// `a *bold** c` — the styling gone and a literal asterisk in its place. The
2000 /// delete steps over the whole delimiter to the visible character in front of
2001 /// it instead. Walks out to the *outermost* mark opening there, so
2002 /// `**_x_**` clears every delimiter at once, and is a no-op anywhere else.
2003 fn skip_leading_open_delims(&mut self, off: usize) -> usize {
2004 let off = off.min(self.source.len());
2005 self.editor
2006 .ancestors_at(off)
2007 .unwrap_or_default()
2008 .into_iter()
2009 .filter(|m| inline_kind(&m.kind).is_some())
2010 .filter(|m| {
2011 m.span.start < off && m.content_span.as_ref().is_some_and(|c| c.start == off)
2012 })
2013 .map(|m| m.span.start)
2014 .min()
2015 .unwrap_or(off)
2016 }
2017
2018 /// `off` moved *inside* the run whose closing delimiters end there — the
2019 /// other offset the rich view draws in the same place, since a `**` renders
2020 /// no glyph of its own. `**bold**` has a caret home on each side of its
2021 /// closing delimiter, one column apart on screen and eight bytes and a whole
2022 /// run apart in the file, and a plain ← lands on the outer one whenever a
2023 /// space follows the phrase. The inner one is what the writer is pointing at
2024 /// there: the end of their bold word. Walks in through every mark closing at
2025 /// that point, innermost last, so `***both***` lands inside both. A no-op
2026 /// anywhere else — mid-run, or in prose, no mark's span ends at `off`.
2027 fn step_inside_close_delims(&mut self, off: usize) -> usize {
2028 let mut off = off.min(self.source.len());
2029 loop {
2030 let inner = self
2031 .editor
2032 .ancestors_at(prev_boundary(&self.source, off))
2033 .unwrap_or_default()
2034 .into_iter()
2035 .filter(|m| inline_kind(&m.kind).is_some() && m.span.end == off)
2036 .filter_map(|m| m.content_span.clone().map(|c| c.end))
2037 .filter(|&end| end < off)
2038 .max();
2039 match inner {
2040 Some(end) => off = end,
2041 None => return off,
2042 }
2043 }
2044 }
2045
2046 /// The mirror at the opening edge: `off` moved inside the run whose
2047 /// delimiters *start* there, onto the first character of its text. See
2048 /// [`step_inside_close_delims`](Self::step_inside_close_delims).
2049 fn step_inside_open_delims(&mut self, off: usize) -> usize {
2050 let mut off = off.min(self.source.len());
2051 loop {
2052 let inner = self
2053 .editor
2054 .ancestors_at(off)
2055 .unwrap_or_default()
2056 .into_iter()
2057 .filter(|m| inline_kind(&m.kind).is_some() && m.span.start == off)
2058 .filter_map(|m| m.content_span.clone().map(|c| c.start))
2059 .filter(|&start| start > off)
2060 .min();
2061 match inner {
2062 Some(start) => off = start,
2063 None => return off,
2064 }
2065 }
2066 }
2067
2068 /// The inline mark kinds whose span covers `off`, each with that span — the
2069 /// span-carrying sibling of [`marks_at`](Self::marks_at), which reports node
2070 /// ids instead. Used to shed a mark by stepping past the end of its run.
2071 fn mark_spans_at(&mut self, off: usize) -> Vec<(InlineKind, std::ops::Range<usize>)> {
2072 let off = off.min(self.source.len());
2073 self.editor
2074 .ancestors_at(off)
2075 .unwrap_or_default()
2076 .into_iter()
2077 .filter(|m| off < m.span.end)
2078 .filter_map(|m| inline_kind(&m.kind).map(|k| (k, m.span.clone())))
2079 .collect()
2080 }
2081
2082 /// Insert clipboard `text` at the caret, replacing the selection if there is
2083 /// one — always its own undo step, whatever its length.
2084 ///
2085 /// Provenance is the whole point, and only the caller has it. `insert` reads
2086 /// a lone character as a keystroke and folds it into the run around it,
2087 /// which is right for typing and wrong for a one-character paste: that paste
2088 /// would vanish mid-run on an undo it was never part of, and the characters
2089 /// the user actually typed would go with it. Length can't tell the two
2090 /// apart — `⌘V` of `x` and typing `x` are the same string — so the door the
2091 /// caller comes through is what says which happened.
2092 pub fn paste(&mut self, text: &str) {
2093 // Pasting against a block picture dissolves it exactly as typing does,
2094 // and for the same reason — see `open_paragraph_at_block_media`.
2095 self.open_paragraph_at_block_media(text);
2096 let (s, e) = self.selection().unwrap_or((self.caret, self.caret));
2097 self.splice(s, e, text, EditKind::Other);
2098 }
2099
2100 /// Replace `[start, end)` with `text` as one step of an IME composition —
2101 /// the same splice as [`edit`](Self::edit), but marked so the run of steps
2102 /// folds into a single undo.
2103 ///
2104 /// A composition is *one* act of writing. Typing `かんじ` and picking 感じ is a
2105 /// dozen calls here, each replacing the last one's provisional bytes, and an
2106 /// undo step per call means undoing a word means pressing ⌘Z until the reading
2107 /// unspools backwards through kana — the intermediate states were never text
2108 /// the user wrote. Only the frontend knows a call is provisional (the bytes
2109 /// look like any other edit), so the door the caller comes through is what
2110 /// says so, exactly as it is for [`paste`](Self::paste) versus
2111 /// [`insert`](Self::insert).
2112 ///
2113 /// Pair with [`end_composition`](Self::end_composition), or the *next*
2114 /// composition folds into this one.
2115 pub fn edit_composing(&mut self, start: usize, end: usize, text: &str) {
2116 self.splice(start, end, text, EditKind::Compose);
2117 }
2118
2119 /// Close the open composition run, so the next one is its own undo step.
2120 /// Call when the IME commits or withdraws a composition.
2121 ///
2122 /// Only clears a *composition* run: a frontend that reports an end it never
2123 /// began (some IMEs unmark unprompted) would otherwise split the run of
2124 /// typing around it into two undo steps for no reason the user can see.
2125 pub fn end_composition(&mut self) {
2126 if self.last_edit_kind == Some(EditKind::Compose) {
2127 self.last_edit_kind = None;
2128 }
2129 }
2130
2131 // ── the clipboard's rich flavor ──────────────────────────────────────────
2132
2133 /// The selection rendered as HTML, for the clipboard's `text/html` flavor —
2134 /// what lets a paste into Docs/Mail/Slack keep its formatting. `None` when
2135 /// nothing is selected, or when the selection doesn't render (the caller
2136 /// still has [`selected_text`](Self::selected_text), which is what to publish
2137 /// as `text/plain` either way).
2138 ///
2139 /// **The fragment is a source substring, and that is the honest limit here.**
2140 /// It's parsed standalone, so a selection whose meaning depends on its
2141 /// surroundings converts as what it literally says rather than what it looks
2142 /// like on screen: half a list item is a paragraph, a row torn out of a table
2143 /// is the text of a row, the `**` of a bold run selected without its closing
2144 /// `**` is two asterisks. Every one of those still *renders* — there's no
2145 /// error to report — it just renders as the fragment and not as the document.
2146 /// Widening the range to whole blocks would publish text the user didn't
2147 /// select, which is a worse lie than a fragment being a fragment; the plain
2148 /// flavor has the same substring, so the two flavors at least agree.
2149 pub fn selection_html(&mut self) -> Option<String> {
2150 let (start, end) = self.selection()?;
2151 let inline = self.selection_is_inline(start, end);
2152 let html = html::render_fragment(&self.source[start..end], self.format)?;
2153 Some(match inline {
2154 true => html::strip_sole_paragraph(html),
2155 false => html,
2156 })
2157 }
2158
2159 /// Paste the clipboard's `text/html` flavor, converting it to this document's
2160 /// format first. Its own undo step, like any [`paste`](Self::paste).
2161 ///
2162 /// Returns whether it landed. `false` means the HTML didn't convert to
2163 /// anything worth pasting — the caller should fall back to the plain flavor
2164 /// rather than treat it as an error. The `html` module has the full list of
2165 /// what that covers: a table twig won't build, markup it doesn't recognise,
2166 /// an empty result.
2167 pub fn paste_html(&mut self, html: &str) -> bool {
2168 match html::parse_fragment(html, self.format) {
2169 Some(source) => {
2170 self.paste(&source);
2171 true
2172 }
2173 None => false,
2174 }
2175 }
2176
2177 /// Does the selection live *inside* a single top-level block?
2178 ///
2179 /// The question [`selection_html`](Self::selection_html) needs and the
2180 /// fragment can't answer: `**bold**` renders as `<p><strong>bold</strong></p>`
2181 /// whether the user selected one word of a sentence or a whole paragraph, and
2182 /// only the document knows which. Selecting a word and pasting into Docs
2183 /// should extend the line you paste into; selecting the paragraph should make
2184 /// a paragraph. So a selection strictly within one block is inline (its `<p>`
2185 /// is an artifact of standalone parsing), and one that covers a whole block —
2186 /// or spans two — keeps its structure.
2187 ///
2188 /// Reads the block from twig rather than guessing from the bytes:
2189 /// `ancestors_at` is `[doc, block, …inline]`, so index 1 is the top-level
2190 /// block containing an offset, and two ends inside the same one cannot have
2191 /// crossed a block boundary.
2192 fn selection_is_inline(&mut self, start: usize, end: usize) -> bool {
2193 // The last *character*, not `end - 1`: the selection's end is exclusive
2194 // and may sit mid-codepoint's-worth of bytes past the last char.
2195 let Some((off, _)) = self.source[start..end].char_indices().next_back() else {
2196 return false;
2197 };
2198 let (Some(head), Some(tail)) =
2199 (self.top_block_span(start), self.top_block_span(start + off))
2200 else {
2201 return false;
2202 };
2203 head == tail && !(start <= head.start && end >= head.end)
2204 }
2205
2206 /// The byte span of the top-level block containing `offset`, or `None` at an
2207 /// offset that belongs to no block (the blank line between two of them).
2208 fn top_block_span(&mut self, offset: usize) -> Option<std::ops::Range<usize>> {
2209 self.editor
2210 .ancestors_at(offset)
2211 .ok()?
2212 .get(1)
2213 .map(|m| m.span.clone())
2214 }
2215
2216 // ── indentation ──────────────────────────────────────────────────────────
2217
2218 /// One indent level.
2219 ///
2220 /// Two spaces, not the four both frontends type for Tab today, because in a
2221 /// markdown document four columns isn't a width — it's a *meaning*. Four
2222 /// spaces at the head of a line is markdown's indented-code-block marker, so
2223 /// one Tab on a paragraph would reparse it into code and style it as such;
2224 /// two cannot, and the line stays the prose it was. Two is also exactly
2225 /// where a `- ` bullet's content starts, so an indented line lands under its
2226 /// parent item's text instead of beside it — the column a list-aware indent
2227 /// has to hit anyway, which keeps this width from being relitigated later.
2228 const INDENT: &'static str = " ";
2229
2230 /// Indent the selected lines — or the caret's line, with no selection — by
2231 /// one level (Tab).
2232 pub fn indent(&mut self) {
2233 self.reindent(true);
2234 // Nesting changes an ordered list's numbering (the nested item restarts,
2235 // its old siblings resume) — keep the source markers in step.
2236 self.renumber_here();
2237 // Nesting an empty `-` item under a text line reparses that text as a
2238 // setext heading; swap the dash for a `*` before it can (a no-op unless
2239 // the collapse actually happened).
2240 self.avoid_setext_collapse();
2241 }
2242
2243 /// Take one indent level back off the selected lines, or the caret's line
2244 /// (Shift+Tab). A line with no indentation is left exactly as it is.
2245 ///
2246 /// A line with *less* than a full level gives back what it has rather than
2247 /// refusing: outdent's job is to walk a line left, and real documents — hand
2248 /// written, or reflowed by some other editor — are full of indentation that
2249 /// was never a clean multiple of anything. Refusing there would strand the
2250 /// line at a depth Shift+Tab couldn't undo.
2251 pub fn outdent(&mut self) {
2252 self.reindent(false);
2253 self.renumber_here();
2254 }
2255
2256 /// The body of [`indent`](Self::indent) / [`outdent`](Self::outdent).
2257 ///
2258 /// One splice across the whole line range, never one per line: a Tab is one
2259 /// thing the user did, so it has to be one undo step and one reparse. Per
2260 /// line, twig would reparse the document once per line and leave a stack of
2261 /// steps that Shift+⌘Z walks back one line at a time.
2262 fn reindent(&mut self, add: bool) {
2263 let (sel_start, sel_end) = self.selection().unwrap_or((self.caret, self.caret));
2264 let start = source_line_range(&self.source, sel_start).start;
2265 let end = source_line_range(&self.source, sel_end).end;
2266 let region = self.source[start..end].to_string();
2267 let lines: Vec<&str> = region.split('\n').collect();
2268 // A blank line has no text to move, and padding it would leave nothing
2269 // but trailing whitespace — but Tab on a blank line *is* a request for
2270 // indentation to type into, so the skip only applies where the op has
2271 // other lines to do real work on.
2272 let skip_blank = add && lines.len() > 1;
2273
2274 let mut out = String::with_capacity(region.len() + lines.len() * Self::INDENT.len());
2275 let mut deltas: Vec<isize> = Vec::with_capacity(lines.len());
2276 let mut line_off = start;
2277 for (i, full) in lines.iter().enumerate() {
2278 if i > 0 {
2279 out.push('\n');
2280 }
2281 // A list item moves by having its whole leading prefix *replaced*,
2282 // never by having spaces pushed in front of the line. twig spells
2283 // both prefixes, so the quote markers, the parent's indent and an
2284 // ordered marker's extra column all come out right without leaf
2285 // measuring any of them — and a line that only looks like an item
2286 // (a Djot continuation) reports no marker and is left to the plain
2287 // path, where a Tab is just a Tab.
2288 let marker = self.list_marker_on_line(line_off);
2289 let own = marker
2290 .as_ref()
2291 .map(|m| m.marker_start - m.line_start)
2292 .unwrap_or(0);
2293 let delta = if add {
2294 if skip_blank && full.trim().is_empty() {
2295 out.push_str(full);
2296 0
2297 } else if marker.is_some() && self.first_item_of_list(line_off) {
2298 // The first item of a list has no preceding sibling to nest
2299 // under, so a Tab here can't spell a sub-list — twig would
2300 // reparse the shoved-over marker as the same list, only
2301 // indented, which Shift+Tab then can't cleanly undo. Leave the
2302 // item where it is, the way every list editor refuses to
2303 // over-indent a list's first line.
2304 out.push_str(full);
2305 0
2306 } else if marker.is_some() {
2307 // Nesting means standing where a *continuation* of this line
2308 // would stand: past the parent's marker, inside its content
2309 // column. That is `continuation_prefix`, less a checkbox.
2310 let new = self.nesting_prefix_at(line_off);
2311 let delta = new.len() as isize - own as isize;
2312 out.push_str(&new);
2313 out.push_str(&full[own..]);
2314 delta
2315 } else {
2316 out.push_str(Self::INDENT);
2317 out.push_str(full);
2318 Self::INDENT.len() as isize
2319 }
2320 } else if marker.is_some() {
2321 // Unnesting is the mirror: stand where the parent item's own
2322 // line starts, which drops exactly the level it contributed.
2323 let new = self.outdent_prefix_at(line_off);
2324 let delta = new.len() as isize - own as isize;
2325 out.push_str(&new);
2326 out.push_str(&full[own..]);
2327 delta
2328 } else {
2329 // A plain line gives back the ordinary step.
2330 let strip = outdent_width(full, Self::INDENT.len());
2331 out.push_str(&full[strip..]);
2332 -(strip as isize)
2333 };
2334 deltas.push(delta);
2335 line_off += full.len() + 1;
2336 }
2337 // Nothing to give back. Returning before the splice keeps an outdent at
2338 // column zero from spending an undo step on a document it never changed.
2339 if deltas.iter().all(|d| *d == 0) {
2340 return;
2341 }
2342
2343 // Every line's text keeps its offset *within the line*, so the caret is
2344 // remapped by its column, not by its byte offset — which the prefixes on
2345 // the lines above it have already invalidated.
2346 let remap = |off: usize| -> usize {
2347 let (mut old_ls, mut new_ls) = (start, start);
2348 for (line, delta) in lines.iter().zip(&deltas) {
2349 let old_le = old_ls + line.len();
2350 let new_len = (line.len() as isize + delta) as usize;
2351 if off <= old_le {
2352 let col = (off - old_ls) as isize;
2353 return new_ls + ((col + delta).max(0) as usize).min(new_len);
2354 }
2355 old_ls = old_le + 1;
2356 new_ls += new_len + 1;
2357 }
2358 start + out.len()
2359 };
2360 let placed = match self.selection() {
2361 // Keep the rewritten region selected, the way a container toggle
2362 // keeps its own: it leaves a second Tab aimed at the same lines
2363 // rather than at whatever the shifted offsets now happen to cover.
2364 Some(_) => (start + out.len(), Some(start)),
2365 None => (remap(self.caret), None),
2366 };
2367
2368 // A rolled-back splice leaves the old source in place, where every offset
2369 // computed above addresses text that was never written.
2370 if !self.splice(start, end, &out, EditKind::Other) {
2371 return;
2372 }
2373 // `splice` re-anchors to the end of the `Change`, which for a whole-region
2374 // rewrite is the last line's end — nowhere the caret was. Place it, then
2375 // re-record the caret so this is the state redo restores, not the one
2376 // `splice` left behind from the `Change`.
2377 self.caret = placed.0.min(self.source.len());
2378 self.anchor = placed.1;
2379 self.clamp_caret();
2380 self.record_caret();
2381 }
2382
2383 /// The Enter key.
2384 ///
2385 /// In source view it's a literal newline. In WYSIWYG it's **AST-aware**: a
2386 /// bare `\n` is only a markdown soft break (same paragraph), so the block the
2387 /// caret is in decides what actually gets written.
2388 ///
2389 /// - paragraph → twig's [`Editor::split_block`], which parts the
2390 /// block at the caret and reopens its container
2391 /// - list item → likewise: the next item, its indent, quote
2392 /// prefix and `[ ]` box all reproduced by twig —
2393 /// except an *empty* item, which exits the list
2394 /// - block quote → likewise: a new paragraph inside the quote
2395 /// - heading → a new *paragraph*, not another heading
2396 /// - code block → a literal newline (stay in the block)
2397 /// - blank line → a literal newline (one Backspace undoes it)
2398 /// - [`LineFlow::Preserve`] → a single soft break, which renders as a
2399 /// visible line
2400 ///
2401 /// Where `split_block` is used it replaces markup leaf used to spell by hand,
2402 /// and it is better at it: it drops the whitespace the caret was sitting in
2403 /// front of instead of stranding it at the head of the second half, and it
2404 /// knows continuations leaf's marker scan never covered — a checklist item
2405 /// continues as an *unchecked* checklist item rather than a plain bullet.
2406 ///
2407 /// The exceptions above are exceptions because `split_block` is either wrong
2408 /// there or refuses: parting a fence yields two fences with the code split
2409 /// between them, parting a heading yields a second heading where every editor
2410 /// gives a paragraph, and a blank line, an empty item, a setext heading and a
2411 /// table all report an error rather than a split.
2412 pub fn newline(&mut self) {
2413 if self.view == View::Source {
2414 self.insert_raw("\n");
2415 return;
2416 }
2417 // Enter over a selection replaces it with a paragraph break.
2418 if let Some((s, e)) = self.selection() {
2419 self.splice(s, e, "\n\n", EditKind::Other);
2420 return;
2421 }
2422 // A caret resting exactly between an inline mark's content and its own
2423 // closing delimiter (`**bold**` with nothing after it on the line —
2424 // the WYSIWYG caret's natural end-of-line position) must not splice a
2425 // block break there: every path below eventually does via
2426 // `insert_raw`/`self.caret`, and splicing before the hidden closing
2427 // delimiter would strand it alone on the new line.
2428 self.caret = self.skip_trailing_close_delims(self.caret);
2429 // The block the caret is in. `block_offset_for_caret` nudges off a line
2430 // end (where the caret sits at the doc level); on a bare line (e.g. an
2431 // empty list item) fall back to the caret so the enclosing list/quote is
2432 // still visible in the ancestors.
2433 let off = self.block_offset_for_caret().unwrap_or(self.caret);
2434 let kinds: Vec<Kind> = self
2435 .editor
2436 .ancestors_at(off)
2437 .map(|c| c.into_iter().map(|m| m.kind).collect())
2438 .unwrap_or_default();
2439 let has = |k: Kind| kinds.contains(&k);
2440
2441 if has(Kind::CodeBlock) {
2442 self.insert_raw("\n");
2443 return;
2444 }
2445 // An *empty* list item exits the list — the standard double-Enter — which
2446 // `split_block` reports as an error rather than a split (there is no
2447 // content to part), so it stays leaf's. `list_marker_on_line` is itself
2448 // the AST gate — it answers from the tree, so a `- ` that reads as a
2449 // marker byte-for-byte but opens no item (a setext underline, a Djot
2450 // continuation line) never reaches here.
2451 if let Some(marker) = self.list_marker_on_line(self.caret)
2452 && self.item_is_empty(&marker)
2453 {
2454 self.exit_list(&marker);
2455 return;
2456 }
2457 // On an *empty* paragraph line, a lone Enter should add a single blank line,
2458 // not another full paragraph break — so it moves down one line and one
2459 // Backspace undoes it, not two. (`split_block` errors here too.)
2460 let line_start = self.source[..self.caret].rfind('\n').map_or(0, |i| i + 1);
2461 let line_end = self.source[self.caret..]
2462 .find('\n')
2463 .map_or(self.source.len(), |i| self.caret + i);
2464 if self.source[line_start..line_end].trim().is_empty() {
2465 self.insert_raw("\n");
2466 return;
2467 }
2468 // In `Preserve` flow a soft break is a *visible* line the author means to
2469 // make, so Enter writes a single `\n` and typing continues the same
2470 // paragraph on the next line — the behaviour of an ordinary text editor.
2471 // A second Enter then lands on the blank line above and takes the
2472 // empty-line branch, so double-Enter still promotes to a full paragraph
2473 // break; and Backspace, which deletes a lone `\n` over a soft break,
2474 // undoes a single Enter symmetrically. In `Fold` flow a lone `\n` would
2475 // render as an invisible space, so Enter keeps making the paragraph break
2476 // that actually shows.
2477 //
2478 // Only in running prose. A list or a quote has a continuation of its own
2479 // to write, and a `\n` there is not a soft line but a lost container.
2480 let in_container = has(Kind::ListItem) || has(Kind::TaskListItem) || has(Kind::BlockQuote);
2481 if self.line_flow == LineFlow::Preserve && !in_container {
2482 self.insert_raw("\n");
2483 return;
2484 }
2485 // A heading gets a *paragraph*, never a second heading: Enter at the end
2486 // of a title is how every editor is asked for the body under it, and
2487 // `split_block` would repeat the `#` instead. Whitespace at the split
2488 // point goes with the break rather than opening the new paragraph, which
2489 // is what `split_block` does everywhere else.
2490 if has(Kind::Heading) {
2491 let mut end = self.caret;
2492 while self.source.as_bytes().get(end) == Some(&b' ') {
2493 end += 1;
2494 }
2495 self.splice(self.caret, end, "\n\n", EditKind::Other);
2496 return;
2497 }
2498 self.split_block_here();
2499 }
2500
2501 /// Part the block at the caret with twig's [`Editor::split_block`], leaving
2502 /// the caret in the second half.
2503 ///
2504 /// twig reopens whatever the first half was inside of — the bullet with its
2505 /// indent, the quote's `>`, a checklist item's `[ ]` — which is the whole
2506 /// reason this replaced the markup leaf used to spell from the line's bytes.
2507 /// It renumbers nothing, though: a new item mid-list is written with its
2508 /// neighbour's number, so [`renumber_here`](Self::renumber_here) still runs
2509 /// behind it, folded into the same undo step.
2510 ///
2511 /// Falls back to a plain paragraph break if twig declines, so an unhandled
2512 /// shape still moves the caret down rather than swallowing the keystroke.
2513 fn split_block_here(&mut self) {
2514 // The read-only gate — this door reaches twig without the splice.
2515 if self.read_only {
2516 return;
2517 }
2518 match self.editor.split_block(self.caret) {
2519 Ok(change) => {
2520 self.last_edit_kind = None;
2521 self.refresh();
2522 self.anchor = None;
2523 self.caret = change.new.end;
2524 self.dirty = self.source != self.clean_source;
2525 self.status = None;
2526 self.clamp_caret();
2527 self.record_caret();
2528 // Aimed at the new block's *start*: the caret twig leaves is one
2529 // past the marker it wrote, where there is no list in reach.
2530 self.renumber_at(change.new.start);
2531 }
2532 Err(_) => self.insert_raw("\n\n"),
2533 }
2534 }
2535
2536 /// Whether the item on the marker's line carries no content — the shape
2537 /// double-Enter reads as "I'm done with this list."
2538 fn item_is_empty(&self, line: &ListMarker) -> bool {
2539 let content_start = line.content_start().min(self.source.len());
2540 let line_end = self.source[self.caret..]
2541 .find('\n')
2542 .map(|i| self.caret + i)
2543 .unwrap_or(self.source.len());
2544 self.source[content_start..line_end.max(content_start)]
2545 .trim()
2546 .is_empty()
2547 }
2548
2549 /// Leave the list: replace the empty item's marker with a blank line, so the
2550 /// caret lands in a fresh paragraph below it.
2551 ///
2552 /// Inside a quote the blank line has to stay quoted (a bare one would end the
2553 /// quote), and the caret's new line keeps the `> ` it was already behind —
2554 /// leaving the list without also leaving the quote.
2555 fn exit_list(&mut self, line: &ListMarker) {
2556 let prefix = self.quote_prefix_at(line.marker_start);
2557 let blank = prefix.trim_end();
2558 self.splice(
2559 line.line_start,
2560 self.caret,
2561 &format!("{blank}\n{prefix}"),
2562 EditKind::Other,
2563 );
2564 }
2565
2566 /// What a line continuing the containers at `off` has to open with — the
2567 /// quote markers reproduced, each enclosing item's marker as its width in
2568 /// spaces. Also the column a nested item's marker stands in, which is what
2569 /// makes it Tab's answer.
2570 fn continuation_prefix_at(&mut self, off: usize) -> String {
2571 self.editor
2572 .document()
2573 .and_then(|mut d| d.continuation_prefix(off))
2574 .map(|p| p.text)
2575 .unwrap_or_default()
2576 }
2577
2578 /// The column a *nested list* may open at inside the item at `off` — which
2579 /// is not always where the item's own text continues.
2580 ///
2581 /// twig counts a task item's `[ ] ` box as part of its marker, correctly:
2582 /// it is markup a rich view hides, and the item's own wrapped text does
2583 /// stand past it. But a nested list may only open at the *list* marker's
2584 /// column, and four columns further in is an indented continuation of the
2585 /// paragraph instead — `- [ ] a` + ` - [ ] b` is one item, not two.
2586 /// So the box's own width goes back.
2587 ///
2588 /// The one place leaf still reads a checkbox's spelling. It goes when twig
2589 /// reports the list marker's column apart from the box; `checked` is what
2590 /// says a box is there at all, so only its width is being measured here.
2591 fn nesting_prefix_at(&mut self, off: usize) -> String {
2592 let cont = self.continuation_prefix_at(off);
2593 let Some(item) = self.innermost_list_item(off) else {
2594 return cont;
2595 };
2596 if item.checked.is_none() {
2597 return cont;
2598 }
2599 let box_width = item
2600 .marker_span
2601 .and_then(|m| self.source.get(m))
2602 .and_then(|marker| marker.rfind('[').map(|i| marker.len() - i))
2603 .unwrap_or(0);
2604 // The trailing columns are the ones the item's own marker contributed,
2605 // so trimming from the end leaves any quote prefix standing.
2606 cont[..cont.len().saturating_sub(box_width)].to_string()
2607 }
2608
2609 /// Where the line of the item *containing* the item at `off` begins — the
2610 /// prefix Shift+Tab moves back to, which gives up exactly the level the
2611 /// parent contributed. The quote prefix alone for a top-level item, which
2612 /// has no level left to give.
2613 fn outdent_prefix_at(&mut self, off: usize) -> String {
2614 let items: Vec<usize> = self
2615 .editor
2616 .document()
2617 .and_then(|mut d| d.ancestors_at_caret(off))
2618 .map(|c| {
2619 c.into_iter()
2620 .filter(|m| m.kind == Kind::ListItem || m.kind == Kind::TaskListItem)
2621 .map(|m| m.span.start)
2622 .collect()
2623 })
2624 .unwrap_or_default();
2625 // The second-innermost item is the parent; its own line's indent is the
2626 // target. `list_marker_on_line` gives that line's prefix directly.
2627 let parent = items.len().checked_sub(2).map(|i| items[i]);
2628 match parent.and_then(|p| self.list_marker_on_line(p)) {
2629 Some(m) => self.source[m.line_start..m.marker_start].to_string(),
2630 None => self.quote_prefix_at(off),
2631 }
2632 }
2633
2634 /// The block-quote prefix in force at `off` — `""` outside a quote, `"> "`
2635 /// inside one, `"> > "` inside two.
2636 ///
2637 /// Assembled from each enclosing quote's own [`FlatNode::marker_span`], so
2638 /// the `>` and the space after it are twig's spelling rather than leaf's.
2639 /// The whole line prefix can't answer this: it also carries the indent of
2640 /// whatever the quote holds, which a blank separator line must *not* repeat.
2641 fn quote_prefix_at(&mut self, off: usize) -> String {
2642 let Ok(chain) = self
2643 .editor
2644 .document()
2645 .and_then(|mut d| d.ancestors_at_caret(off))
2646 else {
2647 return String::new();
2648 };
2649 let quotes: Vec<usize> = chain
2650 .iter()
2651 .filter(|m| m.kind == Kind::BlockQuote)
2652 .map(|m| m.node_id as usize)
2653 .collect();
2654 let Ok(nodes) = self.editor.nodes() else {
2655 return String::new();
2656 };
2657 quotes
2658 .iter()
2659 .filter_map(|id| nodes.get(*id)?.marker_span.clone())
2660 .filter_map(|s| self.source.get(s))
2661 .collect()
2662 }
2663
2664 /// Whether the item at `off` sits inside another one — the test Backspace
2665 /// uses to choose between outdenting and dropping the marker.
2666 ///
2667 /// Counted from the AST rather than from the line's leading whitespace,
2668 /// which is indentation in Markdown and, in Djot, may be nothing at all.
2669 fn item_is_nested(&mut self, off: usize) -> bool {
2670 self.editor
2671 .document()
2672 .and_then(|mut d| d.ancestors_at_caret(off))
2673 .map(|c| {
2674 c.into_iter()
2675 .filter(|m| m.kind == Kind::ListItem || m.kind == Kind::TaskListItem)
2676 .count()
2677 > 1
2678 })
2679 .unwrap_or(false)
2680 }
2681
2682 /// The innermost list item containing `probe`, under twig's **caret**
2683 /// containment rule — a block's end is inside it.
2684 ///
2685 /// Half-open containment can't answer this. An empty item's span is exactly
2686 /// its marker, so the caret sitting after `- ` is one past the end and the
2687 /// item it is plainly in tests as out of reach; that is the shape
2688 /// double-Enter has to recognise to leave the list.
2689 fn innermost_list_item(&mut self, probe: usize) -> Option<FlatNode> {
2690 let chain = self
2691 .editor
2692 .document()
2693 .and_then(|mut d| d.ancestors_at_caret(probe))
2694 .ok()?;
2695 let id = chain
2696 .iter()
2697 .rev()
2698 .find(|m| m.kind == Kind::ListItem || m.kind == Kind::TaskListItem)?
2699 .node_id as usize;
2700 self.editor.nodes().ok()?.get(id).cloned()
2701 }
2702
2703 /// The list marker opening `off`'s line, per twig — `None` when that line
2704 /// opens no list item.
2705 ///
2706 /// [`Document::line_prefix`] is the whole hidden run from the line start:
2707 /// `> 1. ` is a quote's marker, an indent, and an item's marker together,
2708 /// and it is `None` on a *continuation* line, which opens nothing. That last
2709 /// case is the one leaf could never get right by reading bytes. `- a\n - b`
2710 /// is two items in Markdown and one in Djot, where a marker cannot interrupt
2711 /// a paragraph and ` - b` is literal text — identical bytes, and only the
2712 /// parser knows which document it is looking at.
2713 ///
2714 /// The item's own marker is separated out via its
2715 /// [`FlatNode::marker_span`], so `marker_start` splits the prefix into what
2716 /// the containers around it contribute and what the item does.
2717 fn list_marker_on_line(&mut self, off: usize) -> Option<ListMarker> {
2718 let off = off.min(self.source.len());
2719 let prefix = self.editor.document().ok()?.line_prefix(off).ok()??;
2720 // The prefix belongs to a list only when an item's marker closes it —
2721 // a heading's `# ` or a bare quote's `> ` is a prefix too.
2722 let item = self.innermost_list_item(prefix.end.min(self.source.len()))?;
2723 let marker = item.marker_span.clone()?;
2724 if marker.end != prefix.end {
2725 return None;
2726 }
2727 Some(ListMarker {
2728 line_start: prefix.start,
2729 marker_start: marker.start,
2730 text: self.source.get(prefix)?.to_string(),
2731 })
2732 }
2733
2734 /// Whether the list item on `line_start`'s line is the **first item** of its
2735 /// list — the one Tab must not nest, because nesting needs a preceding
2736 /// sibling to become the new parent and a first item has none. `false` for a
2737 /// line that isn't a list item, and for an item with a sibling above it (the
2738 /// one Tab *can* nest). Gated on the AST, not the marker bytes: `- ` reads
2739 /// the same in a setext underline that opens no list at all.
2740 fn first_item_of_list(&mut self, line_start: usize) -> bool {
2741 let Some(marker) = self.list_marker_on_line(line_start) else {
2742 return false;
2743 };
2744 // Probe just inside the marker, where the item's own node is in reach —
2745 // the marker offset itself can resolve to the enclosing list, not the
2746 // `list_item`, whose span starts at the marker.
2747 let probe = marker.content_start().min(self.source.len());
2748 let Some(item) = self.innermost_list_item(probe) else {
2749 return false;
2750 };
2751 let Ok(nodes) = self.editor.nodes() else {
2752 return false;
2753 };
2754 match item.parent {
2755 // First when the parent list opens with this very item.
2756 Some(pid) => nodes
2757 .get(pid.0 as usize)
2758 .is_some_and(|p| p.first_child == Some(item.id)),
2759 // A parentless item is trivially the first (and only) one.
2760 None => true,
2761 }
2762 }
2763
2764 pub fn backspace(&mut self) {
2765 if let Some((s, e)) = self.selection() {
2766 self.splice(s, e, "", EditKind::Other);
2767 return;
2768 }
2769 // WYSIWYG: Backspace at the very start of a list item's content is a
2770 // structural key, not a character delete — it walks the "un-indent, then
2771 // un-list" ladder every list editor gives that keystroke (outdent a
2772 // nested item, strip a top-level one's marker to a paragraph). In source
2773 // view the `- ` is visible text the user is deleting a byte of, so it
2774 // keeps its literal meaning there, like Enter does.
2775 if self.view != View::Source && self.backspace_list_start() {
2776 return;
2777 }
2778 // WYSIWYG: and the same at the start of a heading's content — the `# `
2779 // there is markup the rich view hides, not text the user typed.
2780 if self.view != View::Source && self.backspace_heading_start() {
2781 return;
2782 }
2783 // WYSIWYG: at a block picture's stops, a byte-at-a-time delete would take
2784 // the markup apart under a caret that cannot see it — see
2785 // `delete_around_block_media`.
2786 if self.view != View::Source && self.delete_around_block_media(false) {
2787 return;
2788 }
2789 // WYSIWYG: Backspace on a *blank line* deletes back to the previous caret
2790 // stop, not a single newline. On a line with no text of its own, the byte
2791 // before the caret is a `\n` that spells part of a block boundary — the gap
2792 // between two blocks, drawn but never a caret home. Removing just it strands
2793 // the caret in that gap and leaves an odd blank line the eye reads as one
2794 // separator but the caret can't land on: the "extra newline" left behind
2795 // after leaving a list (Enter, Enter) or a paragraph and pressing Backspace.
2796 // Deleting to the previous stop instead collapses the whole break at once,
2797 // landing the caret at the end of the block above. Two blank lines in a row
2798 // are one stop apart, so this still removes exactly one — the lone-Enter /
2799 // lone-Backspace symmetry the empty-line case is built on is untouched.
2800 if self.view != View::Source
2801 && self.caret > self.caret_floor()
2802 && self.caret_on_blank_line()
2803 && let Some(stop) = self.vmap.stop_before(self.caret)
2804 {
2805 let stop = stop.max(self.caret_floor());
2806 if stop < self.caret {
2807 self.splice(stop, self.caret, "", EditKind::Delete);
2808 return;
2809 }
2810 }
2811 if self.caret > self.caret_floor() {
2812 // An in-cell `<br>` draws as one newline glyph, so Backspace over it
2813 // takes the whole tag — a single-byte step would leave a broken `<br`
2814 // showing in the cell. Rich view only (source view edits the literal).
2815 if self.view != View::Source
2816 && let Some((start, end)) = self.cell_break_at(BreakEdge::Backward)
2817 {
2818 let start = start.max(self.caret_floor());
2819 if start < end {
2820 self.splice(start, end, "", EditKind::Delete);
2821 return;
2822 }
2823 }
2824 // Aim the delete at the character the writer can *see* behind the
2825 // caret, never at a delimiter the rich view drew nothing for. Two
2826 // steps, and either can apply: from the far side of a run's closing
2827 // `**` step back into the run (the caret is drawn at the end of its
2828 // word), and at the start of a run's text step out past its opening
2829 // `**` to the character in front of it, leaving the run standing.
2830 // Without them a plain Backspace unspells the phrase it is editing
2831 // and leaves a literal asterisk on screen.
2832 let end = if self.view == View::Source {
2833 self.caret
2834 } else {
2835 let inside = self.step_inside_close_delims(self.caret);
2836 self.skip_leading_open_delims(inside)
2837 .max(self.caret_floor())
2838 };
2839 // Never delete back across the floor — that would eat hidden
2840 // frontmatter the WYSIWYG caret can't even see.
2841 let mut prev = prev_boundary(&self.source, end).max(self.caret_floor());
2842 // Take a hidden escape backslash with the char it escapes: the rich
2843 // view draws `\*` as a single `*`, so Backspace over it must delete
2844 // both bytes, never strand the `\` as a lone visible backslash (the
2845 // mirror of the Hidden-mode typing that wrote the escape). Source view
2846 // shows the `\`, so there it is an ordinary character.
2847 if self.view != View::Source
2848 && prev > self.caret_floor()
2849 && self.is_hidden_escape(prev - 1)
2850 {
2851 prev -= 1;
2852 }
2853 if prev < end {
2854 self.splice(prev, end, "", EditKind::Delete);
2855 }
2856 }
2857 }
2858
2859 /// Whether the caret's own source line holds nothing but whitespace — an
2860 /// empty paragraph, or the blank line a block boundary is spelled with. The
2861 /// test for [`backspace`](Self::backspace)'s stop-wise delete: such a line has
2862 /// no text of its own, so the newline before the caret belongs to the gap
2863 /// between blocks rather than to any word the caret is editing.
2864 fn caret_on_blank_line(&self) -> bool {
2865 let line_start = self.source[..self.caret].rfind('\n').map_or(0, |i| i + 1);
2866 let line_end = self.source[self.caret..]
2867 .find('\n')
2868 .map_or(self.source.len(), |i| self.caret + i);
2869 self.source[line_start..line_end].trim().is_empty()
2870 }
2871
2872 /// The source span of an in-cell hard break (`<br>`) touching the caret on the
2873 /// `edge` side — the byte range to delete whole. A table row is one source
2874 /// line, so its break is spelled `<br>` yet drawn as a single newline glyph
2875 /// (see `wysiwyg.rs`); a delete over it must take every byte, or a one-byte
2876 /// step strands a broken `<br` in the cell. `Backward` matches a break ending
2877 /// at the caret (Backspace), `Forward` one starting at it (Delete). `None`
2878 /// when no such break is adjacent. Only the in-cell break is spelled `<br>`
2879 /// (an ordinary hard break is ` \n`), so the leading `<` alone tells them
2880 /// apart — no ancestor walk needed. Rich view only; source view shows the
2881 /// literal tag and deletes it a byte at a time.
2882 fn cell_break_at(&mut self, edge: BreakEdge) -> Option<(usize, usize)> {
2883 let caret = self.caret;
2884 let nodes = self.nodes();
2885 let src = self.source.as_bytes();
2886 nodes
2887 .iter()
2888 .find(|n| {
2889 n.kind == Kind::HardBreak
2890 && n.span.start < n.span.end
2891 && src.get(n.span.start) == Some(&b'<')
2892 && match edge {
2893 BreakEdge::Backward => n.span.end == caret,
2894 BreakEdge::Forward => n.span.start == caret,
2895 }
2896 })
2897 .map(|n| (n.span.start, n.span.end))
2898 }
2899
2900 /// Whether the source byte at `off` is a backslash twig consumed as an escape
2901 /// (hidden in the rich view), as against a literal backslash (drawn). A
2902 /// backslash escapes exactly an ASCII-punctuation character (the CommonMark /
2903 /// Djot rule twig follows), so `\` + punctuation is the whole test — no AST
2904 /// round-trip needed.
2905 fn is_hidden_escape(&self, off: usize) -> bool {
2906 let b = self.source.as_bytes();
2907 b.get(off) == Some(&b'\\') && b.get(off + 1).is_some_and(u8::is_ascii_punctuation)
2908 }
2909
2910 /// Backspace's list behaviour: when the caret sits exactly at the start of a
2911 /// list item's content (right after its marker), outdent the item if it's
2912 /// nested, else strip the marker so it becomes a paragraph. Returns whether
2913 /// it acted — `false` leaves Backspace its ordinary character delete.
2914 fn backspace_list_start(&mut self) -> bool {
2915 let Some(marker) = self.list_marker_on_line(self.caret) else {
2916 return false;
2917 };
2918 // Only right after the marker. That the line opens a real item is
2919 // already settled: `list_marker_on_line` answers from the tree.
2920 if self.caret != marker.content_start() {
2921 return false;
2922 }
2923 if self.item_is_nested(marker.marker_start) {
2924 // Nested: give back one level, keeping the marker and carrying the
2925 // caret with it.
2926 self.outdent();
2927 } else {
2928 // Top level: drop the marker, leaving a paragraph, then renumber the
2929 // siblings the removed item was counted among. Only the marker goes —
2930 // a quote prefix in front of it still has a quote to hold up.
2931 self.splice(marker.marker_start, self.caret, "", EditKind::Other);
2932 self.renumber_here();
2933 }
2934 true
2935 }
2936
2937 /// Backspace's heading behaviour: with the caret exactly at the start of an
2938 /// ATX heading's content — right after the `#` marker the rich view hides —
2939 /// strip the marker so the line becomes a paragraph. The peer of
2940 /// [`backspace_list_start`](Self::backspace_list_start)'s ladder, and the same
2941 /// reasoning: hidden block markup is structure, so the keystroke over it is
2942 /// structural.
2943 ///
2944 /// Without this the ordinary delete takes the space out of `# Title` and
2945 /// leaves `#Title`, which is no longer a heading at all — the hash the view
2946 /// had been hiding surfaces as literal text the user has to delete a second
2947 /// time, having never typed it. A closing sequence (`# Title #`, hidden at the
2948 /// other end) goes with the marker for the same reason.
2949 ///
2950 /// Returns whether it acted; `false` leaves Backspace its character delete.
2951 fn backspace_heading_start(&mut self) -> bool {
2952 let caret = self.caret;
2953 // The heading whose content opens exactly at the caret. A bare `#` has no
2954 // content span at all — its content starts (and ends) where the line does.
2955 let Some((span, content_end, marker)) = self.nodes().iter().find_map(|n| {
2956 let (start, end) = match &n.content_span {
2957 Some(c) => (c.start, c.end),
2958 None => (n.span.end, n.span.end),
2959 };
2960 (n.kind == Kind::Heading && start == caret)
2961 .then(|| (n.span.clone(), end, n.marker_span.clone()))
2962 }) else {
2963 return false;
2964 };
2965 // twig reports the marker's own extent, so there is nothing to walk back
2966 // over and no `#` in this file. A setext heading has no marker — its
2967 // content opens the line — so it falls through to the ordinary delete,
2968 // as does anything else sitting at a content start.
2969 // `m.end == caret` is what excludes a setext heading, whose marker is the
2970 // underline *after* the content rather than a prefix before it.
2971 let Some(marker) = marker.filter(|m| m.end == caret) else {
2972 return false;
2973 };
2974 let start = marker.start;
2975 // A closing `#` sequence is hidden too, so it can't be left behind. Only
2976 // when the tail really is one: trailing spaces alone are nothing to strip.
2977 let tail = &self.source[content_end..span.end];
2978 if tail.contains('#') && tail.chars().all(|c| c == '#' || c.is_whitespace()) {
2979 let kept = self.source[caret..content_end].to_string();
2980 self.splice(start, span.end, &kept, EditKind::Other);
2981 // The splice leaves the caret past the text it re-wrote; the caret
2982 // belongs where the content now starts, which is where it already was.
2983 self.caret = start;
2984 self.record_caret();
2985 } else {
2986 self.splice(start, caret, "", EditKind::Other);
2987 }
2988 true
2989 }
2990
2991 pub fn delete_forward(&mut self) {
2992 if let Some((s, e)) = self.selection() {
2993 self.splice(s, e, "", EditKind::Other);
2994 } else if self.caret < self.source.len() {
2995 // The mirror of Backspace's: forward-delete in front of a picture
2996 // would eat the `!` off its markup and leave a link where a photo was.
2997 if self.view != View::Source && self.delete_around_block_media(true) {
2998 return;
2999 }
3000 // Delete forward over an in-cell `<br>` takes the whole tag, the mirror
3001 // of Backspace's swallow (see `cell_break_at`) — else a byte-step
3002 // strands a broken `<br` in the cell.
3003 if self.view != View::Source
3004 && let Some((start, end)) = self.cell_break_at(BreakEdge::Forward)
3005 {
3006 self.splice(start, end, "", EditKind::Delete);
3007 return;
3008 }
3009 // The mirror of Backspace's two steps: from in front of a run's
3010 // opening `**` step into it, onto the first letter of its text, and
3011 // at the end of a run's text step out past its closing `**` to the
3012 // character beyond. Either way Delete takes the character it looks
3013 // like it is pointing at, and never a delimiter drawn as nothing.
3014 // The caret then settles back inside the run it was standing in —
3015 // see `settle_inside_close_delims`.
3016 let from = if self.view == View::Source {
3017 self.caret
3018 } else {
3019 let inside = self.step_inside_open_delims(self.caret);
3020 self.skip_trailing_close_delims(inside)
3021 };
3022 let next = next_boundary(&self.source, from);
3023 if from < next {
3024 self.splice(from, next, "", EditKind::Delete);
3025 }
3026 }
3027 }
3028
3029 /// Delete from the caret back to the start of the previous word (⌥⌫ /
3030 /// Ctrl+⌫). Deletes the selection instead when one is active.
3031 pub fn delete_word_back(&mut self) {
3032 if let Some((s, e)) = self.selection() {
3033 self.splice(s, e, "", EditKind::Other);
3034 } else {
3035 // A word back from just past a picture is a word *of its markup*, and
3036 // a word back from in front of one runs through the paragraph break
3037 // into the prose above — dissolving the picture either way. See
3038 // `delete_around_block_media`.
3039 if self.view != View::Source && self.delete_around_block_media(false) {
3040 return;
3041 }
3042 let start = self.word_left_from(self.caret).max(self.caret_floor());
3043 if start < self.caret {
3044 let (s, e) = self.widen_over_emptied_inlines(start, self.caret);
3045 self.splice(s, e, "", EditKind::Delete);
3046 }
3047 }
3048 }
3049
3050 /// Delete from the caret forward to the end of the next word (⌥⌦ /
3051 /// Ctrl+Del). Deletes the selection instead when one is active.
3052 pub fn delete_word_forward(&mut self) {
3053 if let Some((s, e)) = self.selection() {
3054 self.splice(s, e, "", EditKind::Other);
3055 } else {
3056 // The mirror: a word forward from in front of a picture is its markup.
3057 if self.view != View::Source && self.delete_around_block_media(true) {
3058 return;
3059 }
3060 let end = self.word_right_from(self.caret);
3061 if end > self.caret {
3062 let (s, e) = self.widen_over_emptied_inlines(self.caret, end);
3063 self.splice(s, e, "", EditKind::Delete);
3064 }
3065 }
3066 }
3067
3068 /// Delete from the caret back to the start of its line (⌘⌫). Deletes the
3069 /// selection instead when one is active, as every other delete here does.
3070 ///
3071 /// The line is the view's own — the one Home and End work on, so in WYSIWYG
3072 /// a soft-wrapped row is a line. It is not Home's *target*, though: Home
3073 /// stops at the first character and this takes the indentation with it, the
3074 /// way Cocoa's `deleteToBeginningOfLine:` does. Stopping at the text would
3075 /// leave an indent behind that nothing can then ask to delete, where a caret
3076 /// left at column 0 is one press of Home away from either.
3077 pub fn delete_to_line_start(&mut self) {
3078 if let Some((s, e)) = self.selection() {
3079 self.splice(s, e, "", EditKind::Other);
3080 return;
3081 }
3082 // Never back across the floor: hidden frontmatter isn't on this line, or
3083 // on any line the WYSIWYG caret can see.
3084 let (start, _) = self.line_span();
3085 let start = start.max(self.caret_floor());
3086 if start < self.caret {
3087 let (s, e) = self.widen_over_emptied_inlines(start, self.caret);
3088 self.splice(s, e, "", EditKind::Delete);
3089 }
3090 }
3091
3092 /// Kill from the caret to the end of its line (^K). Deletes the selection
3093 /// instead when one is active.
3094 ///
3095 /// At the end of the line it does nothing, rather than pulling the line
3096 /// below up into this one. Joining has no meaning to give it in both views
3097 /// at once: a WYSIWYG line ends at a soft wrap as often as at a newline, and
3098 /// there is nothing there to delete, while the newline a *source* line ends
3099 /// with is only half of the blank line that separates two paragraphs —
3100 /// deleting one leaves a soft break, which is not the join it looks like.
3101 /// The views agreeing is worth more than emacs' second press, and Delete is
3102 /// already the key that joins.
3103 pub fn delete_to_line_end(&mut self) {
3104 if let Some((s, e)) = self.selection() {
3105 self.splice(s, e, "", EditKind::Other);
3106 return;
3107 }
3108 let (_, end) = self.line_span();
3109 if end > self.caret {
3110 let (s, e) = self.widen_over_emptied_inlines(self.caret, end);
3111 self.splice(s, e, "", EditKind::Delete);
3112 }
3113 }
3114
3115 /// Grow a WYSIWYG word-delete to swallow any inline node it empties.
3116 ///
3117 /// A glyph-space range covers what the user can see, which for `**bold**` is
3118 /// the word and never the delimiters around it — so deleting the word on its
3119 /// own leaves `a **** c`, markup wrapped around nothing. They asked for the
3120 /// word, and the styling was the word's; the two go together. Only the
3121 /// node's delimiters are taken, and those are hidden here anyway, so nothing
3122 /// visible outside the range is lost.
3123 ///
3124 /// Repeated to a fixed point: emptying `***bold***` empties the emph inside
3125 /// the strong, and only then is the strong empty too.
3126 fn widen_over_emptied_inlines(&mut self, start: usize, end: usize) -> (usize, usize) {
3127 if self.view == View::Source {
3128 return (start, end);
3129 }
3130 let nodes = self.nodes();
3131 let (mut s, mut e) = (start, end);
3132 loop {
3133 let mut grew = false;
3134 for n in nodes.iter().filter(|n| wysiwyg::is_inline(n)) {
3135 let Some(text) = inline_content_span(n, &self.source) else {
3136 continue;
3137 };
3138 // Some of its text survives, so the node still has a job.
3139 if text.start < s || text.end > e {
3140 continue;
3141 }
3142 if n.span.start < s || n.span.end > e {
3143 s = s.min(n.span.start);
3144 e = e.max(n.span.end);
3145 grew = true;
3146 }
3147 }
3148 if !grew {
3149 return (s, e);
3150 }
3151 }
3152 }
3153
3154 /// One splice of document text, keeping the **mark-edge rule**: an inline
3155 /// mark's content never begins or ends with whitespace. In Markdown and Djot
3156 /// a delimiter standing against a space is not a delimiter at all — `**bold **`
3157 /// is four literal asterisks around a word, and a rich view drawing the
3158 /// document faithfully has no choice but to show them. That is correct
3159 /// rendering of what the file says, and nobody typing a space after a bold
3160 /// word meant to say it.
3161 ///
3162 /// So the space goes *outside* the run instead — `**bold** ` — which is the
3163 /// same document to a reader and a live one to a parser. The caret follows it
3164 /// out and keeps the marks armed (see [`rearm`](Self::rearm)), so the next
3165 /// character rejoins the run (see [`rejoin_run`](Self::rejoin_run)) and the
3166 /// writer sees one unbroken bold phrase, never a flash of raw syntax.
3167 ///
3168 /// Every ordinary edit — typing, deleting, pasting, an IME step — comes
3169 /// through here, so the rule holds however the whitespace arrives at the
3170 /// edge. The repair is decided *after* the plain edit, by asking whether the
3171 /// mark actually died: a code span's backticks aren't whitespace-sensitive
3172 /// (`` `code ` `` is still code), and nothing is re-spelled when nothing broke.
3173 fn splice(&mut self, start: usize, end: usize, text: &str, kind: EditKind) -> bool {
3174 let fix = self.mark_edge_fix(start, end, text);
3175 if !self.splice_exact(start, end, text, kind) {
3176 return false;
3177 }
3178 if let Some(fix) = fix {
3179 self.repair_mark_edges(fix);
3180 }
3181 if text.is_empty() && end > start {
3182 self.settle_inside_close_delims();
3183 }
3184 true
3185 }
3186
3187 /// After a delete, take a caret left standing past a run's closing delimiters
3188 /// back inside the run.
3189 ///
3190 /// A delete leaves the caret where the deleted bytes began, and when those
3191 /// bytes were the last thing after a marked phrase — the space the mark-edge
3192 /// rule pushed out of `**bold** `, say — that spot is the far side of the
3193 /// closing `**`. The rich view has nothing to draw there: the delimiters are
3194 /// hidden, so the caret shows at the end of the word either way, and the two
3195 /// offsets are one place on screen with two different meanings. Typing at the
3196 /// outer one lands past the run, so the writer who backspaced a space out of
3197 /// their bold phrase watches the next character come out plain, and the
3198 /// toolbar button go dark, with the caret never appearing to move.
3199 ///
3200 /// The end of the run's text is the caret's home there — a delete that took
3201 /// away everything after a phrase leaves the caret at the end of that phrase,
3202 /// which is inside it — so it settles onto that
3203 /// ([`step_inside_close_delims`](Self::step_inside_close_delims) does the
3204 /// walk, through every mark closing at the point): the word stays bold, the
3205 /// button stays lit, and the next character carries on the phrase.
3206 ///
3207 /// Rich view only, and only where a mark really closes at the caret — mid-run
3208 /// or in plain prose no span ends there and the caret stays put. The opening
3209 /// edge is left alone on purpose: a caret in front of a run inherits from the
3210 /// text on its left, which is the plain text outside.
3211 fn settle_inside_close_delims(&mut self) {
3212 if self.view != View::Wysiwyg {
3213 return;
3214 }
3215 let at = self.step_inside_close_delims(self.caret);
3216 if at != self.caret {
3217 self.caret = at;
3218 self.clear_pending();
3219 self.record_caret();
3220 }
3221 }
3222
3223 /// The splice exactly as asked, with no mark-edge repair — for the callers
3224 /// that are *writing* the delimiters themselves ([`insert_with_marks`](Self::insert_with_marks)
3225 /// and [`rejoin_run`](Self::rejoin_run)) and place their own offsets around
3226 /// the bytes they inserted.
3227 ///
3228 /// One `edit_range` through twig, then re-anchor the caret from the returned
3229 /// `Change` and refresh the cached source. A reparse-breaking edit (rare for
3230 /// Markdown/Djot) leaves the document untouched and reports.
3231 ///
3232 /// Returns whether the edit landed — for a caller that has offsets of its
3233 /// own to place afterwards, which a rolled-back splice would leave pointing
3234 /// into text that never came to exist.
3235 fn splice_exact(&mut self, start: usize, end: usize, text: &str, kind: EditKind) -> bool {
3236 // The read-only gate, for every edit at once — see the field.
3237 if self.read_only {
3238 return false;
3239 }
3240 // twig records an undo step for every edit; when this one continues a
3241 // run of the same kind (typing, deleting), tell twig to fold it into the
3242 // step before it so the whole run undoes at once.
3243 let coalesce = kind != EditKind::Other && self.last_edit_kind == Some(kind);
3244 // Hand twig the pre-edit caret before the splice, so the undo step it
3245 // retires carries where the caret was standing.
3246 self.record_caret();
3247 match self.editor.edit_range(start, end, text) {
3248 Ok(change) => {
3249 if coalesce {
3250 let _ = self.editor.coalesce_last_undo();
3251 }
3252 self.last_edit_kind = Some(kind);
3253 self.refresh();
3254 self.caret = change.new.end;
3255 self.anchor = None;
3256 self.goal_col = None;
3257 self.clear_pending();
3258 self.dirty = self.source != self.clean_source;
3259 self.status = None;
3260 // And the post-edit caret, so a later redo restores it.
3261 self.record_caret();
3262 true
3263 }
3264 // The edit was rolled back, so twig's history did not move and
3265 // neither may ours: pushing here would leave a step with no edit
3266 // under it and shift every later undo onto the wrong caret.
3267 Err(e) => {
3268 self.status = Some(format!("edit: {e}"));
3269 false
3270 }
3271 }
3272 }
3273
3274 /// The re-spelling that would keep the mark-edge rule for the edit
3275 /// `[start, end)` → `text`, or `None` when the edit leaves no whitespace
3276 /// against a delimiter and the plain splice is already right. Computed
3277 /// *before* the edit, while the run's spans and delimiters can still be read
3278 /// off the document; applied afterwards, and only if the mark really died —
3279 /// see [`repair_mark_edges`](Self::repair_mark_edges).
3280 ///
3281 /// Rich view only. Source view is for typing raw markup, where a space put
3282 /// against a `**` is exactly the character it looks like.
3283 fn mark_edge_fix(&mut self, start: usize, end: usize, text: &str) -> Option<MarkEdgeFix> {
3284 if self.view != View::Wysiwyg || start > end || end > self.source.len() {
3285 return None;
3286 }
3287 // Every inline mark standing over the edit, outermost first, with the
3288 // content span that says where its delimiters are.
3289 let chain: Vec<(InlineKind, std::ops::Range<usize>, std::ops::Range<usize>)> = self
3290 .editor
3291 .ancestors_at(start)
3292 .unwrap_or_default()
3293 .into_iter()
3294 .filter_map(|m| {
3295 let kind = inline_kind(&m.kind)?;
3296 let content = m.content_span.clone()?;
3297 Some((kind, m.span.clone(), content))
3298 })
3299 .collect();
3300 // The innermost run whose *content* holds the whole edit: the one whose
3301 // text is being changed, rather than one the edit merely sits under.
3302 let (kind, span, content) = chain
3303 .iter()
3304 .rev()
3305 .find(|(_, _, c)| c.start <= start && end <= c.end)?
3306 .clone();
3307 // What that content becomes. Whitespace at either end of it is what
3308 // would put out the mark.
3309 let body = format!(
3310 "{}{text}{}",
3311 &self.source[content.start..start],
3312 &self.source[end..content.end]
3313 );
3314 let (lead, trail) = if body.trim().is_empty() {
3315 // Nothing but whitespace left: there is no content to mark at all,
3316 // and the delimiters go with it rather than closing on a space.
3317 (body.len(), 0)
3318 } else {
3319 (
3320 body.len() - body.trim_start().len(),
3321 body.len() - body.trim_end().len(),
3322 )
3323 };
3324 // Nothing against a delimiter, and something still between them: the
3325 // plain edit stands. An emptied run is broken just as surely (`**b**`
3326 // with the `b` deleted is the literal `****`) and is re-spelt as the
3327 // nothing it now says.
3328 if lead == 0 && trail == 0 && !body.is_empty() {
3329 return None;
3330 }
3331 // Marks that open or close exactly where this one does — `***both***` is
3332 // two runs sharing an edge — spell their delimiters as one run of bytes,
3333 // so the whitespace has to clear all of them together.
3334 let (mut open_at, mut close_at) = (span.start, span.end);
3335 for _ in 0..chain.len() {
3336 match chain.iter().find(|(_, _, c)| c.start == open_at) {
3337 Some((_, s, _)) => open_at = s.start,
3338 None => break,
3339 }
3340 }
3341 for _ in 0..chain.len() {
3342 match chain.iter().find(|(_, _, c)| c.end == close_at) {
3343 Some((_, s, _)) => close_at = s.end,
3344 None => break,
3345 }
3346 }
3347 let open = &self.source[open_at..content.start];
3348 let close = &self.source[content.end..close_at];
3349 let core = &body[lead..body.len() - trail];
3350 let respelt = if core.is_empty() {
3351 body.clone()
3352 } else {
3353 format!(
3354 "{}{open}{core}{close}{}",
3355 &body[..lead],
3356 &body[body.len() - trail..]
3357 )
3358 };
3359 // The caret sits just past the inserted text within the new content —
3360 // which, when that lands in the whitespace, is now outside the delimiters.
3361 let pos = (start - content.start) + text.len();
3362 let caret = if core.is_empty() || pos <= lead {
3363 open_at + pos
3364 } else if pos >= lead + core.len() {
3365 open_at + lead + open.len() + core.len() + close.len() + (pos - lead - core.len())
3366 } else {
3367 open_at + lead + open.len() + (pos - lead)
3368 };
3369 Some(MarkEdgeFix {
3370 kind,
3371 probe: content.start,
3372 start: open_at,
3373 end: close_at + text.len() - (end - start),
3374 text: respelt,
3375 caret,
3376 // The marks in force here, resolved against any armed sticky delta —
3377 // what the writer is typing in, and so what has to still be true on
3378 // the far side of the delimiter the caret just stepped over.
3379 want: chain
3380 .iter()
3381 .filter(|(_, s, _)| start < s.end)
3382 .map(|(k, _, _)| *k)
3383 .collect::<InlineMarks>()
3384 .xor(self.pending_here()),
3385 })
3386 }
3387
3388 /// Apply a [`MarkEdgeFix`] — but only if the edit it was computed for really
3389 /// did break the mark. Whether whitespace at a delimiter is fatal is the
3390 /// format's business, not leaf's: `**bold **` is no longer strong, while
3391 /// `` `code ` `` is still perfectly good verbatim, and Djot's braced spellings
3392 /// don't care either. Asking the parser afterwards settles it for every kind
3393 /// and format at once, and costs a re-spelling only where one is due.
3394 ///
3395 /// The repair rides along with the edit that caused it — one undo step puts
3396 /// back what the writer typed, not a delimiter shuffle they never saw.
3397 fn repair_mark_edges(&mut self, fix: MarkEdgeFix) {
3398 if fix.end > self.source.len() {
3399 return;
3400 }
3401 if self.marks_at(fix.probe).iter().any(|(k, _)| *k == fix.kind) {
3402 return; // still a mark: these delimiters don't mind the whitespace
3403 }
3404 let resumed = self.last_edit_kind;
3405 if !self.splice_exact(fix.start, fix.end, &fix.text, EditKind::Other) {
3406 return;
3407 }
3408 let _ = self.editor.coalesce_last_undo();
3409 // The keystroke owns the undo step, so the run of typing it belongs to
3410 // keeps coalescing over the repair rather than breaking in two here.
3411 self.last_edit_kind = resumed;
3412 self.caret = fix.caret.min(self.source.len());
3413 self.anchor = None;
3414 self.goal_col = None;
3415 self.rearm(fix.want);
3416 self.clamp_caret();
3417 self.record_caret();
3418 }
3419
3420 /// Arm whatever sticky delta reproduces `want` at the caret — the marks the
3421 /// writer is typing in, carried across an edit that moved the caret out of
3422 /// the run holding them. Arms nothing when the caret already stands in
3423 /// exactly those marks, but still remembers the spot, so a further ⌘b starts
3424 /// a clean delta here (see [`toggle`](Self::toggle)).
3425 fn rearm(&mut self, want: InlineMarks) {
3426 let here: InlineMarks = self
3427 .marks_at(self.caret)
3428 .into_iter()
3429 .map(|(k, _)| k)
3430 .collect();
3431 self.pending_marks = want.xor(here);
3432 self.pending_at = Some(self.caret);
3433 }
3434
3435 /// Insert `text` at `at` as a *literal* run via twig's `insert_literal`,
3436 /// which backslash-escapes any character that would otherwise open markup in
3437 /// this format and position (`*` → `\*`, a line-start `#` → `\#`). The mirror
3438 /// of [`splice`](Self::splice) for the Hidden reveal mode's typing path, with
3439 /// the same caret re-anchor, coalescing, and rollback contract. `at` must be
3440 /// a collapsed point — a selection is deleted by the caller first, since
3441 /// `insert_literal` inserts rather than replaces.
3442 fn insert_literal_at(
3443 &mut self,
3444 at: usize,
3445 text: &str,
3446 kind: EditKind,
3447 force_coalesce: bool,
3448 ) -> bool {
3449 // The read-only gate: this door goes to twig directly, not through
3450 // `splice_exact`, so it guards itself — see the field.
3451 if self.read_only {
3452 return false;
3453 }
3454 // `force_coalesce` folds this into the immediately preceding edit (the
3455 // selection-delete of an overwrite) so the pair is one undo step; else it
3456 // coalesces only when it continues a run of the same-kind typing.
3457 let coalesce =
3458 force_coalesce || (kind != EditKind::Other && self.last_edit_kind == Some(kind));
3459 // The mark-edge rule holds for typed text however it is spelled — see
3460 // `splice`. Only an insert twig passed through unchanged can use it,
3461 // since a fix is measured in the bytes that actually land, and an escape
3462 // adds bytes this couldn't have counted.
3463 let fix = self.mark_edge_fix(at, at, text);
3464 self.record_caret();
3465 match self.editor.insert_literal(at, text) {
3466 Ok(change) => {
3467 if coalesce {
3468 let _ = self.editor.coalesce_last_undo();
3469 }
3470 self.last_edit_kind = Some(kind);
3471 self.refresh();
3472 self.caret = change.new.end;
3473 self.anchor = None;
3474 self.goal_col = None;
3475 self.clear_pending();
3476 self.dirty = self.source != self.clean_source;
3477 self.status = None;
3478 self.record_caret();
3479 if let Some(fix) = fix.filter(|_| change.new.end - change.new.start == text.len()) {
3480 self.repair_mark_edges(fix);
3481 }
3482 true
3483 }
3484 Err(e) => {
3485 self.status = Some(format!("edit: {e}"));
3486 false
3487 }
3488 }
3489 }
3490
3491 /// After a structural list edit (a new item, a nest/unnest), renumber the
3492 /// ordered list the caret sits in so its source markers run `1, 2, 3, …`
3493 /// again — a raw splice leaves them stale (`1. 2. 2. 3.`). twig does the
3494 /// renumber as its own edit; fold it into the edit that triggered it so the
3495 /// two undo as one, and only when it actually changed the source (a no-op or
3496 /// a caret outside any ordered list must not coalesce the real edit into the
3497 /// step before it).
3498 fn renumber_here(&mut self) {
3499 self.renumber_at(self.caret);
3500 }
3501
3502 /// [`renumber_here`](Self::renumber_here) aimed somewhere other than the
3503 /// caret — for an edit that leaves the caret one past the item it just wrote,
3504 /// where twig resolves no list to renumber.
3505 fn renumber_at(&mut self, off: usize) {
3506 // The read-only gate — this door reaches twig without the splice.
3507 if self.read_only {
3508 return;
3509 }
3510 let before = self.source.clone();
3511 if self.editor.renumber_ordered_lists(off).is_err() {
3512 return; // not inside an ordered list — nothing to renumber
3513 }
3514 self.refresh();
3515 if self.source != before {
3516 let _ = self.editor.coalesce_last_undo();
3517 self.dirty = self.source != self.clean_source;
3518 self.clamp_caret();
3519 self.record_caret();
3520 }
3521 }
3522
3523 /// Repair the one trap a list edit can spring on itself. An *empty* `-`
3524 /// sub-item written directly beneath a text line reparses that text as a
3525 /// setext heading — `- hello\n - ` is `<h2>hello</h2>`, because a lone `-`
3526 /// is also a setext-H2 underline (twig is right; pandoc agrees). `*` and `+`
3527 /// bullets can't underline anything, so swap the dash for a `*`: the item
3528 /// stays an empty nested bullet, the parent stays prose, and the source
3529 /// round-trips instead of hiding a heading the user never asked for. Folded
3530 /// into the triggering edit's undo step, the way renumbering is.
3531 ///
3532 /// Gated on the collapse having actually happened (the swapped dash was
3533 /// swallowed into a `heading`), so a real setext heading the author wrote —
3534 /// or a `- x` with content, which can't underline anything — is never
3535 /// touched. This has to live in the *edit*, not the renderer: leaving the
3536 /// hazardous bytes on disk and only painting over them would ship a file
3537 /// every other CommonMark tool reads as a heading.
3538 ///
3539 /// This one keeps its own byte scan, and has to: the hazard is precisely
3540 /// that the dash stopped being a list marker, so [`list_marker_on_line`] —
3541 /// which asks twig which lines open an item — reports nothing here. There is
3542 /// no node to ask about. It is also the last Markdown spelling leaf writes on
3543 /// purpose rather than for want of an answer; once twig spells continuations
3544 /// itself, avoiding the trap becomes twig's, and this goes.
3545 ///
3546 /// [`list_marker_on_line`]: Self::list_marker_on_line
3547 fn avoid_setext_collapse(&mut self) {
3548 let caret = self.caret.min(self.source.len());
3549 let line_start = self.source[..caret].rfind('\n').map_or(0, |i| i + 1);
3550 let bytes = self.source.as_bytes();
3551 let mut dash = line_start;
3552 while matches!(bytes.get(dash), Some(b' ' | b'\t')) {
3553 dash += 1;
3554 }
3555 // A dash bullet is the only marker that doubles as a setext underline.
3556 if bytes.get(dash) != Some(&b'-') {
3557 return;
3558 }
3559 // Only an *empty* item is a bare underline; `- x` carries content and
3560 // can't fold the line above into a heading.
3561 let line_end = self.source[dash..]
3562 .find('\n')
3563 .map_or(self.source.len(), |i| dash + i);
3564 if !self.source[dash + 1..line_end].trim().is_empty() {
3565 return;
3566 }
3567 // The tell: that dash was swallowed into a `heading`. A properly nested
3568 // empty item sits under a `list_item`, with no heading in reach. Probe
3569 // the dash byte itself (well inside the heading), not the caret, whose
3570 // end-of-line offset can fall on the half-open span boundary.
3571 let collapsed = self
3572 .editor
3573 .ancestors_at(dash)
3574 .map(|c| c.into_iter().any(|m| m.kind == Kind::Heading))
3575 .unwrap_or(false);
3576 if !collapsed {
3577 return;
3578 }
3579 let caret = self.caret;
3580 if self.splice(dash, dash + 1, "*", EditKind::Other) {
3581 // Same width, so the caret keeps its column; fold into the edit that
3582 // triggered this so Tab stays one undo step.
3583 let _ = self.editor.coalesce_last_undo();
3584 self.caret = caret.min(self.source.len());
3585 self.clamp_caret();
3586 self.record_caret();
3587 }
3588 }
3589
3590 fn snapshot(&self) -> CaretState {
3591 CaretState {
3592 caret: self.caret,
3593 anchor: self.anchor,
3594 }
3595 }
3596
3597 /// Hand twig the current caret and selection as the blob for the live
3598 /// document state. Called before an edit — so the step twig retires records
3599 /// where the caret was, and undo can restore it — and again once the op has
3600 /// placed the caret, so redo restores where the edit left it.
3601 ///
3602 /// This is the whole of leaf's undo-caret bookkeeping now. twig carries the
3603 /// caret through its own history, so coalescing falls out for free (folding
3604 /// two twig steps into one drops the intermediate blob, keeping the run's
3605 /// first) and the parallel stacks that had to march in lockstep — and could
3606 /// silently drift out of it — are gone.
3607 fn record_caret(&mut self) {
3608 let _ = self.editor.set_caret_blob(&self.snapshot().to_blob());
3609 }
3610
3611 /// Toggle an inline mark over the selection (Bold / Italic / Code / …). Keeps
3612 /// the toggled region selected so a second press cleanly reverses it.
3613 pub fn toggle(&mut self, kind: InlineKind) {
3614 // The read-only gate — this door reaches twig without the splice.
3615 if self.read_only {
3616 return;
3617 }
3618 // Ahead of the no-selection branch below: arming a mark for text not yet
3619 // typed is a promise `insert` cannot keep in a format with no delimiters
3620 // to spell it with. Per *kind*, not per format — Markdown spells five
3621 // of the eight marks (highlight among them, under the `highlight`
3622 // extension leaf parses with), djot all eight, HTML seven.
3623 if self.refuse_unsupported(&format!("{kind:?}"), Gesture::ToggleInline(kind)) {
3624 return;
3625 }
3626 let Some((s, e)) = self.selection() else {
3627 // No selection: arm the mark for the next text typed here, the way a
3628 // word processor does. `⌘b`, type, `⌘b` again toggles bold on and off
3629 // in the flow of typing without ever selecting anything — the delta
3630 // is realised onto the freshly typed text by `insert`. A fresh caret
3631 // position starts the delta over from the marks actually in force.
3632 if self.pending_at != Some(self.caret) {
3633 self.pending_marks = InlineMarks::empty();
3634 self.pending_at = Some(self.caret);
3635 }
3636 self.pending_marks.flip(kind);
3637 self.status = None;
3638 return;
3639 };
3640 // Whitespace at the edge of a selection is not part of what was chosen —
3641 // a double-click takes the space after the word with it — and a mark
3642 // cannot close against one anyway: `**word **` is four literal asterisks
3643 // (the mark-edge rule, see `splice`). Mark the words, leave the spaces.
3644 let picked = &self.source[s..e];
3645 let (s, e) = (
3646 s + (picked.len() - picked.trim_start().len()),
3647 e - (picked.len() - picked.trim_end().len()),
3648 );
3649 if s >= e {
3650 self.status = Some(format!("{kind:?}: nothing selected to mark"));
3651 return;
3652 }
3653 // Styling a selection is a one-shot act, not a sticky mode.
3654 self.clear_pending();
3655 self.record_caret();
3656 match self.editor.toggle_inline(s, e, kind) {
3657 Ok(change) => {
3658 self.last_edit_kind = None; // structural edit is its own undo step
3659 self.refresh();
3660 self.anchor = Some(change.new.start);
3661 self.caret = change.new.end;
3662 self.dirty = self.source != self.clean_source;
3663 self.status = None;
3664 self.record_caret();
3665 }
3666 Err(e) => self.status = Some(format!("{kind:?}: {e}")),
3667 }
3668 }
3669
3670 /// Whether the caret stands in a highlight — what a frontend asks to enable
3671 /// or disable its highlight-colour controls, the way
3672 /// [`caret_in_table`](Self::caret_in_table) gates the grid ones.
3673 ///
3674 /// A fact about the *caret*, and the other half of
3675 /// [`Capabilities::mark_color`], which is the fact about the format. A
3676 /// frontend needs both: djot spells a highlight and no colour for it, so a
3677 /// caret standing in `{=word=}` answers `true` here and still has no palette
3678 /// to offer.
3679 ///
3680 /// The rule is [`active_inline_marks`](Self::active_inline_marks)' rule, so
3681 /// the palette appears exactly where the Highlight button is lit — with one
3682 /// deliberate exception: a mark *armed* at a bare caret and not yet typed
3683 /// into lights the button and answers `false` here, because there is no node
3684 /// to colour until the text exists.
3685 pub fn caret_in_mark(&mut self) -> bool {
3686 self.mark_offset().is_some()
3687 }
3688
3689 /// The offset [`set_mark_color`](Self::set_mark_color) speaks for — the one
3690 /// standing in the highlight the gesture means — or `None` when neither end
3691 /// of what is selected is in one.
3692 ///
3693 /// The caret first, and the selection's *start* after it, because of what
3694 /// [`toggle`](Self::toggle) leaves behind: a fresh `==word==` is selected
3695 /// whole, with the caret at its far edge, one past the closing `==` and so
3696 /// (by `marks_at`' half-open rule) not in the mark at all. Highlight a word
3697 /// and colour it — the two presses a coloured highlight is made of — would
3698 /// otherwise refuse on the second, having just written the highlight the
3699 /// author is pointing at.
3700 fn mark_offset(&mut self) -> Option<usize> {
3701 let in_mark = |d: &mut Self, off: usize| {
3702 d.marks_at(off)
3703 .into_iter()
3704 .any(|(k, _)| k == InlineKind::Mark)
3705 .then_some(off)
3706 };
3707 let caret = self.caret.min(self.source.len());
3708 in_mark(self, caret).or_else(|| {
3709 let start = self.selection()?.0;
3710 in_mark(self, start)
3711 })
3712 }
3713
3714 /// The colour of the highlight at the caret — `None` both when the caret is
3715 /// in no highlight and when the highlight it is in names no colour, which
3716 /// are the same answer to "which swatch is lit".
3717 ///
3718 /// The innermost mark, by span, for the same reason
3719 /// [`current_heading_level`](Self::current_heading_level) walks the tree:
3720 /// what the caret is *in* is the deepest node containing it. A `data-color`
3721 /// naming a colour this build has no variant for reads as `None` — the
3722 /// renderer already draws that as a plain highlight rather than guessing,
3723 /// and the toolbar agrees with the renderer.
3724 pub fn mark_color_at_caret(&mut self) -> Option<MarkColor> {
3725 let at = self.mark_offset()?;
3726 self.mark_color_at(at)
3727 }
3728
3729 /// [`mark_color_at_caret`](Self::mark_color_at_caret) at a given offset —
3730 /// the innermost `mark` covering it, and the colour it names.
3731 fn mark_color_at(&mut self, off: usize) -> Option<MarkColor> {
3732 self.nodes()
3733 .into_iter()
3734 .filter(|n| n.kind == Kind::Mark)
3735 .filter(|n| n.span.start <= off && off < n.span.end)
3736 .min_by_key(|n| n.span.end - n.span.start)
3737 .and_then(|n| MarkColor::from_attrs(&n.attrs))
3738 }
3739
3740 /// Colour the highlight at the caret, or clear its colour with `None` — the
3741 /// palette behind a toolbar's Highlight button.
3742 ///
3743 /// Markdown only, and the one gesture whose availability is a fact about the
3744 /// *parse extensions* rather than about the format alone: the colour is
3745 /// spelled `==🔴 text==`, an emoji twig reads back out of the content and
3746 /// records as the mark's `data-color`, and only an editor parsing with
3747 /// `highlight_colors` (which [`parse_extensions`] turns on for every leaf
3748 /// document) reads it back that way. Djot spells the highlight and no colour
3749 /// for it, so this refuses there — see [`Capabilities::mark_color`].
3750 ///
3751 /// **A colour is a property of a highlight that already exists.** There is
3752 /// no "highlight this in red" here, because that is two splices and would be
3753 /// two undo steps under one press; a frontend that wants it calls
3754 /// [`toggle`](Self::toggle) with [`InlineKind::Mark`] first, which is the
3755 /// order the two buttons already sit in. With no highlight at the caret this
3756 /// says so in the status line and writes nothing.
3757 ///
3758 /// The caret keeps its place in the *text*: the splice is entirely in the
3759 /// prefix between the opening `==` and the first word, so an offset past it
3760 /// rides the emoji's width, and one standing on the prefix itself lands
3761 /// where the prefix now ends.
3762 pub fn set_mark_color(&mut self, color: Option<MarkColor>) {
3763 // The read-only gate — this door reaches twig without the splice.
3764 if self.read_only {
3765 return;
3766 }
3767 if self.refuse_unsupported("highlight colour", Gesture::SetMarkColor) {
3768 return;
3769 }
3770 let Some(at) = self.mark_offset() else {
3771 self.status = Some("highlight colour: no highlight at the caret".into());
3772 return;
3773 };
3774 // Clearing a colour a highlight hasn't got is twig's one *successful*
3775 // no-op, and the `Change` it hands back then describes whatever edit came
3776 // before it — a stale span that would drag the caret somewhere it never
3777 // was. Answer it here, where the question is cheap, rather than trusting
3778 // a change that isn't one.
3779 if color.is_none() && self.mark_color_at(at).is_none() {
3780 self.status = None;
3781 return;
3782 }
3783 self.record_caret();
3784 match self.editor.set_mark_color(at, color.map(twig_mark_color)) {
3785 Ok(change) => {
3786 // Re-anchored from the offsets as they were, *before* `refresh`
3787 // sees the new bytes: the caret it clamps is one standing inside
3788 // a prefix that didn't exist a moment ago, and walking it back to
3789 // a char boundary of the emoji loses the place this is restoring.
3790 let caret = reanchor(self.caret, &change);
3791 let anchor = self.anchor.map(|a| reanchor(a, &change));
3792 self.last_edit_kind = None; // structural edit is its own undo step
3793 self.refresh();
3794 self.caret = caret;
3795 self.anchor = anchor;
3796 self.dirty = self.source != self.clean_source;
3797 self.status = None;
3798 self.clamp_caret();
3799 self.record_caret();
3800 }
3801 Err(e) => self.status = Some(format!("highlight colour: {e}")),
3802 }
3803 }
3804
3805 /// One press of a colour swatch: colour the highlight at the caret, or —
3806 /// over a selection that isn't highlighted yet — highlight it and colour it,
3807 /// as **one** undo step.
3808 ///
3809 /// [`set_mark_color`](Self::set_mark_color) is the exact gesture and stays
3810 /// one splice; this is the compound every toolbar actually presses, and it
3811 /// lives here rather than in each frontend because the rule it encodes —
3812 /// what a swatch means when there is no highlight under it yet — is one
3813 /// answer, not one per frontend. The two splices are folded into a single
3814 /// history step, so the press that made a red highlight is taken back by a
3815 /// single undo rather than leaving an uncoloured one behind.
3816 ///
3817 /// `None` clears the colour, and over an unhighlighted selection means
3818 /// simply "highlight this" — the same thing the Highlight button does.
3819 /// A bare caret in no highlight is left alone with a status line, because
3820 /// [`toggle`](Self::toggle) there arms a mark for text not yet typed and a
3821 /// colour cannot be armed with it.
3822 pub fn highlight(&mut self, color: Option<MarkColor>) {
3823 if self.caret_in_mark() || self.selection().is_none() {
3824 self.set_mark_color(color);
3825 return;
3826 }
3827 self.toggle(InlineKind::Mark);
3828 // The format may not spell a highlight at all (`toggle` said so), and
3829 // there is nothing to colour if it doesn't.
3830 if self.status.is_some() {
3831 return;
3832 }
3833 let before = self.revision;
3834 self.set_mark_color(color);
3835 // Only fold when the colour really spliced. `highlight(None)` over a
3836 // fresh highlight is a no-op by design, and coalescing there would eat
3837 // the *previous* edit into the toggle instead.
3838 if self.revision != before {
3839 let _ = self.editor.coalesce_last_undo();
3840 }
3841 }
3842
3843 /// Convert the block at the caret to a heading level or paragraph.
3844 pub fn set_block(&mut self, kind: BlockKind) {
3845 // The read-only gate — this door reaches twig without the splice.
3846 if self.read_only {
3847 return;
3848 }
3849 if self.refuse_unsupported(&format!("{kind:?}"), Gesture::SetBlock) {
3850 return;
3851 }
3852 self.record_caret();
3853 // A blank line has no node to convert, and twig opens a block there
3854 // rather than declining — so the caret's own offset is the right thing
3855 // to hand it when `block_offset_for_caret` finds nothing.
3856 let offset = self.block_offset_for_caret().unwrap_or(self.caret);
3857 match self.editor.set_block(offset, kind) {
3858 Ok(change) => {
3859 self.last_edit_kind = None;
3860 self.refresh();
3861 // Opening a block on a blank line writes a marker the caret
3862 // belongs *after*; converting an existing one moves nothing.
3863 self.caret = self.caret.max(change.new.end);
3864 self.clamp_caret();
3865 self.anchor = None;
3866 self.dirty = self.source != self.clean_source;
3867 self.status = None;
3868 self.record_caret();
3869 }
3870 Err(e) => self.status = Some(format!("{kind:?}: {e}")),
3871 }
3872 }
3873
3874 /// Whether `off` is inside a text block (paragraph, heading, code block…).
3875 fn has_block_at(&mut self, off: usize) -> bool {
3876 self.editor.ancestors_at(off).ok().is_some_and(|chain| {
3877 chain
3878 .iter()
3879 .any(|m| !wysiwyg::is_inline_kind(&m.kind) && !is_block_container(&m.kind))
3880 })
3881 }
3882
3883 /// The offset to hand twig's `set_block`: the caret when it is already inside
3884 /// a block, otherwise nudged onto the previous character (a caret at a line
3885 /// end sits at the doc level, outside the block). `None` when the caret is on
3886 /// a blank line — a new paragraph with no block node to convert.
3887 fn block_offset_for_caret(&mut self) -> Option<usize> {
3888 let caret = self.caret.min(self.source.len());
3889 if self.has_block_at(caret) {
3890 return Some(caret);
3891 }
3892 // Nudge to the previous character — but never across a newline: that would
3893 // target the previous block, and a blank line genuinely has no block.
3894 if let Some((i, ch)) = self.source[..caret].char_indices().next_back()
3895 && ch != '\n'
3896 && self.has_block_at(i)
3897 {
3898 return Some(i);
3899 }
3900 None
3901 }
3902
3903 /// The heading level of the text block at the caret, or `None` when that
3904 /// block is not a heading.
3905 pub fn current_heading_level(&mut self) -> Option<u32> {
3906 let caret = self.caret;
3907 self.nodes()
3908 .into_iter()
3909 .filter(|n| n.kind == Kind::Heading)
3910 .find(|n| n.span.start <= caret && caret <= n.span.end)
3911 .and_then(|n| n.level)
3912 }
3913
3914 /// The inline marks in force at the caret (or over the selection) — what a
3915 /// toolbar draws lit, and the block-level [`Doc::current_heading_level`]'s
3916 /// inline counterpart. Cheap enough to call every frame: one twig
3917 /// `ancestors_at` query per caret (two with a selection), each walking root
3918 /// → deepest node at one offset. It never snapshots the tree the way
3919 /// `current_heading_level` does, and the returned set is a `Copy` bitset, so
3920 /// the only allocation is twig's own small ancestor `Vec`.
3921 ///
3922 /// **A selection reports a mark only when the mark covers *all* of it.**
3923 /// That's what every real toolbar means by an active button — Bold lit over
3924 /// a half-bold selection would claim a press turns bold *off*, when
3925 /// [`Doc::toggle`] hands the range to twig and gets the whole thing bolded.
3926 /// Whole-coverage is asked as "is the same mark node standing over both the
3927 /// first and the last character?": inline nodes are contiguous, so one node
3928 /// covering both ends covers every byte between them. Two touching runs
3929 /// (`**a****b**`) are two nodes, and correctly light nothing.
3930 ///
3931 /// At a bare caret a mark is active when the caret stands inside the mark's
3932 /// span — `span.start <= caret < span.end`, delimiters included, which is
3933 /// what makes the boundaries behave. In `a **bold** b` the offsets from the
3934 /// opening `*` (2) through the last byte of the closing `**` (9) are all
3935 /// bold, so the WYSIWYG caret both before `b` and after `d` (the delimiters
3936 /// are hidden, and those offsets are 4 and 8) reports bold — matching where
3937 /// typing would actually land inside the marked run. The offset one past the
3938 /// mark (10) is the text after it and reports nothing, at the end of the
3939 /// buffer exactly as in the middle.
3940 pub fn active_inline_marks(&mut self) -> InlineMarks {
3941 let Some((start, end)) = self.selection() else {
3942 // The marks actually in force at the caret, flipped by any armed
3943 // sticky delta — so `⌘b` at a bare caret lights the Bold button
3944 // immediately, before a single character is typed.
3945 let base: InlineMarks = self
3946 .marks_at(self.caret)
3947 .into_iter()
3948 .map(|(k, _)| k)
3949 .collect();
3950 return base.xor(self.pending_here());
3951 };
3952 // The selection's *last character*, not its exclusive end: `end` is the
3953 // offset one past the selection, which for a selection ending exactly at
3954 // a mark's close is already outside it (`[4,10)` of `a **bold** b` is
3955 // entirely bold, but offset 10 is the space after).
3956 let last = prev_boundary(&self.source, end);
3957 let head = self.marks_at(start);
3958 let tail = self.marks_at(last);
3959 head.into_iter()
3960 .filter(|m| tail.contains(m))
3961 .map(|(k, _)| k)
3962 .collect()
3963 }
3964
3965 /// The inline marks whose span covers `off`, each with the id of the node
3966 /// carrying it — the id is what lets a selection tell one mark node from
3967 /// another of the same kind.
3968 fn marks_at(&mut self, off: usize) -> Vec<(InlineKind, u32)> {
3969 let off = off.min(self.source.len());
3970 self.editor
3971 .ancestors_at(off)
3972 .unwrap_or_default()
3973 .into_iter()
3974 // `span.end` is the offset one *past* the mark, so it isn't in it.
3975 // twig already resolves a boundary to whatever starts there — in
3976 // `**bold** x` offset 8 is the following text, not the strong — but
3977 // when nothing follows, the tie has nobody to break for and the
3978 // chain still ends at the mark. That would make the answer at the
3979 // last offset of the document depend on whether the file happens to
3980 // end in a newline; the rule is `span.start <= off < span.end`, and
3981 // it's the same rule at the end of a buffer as in the middle.
3982 .filter(|m| off < m.span.end)
3983 .filter_map(|m| inline_kind(&m.kind).map(|k| (k, m.node_id)))
3984 .collect()
3985 }
3986
3987 /// Toggle a heading at the caret: if the block is already this heading level,
3988 /// revert it to a paragraph; otherwise convert it to this heading level.
3989 /// This gives the heading commands the same toggle feel as bold/italic/code —
3990 /// re-applying a heading a line already has turns it back into body text.
3991 pub fn toggle_heading(&mut self, level: u32) {
3992 if self.current_heading_level() == Some(level) {
3993 self.set_block(BlockKind::Paragraph);
3994 } else {
3995 self.set_block(BlockKind::Heading(level));
3996 }
3997 }
3998
3999 /// Toggle a block quote around the selection, or around the block at the
4000 /// caret — the toolbar's Quote button.
4001 pub fn toggle_blockquote(&mut self) {
4002 self.toggle_container(BlockContainerKind::BlockQuote);
4003 }
4004
4005 /// Toggle a numbered (`ordered`) or bulleted list over the selection, or
4006 /// over the block at the caret — one op with the kind as a flag, the way
4007 /// `toggle_heading` takes its level, so a frontend needs no twig type to
4008 /// name the two buttons.
4009 ///
4010 /// Pressing the *other* list's button while in a list converts in place
4011 /// rather than nesting, so the pair reads as one three-state control
4012 /// (bulleted / numbered / neither) rather than two independent wrappers.
4013 pub fn toggle_list(&mut self, ordered: bool) {
4014 self.toggle_container(if ordered {
4015 BlockContainerKind::OrderedList
4016 } else {
4017 BlockContainerKind::BulletList
4018 });
4019 }
4020
4021 // ── Task list items ──────────────────────────────────────────────────────
4022 // The checkbox in `- [x] done`. twig owns all three gestures: the box is
4023 // inline content of the item's first paragraph rather than part of its
4024 // marker, so adding or removing one must leave the item's continuation
4025 // indentation alone, and an item inside a quote is found past the quote
4026 // markers. leaf names the gesture and the offset; the spelling is twig's.
4027
4028 /// Whether the list item at the caret carries a checkbox, and which way it
4029 /// faces — `Some(true)` ticked, `Some(false)` empty, `None` for a plain list
4030 /// item or no item at all. What a toolbar reads to light its checkbox button.
4031 pub fn task_checked_at_caret(&mut self) -> Option<bool> {
4032 self.task_checked_at(self.caret)
4033 }
4034
4035 /// [`task_checked_at_caret`](Self::task_checked_at_caret) for an arbitrary
4036 /// offset — what a frontend asks before deciding a click landed on a box.
4037 pub fn task_checked_at(&mut self, offset: usize) -> Option<bool> {
4038 self.innermost_list_item(offset.min(self.source.len()))?
4039 .checked
4040 }
4041
4042 /// Tick or untick the task item at the caret (the checkbox's keyboard half).
4043 /// A no-op with a reported reason when the caret is in no task item — minting
4044 /// a box here is [`toggle_task_item`](Self::toggle_task_item)'s job.
4045 pub fn toggle_task_checked(&mut self) {
4046 self.toggle_task_at(self.caret);
4047 }
4048
4049 /// Tick or untick the task item covering `offset` — what a *click* on a
4050 /// rendered checkbox is. Separate from the caret form because a click carries
4051 /// its own offset and must not first move the caret there: ticking a box
4052 /// three paragraphs away should not take the cursor with it.
4053 pub fn toggle_task_at(&mut self, offset: usize) {
4054 // The read-only gate — this door reaches twig without the splice.
4055 if self.read_only {
4056 return;
4057 }
4058 if self.refuse_unsupported("task", Gesture::ToggleTaskChecked) {
4059 return;
4060 }
4061 let offset = offset.min(self.source.len());
4062 self.record_caret();
4063 match self.editor.toggle_task_checked(offset) {
4064 Ok(_) => self.after_task_edit(),
4065 Err(e) => self.status = Some(format!("task: {e}")),
4066 }
4067 }
4068
4069 /// Give the list item at the caret a checkbox, or take its checkbox away —
4070 /// the gesture that converts between a plain bullet and a task. A new box
4071 /// arrives unticked.
4072 pub fn toggle_task_item(&mut self) {
4073 // The read-only gate — this door reaches twig without the splice.
4074 if self.read_only {
4075 return;
4076 }
4077 if self.refuse_unsupported("task", Gesture::ToggleTaskItem) {
4078 return;
4079 }
4080 let caret = self.caret.min(self.source.len());
4081 self.record_caret();
4082 match self.editor.toggle_task_item(caret) {
4083 Ok(_) => self.after_task_edit(),
4084 Err(e) => self.status = Some(format!("task: {e}")),
4085 }
4086 }
4087
4088 /// Settle after a task gesture. The caret rides its old byte offset and is
4089 /// clamped back in: a box is three or four bytes on the item's first line, so
4090 /// text after it shifts by that much at most, and `clamp_caret` lands it on a
4091 /// real stop either way.
4092 fn after_task_edit(&mut self) {
4093 self.last_edit_kind = None;
4094 self.refresh();
4095 self.anchor = None;
4096 self.dirty = self.source != self.clean_source;
4097 self.status = None;
4098 self.clamp_caret();
4099 self.record_caret();
4100 }
4101
4102 // ── Tables ───────────────────────────────────────────────────────────────
4103 // A table is a grid, and twig edits it as one — add/remove/move a row or
4104 // column, set a column's alignment — re-spelling the whole table in a single
4105 // splice. Every gesture is anchored at the caret's cell. leaf just names the
4106 // gesture and re-reads the result; the whole table's numbering, borders, and
4107 // delimiter are twig's to keep straight.
4108
4109 /// Whether the caret is inside a table — what a frontend asks to enable or
4110 /// disable its table controls.
4111 ///
4112 /// An HTML `<table>` still answers `true`: the caret really is in a table,
4113 /// and the reason the grid controls stay dark there is
4114 /// [`Capabilities::table`], which is a fact about the document's format
4115 /// rather than about the caret. A frontend needs both.
4116 pub fn caret_in_table(&mut self) -> bool {
4117 let caret = self.caret.min(self.source.len());
4118 self.editor
4119 .ancestors_at(caret)
4120 .map(|c| c.into_iter().any(|m| m.kind == Kind::Table))
4121 .unwrap_or(false)
4122 }
4123
4124 /// One grid op, guarded and settled — the shared body of the seven below.
4125 ///
4126 /// The guard is why this exists rather than seven copies of the same three
4127 /// lines, and it is the one guard leaf cannot delegate to twig. The table
4128 /// editor is the gesture family that consults no `Syntax` table (it spells a
4129 /// grid, not a delimiter) and therefore the one twig's `Format::supports`
4130 /// deliberately has no variant for: handed an HTML `<table>` it rebuilds the
4131 /// grid as a *pipe table* and reports success, swapping the element out for
4132 /// `| a | b |` and taking the rest of the document's markup with it. Nothing
4133 /// downstream could tell that from a successful edit — the splice is real,
4134 /// the reparse succeeds, `dirty` is honest — which is what makes it worth
4135 /// stopping at the door rather than detecting after the fact. See
4136 /// [`spells_pipe_tables`].
4137 fn table_op(
4138 &mut self,
4139 what: &str,
4140 op: impl FnOnce(&mut Editor, usize) -> Result<(), twig::Error>,
4141 ) {
4142 if self.refuse_unless(what, spells_pipe_tables(self.format)) {
4143 return;
4144 }
4145 self.record_caret();
4146 let at = self.caret;
4147 let r = op(&mut self.editor, at);
4148 self.apply_table(r, what);
4149 }
4150
4151 /// Insert an empty row below (`below`) or above the caret's row.
4152 pub fn table_insert_row(&mut self, below: bool) {
4153 self.table_op("table row", |e, at| e.table_insert_row(at, below));
4154 }
4155
4156 /// Delete the caret's row (not the header, not the last body row).
4157 pub fn table_delete_row(&mut self) {
4158 self.table_op("table row", |e, at| e.table_delete_row(at));
4159 }
4160
4161 /// Insert an empty column right (`right`) or left of the caret's column.
4162 pub fn table_insert_column(&mut self, right: bool) {
4163 self.table_op("table column", |e, at| e.table_insert_column(at, right));
4164 }
4165
4166 /// Delete the caret's column (unless it is the only one).
4167 pub fn table_delete_column(&mut self) {
4168 self.table_op("table column", |e, at| e.table_delete_column(at));
4169 }
4170
4171 /// Set the caret's column to `alignment`.
4172 pub fn table_set_alignment(&mut self, alignment: Alignment) {
4173 self.table_op("table alignment", |e, at| {
4174 e.table_set_alignment(at, alignment)
4175 });
4176 }
4177
4178 /// Move the caret's row one place down (`down`) or up, within the body rows.
4179 pub fn table_move_row(&mut self, down: bool) {
4180 self.table_op("table row", |e, at| e.table_move_row(at, down));
4181 }
4182
4183 /// Move the caret's column one place right (`right`) or left.
4184 pub fn table_move_column(&mut self, right: bool) {
4185 self.table_op("table column", |e, at| e.table_move_column(at, right));
4186 }
4187
4188 /// Settle the caret and document flags after a table op (or report its
4189 /// error). twig re-spells the whole table, so the caret rides its old byte
4190 /// offset and is clamped back into the rebuilt bytes — near enough to where
4191 /// it was, since the op preserves the cells' content and order around it.
4192 fn apply_table(&mut self, result: Result<(), twig::Error>, what: &str) {
4193 match result {
4194 Ok(()) => {
4195 self.last_edit_kind = None;
4196 self.refresh();
4197 self.anchor = None;
4198 self.clamp_caret();
4199 self.dirty = self.source != self.clean_source;
4200 self.status = None;
4201 self.record_caret();
4202 }
4203 Err(e) => self.status = Some(format!("{what}: {e}")),
4204 }
4205 }
4206
4207 /// One `toggle_block_container` over the block-level target.
4208 ///
4209 /// leaf says *where*; twig decides everything else — which blocks the range
4210 /// covers, whether that means wrapping, unwrapping, nesting or converting,
4211 /// and how this document's format spells the prefix. The rule that a
4212 /// container only comes off when the range covers every block it holds is
4213 /// what the re-anchoring below is built around.
4214 fn toggle_container(&mut self, kind: BlockContainerKind) {
4215 // The read-only gate — this door reaches twig without the splice.
4216 if self.read_only {
4217 return;
4218 }
4219 if self.refuse_unsupported(&format!("{kind:?}"), Gesture::ToggleBlockContainer(kind)) {
4220 return;
4221 }
4222 let selected = self.selection();
4223 // A blank line holds no block, and twig opens an *empty* container on one
4224 // — since 3.2.0; it used to decline the range with `NotFound`, which is
4225 // why this used to lend it a scratch paragraph to wrap. Worth knowing
4226 // here because the line-for-line caret mapping below cannot describe it:
4227 // opening one under a paragraph writes the blank line the format needs
4228 // above the marker too, so the rewritten region has a line the old one
4229 // didn't, and "the same line, the same distance from its end" lands on
4230 // that new blank instead of in the container.
4231 let opened_empty = selected.is_none() && self.block_offset_for_caret().is_none();
4232 // Without a selection the target is the caret's own block, resolved the
4233 // way `set_block` resolves it — a caret at a line end sits at the doc
4234 // level and has to be nudged back onto the block it looks like it's in.
4235 // An empty range is enough: twig widens to the whole lines it touches.
4236 let (start, end) = match selected {
4237 Some(range) => range,
4238 None => {
4239 let off = self.block_offset_for_caret().unwrap_or(self.caret);
4240 (off, off)
4241 }
4242 };
4243 self.record_caret();
4244 match self.editor.toggle_block_container(start, end, kind) {
4245 Ok(change) => {
4246 // Read the caret's place out of the *pre-edit* source, before
4247 // `refresh` swaps that source out from under it.
4248 let place = (selected.is_none() && !opened_empty)
4249 .then(|| self.caret_line_tail(&change.old));
4250 self.last_edit_kind = None; // structural edit is its own undo step
4251 self.refresh();
4252 match place {
4253 // Both land the caret at the far end of what twig wrote, and
4254 // differ only in what they leave selected.
4255 //
4256 // From a selection: select what the container now holds, the
4257 // way `toggle` keeps its marked region selected — and for a
4258 // stronger reason than symmetry: a container comes *off* only
4259 // a range covering every block it holds, so a selection left
4260 // on its old bytes (now short by a prefix per line) would nest
4261 // on the second press instead of reversing the first.
4262 //
4263 // From a blank line: nothing to select, and the end of the
4264 // region is exactly past the bare `> ` / `- ` twig wrote —
4265 // the caret standing inside the container that was asked for.
4266 None => {
4267 self.anchor = (!opened_empty).then_some(change.new.start);
4268 self.caret = change.new.end;
4269 }
4270 Some(place) => {
4271 self.anchor = None;
4272 self.caret = self.line_tail_offset(&change.new, place);
4273 }
4274 }
4275 self.dirty = self.source != self.clean_source;
4276 self.status = None;
4277 self.clamp_caret();
4278 self.record_caret();
4279 }
4280 Err(e) => self.status = Some(format!("{kind:?}: {e}")),
4281 }
4282 }
4283
4284 /// The caret's place inside the region a container toggle is rewriting, in
4285 /// the only terms the rewrite preserves: which of the region's lines it sits
4286 /// on, and how many bytes of that line lie ahead of it.
4287 ///
4288 /// A container's markup goes in at column 0 and never touches what follows
4289 /// on the line, so that pair survives the edit exactly where a byte offset
4290 /// does not — a caret left on its old offset slides back by one prefix per
4291 /// line above it, which on a hard-wrapped paragraph parks it *inside* the
4292 /// `> ` it just asked for.
4293 fn caret_line_tail(&self, old: &std::ops::Range<usize>) -> (usize, usize) {
4294 let caret = self.caret.clamp(old.start, old.end);
4295 let line = self.source[old.start..caret].matches('\n').count();
4296 let end = self.source[caret..old.end]
4297 .find('\n')
4298 .map_or(old.end, |i| caret + i);
4299 (line, end - caret)
4300 }
4301
4302 /// [`caret_line_tail`](Self::caret_line_tail) undone against the rewritten
4303 /// region: the offset `tail` bytes back from the end of the region's `line`.
4304 ///
4305 /// Both walks are clamped rather than trusted, because the one op that does
4306 /// *not* keep a region's lines one-to-one is stripping a list — twig blows
4307 /// the items back apart with blank lines between them — and a caret landing
4308 /// on the nearest line of the right item beats one landing out of the region
4309 /// entirely.
4310 fn line_tail_offset(
4311 &self,
4312 new: &std::ops::Range<usize>,
4313 (line, tail): (usize, usize),
4314 ) -> usize {
4315 let region = &self.source[new.start.min(self.source.len())..new.end.min(self.source.len())];
4316 let mut start = 0;
4317 for _ in 0..line {
4318 match region[start..].find('\n') {
4319 Some(i) => start += i + 1,
4320 None => break,
4321 }
4322 }
4323 let end = region[start..]
4324 .find('\n')
4325 .map_or(region.len(), |i| start + i);
4326 new.start + end.saturating_sub(tail).max(start)
4327 }
4328
4329 /// Link the selection to `destination` — the toolbar's Link button. With no
4330 /// selection it acts at the caret, which re-points a link the caret is
4331 /// already standing in (twig replaces an existing link's destination and
4332 /// keeps its text) and otherwise spells a link that has no text of its own:
4333 /// an autolink (`<https://x.dev>`) where the destination is one, and
4334 /// `[destination](destination)` where it isn't.
4335 ///
4336 /// `destination` reaches twig raw. Escaping it is format knowledge and the
4337 /// two formats genuinely disagree — Markdown ends a destination at the first
4338 /// space and moves it into `<…>`, djot reads that `<…>` as part of the URL
4339 /// itself — so the side holding the document is the side that gets to spell
4340 /// it. A destination twig can't carry at all (one with a newline) comes back
4341 /// as an error rather than a quietly rewritten URL.
4342 pub fn insert_link(&mut self, destination: &str) {
4343 if self.read_only || self.refuse_unsupported("link", Gesture::InsertLink) {
4344 return;
4345 }
4346 let (start, end) = self.selection().unwrap_or((self.caret, self.caret));
4347 self.record_caret();
4348 match self.editor.insert_link(start, end, destination) {
4349 Ok(change) => {
4350 self.last_edit_kind = None;
4351 self.refresh();
4352 match self.link_text_span(change.new.start) {
4353 // A link with text of its own: select it, so typing replaces
4354 // a `[dest](dest)`'s stand-in label and a second press
4355 // re-points what the first one linked.
4356 Some(text) => {
4357 self.anchor = (text.start != text.end).then_some(text.start);
4358 self.caret = text.end;
4359 }
4360 // An autolink is finished the moment it's written — its text
4361 // *is* the URL. Leaving it selected would aim the next press
4362 // at the one shape twig still wraps instead of re-points.
4363 None => {
4364 self.anchor = None;
4365 self.caret = change.new.end;
4366 }
4367 }
4368 self.dirty = self.source != self.clean_source;
4369 self.status = None;
4370 self.clamp_caret();
4371 self.record_caret();
4372 }
4373 Err(e) => self.status = Some(format!("link: {e}")),
4374 }
4375 }
4376
4377 /// Insert a block-level image at the caret: ``. Any
4378 /// selection becomes the alt text (so "select a caption, insert image" labels
4379 /// it); with no selection, `alt` is used — empty for none. The caret lands
4380 /// just past the inserted image.
4381 ///
4382 /// Both halves go through twig (`insert_literal` for the alt text,
4383 /// `insert_image` for the image), so neither is spelled here. That used to be a
4384 /// `format!`, and it was wrong the first time an app inserted a real filename:
4385 /// Markdown ends a destination at the first space, so `` is
4386 /// not an image at all — and the fix is per-format, since moving into the
4387 /// `<…>` form is exactly wrong for Djot, where `<…>` becomes the URL itself.
4388 pub fn insert_image(&mut self, destination: &str, alt: &str) {
4389 if self.read_only || self.refuse_unsupported("image", Gesture::InsertImage) {
4390 return;
4391 }
4392 let (start, end) = self.selection().unwrap_or((self.caret, self.caret));
4393 self.record_caret();
4394 // With no selection and an explicit `alt`, the alt text has to exist in the
4395 // document before it can be the image's — and it is raw caller input, so
4396 // it goes in through `insert_literal`, which escapes it for the format
4397 // rather than letting a `]` in someone's caption close the image early.
4398 let (start, end) = if start == end && !alt.is_empty() {
4399 match self.editor.insert_literal(start, alt) {
4400 Ok(change) => (change.new.start, change.new.end),
4401 Err(e) => {
4402 self.status = Some(format!("image: {e}"));
4403 return;
4404 }
4405 }
4406 } else {
4407 (start, end)
4408 };
4409 match self.editor.insert_image(start, end, destination) {
4410 Ok(change) => {
4411 self.last_edit_kind = None;
4412 self.refresh();
4413 // Just past the image, nothing selected — where a caret belongs
4414 // after inserting one.
4415 self.anchor = None;
4416 self.caret = change.new.end;
4417 self.dirty = self.source != self.clean_source;
4418 self.status = None;
4419 self.clamp_caret();
4420 self.record_caret();
4421 }
4422 Err(e) => self.status = Some(format!("image: {e}")),
4423 }
4424 }
4425
4426 /// Insert a block-level image, video, or audio at the caret. The image case
4427 /// is [`insert_image`](Self::insert_image); video and audio are spelled as
4428 /// HTML elements, which is the only spelling Markdown and Djot have for them:
4429 ///
4430 /// ```text
4431 /// <video src="clip.mp4" controls>alt</video>
4432 /// <audio src="take.mp3" controls>alt</audio>
4433 /// ```
4434 ///
4435 /// HTML rather than a `::video{…}` directive deliberately. A directive means
4436 /// something only to an app that knows the vocabulary, so the document would
4437 /// read as literal punctuation everywhere else; `<video>` is what every other
4438 /// renderer already understands, and what leaf's own reader picks back up
4439 /// through `html_elements` promotion (see [`parse_extensions`]).
4440 ///
4441 /// The one-line spelling needs twig ≥ 2.5.1, which widened CommonMark's
4442 /// HTML-block tag list to cover `<video>`/`<audio>`/`<picture>` under
4443 /// `html_elements`. Before that only the multi-line form parsed as a block at
4444 /// all, and this wrote three lines to work around it.
4445 ///
4446 /// `controls` is always written: a player with no transport is a still frame
4447 /// the reader can't do anything with. Any selection becomes the element's
4448 /// fallback text, exactly as it becomes an image's alt.
4449 ///
4450 /// The same verbatim-insertion caveat as [`insert_image`](Self::insert_image)
4451 /// applies, and bites harder here: a `"` in `destination` closes the
4452 /// attribute. A frontend taking these from a file picker is fine; one taking
4453 /// them from free text should keep them tame.
4454 ///
4455 /// [`MediaInfo`]: crate::MediaInfo
4456 pub fn insert_media(&mut self, kind: MediaKind, destination: &str, alt: &str) {
4457 if kind == MediaKind::Image {
4458 return self.insert_image(destination, alt);
4459 }
4460 // Gated on the *image* gesture, not on one of its own — there isn't one,
4461 // since the bytes below are spelled here rather than by twig, and an HTML
4462 // document would in fact parse them. The button is one control with three
4463 // kinds behind it, and two of them working in a format where the third
4464 // cannot is a worse surface than three that agree — especially as
4465 // `insert_image` is the kind anyone reaches for first.
4466 if self.refuse_unsupported("media", Gesture::InsertImage) {
4467 return;
4468 }
4469 let (start, end) = self.selection().unwrap_or((self.caret, self.caret));
4470 let alt_text = self
4471 .selected_text()
4472 .map(str::to_string)
4473 .unwrap_or_else(|| alt.to_string());
4474 let tag = match kind {
4475 MediaKind::Audio => "audio",
4476 _ => "video",
4477 };
4478 let markup = format!("<{tag} src=\"{destination}\" controls>{alt_text}</{tag}>");
4479 self.edit(start, end, &markup);
4480 }
4481
4482 /// Insert a thematic break at the caret — the toolbar's Horizontal Rule
4483 /// button. Spelling and placement are both twig's; leaf used to write `---`
4484 /// itself, which was the Markdown spelling in a djot document too.
4485 ///
4486 /// A rule is a block, so `insert_thematic_break` alone has nowhere to put one
4487 /// mid-paragraph and lands it after the caret's whole block. To get a rule
4488 /// *at* the caret — the paragraph parted in two around it, which is what a
4489 /// rule button is understood to do — the paragraph is first divided with
4490 /// `split_block` and the rule then aimed at the **first** half. Aiming it at
4491 /// the offset `split_block` returns puts the rule after the *second* half
4492 /// instead, which is a rule in the right document and the wrong place.
4493 ///
4494 /// Only a plain paragraph is split, and only where there is something to
4495 /// part: at the paragraph's end the split has no second half to mint and
4496 /// would write the separator anyway — a blank line and the empty slot Enter
4497 /// leaves for the next paragraph, which the rule then lands above and
4498 /// nothing fills — so there the rule goes straight after the paragraph,
4499 /// which is where the split-and-aim was sending it regardless. At the
4500 /// paragraph's *start* the split is kept, though it parts nothing either:
4501 /// `|para` becomes `\npara` with the caret on the new blank line, and a
4502 /// rule aimed at a blank line is written on it (twig ≥ 3.5.2), which is how
4503 /// "before the paragraph" is said through a gesture that only knows
4504 /// "after" — `---\n\npara`, and `prev\n\n---\n\npara` mid-document. Everywhere
4505 /// else the rule simply lands after the block, which is both twig's own
4506 /// answer and the better one: splitting a fenced code block would leave two
4507 /// fences with a rule between them, and splitting a list item would mint an
4508 /// item nobody asked for on the way to a rule that lands after the list
4509 /// regardless. A table and a setext heading refuse the split outright, so
4510 /// they take the same path by themselves.
4511 pub fn insert_thematic_break(&mut self) {
4512 if self.read_only || self.refuse_unsupported("thematic break", Gesture::InsertThematicBreak)
4513 {
4514 return;
4515 }
4516 self.caret = self.skip_trailing_close_delims(self.caret);
4517 // A selection is replaced by the rule, so collapse it first and let the
4518 // split-and-rule below run from the caret it leaves behind.
4519 if let Some((s, e)) = self.selection() {
4520 self.splice(s, e, "", EditKind::Other);
4521 }
4522 self.anchor = None;
4523 self.record_caret();
4524 let at = self.caret;
4525 if self.caret_parts_bare_paragraph() {
4526 // A failure here is not fatal: the rule still lands after the block,
4527 // which is exactly what this call was trying to improve on.
4528 let _ = self.editor.split_block(at);
4529 }
4530 match self.editor.insert_thematic_break(at) {
4531 Ok(change) => {
4532 self.last_edit_kind = None;
4533 self.refresh();
4534 self.anchor = None;
4535 self.caret = change.new.end;
4536 self.dirty = self.source != self.clean_source;
4537 self.status = None;
4538 self.clamp_caret();
4539 self.record_caret();
4540 }
4541 Err(e) => self.status = Some(format!("thematic break: {e}")),
4542 }
4543 }
4544
4545 /// Insert a fresh table at the caret — the toolbar's Table button. One
4546 /// header row, `rows` empty body rows, `cols` columns, spelled by twig in
4547 /// the document's own dialect and placed the way its thematic break is:
4548 /// after the caret's block, blank-separated. A bare paragraph is parted
4549 /// around the caret first, exactly as
4550 /// [`insert_thematic_break`](Self::insert_thematic_break) parts it, so the
4551 /// table lands *at* the caret rather than after everything the caret's
4552 /// paragraph says.
4553 ///
4554 /// The caret ends in the first header cell, selected the way Tab selects
4555 /// a cell — the natural next act is to type the heading, and Tab then
4556 /// walks the grid. That cell is read back from the rebuilt table map
4557 /// rather than computed from the splice, because twig's blank line and
4558 /// quote prefix put the first bar at an offset only the reparse knows.
4559 ///
4560 /// The shape is the caller's: a menu offers a few, a dialog asks. Zero
4561 /// rows or columns is twig's refusal (a header with nothing under it is
4562 /// what its row delete refuses to leave), reported through `status`.
4563 pub fn insert_table(&mut self, rows: usize, cols: usize) {
4564 if self.read_only || self.refuse_unsupported("table", Gesture::InsertTable) {
4565 return;
4566 }
4567 self.caret = self.skip_trailing_close_delims(self.caret);
4568 if let Some((s, e)) = self.selection() {
4569 self.splice(s, e, "", EditKind::Other);
4570 }
4571 self.anchor = None;
4572 self.record_caret();
4573 let at = self.caret;
4574 if self.caret_parts_bare_paragraph() {
4575 let _ = self.editor.split_block(at);
4576 }
4577 match self.editor.insert_table(at, rows, cols) {
4578 Ok(change) => {
4579 self.last_edit_kind = None;
4580 self.refresh();
4581 self.anchor = None;
4582 self.caret = change.new.end;
4583 self.dirty = self.source != self.clean_source;
4584 self.status = None;
4585 self.clamp_caret();
4586 // Into the first header cell of the table just written: the
4587 // first table whose grid begins inside the splice.
4588 self.rebuild_map();
4589 let first_cell = self
4590 .vmap
4591 .tables
4592 .iter()
4593 .filter_map(|t| t.grid.first().and_then(|row| row.cells.first()))
4594 .find(|cell| cell.start >= change.new.start && cell.start < change.new.end)
4595 .map(|cell| (cell.start, cell.end));
4596 if let Some((start, end)) = first_cell {
4597 self.select_cell(start, end);
4598 }
4599 self.record_caret();
4600 }
4601 Err(e) => self.status = Some(format!("table: {e}")),
4602 }
4603 }
4604
4605 /// Whether the caret sits in a paragraph and nothing else — no list item, no
4606 /// quote, no fence, no table — with paragraph text still ahead of it. The
4607 /// one shape where parting the block around the caret is unambiguously what
4608 /// a rule button means; see
4609 /// [`insert_thematic_break`](Self::insert_thematic_break) for why every other
4610 /// container is left to take the rule after itself.
4611 ///
4612 /// The "text ahead" half is what keeps `split_block` from running at the
4613 /// one edge where its output composes badly. At a paragraph's end twig
4614 /// cannot mint the empty second half (no format spells an empty
4615 /// paragraph), so it writes only the separator — a blank line and the
4616 /// slot Enter leaves for the paragraph to come — and a block then aimed at
4617 /// the first half lands above a slot that nothing fills: `para\n` with the
4618 /// caret at 4 came out as `para\n\n* * *\n\n\n`. Trailing whitespace counts
4619 /// as nothing ahead, since the split would shed it as the second half's
4620 /// leading indent and leave the same slot. Which end of the newline a
4621 /// paragraph's span stops at differs between the formats (Markdown before
4622 /// it, djot after), which is why this reads the remaining bytes rather
4623 /// than comparing offsets. The paragraph's
4624 /// start is deliberately not the same case — see
4625 /// [`insert_thematic_break`](Self::insert_thematic_break) for why that
4626 /// split is kept.
4627 fn caret_parts_bare_paragraph(&mut self) -> bool {
4628 let caret = self.caret.min(self.source.len());
4629 let Ok(chain) = self.editor.ancestors_at(caret) else {
4630 return false;
4631 };
4632 let mut para_end = None;
4633 for m in chain {
4634 match m.kind {
4635 Kind::Para => para_end = Some(m.span.end.min(self.source.len())),
4636 Kind::ListItem
4637 | Kind::TaskListItem
4638 | Kind::BlockQuote
4639 | Kind::CodeBlock
4640 | Kind::Table => return false,
4641 _ => {}
4642 }
4643 }
4644 match para_end {
4645 Some(end) if end > caret => !self.source[caret..end].trim().is_empty(),
4646 _ => false,
4647 }
4648 }
4649
4650 /// The destination of the link under the caret — what a Link prompt shows so
4651 /// ⌘K on an existing link edits its URL instead of asking for it again.
4652 /// `None` when the caret stands in no link.
4653 ///
4654 /// An autolink carries no separate destination: its text *is* the URL, so
4655 /// that's what comes back for one.
4656 pub fn link_destination_at_caret(&mut self) -> Option<String> {
4657 self.link_destination_at(self.caret)
4658 }
4659
4660 /// The destination of the link at `off`.
4661 /// [`link_destination_at_caret`](Self::link_destination_at_caret) for a place
4662 /// the caret isn't.
4663 ///
4664 /// The offset form exists for the same reason
4665 /// [`footnote_at`](Self::footnote_at)'s does: a frontend drawing a *piece* of
4666 /// the document somewhere else — a footnote's text in a popover, say — has
4667 /// rows and runs but no caret in them, and still needs to know which of those
4668 /// runs a reader can follow.
4669 pub fn link_destination_at(&mut self, off: usize) -> Option<String> {
4670 self.nodes()
4671 .into_iter()
4672 .filter(|n| matches!(n.kind.as_str(), "link" | "url" | "email"))
4673 .filter(|n| n.span.start <= off && off < n.span.end)
4674 .max_by_key(|n| n.span.start)
4675 .and_then(|n| n.destination.or(n.text))
4676 }
4677
4678 /// Where the locator `id` lands in this document — the `#v2` half of a
4679 /// `chapter.dj#v2`, resolved to the block it names. `None` when nothing here
4680 /// answers to it.
4681 ///
4682 /// The other end of a link, and the reason this exists: without it a
4683 /// destination has only file granularity, so following a citation into a
4684 /// chapter drops the reader at the top of it to hunt for the verse. Which is
4685 /// also why it is a *document* query rather than a caret one — the document
4686 /// being asked is usually not the one the reader is in.
4687 ///
4688 /// Three readings, tried in order, because the same `#some-heading` is
4689 /// written three ways across the formats leaf opens:
4690 ///
4691 /// 1. **A declared id**, exactly as written: djot's `{#v1}` on a block, and
4692 /// the auto-ids djot mints for its headings. The only exact answer, so it
4693 /// goes first — a document that says `{#v1}` has settled the question.
4694 /// 2. **A declared id, slugged.** djot spells a heading's auto-id
4695 /// `Some-Heading-Here`; nearly every tool that *writes* a link to one
4696 /// spells it `#some-heading-here`. Comparing slugs is what lets a link
4697 /// authored anywhere land on a djot heading.
4698 /// 3. **A heading's text, slugged.** Markdown has no ids at all — twig mints
4699 /// none and `{#custom}` is literal text in a Markdown heading — so for
4700 /// the format most vaults are written in, the heading's own words are the
4701 /// only thing a fragment can name. This is the rule every Markdown
4702 /// renderer already follows, which is what makes `#a-heading` mean in
4703 /// diaryx what it means on the web.
4704 ///
4705 /// Ties go to the earliest match, then to the widest: a duplicated id is the
4706 /// document's mistake and the first one is the answer every anchor
4707 /// implementation gives, while preferring the wider span picks the section
4708 /// over the heading that opens it — more for a peek to show, same place to
4709 /// land.
4710 pub fn locate(&mut self, id: &str) -> Option<Landing> {
4711 let id = id.trim();
4712 if id.is_empty() {
4713 return None;
4714 }
4715 let nodes = self.nodes();
4716
4717 // Earliest wins, then widest. `Reverse` on the end because `min_by_key`
4718 // is picking, among nodes that start together, the one that ends last.
4719 let pick = |matches: &mut dyn Iterator<Item = &FlatNode>| {
4720 matches
4721 .min_by_key(|n| (n.span.start, std::cmp::Reverse(n.span.end)))
4722 .map(|n| Landing {
4723 start: n.span.start,
4724 end: n.span.end,
4725 })
4726 };
4727
4728 if let Some(landing) = pick(&mut nodes.iter().filter(|n| declared_id(n) == Some(id))) {
4729 return Some(landing);
4730 }
4731 let want = slug(id);
4732 if want.is_empty() {
4733 return None;
4734 }
4735 if let Some(landing) = pick(
4736 &mut nodes
4737 .iter()
4738 .filter(|n| declared_id(n).map(slug).as_deref() == Some(&*want)),
4739 ) {
4740 return Some(landing);
4741 }
4742
4743 // A heading by its words. Its span is one line, so the end comes from
4744 // where the *section* it opens gives out — the next heading that is not
4745 // under it, or the end of the document. A Markdown heading has no
4746 // section node to ask (twig only builds those for djot), and a peek that
4747 // showed the heading alone would answer "what does that say" with the
4748 // title of the thing it says.
4749 let heading = nodes
4750 .iter()
4751 .filter(|n| n.kind == Kind::Heading)
4752 .filter(|n| {
4753 n.content_span
4754 .clone()
4755 .and_then(|s| self.source.get(s))
4756 .is_some_and(|text| slug(text) == want)
4757 })
4758 .min_by_key(|n| n.span.start)?;
4759 let level = heading.level.unwrap_or(u32::MAX);
4760 let end = nodes
4761 .iter()
4762 .filter(|n| n.kind == Kind::Heading)
4763 .filter(|n| n.span.start > heading.span.start)
4764 .filter(|n| n.level.unwrap_or(u32::MAX) <= level)
4765 .map(|n| n.span.start)
4766 .min()
4767 .unwrap_or(self.source.len());
4768 Some(Landing {
4769 start: heading.span.start,
4770 end,
4771 })
4772 }
4773
4774 /// Write a footnote at the caret — the toolbar's Footnote button, and the
4775 /// one gesture in the footnote story that *authors* rather than follows.
4776 ///
4777 /// Both halves go in as one twig edit: the `[^1]` where the caret is, and
4778 /// the `[^1]:` definition at the end of the document. Half a footnote is not
4779 /// a footnote — a bare reference with nothing defining it renders as literal
4780 /// brackets — so a single button that wrote only the reference would leave
4781 /// the author to hand-spell the other half in a document that had just
4782 /// stopped showing them what the first half meant. One edit also means one
4783 /// undo takes both back.
4784 ///
4785 /// The definition's body is left empty and **the caret lands in it**, which
4786 /// is the whole point of pressing the button: nobody wants a reference to a
4787 /// note they have not written yet. Getting back to where they were writing
4788 /// is [`footnote_definition_at_caret`](Self::footnote_definition_at_caret) —
4789 /// the same return leg a reader following a reference already uses, so the
4790 /// author is left standing on the near end of a round trip that works.
4791 ///
4792 /// A selection collapses to its *end* rather than being replaced: a
4793 /// reference annotates the words before it, so "select the claim, add a
4794 /// footnote" should mark that claim, not consume it.
4795 pub fn insert_footnote(&mut self) {
4796 if self.read_only || self.refuse_unsupported("footnote", Gesture::InsertFootnote) {
4797 return;
4798 }
4799 let at = self.selection().map_or(self.caret, |(_, end)| end);
4800 self.anchor = None;
4801 self.caret = at;
4802 self.record_caret();
4803 let label = self.next_footnote_label();
4804 match self.editor.insert_footnote(at, &label) {
4805 Ok(change) => {
4806 self.last_edit_kind = None;
4807 self.refresh();
4808 self.anchor = None;
4809 // `change.new` runs from the reference to the end of the
4810 // document, so its start is the `[^1]` just written and
4811 // `footnote_at` resolves it to the note the same way a reader's
4812 // tap does — and to the note's *body*, which is already a caret
4813 // stop even when it is empty (the `[^1]:` marker draws as `[1] `
4814 // and has none), so this needs no snap on top. The fallback is
4815 // the reference's own offset: a format that spelled the pair some
4816 // way leaf can't read back should still leave the caret on the
4817 // edit rather than at the far end of a document it just grew.
4818 self.caret = self
4819 .footnote_at(change.new.start)
4820 .and_then(|note| note.offset)
4821 .unwrap_or(change.new.start);
4822 self.dirty = self.source != self.clean_source;
4823 self.status = None;
4824 self.clamp_caret();
4825 self.record_caret();
4826 }
4827 Err(e) => self.status = Some(format!("footnote: {e}")),
4828 }
4829 }
4830
4831 /// The label to give a footnote the author has not named: the lowest counting
4832 /// number no footnote in the document is already wearing.
4833 ///
4834 /// twig takes the label rather than minting one, because it holds no opinion
4835 /// about what a document's footnotes should be called — and it is right not
4836 /// to. Numbering them is what every author of a numbered note expects, and
4837 /// re-using a taken number would silently point the new reference at somebody
4838 /// else's note (twig reuses an existing definition rather than appending a
4839 /// second one, which is the right rule for citing a note twice on purpose and
4840 /// exactly the wrong accident to have by default).
4841 ///
4842 /// *References* are counted alongside definitions, not just definitions: a
4843 /// document carrying a dangling `[^2]` has a 2 that means something to
4844 /// whoever wrote it, and minting a definition for it here would answer a
4845 /// question nobody asked. Non-numeric labels (`[^why]`) are left out of the
4846 /// count entirely — they take no number, so they block none.
4847 fn next_footnote_label(&mut self) -> String {
4848 let mut taken: Vec<u32> = wysiwyg::footnote_definitions(&mut self.editor)
4849 .into_iter()
4850 .filter_map(|note| wysiwyg::footnote_label(&self.source, note.span.start))
4851 .filter_map(|label| label.parse().ok())
4852 .collect();
4853 taken.extend(
4854 self.nodes()
4855 .into_iter()
4856 .filter(|n| n.kind == Kind::FootnoteReference)
4857 .filter_map(|n| wysiwyg::footnote_reference_label(&self.source, n.span))
4858 .filter_map(|label| label.parse::<u32>().ok()),
4859 );
4860 (1..).find(|n| !taken.contains(n)).unwrap_or(1).to_string()
4861 }
4862
4863 /// The footnote reference under the caret, resolved to the note it names.
4864 /// [`footnote_at`](Self::footnote_at) at the caret's offset.
4865 pub fn footnote_at_caret(&mut self) -> Option<FootnoteRef> {
4866 self.footnote_at(self.caret)
4867 }
4868
4869 /// The footnote reference at `off`, resolved to the note it names — what a
4870 /// frontend shows when a reader activates a `[^1]`.
4871 ///
4872 /// A reference is not a link node, so
4873 /// [`link_destination_at_caret`](Self::link_destination_at_caret) does not
4874 /// (and should not) answer for one: a link names a destination to leave for,
4875 /// a reference names a note that is already in this document. Following one
4876 /// is a move within the page, which is why this hands back an `offset`
4877 /// rather than something to open.
4878 ///
4879 /// Offset-based rather than caret-only because the gesture that wants this
4880 /// most is the one that must not move the caret: a pointer hovering a `[1]`
4881 /// asks what note it names without disturbing where the reader was typing.
4882 /// The caret is just the offset a click already placed —
4883 /// [`footnote_at_caret`](Self::footnote_at_caret) passes it.
4884 ///
4885 /// `None` when `off` stands in no reference. A reference whose note the
4886 /// document never defines is *not* `None` — it answers with the label it
4887 /// looked for and no text, which is what lets a frontend say so instead of
4888 /// silently doing nothing.
4889 pub fn footnote_at(&mut self, off: usize) -> Option<FootnoteRef> {
4890 // Innermost-wins by latest start, the rule its link sibling uses.
4891 let span = self
4892 .nodes()
4893 .into_iter()
4894 .filter(|n| n.kind == Kind::FootnoteReference)
4895 .filter(|n| n.span.start <= off && off < n.span.end)
4896 .max_by_key(|n| n.span.start)?
4897 .span;
4898 let label = wysiwyg::footnote_reference_label(&self.source, span)?.to_string();
4899
4900 // The note itself. Definitions are roots beside `doc` rather than
4901 // children of it, so they're asked for directly — see
4902 // `wysiwyg::footnote_definitions`.
4903 let note = wysiwyg::footnote_definitions(&mut self.editor)
4904 .into_iter()
4905 .find(|m| wysiwyg::footnote_label(&self.source, m.span.start) == Some(&label));
4906 let Some(note) = note else {
4907 return Some(FootnoteRef {
4908 label,
4909 text: None,
4910 offset: None,
4911 end: None,
4912 });
4913 };
4914 let body = wysiwyg::footnote_body_span(&self.source, note.span.clone());
4915 Some(FootnoteRef {
4916 label,
4917 text: body
4918 .clone()
4919 .and_then(|b| self.source.get(b))
4920 .map(str::to_string),
4921 // The body's start, not the definition's — see `FootnoteRef::offset`.
4922 offset: body.clone().map(|b| b.start),
4923 end: body.map(|b| b.end),
4924 })
4925 }
4926
4927 /// The footnote *definition* the caret stands in, and where the reference
4928 /// that names it is. [`footnote_definition_at`](Self::footnote_definition_at)
4929 /// at the caret's offset.
4930 pub fn footnote_definition_at_caret(&mut self) -> Option<FootnoteDef> {
4931 self.footnote_definition_at(self.caret)
4932 }
4933
4934 /// The footnote definition spanning `off`, and where the reference that
4935 /// names it is — the return leg of [`footnote_at`](Self::footnote_at).
4936 ///
4937 /// The mirror image, deliberately: the same gesture that takes a reader from
4938 /// `[1]` down to the note takes them from the note back up to `[1]`, so
4939 /// following a footnote is a round trip rather than a fall. It needs no
4940 /// memory of how the reader arrived — the document says where the reference
4941 /// is — which is what makes it work for a reader who scrolled to the notes
4942 /// themselves, and what keeps it right after an edit moves either end.
4943 ///
4944 /// `None` when `off` stands in no definition. A definition nothing cites is
4945 /// *not* `None`, for [`FootnoteRef`]'s reason in reverse: it answers with
4946 /// its label and no offset, so a frontend can say "nothing refers to this"
4947 /// rather than offer a jump that goes nowhere.
4948 pub fn footnote_definition_at(&mut self, off: usize) -> Option<FootnoteDef> {
4949 // Definitions are roots beside `doc`, so `nodes()` — which walks the
4950 // document body — never reports one. They're asked for directly, the way
4951 // `footnote_at` asks for the note it resolves to.
4952 //
4953 // Closed at the end, unlike the half-open test its neighbours use. A
4954 // definition's span stops at its last content byte — the newline ending
4955 // the line is outside it — so `span.end` is the caret stop at the end of
4956 // the note's own row, not the first byte of anything after. Excluding it
4957 // meant the one caret an author is guaranteed to have, the one left
4958 // sitting at the end of the note they just typed, was in no definition at
4959 // all: writing a note and then asking to go back to its reference
4960 // answered nothing. Two definitions in a row still can't both match —
4961 // there is a blank line between them — and `max_by_key` decides anyway.
4962 let note = wysiwyg::footnote_definitions(&mut self.editor)
4963 .into_iter()
4964 .filter(|m| m.span.start <= off && off <= m.span.end)
4965 .max_by_key(|m| m.span.start)?;
4966 let label = wysiwyg::footnote_label(&self.source, note.span.start)?.to_string();
4967
4968 // The earliest reference carrying this label. `min` rather than a `find`,
4969 // because `nodes()` reports a flattened walk whose order is twig's
4970 // business, not document order. Bound first: the walk needs `&mut self`
4971 // and reading the labels back out needs `&self.source`.
4972 let nodes = self.nodes();
4973 let offset = nodes
4974 .into_iter()
4975 .filter(|n| n.kind == Kind::FootnoteReference)
4976 .filter(|n| {
4977 wysiwyg::footnote_reference_label(&self.source, n.span.clone()) == Some(&*label)
4978 })
4979 // Past the `[^`, onto the label — see `FootnoteDef::offset`.
4980 .map(|n| n.span.start + 2)
4981 .min();
4982 Some(FootnoteDef { label, offset })
4983 }
4984
4985 /// The destination of the image under the caret — what an image prompt shows
4986 /// so editing an existing image starts from its current URL instead of blank,
4987 /// the image analogue of [`link_destination_at_caret`](Self::link_destination_at_caret).
4988 /// `None` when the caret stands in no image. A caret resting just after a
4989 /// block image (its trailing stop) is still "in" it — the half-open span test
4990 /// excludes that offset, which is the intended precision: past the image is
4991 /// past it.
4992 pub fn image_destination_at_caret(&mut self) -> Option<String> {
4993 let off = self.caret;
4994 self.nodes()
4995 .into_iter()
4996 .filter(|n| n.kind == Kind::Image)
4997 .filter(|n| n.span.start <= off && off < n.span.end)
4998 .max_by_key(|n| n.span.start)
4999 .and_then(|n| n.destination)
5000 }
5001
5002 /// The language of the fenced code block the caret stands in — what a
5003 /// language prompt shows so editing it starts from the current value rather
5004 /// than blank. `None` when the caret is in no code block, or in one whose
5005 /// fence carries no language (or an indented block, which has no fence).
5006 pub fn code_language_at_caret(&mut self) -> Option<String> {
5007 let start = self.code_block_start_at_caret()?;
5008 wysiwyg::code_language(&self.source, start)
5009 }
5010
5011 /// Whether the caret stands in a fenced code block — the one a language
5012 /// prompt could edit. A frontend gates its "set language" affordance on this
5013 /// (an indented block, which can't carry a language, reports `false`).
5014 pub fn caret_in_fenced_code(&mut self) -> bool {
5015 self.code_block_start_at_caret()
5016 .is_some_and(|start| wysiwyg::code_info_span(&self.source, start).is_some())
5017 }
5018
5019 /// Set (or clear, with `""`) the language of the fenced code block the caret
5020 /// is in — the prompt's confirm. A no-op when the caret is in no fenced
5021 /// block, and a reported error for a language the format's fence cannot
5022 /// carry.
5023 ///
5024 /// twig rewrites the info string, so the fence's own width — measured
5025 /// against a body neither side touches — is kept, and a language holding a
5026 /// space, a line end or the fence character is refused rather than written
5027 /// out to reparse as something else. Leaf used to splice over the info span
5028 /// itself and `trim()` the input, which handled the one bad case it had
5029 /// thought of.
5030 pub fn set_code_language(&mut self, lang: &str) {
5031 // The read-only gate — this door reaches twig without the splice.
5032 if self.read_only {
5033 return;
5034 }
5035 if self.refuse_unsupported("code language", Gesture::SetCodeLanguage) {
5036 return;
5037 }
5038 if self.code_block_start_at_caret().is_none() {
5039 return;
5040 }
5041 let lang = lang.trim();
5042 // `None` clears the info string; `Some("")` asks for an empty one. Both
5043 // write a bare fence, and the prompt's empty value means "clear".
5044 let want = (!lang.is_empty()).then_some(lang);
5045 self.record_caret();
5046 match self.editor.set_code_language(self.caret, want) {
5047 Ok(_) => {
5048 self.last_edit_kind = None;
5049 self.refresh();
5050 self.anchor = None;
5051 self.dirty = self.source != self.clean_source;
5052 self.status = None;
5053 self.clamp_caret();
5054 self.record_caret();
5055 }
5056 Err(e) => self.status = Some(format!("code language: {e}")),
5057 }
5058 }
5059
5060 /// The `span.start` of the code block covering the caret — the anchor
5061 /// [`wysiwyg::code_info_span`] reads the fence from. `None` when the caret is
5062 /// in none.
5063 fn code_block_start_at_caret(&mut self) -> Option<usize> {
5064 let off = self.caret;
5065 self.nodes()
5066 .into_iter()
5067 .filter(|n| n.kind == Kind::CodeBlock && n.span.start <= off && off <= n.span.end)
5068 .max_by_key(|n| n.span.start)
5069 .map(|n| n.span.start)
5070 }
5071
5072 /// The source range of the text inside the link covering `off` — what sits
5073 /// between its `[` and `]`. `None` when twig reports no link there.
5074 fn link_text_span(&mut self, off: usize) -> Option<std::ops::Range<usize>> {
5075 self.nodes()
5076 .into_iter()
5077 // Two links can touch (`[a](x)[b](y)`), and then one's `span.end` is
5078 // the other's `span.start`; the link that starts latest at or before
5079 // `off` is the one `off` is actually in.
5080 .filter(|n| n.kind == Kind::Link && n.span.start <= off && off < n.span.end)
5081 .max_by_key(|n| n.span.start)
5082 .and_then(|n| n.content_span)
5083 }
5084
5085 // ── undo / redo ───────────────────────────────────────────────────────────
5086 // twig owns the history of *bytes* (it owns the buffer) and now carries the
5087 // caret through it too: `record_caret` stashes each state's caret in twig's
5088 // opaque per-step blob, and undo/redo hand it back with the source they
5089 // restore. So leaf keeps no history of its own — no parallel stacks to march
5090 // in lockstep and silently drift out of it.
5091
5092 /// Undo the last edit step (⌘Z / ^Z), putting the caret and selection back
5093 /// where they were when that step began.
5094 pub fn undo(&mut self) {
5095 if self.read_only {
5096 return;
5097 }
5098 let (undone, redoable) = (self.undo_steps, self.redo_steps);
5099 match self.editor.undo() {
5100 Ok(Some(change)) => {
5101 self.after_history(change);
5102 // `refresh` counted the restore as an edit; it was a step back.
5103 self.undo_steps = undone.saturating_sub(1);
5104 self.redo_steps = redoable + 1;
5105 }
5106 Ok(None) => {
5107 self.undo_steps = 0;
5108 self.status = Some("nothing to undo".into());
5109 }
5110 Err(e) => self.status = Some(format!("undo: {e}")),
5111 }
5112 }
5113
5114 /// Redo the last undone edit step (⇧⌘Z / ^Y), putting the caret and
5115 /// selection back where that step originally left them.
5116 pub fn redo(&mut self) {
5117 if self.read_only {
5118 return;
5119 }
5120 let (undone, redoable) = (self.undo_steps, self.redo_steps);
5121 match self.editor.redo() {
5122 Ok(Some(change)) => {
5123 self.after_history(change);
5124 // `refresh` counted the restore as an edit; it was a step forward.
5125 self.undo_steps = undone + 1;
5126 self.redo_steps = redoable.saturating_sub(1);
5127 }
5128 Ok(None) => {
5129 self.redo_steps = 0;
5130 self.status = Some("nothing to redo".into());
5131 }
5132 Err(e) => self.status = Some(format!("redo: {e}")),
5133 }
5134 }
5135
5136 /// Refresh the cached source and put the caret back where the step being
5137 /// undone/redone had it, clearing any active run.
5138 ///
5139 /// The caret comes from twig's blob for the restored state (what
5140 /// `record_caret` stored). `change` is only the fallback for a state with no
5141 /// blob — a caret at the end of the restored text, which is where this always
5142 /// landed before the blobs were kept. It is the edit site, not where the user
5143 /// was standing, so it's a floor and not the behaviour: undoing should hand
5144 /// back the document *and* the place you were working, which for an edit made
5145 /// anywhere but under the caret are two different places.
5146 fn after_history(&mut self, change: Change) {
5147 self.refresh();
5148 match self
5149 .editor
5150 .caret_blob()
5151 .ok()
5152 .and_then(|b| CaretState::from_blob(&b))
5153 {
5154 Some(state) => {
5155 self.caret = state.caret.min(self.source.len());
5156 self.anchor = state.anchor.map(|a| a.min(self.source.len()));
5157 }
5158 None => {
5159 self.caret = change.new.end.min(self.source.len());
5160 self.anchor = None;
5161 }
5162 }
5163 self.goal_col = None;
5164 self.last_edit_kind = None;
5165 self.dirty = self.source != self.clean_source;
5166 self.status = None;
5167 self.clamp_caret();
5168 }
5169
5170 // ── the file ──────────────────────────────────────────────────────────────
5171
5172 #[cfg(feature = "fs")]
5173 pub fn save(&mut self) {
5174 if self.is_untitled() {
5175 // No path to write and no name to invent: ⌘S on an untitled document
5176 // is a Save As, and only a frontend has a picker to ask with. Say so
5177 // rather than failing at the filesystem with an empty path.
5178 self.status = Some("untitled — save as…".into());
5179 return;
5180 }
5181 let path = self.path.clone();
5182 if self.write(&path) {
5183 self.mark_saved();
5184 }
5185 }
5186
5187 /// Save As: write the document to `path` and *move* it there — `self.path`
5188 /// becomes `path`, and every later [`Doc::save`] writes the new file. That's
5189 /// what Save As means; a copy would leave the user editing a document whose
5190 /// name is no longer where their keystrokes go.
5191 ///
5192 /// The move only happens if the bytes actually landed. A failed write leaves
5193 /// the path, `dirty`, and the disk watermark exactly as they were, with the
5194 /// same `save failed: …` status a failed [`Doc::save`] sets — the document
5195 /// must never come away believing it was saved.
5196 ///
5197 /// An existing `path` is overwritten, and the caller is the one that knows
5198 /// whether to ask first: a Save As picker has already run that prompt, and a
5199 /// second confirmation from down here would be the same question twice.
5200 ///
5201 /// `format` does **not** follow the new extension. The buffer is parsed as
5202 /// the format it was opened with, and re-reading it as another one is a
5203 /// conversion — a different, lossy operation that would throw away the undo
5204 /// history — not a rename. So `notes.md` saved as `notes.dj` holds Markdown
5205 /// in a `.dj` file, and `format_name()` keeps honestly saying `markdown`
5206 /// until it's reopened.
5207 #[cfg(feature = "fs")]
5208 pub fn save_as(&mut self, path: PathBuf) {
5209 if !self.write(&path) {
5210 return;
5211 }
5212 self.path = path;
5213 self.mark_saved();
5214 }
5215
5216 /// Put `source` on disk at `path`, reporting whether it got there. The one
5217 /// place leaf writes a document, so a save and a Save As can't disagree
5218 /// about what a failure looks like.
5219 #[cfg(feature = "fs")]
5220 fn write(&mut self, path: &Path) -> bool {
5221 match std::fs::write(path, self.source.as_bytes()) {
5222 Ok(()) => true,
5223 Err(e) => {
5224 self.status = Some(format!("save failed: {e}"));
5225 false
5226 }
5227 }
5228 }
5229
5230 /// Re-base the document's saved watermark to the current bytes: clears
5231 /// `dirty`, records `source` as the new clean state (so undoing back to here
5232 /// clears the flag again), and re-stamps the on-disk hash.
5233 ///
5234 /// [`Doc::save`]/[`Doc::save_as`] call this after a write lands. It is also
5235 /// the hook a **filesystem-free host** calls itself once it has persisted
5236 /// [`Doc::source`] its own way (a browser download, `localStorage`, a backend
5237 /// `PUT`) — which is why it is public and touches no filesystem: the bytes
5238 /// are already where that host wants them, and this just tells the model they
5239 /// are safe.
5240 pub fn mark_saved(&mut self) {
5241 self.clean_source = self.source.clone();
5242 self.dirty = false;
5243 // The bytes on disk are now ours, so this is the new watermark: without
5244 // re-stamping it, every save would report its own work as an external
5245 // change forever after.
5246 self.disk_hash = Some(hash_bytes(self.source.as_bytes()));
5247 self.status = Some(format!("saved {}", self.file_name()));
5248 }
5249
5250 /// What the file looks like now against the bytes leaf last read or wrote.
5251 ///
5252 /// Reads the file and hashes it (see `disk_hash` for why it isn't an mtime),
5253 /// so this is a filesystem round-trip, not a per-frame question — ask it
5254 /// when a window regains focus, on a timer, or before a save.
5255 ///
5256 /// This *only* reports the file. Whether the document also has unsaved edits
5257 /// is `dirty`, and the interesting case is the conjunction: `dirty` plus
5258 /// [`DiskState::Changed`] means a save overwrites someone's work and a
5259 /// [`Doc::reload`] discards the user's. leaf-core deliberately won't choose —
5260 /// it has no way to ask — so it hands a frontend both halves and lets it put
5261 /// the question to the person who can answer it.
5262 #[cfg(feature = "fs")]
5263 pub fn disk_state(&self) -> DiskState {
5264 let Some(want) = self.disk_hash else {
5265 return DiskState::Untitled;
5266 };
5267 match std::fs::read(&self.path) {
5268 Ok(bytes) if hash_bytes(&bytes) == want => DiskState::Unchanged,
5269 Ok(_) => DiskState::Changed,
5270 Err(e) if e.kind() == std::io::ErrorKind::NotFound => DiskState::Missing,
5271 Err(_) => DiskState::Unreadable,
5272 }
5273 }
5274
5275 /// Re-read the file and replace the document with what's there — the other
5276 /// answer to a [`DiskState::Changed`].
5277 ///
5278 /// **Discards unsaved changes, unconditionally.** It doesn't check `dirty`
5279 /// first: a frontend that wants to protect unsaved work asks (`dirty` +
5280 /// [`Doc::disk_state`]) *before* calling this, and one reloading a clean
5281 /// document shouldn't have to argue with a guard.
5282 ///
5283 /// **The undo history survives, and the reload is one step in it.** The
5284 /// whole buffer is spliced with the file's bytes through the same door every
5285 /// other edit goes through, as an [`EditKind::Other`] that coalesces with
5286 /// nothing on either side — so ^Z after a formatter or a `git checkout` has
5287 /// swapped the document out from under a reader gives them back what they
5288 /// were looking at, marked dirty, and ^Z again carries on into whatever they
5289 /// had done before it. This used to build a fresh parse and drop the stack,
5290 /// on the reasoning that twig's history belongs to the buffer and these are
5291 /// different bytes; that is true of *rebasing* a step onto them and not of
5292 /// recording the swap itself as one, which is all this is. A splice twig
5293 /// won't take falls back to the fresh parse, and only that path still costs
5294 /// the history.
5295 ///
5296 /// The caret keeps its byte offset, clamped to the new length; the selection
5297 /// is dropped. Anything cleverer would be a lie: leaf doesn't know how the
5298 /// file changed, so it can't know where the caret "still" is. Clamping keeps
5299 /// it where the user left it in the common case (a change further down the
5300 /// file, or none in the text they're sitting in), and never puts it
5301 /// somewhere invalid. A selection has two such offsets and no such excuse —
5302 /// silently reinterpreting one over changed bytes would arm the *next*
5303 /// keystroke to delete something the user never selected.
5304 ///
5305 /// Nothing is touched unless the whole reload succeeds; a failure leaves the
5306 /// document alone with a status.
5307 #[cfg(feature = "fs")]
5308 pub fn reload(&mut self) {
5309 if self.is_untitled() {
5310 self.status = Some("no file to reload".into());
5311 return;
5312 }
5313 let bytes = match std::fs::read(&self.path) {
5314 Ok(b) => b,
5315 Err(e) => {
5316 self.status = Some(format!("reload failed: {e}"));
5317 return;
5318 }
5319 };
5320 let Ok(source) = String::from_utf8(bytes) else {
5321 self.status = Some("reload failed: file is not UTF-8".into());
5322 return;
5323 };
5324 // Already these bytes — someone saved a file back unchanged, or leaf's
5325 // own write is being read back. Re-baseline against it and stop: a
5326 // splice of the text onto itself would put an undo step on the stack for
5327 // something nobody did.
5328 if source == self.source {
5329 self.disk_hash = Some(hash_bytes(source.as_bytes()));
5330 self.clean_source = source;
5331 self.dirty = false;
5332 self.status = Some(format!("reloaded {}", self.file_name()));
5333 return;
5334 }
5335 let caret = self.caret;
5336 // The pre-reload caret, so undoing the swap puts it back where the
5337 // reader was standing — the same bracketing `splice_exact` does.
5338 self.record_caret();
5339 if self
5340 .editor
5341 .edit_range(0, self.source.len(), &source)
5342 .is_ok()
5343 {
5344 self.refresh();
5345 } else {
5346 // twig wouldn't take the splice. Start over from the bytes, which is
5347 // what this always did, and is the one path that still costs the
5348 // history — `format` is the format this document *is*, not what the
5349 // (unchanged) name now says, see `save_as`.
5350 match new_editor(source.as_bytes(), self.format) {
5351 Ok(editor) => {
5352 self.editor = editor;
5353 self.source = source.clone();
5354 // Not going through `refresh`, so the revision has to move
5355 // here or every frontend keeps painting the old file from
5356 // cache.
5357 self.revision += 1;
5358 }
5359 Err(e) => {
5360 self.status = Some(format!("reload failed: {e}"));
5361 return;
5362 }
5363 }
5364 }
5365 self.disk_hash = Some(hash_bytes(source.as_bytes()));
5366 self.clean_source = self.source.clone();
5367 self.caret = caret.min(self.source.len());
5368 self.anchor = None;
5369 self.goal_col = None;
5370 self.last_edit_kind = None;
5371 self.dirty = false;
5372 self.status = Some(format!("reloaded {}", self.file_name()));
5373 self.clamp_caret();
5374 // And the post-reload caret, so a redo restores it.
5375 self.record_caret();
5376 }
5377
5378 /// Re-read the source from twig after it has changed the document. The one
5379 /// funnel every edit, undo, and redo comes through — so it's where the
5380 /// revision moves, and anything cached against the text dies here.
5381 fn refresh(&mut self) {
5382 if let Ok(s) = self.editor.source_str() {
5383 self.source = s;
5384 }
5385 self.revision += 1;
5386 // An edit is a step onto the history and the end of anything undone;
5387 // `undo`/`redo` come through here too and correct this after.
5388 self.undo_steps += 1;
5389 self.redo_steps = 0;
5390 self.clamp_caret();
5391 }
5392
5393 /// Whether [`undo`](Self::undo) has a step to take back — for a native
5394 /// Edit menu to enable its item by. See the note on `undo_steps` for what
5395 /// "has" means here.
5396 pub fn can_undo(&self) -> bool {
5397 !self.read_only && self.undo_steps > 0
5398 }
5399
5400 /// Whether [`redo`](Self::redo) has an undone step to restore.
5401 pub fn can_redo(&self) -> bool {
5402 !self.read_only && self.redo_steps > 0
5403 }
5404
5405 // ── caret movement ─────────────────────────────────────────────────────────
5406 // `extend` grows the selection (Shift+motion): it pins the anchor on the
5407 // first extended step and moves only the caret; an un-extended motion drops
5408 // the selection.
5409
5410 /// Place the caret at byte `offset` (clamped to a char boundary), extending
5411 /// the selection when `extend` is set. The public form of `move_to`, for a
5412 /// frontend that hit-tests pixels straight to a source offset.
5413 pub fn place_caret(&mut self, offset: usize, extend: bool) {
5414 self.goal_col = None;
5415 let before = self.caret;
5416 // A pixel hit-test can land between the visible caret stops — in the
5417 // blank gap a paragraph break is drawn with, or inside a hidden delimiter.
5418 // Snap to the nearest real stop so the caret can't come to rest where it
5419 // would draw in one place and type in another. The `(row, col)` click
5420 // path (`click`) already snaps this way through `offset_of_pos`; the
5421 // source view reaches every byte, so it snaps to nothing.
5422 let target = match self.view {
5423 View::Wysiwyg => self.vmap.snap_to_stop(offset.min(self.source.len())),
5424 // The source view reaches every byte, so there is no stop to snap
5425 // to — but "every byte" still means every *character* boundary. A
5426 // caret resting inside a multi-byte character draws nowhere real
5427 // and panics the next time anything slices there.
5428 View::Source => self.char_boundary_at_or_before(offset),
5429 };
5430 self.move_to(target, extend);
5431 self.clamp_caret();
5432 self.debug_assert_on_a_stop(before);
5433 }
5434
5435 /// Select the whole document (⌘A / Ctrl+A) — everything reachable in the
5436 /// active view, so in WYSIWYG it starts below hidden frontmatter (copy won't
5437 /// grab the metadata) while the source view still selects the literal whole.
5438 pub fn select_all(&mut self) {
5439 self.anchor = Some(self.caret_floor());
5440 self.caret = self.source.len();
5441 self.goal_col = None;
5442 self.last_edit_kind = None;
5443 self.status = None;
5444 }
5445
5446 /// Select the word (or whitespace / punctuation run) at `offset` — the
5447 /// double-click gesture. Anchors on the run's start with the caret at its
5448 /// end so a following Shift-motion extends from the far edge.
5449 pub fn select_word_at(&mut self, offset: usize) {
5450 let (s, e) = word_range_at(&self.source, offset.min(self.source.len()));
5451 self.anchor = Some(s);
5452 self.caret = e;
5453 self.goal_col = None;
5454 self.last_edit_kind = None;
5455 self.status = None;
5456 self.clamp_caret();
5457 }
5458
5459 /// Select the whole enclosing text block (paragraph, heading, list item's
5460 /// text…) at `offset` — the triple-click gesture. Reads the range straight
5461 /// from the AST (twig's `content_span`), so it selects the entire *logical*
5462 /// paragraph even when that paragraph soft-wraps across several visual rows —
5463 /// where a visual-row-based select breaks down, because one source offset at
5464 /// a wrap boundary belongs to two rows at once.
5465 pub fn select_block_at(&mut self, offset: usize) {
5466 let off = offset.min(self.source.len());
5467 let range = self
5468 .editor
5469 .ancestors_at(off)
5470 .ok()
5471 .and_then(|chain| {
5472 // Ancestors run root → deepest; the deepest node that is neither
5473 // an inline span nor a multi-block container is the text block
5474 // the caret sits in (a paragraph, a heading, a code block…).
5475 chain
5476 .into_iter()
5477 .rev()
5478 .find(|m| !wysiwyg::is_inline_kind(&m.kind) && !is_block_container(&m.kind))
5479 .map(|m| m.content_span.unwrap_or(m.span))
5480 })
5481 .unwrap_or_else(|| source_line_range(&self.source, off));
5482 self.anchor = Some(range.start.min(self.source.len()));
5483 self.caret = range.end.min(self.source.len());
5484 self.goal_col = None;
5485 self.last_edit_kind = None;
5486 self.status = None;
5487 self.clamp_caret();
5488 }
5489
5490 /// Select the exact source range `[start, end)` — anchor at `start`, caret
5491 /// at `end` — without snapping either end to a visible caret stop.
5492 ///
5493 /// The one caret verb that takes a range it was *handed* rather than one it
5494 /// worked out, for a host that already knows the bytes it means: a search
5495 /// hit, an annotation's footprint, a quote re-anchored through
5496 /// [`Doc::selection_quote`]. [`place_caret`](Self::place_caret) is the
5497 /// wrong tool for that, and not by a little — it snaps to the nearest
5498 /// *visible* stop, and where a range butts up against a hidden delimiter
5499 /// the nearest stop is the one before it, so selecting the "needle" of
5500 /// `**needle**` comes back with "needl" and an edit against it strands the
5501 /// "e".
5502 ///
5503 /// What `place_caret` does that is bookkeeping rather than snapping still
5504 /// happens here, because a host handing in a range is not asking to opt out
5505 /// of the invariants:
5506 ///
5507 /// - both ends are clamped into the document and up to
5508 /// [`caret_floor`](Self::caret_floor) — in WYSIWYG the leading
5509 /// frontmatter is hidden, and a caret parked in it draws nowhere and
5510 /// types into the metadata;
5511 /// - both land on character boundaries, so nothing slices a `é` in half;
5512 /// - the sticky vertical goal column is dropped, and any armed inline mark
5513 /// disarmed, since a range from outside inherits neither.
5514 ///
5515 /// An empty range is a caret rather than a selection —
5516 /// [`selection`](Self::selection) reports `None` for it, as it does for any
5517 /// anchor that has met the caret.
5518 pub fn select_range(&mut self, start: usize, end: usize) {
5519 let floor = self.caret_floor();
5520 let anchor = self.char_boundary_at_or_before(start.clamp(floor, self.source.len()));
5521 let caret = self.char_boundary_at_or_before(end.clamp(floor, self.source.len()));
5522 self.anchor = Some(anchor);
5523 self.caret = caret;
5524 self.goal_col = None;
5525 self.status = None;
5526 self.last_edit_kind = None;
5527 self.clear_pending();
5528 }
5529
5530 /// `offset` itself if it is a character boundary, else the boundary before
5531 /// it. An offset that isn't one draws nowhere real and panics the next time
5532 /// anything slices there.
5533 fn char_boundary_at_or_before(&self, offset: usize) -> usize {
5534 let mut o = offset.min(self.source.len());
5535 while o > 0 && !self.source.is_char_boundary(o) {
5536 o -= 1;
5537 }
5538 o
5539 }
5540
5541 /// The lowest source offset the caret may occupy in the active view. In
5542 /// WYSIWYG, leading frontmatter is hidden and unreachable, so the floor is
5543 /// the first rendered offset; the source view reaches everything, so it's 0.
5544 fn caret_floor(&self) -> usize {
5545 match self.view {
5546 View::Wysiwyg => self.vmap.content_start.min(self.source.len()),
5547 View::Source => 0,
5548 }
5549 }
5550
5551 /// Land in a table cell with its whole content selected — the anchor at the
5552 /// cell's start, the caret at its end — so a Tab/Return hop into a cell reads
5553 /// like tabbing into a form field: the text comes up selected, so typing
5554 /// replaces it and an arrow collapses to an edge. An empty cell (`start ==
5555 /// end`) collapses to a plain caret home (an empty selection is no selection).
5556 fn select_cell(&mut self, start: usize, end: usize) {
5557 self.select_range(start, end);
5558 }
5559
5560 fn move_to(&mut self, offset: usize, extend: bool) {
5561 if extend {
5562 if self.anchor.is_none() {
5563 self.anchor = Some(self.caret);
5564 }
5565 } else {
5566 self.anchor = None;
5567 }
5568 self.caret = offset.min(self.source.len()).max(self.caret_floor());
5569 self.status = None;
5570 // A caret move ends the current typing/deletion run, so the next edit
5571 // starts a fresh undo group rather than coalescing across the gap.
5572 self.last_edit_kind = None;
5573 // Moving away disarms any sticky mark — "start bold" applies only where
5574 // it was asked for, not wherever the caret next lands.
5575 self.clear_pending();
5576 }
5577
5578 // In the source view, motion walks source bytes / source lines. In the
5579 // WYSIWYG view it walks the rendered glyph grid (the visual map), which is
5580 // what steps the caret cleanly over hidden delimiters.
5581
5582 pub fn move_left(&mut self, extend: bool) {
5583 self.goal_col = None;
5584 if !extend && let Some((s, _e)) = self.selection() {
5585 self.move_to(s, false);
5586 return;
5587 }
5588 let target = match self.view {
5589 View::Source => {
5590 if self.caret > 0 {
5591 prev_boundary(&self.source, self.caret)
5592 } else {
5593 0
5594 }
5595 }
5596 // Walks caret *stops*, not columns: decoration (a table border, a
5597 // cell's padding) is stepped over in one press, and a hidden
5598 // delimiter never holds the caret up — though the end of a mark's
5599 // content is a stop of its own (`VisualMap::mark_ends`), so
5600 // leaving `**bold**` from past its `**` is a press onto the end of
5601 // the bold and another onto the `d`.
5602 View::Wysiwyg => self
5603 .vmap
5604 .caret_stop_before(self.caret)
5605 .unwrap_or(self.caret),
5606 };
5607 let before = self.caret;
5608 self.move_to(target, extend);
5609 self.debug_assert_on_a_stop(before);
5610 }
5611
5612 pub fn move_right(&mut self, extend: bool) {
5613 self.goal_col = None;
5614 if !extend && let Some((_s, e)) = self.selection() {
5615 self.move_to(e, false);
5616 return;
5617 }
5618 let target = match self.view {
5619 View::Source => {
5620 if self.caret < self.source.len() {
5621 next_boundary(&self.source, self.caret)
5622 } else {
5623 self.caret
5624 }
5625 }
5626 View::Wysiwyg => self.vmap.caret_stop_after(self.caret).unwrap_or(self.caret),
5627 };
5628 let before = self.caret;
5629 self.move_to(target, extend);
5630 self.debug_assert_on_a_stop(before);
5631 }
5632
5633 /// Move to the start of the previous word (⌥← / Ctrl+←).
5634 pub fn move_word_left(&mut self, extend: bool) {
5635 self.goal_col = None;
5636 let before = self.caret;
5637 let target = self.word_left_from(self.caret);
5638 self.move_to(target, extend);
5639 self.debug_assert_on_a_stop(before);
5640 }
5641
5642 /// Move to the end of the next word (⌥→ / Ctrl+→).
5643 pub fn move_word_right(&mut self, extend: bool) {
5644 self.goal_col = None;
5645 let before = self.caret;
5646 let target = self.word_right_from(self.caret);
5647 self.move_to(target, extend);
5648 self.debug_assert_on_a_stop(before);
5649 }
5650
5651 // Word boundaries are found in the space the *view* is in. The source view
5652 // walks the source, because there the source is what's rendered. WYSIWYG
5653 // walks the rendered text instead: `**` is invisible to the user, so it has
5654 // to be invisible to word motion too — a caret parked inside one draws in
5655 // the column after `bold` and types two bytes earlier, and a word-delete
5656 // that stops there shreds the markup into `a ** c`.
5657
5658 /// The word boundary to the left of `off` in the active view's space.
5659 fn word_left_from(&self, off: usize) -> usize {
5660 match self.view {
5661 View::Source => prev_word(&self.source, off),
5662 View::Wysiwyg => self.glyph_word_left(off),
5663 }
5664 }
5665
5666 /// The word boundary to the right of `off` in the active view's space.
5667 fn word_right_from(&self, off: usize) -> usize {
5668 match self.view {
5669 View::Source => next_word(&self.source, off),
5670 View::Wysiwyg => self.glyph_word_right(off),
5671 }
5672 }
5673
5674 /// The character class of the glyph drawn at stop `off`.
5675 ///
5676 /// Read from the source, because a stop points at the source byte its glyph
5677 /// came from — the source *is* where the rendered character is written. What
5678 /// makes the walk glyph space rather than source space is that it only ever
5679 /// visits stops, and the hidden bytes between them have none.
5680 fn class_at(&self, off: usize) -> Class {
5681 self.source
5682 .get(off..)
5683 .and_then(|s| s.chars().next())
5684 .map_or(Class::Space, classify)
5685 }
5686
5687 /// [`next_word`] in glyph space: skip any leading separators, then consume
5688 /// the following word run, with the stop table standing in for the source's
5689 /// characters.
5690 fn glyph_word_right(&self, from: usize) -> usize {
5691 let Some(mut off) = self.vmap.stop_at_or_after(from) else {
5692 return from;
5693 };
5694 let mut in_word = false;
5695 loop {
5696 match self.class_at(off) {
5697 Class::Word => in_word = true,
5698 _ if in_word => return off,
5699 _ => {}
5700 }
5701 match self.vmap.stop_after(off) {
5702 Some(next) => off = next,
5703 None => return off,
5704 }
5705 }
5706 }
5707
5708 /// [`prev_word`] in glyph space: skip separators walking left, then consume
5709 /// the preceding word run.
5710 fn glyph_word_left(&self, from: usize) -> usize {
5711 let Some(mut off) = self.vmap.stop_at_or_before(from) else {
5712 return from;
5713 };
5714 let mut in_word = false;
5715 while let Some(prev) = self.vmap.stop_before(off) {
5716 match self.class_at(prev) {
5717 Class::Word => in_word = true,
5718 _ if in_word => return off,
5719 _ => {}
5720 }
5721 off = prev;
5722 }
5723 off
5724 }
5725
5726 /// After a motion that walks the visual map, the caret must be *on* the map.
5727 /// A stop is the only offset where the caret draws and edits in the same
5728 /// place, and it's the invariant both a caret parked inside an emoji and one
5729 /// parked inside a `**` were quietly breaking.
5730 ///
5731 /// Only when the caret actually moved: a walk with nowhere to go leaves it
5732 /// where it was, which is wherever the floor or a frontend put it rather
5733 /// than somewhere this motion chose.
5734 fn debug_assert_on_a_stop(&self, before: usize) {
5735 debug_assert!(
5736 self.view != View::Wysiwyg
5737 || self.vmap.num_rows() == 0
5738 || self.caret == before
5739 || self.vmap.is_stop(self.caret),
5740 "motion left the caret at {}, which is not a caret stop: it would draw in \
5741 one place and type in another",
5742 self.caret
5743 );
5744 }
5745
5746 // Up and Down run off the ends of the document rather than stopping dead at
5747 // them: Up from the first row lands at the document's start, Down from the
5748 // last at its end. That's Cocoa's rule (`moveUp:`/`moveDown:` past the edge
5749 // are `moveToBeginningOfDocument:`/`moveToEndOfDocument:`), and holding ↓
5750 // reaching the end of the text is what a reader means by it.
5751 //
5752 // The views used to disagree here by accident rather than by decision: the
5753 // source view fell into the edge behaviour through `row_col_to_offset`
5754 // clamping an out-of-range row to the end of the string, while WYSIWYG had
5755 // no row below to walk to and did nothing at all. They share the rule now,
5756 // each in its own space — the source view reaches every byte, WYSIWYG only
5757 // the offsets it draws.
5758
5759 pub fn move_up(&mut self, extend: bool) {
5760 let (row, col) = self.caret_pos();
5761 let goal = self.goal_col.unwrap_or(col);
5762 let target = match self.view {
5763 View::Source => match row.checked_sub(1) {
5764 Some(r) => row_col_to_offset(&self.source, r, goal),
5765 None => self.reachable_start(),
5766 },
5767 // A table's border rules are drawn but hold no caret, so Up steps
5768 // over them to the row that does.
5769 View::Wysiwyg => match self.vmap.navigable_above(row) {
5770 Some(r) => self.row_target(r, goal),
5771 None => self.reachable_start(),
5772 },
5773 };
5774 self.step_vertical(target, goal, extend);
5775 }
5776
5777 pub fn move_down(&mut self, extend: bool) {
5778 let (row, col) = self.caret_pos();
5779 let goal = self.goal_col.unwrap_or(col);
5780 let target = match self.view {
5781 View::Source => match self.source_row_below(row) {
5782 Some(r) => row_col_to_offset(&self.source, r, goal),
5783 None => self.reachable_end(),
5784 },
5785 View::Wysiwyg => match self.vmap.navigable_below(row) {
5786 Some(r) => self.row_target(r, goal),
5787 None => self.reachable_end(),
5788 },
5789 };
5790 self.step_vertical(target, goal, extend);
5791 }
5792
5793 /// Land a vertical motion at `target`, latching the `goal` column it aimed
5794 /// with so the rest of the run keeps aiming there.
5795 ///
5796 /// A motion with nowhere to go changes *nothing*, the goal column included:
5797 /// the latch used to run before the early return at the top of the document,
5798 /// so an Up that did nothing still armed a column, and the next Down aimed
5799 /// at one the caret had never been in.
5800 fn step_vertical(&mut self, target: usize, goal: usize, extend: bool) {
5801 let before = self.caret;
5802 if target == before {
5803 return;
5804 }
5805 self.goal_col = Some(goal);
5806 self.move_to(target, extend);
5807 self.debug_assert_on_a_stop(before);
5808 }
5809
5810 /// The source line below `row`, or `None` when `row` is the last one. Lines
5811 /// are counted by newline, so a trailing one leaves a real, empty last line
5812 /// for the caret to sit on — the document ends below it, not on it.
5813 fn source_row_below(&self, row: usize) -> Option<usize> {
5814 let last = self.source.bytes().filter(|&b| b == b'\n').count();
5815 (row < last).then_some(row + 1)
5816 }
5817
5818 /// Where a vertical motion aiming at the `goal` column lands on visual row
5819 /// `r`: the column clamped to the row, mapped to its offset, then held
5820 /// inside the row's own [bounds](Self::row_bounds) — a wrapped row's last
5821 /// column belongs to the row below, and a gutter's column 0 points at the
5822 /// block rather than at this row.
5823 fn row_target(&self, r: usize, goal: usize) -> usize {
5824 let (start, end) = self.row_bounds(r);
5825 self.vmap
5826 .offset_of_pos(r, goal.min(self.vmap.row_width(r)))
5827 .clamp(start, end)
5828 }
5829
5830 /// The first and last offsets the caret can reach in the active view.
5831 ///
5832 /// Not the same span in both: the source view shows every byte, so it can
5833 /// reach every byte. WYSIWYG reaches only what it draws — hidden frontmatter
5834 /// sits below the first stop, and a document's trailing newline is drawn
5835 /// nowhere and so sits past the last.
5836 fn reachable_start(&self) -> usize {
5837 match self.view {
5838 View::Source => 0,
5839 View::Wysiwyg => self.vmap.stop_at_or_after(0).unwrap_or(self.caret),
5840 }
5841 }
5842
5843 fn reachable_end(&self) -> usize {
5844 match self.view {
5845 View::Source => self.source.len(),
5846 View::Wysiwyg => self
5847 .vmap
5848 .stop_at_or_before(self.source.len())
5849 .unwrap_or(self.caret),
5850 }
5851 }
5852
5853 /// The `[start, end]` offsets visual row `r` *draws* — everything on it,
5854 /// including the space a soft wrap ate off its end, which is drawn on this
5855 /// row however much the offset past it belongs to the next one.
5856 fn row_span(&self, r: usize) -> (usize, usize) {
5857 let start = self
5858 .vmap
5859 .row_start(r)
5860 .unwrap_or_else(|| self.vmap.offset_of_pos(r, 0));
5861 let end = self.vmap.offset_of_pos(r, self.vmap.row_width(r));
5862 (start.min(end), end)
5863 }
5864
5865 /// [`row_span`](Self::row_span) narrowed to where the caret can stand: a
5866 /// soft wrap's shared offset opens the row below (see `pos_of_offset`), so
5867 /// this row's last position is the one before it — the offset before the
5868 /// space the wrap ate, where the caret draws just past the row's last word
5869 /// and types there too.
5870 ///
5871 /// Aiming at the shared offset instead is what stalled End: it is the row's
5872 /// last *column*, so End pressed on the row reached it and then read back as
5873 /// the row below's start, where a second press ran on to that row's end and
5874 /// the next to the one after — End walking down the paragraph a row a press.
5875 fn row_bounds(&self, r: usize) -> (usize, usize) {
5876 let (start, end) = self.row_span(r);
5877 let wraps = self
5878 .vmap
5879 .navigable_below(r)
5880 .and_then(|b| self.vmap.row_start(b))
5881 .is_some_and(|off| off == end);
5882 match wraps {
5883 true => (start, self.vmap.stop_before(end).unwrap_or(end).max(start)),
5884 false => (start, end),
5885 }
5886 }
5887
5888 /// The `[start, end]` of the line Home and End aim at: the visual row in
5889 /// WYSIWYG, the logical line in the source view. Both ends are caret stops.
5890 ///
5891 /// A soft-wrapped row is a line here, because it is one to the eye and the
5892 /// eye is what these keys are aimed by — a reader pressing End means the end
5893 /// of the line they can see. (`select_block_at` wants the opposite and reads
5894 /// the AST for it: a triple-click grabs the whole paragraph, however many
5895 /// rows it folds into.)
5896 fn line_bounds(&self) -> (usize, usize) {
5897 let (row, _) = self.caret_pos();
5898 match self.view {
5899 View::Source => {
5900 let start = line_start(&self.source, row);
5901 (start, line_end_from(&self.source, start))
5902 }
5903 View::Wysiwyg => self.row_bounds(row),
5904 }
5905 }
5906
5907 /// The same line as [`line_bounds`](Self::line_bounds), as far as it is
5908 /// *drawn* — what a kill takes.
5909 ///
5910 /// The two part only at a soft wrap, over the space the wrap ate: the caret
5911 /// can't stand after it (that offset opens the row below, and End stopping
5912 /// there would walk), but it is on this row, and a kill that spared it would
5913 /// leave a double space behind where the row's text had been. Deleting it
5914 /// joins nothing — a wrap is drawn, not written.
5915 fn line_span(&self) -> (usize, usize) {
5916 let (row, _) = self.caret_pos();
5917 match self.view {
5918 View::Source => self.line_bounds(),
5919 View::Wysiwyg => self.row_span(row),
5920 }
5921 }
5922
5923 /// The first offset in `[start, end]` holding something other than
5924 /// whitespace, or `end` when the line holds nothing else — where Home aims.
5925 ///
5926 /// Walks the space the view is in, as word motion does: WYSIWYG steps stops,
5927 /// so a hidden delimiter is never taken for the line's first character (nor
5928 /// landed on), and the source view steps the source it is showing.
5929 fn first_non_space(&self, start: usize, end: usize) -> usize {
5930 let mut off = start;
5931 while off < end {
5932 if self.class_at(off) != Class::Space {
5933 return off;
5934 }
5935 off = match self.view {
5936 View::Source => next_boundary(&self.source, off),
5937 View::Wysiwyg => match self.vmap.stop_after(off) {
5938 Some(next) => next,
5939 None => return end,
5940 },
5941 };
5942 }
5943 end
5944 }
5945
5946 /// Home: to the first character on the line, or to column 0 when the caret
5947 /// is already on it — the two-press toggle every editor spells this way.
5948 /// The indentation is somewhere the caret has to be able to reach and almost
5949 /// never where a reader is headed, so it costs the second press.
5950 pub fn move_home(&mut self, extend: bool) {
5951 self.goal_col = None;
5952 let (start, end) = self.line_bounds();
5953 let text = self.first_non_space(start, end);
5954 let target = if self.caret == text { start } else { text };
5955 let before = self.caret;
5956 self.move_to(target, extend);
5957 self.debug_assert_on_a_stop(before);
5958 }
5959
5960 /// End: to the end of the line.
5961 pub fn move_end(&mut self, extend: bool) {
5962 self.goal_col = None;
5963 let (_, end) = self.line_bounds();
5964 let before = self.caret;
5965 self.move_to(end, extend);
5966 self.debug_assert_on_a_stop(before);
5967 }
5968
5969 /// Hop to the next (Tab) or previous (Shift+Tab) table cell, landing with the
5970 /// cell's whole content selected (see [`Self::select_cell`]). Returns `false`
5971 /// when the caret isn't in a table, or is already in the last/first cell — the
5972 /// frontend then does whatever Tab normally does (indent), so Tab keeps its
5973 /// meaning everywhere else.
5974 pub fn cell_hop(&mut self, forward: bool) -> bool {
5975 let Some((grid, r, c)) = self.table_grid_at(self.caret) else {
5976 return false;
5977 };
5978 // Flatten to document (row-major) order and step one cell either way.
5979 let i: usize = grid[..r].iter().map(Vec::len).sum::<usize>() + c;
5980 let flat: Vec<(usize, usize)> = grid.into_iter().flatten().collect();
5981 let next = if forward {
5982 i.checked_add(1)
5983 } else {
5984 i.checked_sub(1)
5985 };
5986 let Some(&(start, end)) = next.and_then(|j| flat.get(j)) else {
5987 return false; // at the table's edge; leave Tab to the frontend
5988 };
5989 self.select_cell(start, end);
5990 true
5991 }
5992
5993 /// Move the caret to the cell directly above (`down == false`) or below in
5994 /// the same column, landing with the cell's whole content selected (see
5995 /// [`Self::select_cell`]). Returns `false` at the grid's top/bottom edge (or
5996 /// when the caret isn't in a table), so the frontend can fall through — the
5997 /// vertical counterpart of [`Self::cell_hop`].
5998 ///
5999 /// A ragged row that is short a column clamps to its last cell, so Down never
6000 /// falls out of the table over a gap the row above happened to have.
6001 pub fn cell_move_vertical(&mut self, down: bool) -> bool {
6002 let Some((grid, r, c)) = self.table_grid_at(self.caret) else {
6003 return false;
6004 };
6005 let target = match down {
6006 true => r + 1,
6007 false if r == 0 => return false,
6008 false => r - 1,
6009 };
6010 let Some(row) = grid.get(target) else {
6011 return false;
6012 };
6013 let Some(&(start, end)) = row.get(c).or_else(|| row.last()) else {
6014 return false;
6015 };
6016 self.select_cell(start, end);
6017 true
6018 }
6019
6020 /// The table containing `off` as a row-major grid of `(start, end)` cell
6021 /// caret homes, plus the `(row, col)` the caret sits in — `None` when `off`
6022 /// isn't in a table. Read straight off the visual map's laid-out grid, so
6023 /// every cell (an empty one included, whose derived home twig gives no
6024 /// `content_span` for) is present and in the order Tab walks them.
6025 // Grid, row, column — three returns that only ever travel together, and a
6026 // named type for the pair of them would be read at one call site.
6027 #[allow(clippy::type_complexity)]
6028 fn table_grid_at(&self, off: usize) -> Option<(Vec<Vec<(usize, usize)>>, usize, usize)> {
6029 for t in &self.vmap.tables {
6030 let mut pos = None;
6031 let grid: Vec<Vec<(usize, usize)>> = t
6032 .grid
6033 .iter()
6034 .enumerate()
6035 .map(|(r, row)| {
6036 row.cells
6037 .iter()
6038 .enumerate()
6039 .map(|(c, cell)| {
6040 if pos.is_none() && off >= cell.start && off <= cell.end {
6041 pos = Some((r, c));
6042 }
6043 (cell.start, cell.end)
6044 })
6045 .collect()
6046 })
6047 .collect();
6048 if let Some((r, c)) = pos {
6049 return Some((grid, r, c));
6050 }
6051 }
6052 None
6053 }
6054
6055 // ── table key policy ──────────────────────────────────────────────────────
6056 // The three keys a table gives its own meaning — Tab, Return, Shift+Return —
6057 // as one policy every frontend shares, rather than each re-deriving it. Each
6058 // reports whether it acted *as a table key*; a `false` hands the key back to
6059 // the frontend's ordinary handling (indent, newline) so it keeps its meaning
6060 // everywhere else.
6061
6062 /// Tab / Shift+Tab inside a table. Tab steps to the next cell, appending a
6063 /// fresh row and entering it when it runs off the last one; Shift+Tab steps
6064 /// back and simply stays put at the very first cell. `false` when the caret
6065 /// isn't in a table.
6066 pub fn cell_tab(&mut self, forward: bool) -> bool {
6067 if !self.caret_in_table() {
6068 return false;
6069 }
6070 if self.cell_hop(forward) {
6071 return true;
6072 }
6073 // Off the last cell: grow the table by a row and step into its first
6074 // cell. (Shift+Tab at the first cell has nowhere to go and just holds.)
6075 if forward {
6076 self.append_row_and_enter(0);
6077 }
6078 true
6079 }
6080
6081 /// Return inside a table: drop to the cell below in the same column,
6082 /// appending a new row when the caret is already in the last one. `false`
6083 /// when the caret isn't in a table, so the frontend inserts a newline.
6084 pub fn cell_return(&mut self) -> bool {
6085 if !self.caret_in_table() {
6086 return false;
6087 }
6088 if self.cell_move_vertical(true) {
6089 return true;
6090 }
6091 // Already on the last row: grow one below and drop into the same column.
6092 let col = self.table_grid_at(self.caret).map_or(0, |(_, _, c)| c);
6093 self.append_row_and_enter(col);
6094 true
6095 }
6096
6097 /// Append a row below the caret's (last) row and land in `col` of it. The
6098 /// caret is in the last row, so twig's "insert below" makes the fresh row the
6099 /// table's new last — but twig re-spells the whole table, moving every byte,
6100 /// so the destination is read back from the rebuilt grid by the table's
6101 /// position (stable across a row insert), not from the pre-edit caret.
6102 fn append_row_and_enter(&mut self, col: usize) {
6103 let table = self.caret_table_index();
6104 self.table_insert_row(true);
6105 self.rebuild_map();
6106 let Some((start, end)) = table
6107 .and_then(|ti| self.vmap.tables.get(ti))
6108 .and_then(|t| t.grid.last())
6109 .and_then(|row| row.cells.get(col.min(row.cells.len().saturating_sub(1))))
6110 .map(|cell| (cell.start, cell.end))
6111 else {
6112 return;
6113 };
6114 self.select_cell(start, end);
6115 }
6116
6117 /// The index, among the document's tables, of the one the caret sits in —
6118 /// `None` when it's in none. Used to re-find a table after an edit re-spells
6119 /// it (a row insert leaves the table order unchanged).
6120 fn caret_table_index(&self) -> Option<usize> {
6121 let off = self.caret;
6122 self.vmap.tables.iter().position(|t| {
6123 t.grid
6124 .iter()
6125 .any(|row| row.cells.iter().any(|c| off >= c.start && off <= c.end))
6126 })
6127 }
6128
6129 /// Shift+Return inside a table: insert a hard line break *within* the current
6130 /// cell, via twig's `insert_line_break`. `false` when the caret isn't in a
6131 /// table, so the frontend inserts an ordinary line break.
6132 ///
6133 /// A table row is a single source line, so the newline-spelled hard break
6134 /// can't live in a cell. twig spells the in-cell break the format's way
6135 /// (`<br>` for Markdown) and reparses it as a *semantic* `hard_break`, so the
6136 /// break round-trips as structure the renderer reads back as a line — not the
6137 /// opaque raw HTML the old raw-splice left behind.
6138 ///
6139 /// Djot has no idiomatic in-cell break, so twig refuses it
6140 /// (`UnsupportedFormat`) rather than emit a `<br>` that any other djot reader
6141 /// would render as the literal text `<br>`. The gesture is still *consumed*
6142 /// there — returning `false` would let the frontend insert a real newline,
6143 /// which splits the one-line row — it just leaves the cell unchanged and says
6144 /// so on the status line. A rollback (`EditConflict`) is swallowed the same.
6145 ///
6146 /// Which formats refuse is [`Capabilities::cell_line_break`], and the two
6147 /// have to be read together: djot is not the only `false`, and naming it in
6148 /// the message was already a guess that HTML — which spells the break as its
6149 /// own `<br>` — would have made wrong.
6150 pub fn cell_line_break(&mut self) -> bool {
6151 if self.read_only || !self.caret_in_table() {
6152 return false;
6153 }
6154 self.record_caret();
6155 match self.editor.insert_line_break(self.caret) {
6156 Ok(change) => {
6157 self.last_edit_kind = None;
6158 self.refresh();
6159 self.caret = change.new.end;
6160 self.anchor = None;
6161 self.goal_col = None;
6162 self.clamp_caret();
6163 self.dirty = self.source != self.clean_source;
6164 self.status = None;
6165 self.record_caret();
6166 }
6167 Err(twig::Error::UnsupportedFormat) => {
6168 self.status = Some(format!(
6169 "in-cell line breaks aren't supported in {}",
6170 self.format_name()
6171 ));
6172 }
6173 Err(_) => {}
6174 }
6175 true
6176 }
6177
6178 /// Rebuild the visual map at the width the last build used. A structural edit
6179 /// bumps the revision and swaps the source in, but leaves the *map* stale;
6180 /// when a single gesture edits and then moves over the result (Tab appending
6181 /// a row, then stepping into it), the move needs the map to already show the
6182 /// edit rather than waiting for the frontend's next frame.
6183 fn rebuild_map(&mut self) {
6184 let wrap = self.vmap_key.as_ref().and_then(|(_, w, _)| *w);
6185 self.build_map(wrap);
6186 }
6187
6188 /// Move the caret to the very start of the document (⌘↑ on macOS,
6189 /// Ctrl+Home on Windows/Linux).
6190 pub fn move_doc_start(&mut self, extend: bool) {
6191 self.goal_col = None;
6192 self.move_to(0, extend);
6193 }
6194
6195 /// Move the caret to the very end of the document (⌘↓ on macOS,
6196 /// Ctrl+End on Windows/Linux).
6197 pub fn move_doc_end(&mut self, extend: bool) {
6198 self.goal_col = None;
6199 let end = self.source.len();
6200 self.move_to(end, extend);
6201 }
6202
6203 /// Point the caret at the body cell `(row, col)` the mouse landed on —
6204 /// `col` being a cell of the terminal grid, which is what a display column
6205 /// is. A click on the far cell of a wide character lands at that
6206 /// character's start; the mapping's own doc-comments carry the rule.
6207 pub fn click(&mut self, row: usize, col: usize, extend: bool) {
6208 self.goal_col = None;
6209 let target = match self.view {
6210 View::Source => row_col_to_offset(&self.source, row, col),
6211 View::Wysiwyg => self.vmap.offset_of_pos(row, col),
6212 };
6213 let before = self.caret;
6214 self.move_to(target, extend);
6215 self.debug_assert_on_a_stop(before);
6216 }
6217
6218 /// Settle `scroll` for a frame about to be drawn: follow the caret onto the
6219 /// screen if it has moved since the last frame, and never scroll past the
6220 /// last of `rows`.
6221 ///
6222 /// Only if it has *moved* — that's the whole point. Revealing the caret on
6223 /// every frame ties the viewport to it, and a scroll wheel that fights the
6224 /// caret for the viewport loses: the view snaps back the instant it tries to
6225 /// pass the caret's row, so the document can't be scrolled beyond what's
6226 /// already on screen. A caret move is the frontend's cue to follow; a scroll
6227 /// with the caret sitting still is the reader's cue to leave it alone.
6228 pub fn follow_caret(&mut self, caret_row: usize, height: usize, rows: usize) {
6229 if self.drawn_caret != Some(self.caret) {
6230 if caret_row < self.scroll {
6231 self.scroll = caret_row;
6232 } else if height > 0 && caret_row >= self.scroll + height {
6233 self.scroll = caret_row + 1 - height;
6234 }
6235 self.drawn_caret = Some(self.caret);
6236 }
6237 self.scroll = self.scroll.min(rows.saturating_sub(1));
6238 }
6239
6240 /// The caret's screen position `(row, col)` in the active view's grid, with
6241 /// `col` a display column: the cell to draw the caret in, which on a line of
6242 /// `你好` or emoji is not the count of characters before it.
6243 pub fn caret_pos(&self) -> (usize, usize) {
6244 match self.view {
6245 View::Source => offset_to_row_col(&self.source, self.caret),
6246 View::Wysiwyg => self.vmap.pos_of_offset(self.caret),
6247 }
6248 }
6249
6250 fn clamp_caret(&mut self) {
6251 if self.caret > self.source.len() {
6252 self.caret = self.source.len();
6253 }
6254 // In WYSIWYG the caret can't sit inside hidden frontmatter; lift it (and
6255 // any selection anchor) to the first rendered offset.
6256 let floor = self.caret_floor();
6257 if self.caret < floor {
6258 self.caret = floor;
6259 }
6260 if let Some(a) = self.anchor
6261 && a < floor
6262 {
6263 self.anchor = Some(floor);
6264 }
6265 while self.caret > 0 && !self.source.is_char_boundary(self.caret) {
6266 self.caret -= 1;
6267 }
6268 }
6269}
6270
6271// ── byte-offset ⇄ (row, col) helpers ─────────────────────────────────────────
6272
6273// Left/right motion and backspace/delete step by *grapheme cluster*, not
6274// codepoint, so an emoji (a ZWJ sequence) or a base letter plus its combining
6275// marks moves and deletes as the single character a user sees. Grapheme
6276// boundaries are a superset of char boundaries, so the caret stays valid for twig.
6277
6278/// How an insert of `text` groups for undo: a single typed character folds into
6279/// the run of typing around it, while a newline or a multi-character insert is a
6280/// step of its own.
6281fn typed_edit_kind(text: &str) -> EditKind {
6282 if text.chars().take(2).count() == 1 && text != "\n" {
6283 EditKind::Insert
6284 } else {
6285 EditKind::Other
6286 }
6287}
6288
6289fn prev_boundary(s: &str, i: usize) -> usize {
6290 let mut cursor = GraphemeCursor::new(i, s.len(), true);
6291 cursor.prev_boundary(s, 0).ok().flatten().unwrap_or(0)
6292}
6293
6294fn next_boundary(s: &str, i: usize) -> usize {
6295 let mut cursor = GraphemeCursor::new(i, s.len(), true);
6296 cursor.next_boundary(s, 0).ok().flatten().unwrap_or(s.len())
6297}
6298
6299// ── word boundaries ──────────────────────────────────────────────────────────
6300// The shared primitive behind word-wise motion, word deletion, and
6301// double-click-to-select-a-word. A "word" is a maximal run of one character
6302// class; whitespace and punctuation are their own classes, so motion skips
6303// cleanly between them the way native text fields do.
6304
6305#[derive(PartialEq, Eq, Clone, Copy)]
6306enum Class {
6307 Word,
6308 Space,
6309 Other,
6310}
6311
6312/// The source range of an inline node's own visible text — the part of it a
6313/// WYSIWYG caret can reach, as against the delimiters that only spell it.
6314/// `None` for a node with no interior to empty (a `str`, a break).
6315///
6316/// twig reports no `content_span` for `verbatim`/`inline_math`, whose text sits
6317/// one delimiter in from the span — the same place the renderer maps it to. A
6318/// longer fence (`` ``a`` ``) breaks that assumption, so the guess is checked
6319/// against the source rather than trusted: a range guessed wrong here is text
6320/// deleted wrong.
6321fn inline_content_span(n: &FlatNode, source: &str) -> Option<std::ops::Range<usize>> {
6322 if let Some(span) = n.content_span.clone() {
6323 return Some(span);
6324 }
6325 match n.kind.as_str() {
6326 "verbatim" | "inline_math" => {
6327 let text = n.text.as_ref()?;
6328 let start = n.span.start + 1;
6329 let range = start..start + text.len();
6330 (source.get(range.clone()) == Some(text.as_str())).then_some(range)
6331 }
6332 _ => None,
6333 }
6334}
6335
6336/// The `id` a node declares, or `None` for one that declares none — the
6337/// attribute djot writes for a `{#v1}` and mints for a heading.
6338///
6339/// A bare attribute (`{#v1 hidden}`'s `hidden`) has no value, and a bare `id`
6340/// names nothing, so it reads as absent rather than as the empty string.
6341fn declared_id(n: &FlatNode) -> Option<&str> {
6342 n.attrs.iter().find(|(k, _)| k == "id")?.1.as_deref()
6343}
6344
6345/// A heading's words reduced to the form a link fragment spells them in:
6346/// lowercase, runs of anything else collapsed to a single `-`, with none left
6347/// dangling at either end. `## Some Heading Here` → `some-heading-here`.
6348///
6349/// The rule every Markdown renderer follows, and applied to djot's own auto-ids
6350/// too so that `#some-heading-here` and `#Some-Heading-Here` are one question.
6351/// Unicode-aware (`is_alphanumeric`, not an ASCII test), because a heading in
6352/// any other language is still a heading someone will link to. Underscores
6353/// survive for the same reason they do on the web: they are word characters
6354/// wherever identifiers are written.
6355fn slug(text: &str) -> String {
6356 let mut out = String::new();
6357 let mut pending = false;
6358 for c in text.chars() {
6359 if c.is_alphanumeric() || c == '_' {
6360 if pending && !out.is_empty() {
6361 out.push('-');
6362 }
6363 pending = false;
6364 out.extend(c.to_lowercase());
6365 } else {
6366 pending = true;
6367 }
6368 }
6369 out
6370}
6371
6372fn is_block_container(kind: &Kind) -> bool {
6373 matches!(
6374 kind,
6375 Kind::Doc
6376 | Kind::Section
6377 | Kind::BlockQuote
6378 | Kind::BulletList
6379 | Kind::OrderedList
6380 | Kind::TaskList
6381 | Kind::ListItem
6382 | Kind::TaskListItem
6383 // Every `container` — a directive in any of its three forms, or a
6384 // promoted HTML element. A *text* directive is really inline, so
6385 // claiming it here is a small overreach, and the deliberate one this
6386 // function's kind-only peer `is_inline_kind` documents: the pair is
6387 // consulted together, and answering "block container" for something
6388 // inline is what keeps an ancestor walk from stopping short of the
6389 // paragraph that actually holds it.
6390 | Kind::Container
6391 )
6392}
6393
6394/// The `[start, end)` byte range of the source line containing `off` (newline
6395/// excluded) — the fallback when `off` sits outside any AST block (e.g. a blank
6396/// line between paragraphs).
6397fn source_line_range(s: &str, off: usize) -> std::ops::Range<usize> {
6398 let off = off.min(s.len());
6399 let start = s[..off].rfind('\n').map(|p| p + 1).unwrap_or(0);
6400 let end = s[off..].find('\n').map(|p| off + p).unwrap_or(s.len());
6401 start..end
6402}
6403
6404/// How many leading bytes an outdent takes off `line`: a whole indent level
6405/// where the line has one, and whatever it has where it has less.
6406///
6407/// A leading tab counts as a level on its own. It's indentation some other
6408/// editor wrote, and one tab is one level everywhere it came from — measuring it
6409/// in spaces it doesn't contain would leave it untouchable.
6410fn outdent_width(line: &str, unit: usize) -> usize {
6411 if line.starts_with('\t') {
6412 return 1;
6413 }
6414 line.bytes().take(unit).take_while(|b| *b == b' ').count()
6415}
6416
6417/// A list marker found at the head of a line, together with everything before it
6418/// that a sibling line has to repeat.
6419///
6420/// The three offsets differ only inside a block quote, where `> - b` opens with
6421/// a `> ` quote marker the line's own text doesn't own. Outside one they collapse:
6422/// `line_start == marker_start`, and `text` is the plain `" - "`.
6423#[derive(Clone, Debug)]
6424struct ListMarker {
6425 /// The line's first byte.
6426 line_start: usize,
6427 /// Where the marker proper begins, past any quote prefix. The offset to hand
6428 /// the AST: a quoted item's span opens at its bullet, not at the `>`.
6429 marker_start: usize,
6430 /// `line_start` through the marker's trailing space — quote prefix, indent
6431 /// and bullet together, which is what the next item's line opens with.
6432 text: String,
6433}
6434
6435impl ListMarker {
6436 /// Where the item's content starts — one past the marker's trailing space.
6437 fn content_start(&self) -> usize {
6438 self.line_start + self.text.len()
6439 }
6440}
6441
6442fn classify(c: char) -> Class {
6443 if c == '_' || c.is_alphanumeric() {
6444 Class::Word
6445 } else if c.is_whitespace() {
6446 Class::Space
6447 } else {
6448 Class::Other
6449 }
6450}
6451
6452/// The offset at the end of the next word to the right of `i` (⌥→ / Ctrl+→):
6453/// skip any leading separators, then consume the following word run.
6454fn next_word(s: &str, i: usize) -> usize {
6455 let mut off = i;
6456 let mut in_word = false;
6457 for c in s[i..].chars() {
6458 if classify(c) == Class::Word {
6459 in_word = true;
6460 } else if in_word {
6461 break;
6462 }
6463 off += c.len_utf8();
6464 }
6465 off
6466}
6467
6468/// The offset at the start of the word to the left of `i` (⌥← / Ctrl+←):
6469/// skip separators walking left, then consume the preceding word run.
6470fn prev_word(s: &str, i: usize) -> usize {
6471 let mut off = i;
6472 let mut in_word = false;
6473 for c in s[..i].chars().rev() {
6474 if classify(c) == Class::Word {
6475 in_word = true;
6476 } else if in_word {
6477 break;
6478 }
6479 off -= c.len_utf8();
6480 }
6481 off
6482}
6483
6484/// The `[start, end)` run of same-class characters surrounding `off` — the
6485/// word (or whitespace/punctuation run) a double-click selects. At end-of-text
6486/// the run ending there is used.
6487fn word_range_at(s: &str, off: usize) -> (usize, usize) {
6488 if s.is_empty() {
6489 return (0, 0);
6490 }
6491 let off = off.min(s.len());
6492 let reference = if off < s.len() {
6493 s[off..].chars().next()
6494 } else {
6495 s[..off].chars().next_back()
6496 };
6497 let Some(rc) = reference else {
6498 return (off, off);
6499 };
6500 let class = classify(rc);
6501
6502 let mut start = off;
6503 for c in s[..start].chars().rev() {
6504 if classify(c) == class {
6505 start -= c.len_utf8();
6506 } else {
6507 break;
6508 }
6509 }
6510 let mut end = off;
6511 for c in s[end..].chars() {
6512 if classify(c) == class {
6513 end += c.len_utf8();
6514 } else {
6515 break;
6516 }
6517 }
6518 (start, end)
6519}
6520
6521/// `(row, col)` of byte offset `off`, `col` counted in *display columns* from
6522/// the line's start — terminal cells, not characters, so the column names the
6523/// cell the caret is drawn in even on a line of `你好` or emoji.
6524fn offset_to_row_col(s: &str, off: usize) -> (usize, usize) {
6525 let off = off.min(s.len());
6526 let mut row = 0;
6527 let mut line_start = 0;
6528 for (i, &b) in s.as_bytes().iter().enumerate() {
6529 if i >= off {
6530 break;
6531 }
6532 if b == b'\n' {
6533 row += 1;
6534 line_start = i + 1;
6535 }
6536 }
6537 (row, wysiwyg::text_width(&s[line_start..off]))
6538}
6539
6540/// The byte offset at display column `col` of `row` (clamped to that line's
6541/// end) — the inverse of [`offset_to_row_col`], which it has to agree with.
6542///
6543/// A column landing *inside* a character — the second cell of `你`, or any cell
6544/// but the first of an emoji — resolves to that character's start, which is the
6545/// column the caret would have been drawn at to begin with. So both cells of a
6546/// wide character mean the character, and every offset survives the round trip
6547/// out to a column and back. The walk steps by grapheme cluster for the same
6548/// reason the caret does: a cluster is the character, and the cells belong to it
6549/// rather than to the codepoints spelling it.
6550fn row_col_to_offset(s: &str, row: usize, col: usize) -> usize {
6551 let start = line_start(s, row);
6552 let end = line_end_from(s, start);
6553 let mut off = start;
6554 let mut at = 0; // the display column `off` sits at
6555 while off < end {
6556 let next = next_boundary(s, off).min(end);
6557 let cells = wysiwyg::text_width(&s[off..next]);
6558 if at + cells > col {
6559 break; // `col` is one of this cluster's own cells
6560 }
6561 at += cells;
6562 off = next;
6563 }
6564 off
6565}
6566
6567fn line_start(s: &str, row: usize) -> usize {
6568 if row == 0 {
6569 return 0;
6570 }
6571 let mut r = 0;
6572 for (i, &b) in s.as_bytes().iter().enumerate() {
6573 if b == b'\n' {
6574 r += 1;
6575 if r == row {
6576 return i + 1;
6577 }
6578 }
6579 }
6580 s.len()
6581}
6582
6583fn line_end_from(s: &str, start: usize) -> usize {
6584 s[start..].find('\n').map(|p| start + p).unwrap_or(s.len())
6585}
6586
6587/// twig's node-kind name for an inline mark, back to the [`InlineKind`] a
6588/// frontend names when it calls [`Doc::toggle`] — the inverse of the mapping
6589/// twig applies writing the mark out, so the toolbar can light the same button
6590/// that made the node.
6591///
6592/// `None` for every other kind, including the inline nodes that aren't marks at
6593/// all (`str`, `link`, `image`, the math and break kinds): they're things a
6594/// caret stands in, not formatting a button toggles.
6595fn inline_kind(kind: &Kind) -> Option<InlineKind> {
6596 Some(match kind {
6597 Kind::Strong => InlineKind::Strong,
6598 Kind::Emph => InlineKind::Emph,
6599 Kind::Verbatim => InlineKind::Verbatim,
6600 Kind::Mark => InlineKind::Mark,
6601 Kind::Superscript => InlineKind::Superscript,
6602 Kind::Subscript => InlineKind::Subscript,
6603 Kind::Insert => InlineKind::Insert,
6604 Kind::Delete => InlineKind::Delete,
6605 _ => return None,
6606 })
6607}
6608
6609/// leaf's [`MarkColor`] as twig's — the palette twig writes as the emoji after
6610/// a highlight's opening `==`.
6611///
6612/// Two enums for one closed vocabulary, and the duplication is the boundary
6613/// working: core's is what a *frontend* names (`style::MarkColor`, beside the
6614/// [`Role`](crate::Role) that carries it into the glyph map) and twig's is what
6615/// the editor writes. Spelled as a match rather than routed through the two
6616/// crates' name strings so that a colour added on either side is a compile
6617/// error here, where the pairing is decided, rather than a runtime `None` that
6618/// would read as "clear the colour".
6619fn twig_mark_color(color: MarkColor) -> twig::MarkColor {
6620 match color {
6621 MarkColor::Red => twig::MarkColor::Red,
6622 MarkColor::Orange => twig::MarkColor::Orange,
6623 MarkColor::Yellow => twig::MarkColor::Yellow,
6624 MarkColor::Green => twig::MarkColor::Green,
6625 MarkColor::Blue => twig::MarkColor::Blue,
6626 MarkColor::Purple => twig::MarkColor::Purple,
6627 MarkColor::Brown => twig::MarkColor::Brown,
6628 }
6629}
6630
6631/// Where an offset lands after a splice it didn't make — twig's own rule, from
6632/// [`Change`]: shift anything at or past the replaced range's end by the length
6633/// the replacement gained or lost, and leave anything before it alone.
6634///
6635/// An offset *inside* the replaced range has no text of its own to ride any
6636/// more, and lands at the end of what replaced it: for
6637/// [`Doc::set_mark_color`] that is a caret standing on the colour prefix when
6638/// the prefix is cleared, which then sits where the highlighted text begins.
6639fn reanchor(off: usize, change: &Change) -> usize {
6640 if off < change.old.start {
6641 return off;
6642 }
6643 if off < change.old.end {
6644 return change.new.end;
6645 }
6646 (off + change.new.end).saturating_sub(change.old.end)
6647}
6648
6649/// A watermark for a file's contents (see `Doc::disk_hash`).
6650///
6651/// `DefaultHasher` is not stable across Rust releases, which doesn't matter: a
6652/// watermark is compared only against one taken by the same process moments
6653/// earlier, and never outlives it. 64 bits leaves a collision — an external edit
6654/// that hashes to exactly what leaf wrote — at odds no filesystem race gets near.
6655fn hash_bytes(bytes: &[u8]) -> u64 {
6656 use std::hash::{Hash, Hasher};
6657 let mut h = std::collections::hash_map::DefaultHasher::new();
6658 bytes.hash(&mut h);
6659 h.finish()
6660}
6661
6662#[cfg(feature = "fs")]
6663fn detect_format(path: &Path) -> Result<Format> {
6664 let ext = path
6665 .extension()
6666 .and_then(|e| e.to_str())
6667 .unwrap_or("")
6668 .to_ascii_lowercase();
6669 Ok(match ext.as_str() {
6670 "dj" | "djot" => Format::Djot,
6671 "md" | "markdown" => Format::Markdown,
6672 "xml" => Format::Xml,
6673 "html" | "htm" => Format::Html,
6674 other => return Err(anyhow!("unknown document extension: .{other}")),
6675 })
6676}
6677
6678#[cfg(test)]
6679mod tests {
6680 use super::*;
6681
6682 /// A document open in `view`. WYSIWYG motion reads the visual map, which the
6683 /// renderer stamps each frame, so the map is built here too — a WYSIWYG doc
6684 /// without one is a view no user is ever in.
6685 fn doc_in(view: View, name: &str, body: &str) -> Doc {
6686 // The fixture name doubles as the temp file's, so two tests picking the
6687 // same one raced under the parallel runner and read each other's body —
6688 // a green suite proving the wrong thing. The counter makes that
6689 // unreachable rather than asking every future caller to notice.
6690 static SEQ: std::sync::atomic::AtomicUsize = std::sync::atomic::AtomicUsize::new(0);
6691 let seq = SEQ.fetch_add(1, std::sync::atomic::Ordering::Relaxed);
6692 let mut p = std::env::temp_dir();
6693 p.push(format!("leaf_test_{name}_{seq}.md"));
6694 std::fs::write(&p, body).unwrap();
6695 let mut d = Doc::open(p).unwrap();
6696 d.view = view;
6697 if view == View::Wysiwyg {
6698 d.build_visual(80);
6699 }
6700 d
6701 }
6702
6703 // Source-view document for the source-behaviour tests. `Doc::open` now
6704 // defaults to WYSIWYG (leaf's default view), so pin the source view here;
6705 // `wysiwyg_doc` builds the rich-text variant on top of this.
6706 fn doc_with(name: &str, body: &str) -> Doc {
6707 doc_in(View::Source, name, body)
6708 }
6709
6710 /// Every visual row's drawn text — what the reader actually sees, which is
6711 /// the only thing the reveal preference is supposed to change.
6712 fn drawn_rows(d: &Doc) -> Vec<String> {
6713 d.vmap
6714 .rows
6715 .iter()
6716 .map(|r| r.glyphs.iter().map(|g| g.ch).collect())
6717 .collect()
6718 }
6719
6720 /// Put the caret at the first byte of `needle` and rebuild, so the row under
6721 /// it becomes the revealed line.
6722 fn caret_at(d: &mut Doc, needle: &str) {
6723 d.caret = d.source.find(needle).expect("needle in source");
6724 d.build_visual(80);
6725 }
6726
6727 #[test]
6728 fn blockquote_after_a_list_is_not_bulleted() {
6729 // twig nests a following top-level block quote under the `bullet_list`
6730 // (a direct child, not a `list_item`). The map must render it de-nested —
6731 // `│ quote`, never `• │ quote` — with a blank separator, like any block
6732 // that follows a list. Regression for the "combined list + blockquote" bug.
6733 let mut d = doc_in(View::Wysiwyg, "bq_after_list", "- item\n\n> quote\n");
6734 d.build_visual(80);
6735 let rows: Vec<String> = d
6736 .vmap
6737 .rows
6738 .iter()
6739 .map(|r| r.glyphs.iter().map(|g| g.ch).collect())
6740 .collect();
6741 assert!(
6742 rows.iter().any(|r| r == "│ quote"),
6743 "block quote should render on its own gutter, got rows: {rows:?}"
6744 );
6745 assert!(
6746 !rows.iter().any(|r| r.contains('•') && r.contains('│')),
6747 "no row should carry both a bullet and a quote gutter, got rows: {rows:?}"
6748 );
6749 }
6750
6751 // ── the map is built at most once per (revision, wrap) ───────────────────
6752 //
6753 // A frontend repaints for reasons that have nothing to do with the text — a
6754 // blinking caret, a scroll — and rebuilding the map is O(document). These
6755 // pin *that the cache fires*, which a passing suite can't tell you: a cache
6756 // that never hits is invisible to every other test in this file.
6757 //
6758 // The probe is to wreck the built map and ask for it again. A rebuild
6759 // repairs it; a cache hit hands the wreckage straight back. Nothing else
6760 // can distinguish the two from outside.
6761
6762 #[test]
6763 fn a_rebuild_with_nothing_changed_reuses_the_map() {
6764 let mut d = doc_in(View::Wysiwyg, "cache_hit", "# Title\n\nbody\n");
6765 d.build_visual(80);
6766 assert!(!d.vmap.rows.is_empty());
6767 d.vmap.rows.clear(); // wreck it
6768 d.build_visual(80);
6769 assert!(
6770 d.vmap.rows.is_empty(),
6771 "the map was rebuilt though nothing changed — the cache never fired"
6772 );
6773 }
6774
6775 #[test]
6776 fn an_edit_rebuilds_the_map() {
6777 let mut d = doc_in(View::Wysiwyg, "cache_edit", "# Title\n\nbody\n");
6778 d.build_visual(80);
6779 let before = d.revision();
6780 d.vmap.rows.clear();
6781 d.insert("x");
6782 d.build_visual(80);
6783 assert!(d.revision() > before, "an edit must move the revision");
6784 assert!(
6785 !d.vmap.rows.is_empty(),
6786 "an edited document must not paint from a stale map"
6787 );
6788 }
6789
6790 #[test]
6791 fn a_width_change_rebuilds_the_map() {
6792 // The map is a function of the wrap width too, so a resize is a miss
6793 // even though the text is untouched.
6794 let mut d = doc_in(
6795 View::Wysiwyg,
6796 "cache_width",
6797 "one two three four five six\n",
6798 );
6799 d.build_visual(80);
6800 d.vmap.rows.clear();
6801 d.build_visual(12);
6802 assert!(!d.vmap.rows.is_empty(), "a resize must rebuild the map");
6803 // And the unwrapped map is its own key, not the same as any width.
6804 d.vmap.rows.clear();
6805 d.build_visual_unwrapped();
6806 assert!(!d.vmap.rows.is_empty(), "unwrapped is a different map");
6807 }
6808
6809 #[test]
6810 fn a_motion_does_not_rebuild_the_map() {
6811 // The whole point: moving the caret changes nothing the map is built
6812 // from. If a motion bumped the revision, every arrow key would cost a
6813 // full rebuild and the cache would be worthless.
6814 let mut d = doc_in(View::Wysiwyg, "cache_motion", "# Title\n\nbody text\n");
6815 d.build_visual(80);
6816 let rev = d.revision();
6817 d.move_right(false);
6818 d.move_right(true);
6819 d.move_down(false);
6820 assert_eq!(d.revision(), rev, "a motion must not move the revision");
6821 d.vmap.rows.clear();
6822 d.build_visual(80);
6823 assert!(
6824 d.vmap.rows.is_empty(),
6825 "a motion should not rebuild the map"
6826 );
6827 }
6828
6829 #[test]
6830 fn saving_does_not_rebuild_the_map() {
6831 // Saving changes `dirty`, not the text.
6832 let mut d = doc_in(View::Wysiwyg, "cache_save", "# Title\n\nbody\n");
6833 d.insert("x");
6834 d.build_visual(80);
6835 let rev = d.revision();
6836 d.save();
6837 assert_eq!(d.revision(), rev, "a save must not move the revision");
6838 assert!(!d.dirty, "the save should have cleaned the document");
6839 }
6840
6841 #[test]
6842 fn a_reload_rebuilds_the_map() {
6843 // Reload replaces the text without going through `refresh`, so it has to
6844 // move the revision itself — else the editor paints the old file.
6845 let mut d = doc_in(View::Wysiwyg, "cache_reload", "# Title\n\nbody\n");
6846 d.build_visual(80);
6847 let rev = d.revision();
6848 std::fs::write(&d.path, "# Other\n\nwholly new\n").unwrap();
6849 d.reload();
6850 assert!(d.revision() > rev, "a reload must move the revision");
6851 d.build_visual(80);
6852 let text: String = d
6853 .vmap
6854 .rows
6855 .iter()
6856 .flat_map(|r| r.glyphs.iter().map(|g| g.ch))
6857 .collect();
6858 assert!(
6859 text.contains("wholly new"),
6860 "the reloaded text should be on screen, got {text:?}"
6861 );
6862 }
6863
6864 // ── golden-case harness ──────────────────────────────────────────────────
6865 // The pattern the whole parity suite can reuse: write a fixture with the
6866 // caret marked by `|`, run one action, and compare the rendered result —
6867 // also caret-marked — against the expected string. One readable line per
6868 // behavior, and it exercises the exact `Doc` ops both frontends call.
6869
6870 /// Split a `|`-marked fixture into `(source, caret_offset)`.
6871 fn parse_caret(marked: &str) -> (String, usize) {
6872 let caret = marked.find('|').expect("fixture needs a `|` caret marker");
6873 (marked.replacen('|', "", 1), caret)
6874 }
6875
6876 /// Render a doc's source with `|` at the caret (and `[`…`]` around any
6877 /// selection) so a result reads like the fixtures.
6878 fn render_caret(d: &Doc) -> String {
6879 // (offset, rank, char); rank keeps coincident markers ordered `[ | ]`
6880 // so the caret always renders inside its own selection.
6881 let mut marks: Vec<(usize, u8, char)> = vec![(d.caret, 1, '|')];
6882 if let Some((s, e)) = d.selection() {
6883 marks.push((s, 0, '['));
6884 marks.push((e, 2, ']'));
6885 }
6886 // Insert right-to-left: descending offset, then descending rank.
6887 marks.sort_by(|a, b| b.0.cmp(&a.0).then(b.1.cmp(&a.1)));
6888 let mut out = d.source.clone();
6889 for (at, _, ch) in marks {
6890 out.insert(at, ch);
6891 }
6892 out
6893 }
6894
6895 /// Load a `|`-marked fixture, run `action`, return the caret-marked result.
6896 fn golden(name: &str, marked: &str, action: impl FnOnce(&mut Doc)) -> String {
6897 golden_in(View::Source, name, marked, action)
6898 }
6899
6900 /// [`golden`] in a chosen view — the editing ops are the view's to share, so
6901 /// the same fixture has to read the same way in both.
6902 fn golden_in(view: View, name: &str, marked: &str, action: impl FnOnce(&mut Doc)) -> String {
6903 let (src, caret) = parse_caret(marked);
6904 let mut d = doc_in(view, name, &src);
6905 d.caret = caret;
6906 action(&mut d);
6907 render_caret(&d)
6908 }
6909
6910 #[test]
6911 fn word_motion_walks_word_by_word() {
6912 let g = |m, f: fn(&mut Doc)| golden("word_motion", m, f);
6913 assert_eq!(
6914 g("hello wor|ld", |d| d.move_word_left(false)),
6915 "hello |world"
6916 );
6917 assert_eq!(
6918 g("hello| world", |d| d.move_word_left(false)),
6919 "|hello world"
6920 );
6921 assert_eq!(
6922 g("hel|lo world", |d| d.move_word_right(false)),
6923 "hello| world"
6924 );
6925 assert_eq!(
6926 g("hello| world", |d| d.move_word_right(false)),
6927 "hello world|"
6928 );
6929 // Punctuation is its own class, so motion stops at the boundary.
6930 assert_eq!(g("|foo.bar", |d| d.move_word_right(false)), "foo|.bar");
6931 }
6932
6933 #[test]
6934 fn word_motion_extends_the_selection_when_asked() {
6935 assert_eq!(
6936 golden("word_sel", "hello |world", |d| d.move_word_right(true)),
6937 "hello [world|]"
6938 );
6939 }
6940
6941 #[test]
6942 fn delete_word_removes_a_whole_word() {
6943 let g = |m, f: fn(&mut Doc)| golden("del_word", m, f);
6944 assert_eq!(g("hello world|", |d| d.delete_word_back()), "hello |");
6945 assert_eq!(g("hello |world", |d| d.delete_word_forward()), "hello |");
6946 assert_eq!(g("foo |bar baz", |d| d.delete_word_back()), "|bar baz");
6947 }
6948
6949 // ── Home / End ───────────────────────────────────────────────────────────
6950
6951 #[test]
6952 fn home_toggles_between_the_line_s_text_and_its_margin() {
6953 // Source: the indentation is what the toggle is for. WYSIWYG resolves an
6954 // indent to the markup it spells everywhere it means one, so the fixture
6955 // with whitespace left to walk is a code block, which is verbatim.
6956 let g = |m, f: fn(&mut Doc)| golden("smart_home", m, f);
6957 assert_eq!(g(" inden|ted", |d| d.move_home(false)), " |indented");
6958 assert_eq!(g(" |indented", |d| d.move_home(false)), "| indented");
6959 assert_eq!(g("| indented", |d| d.move_home(false)), " |indented");
6960 // A line with no indentation has one place to go, so the toggle is a
6961 // no-op rather than a trip to nowhere.
6962 assert_eq!(g("hel|lo", |d| d.move_home(false)), "|hello");
6963 assert_eq!(g("|hello", |d| d.move_home(false)), "|hello");
6964
6965 let mut d = wysiwyg_doc("smart_home_wys", "```\n indented\n```\n");
6966 let indent = d.source.find(" indented").unwrap();
6967 d.caret = indent + 6; // inside "indented"
6968 d.move_home(false);
6969 assert_eq!(
6970 d.caret,
6971 indent + 4,
6972 "wysiwyg: Home aims at the code line's text"
6973 );
6974 d.move_home(false);
6975 assert_eq!(
6976 d.caret, indent,
6977 "wysiwyg: the second press takes the indent"
6978 );
6979 d.move_home(false);
6980 assert_eq!(d.caret, indent + 4, "wysiwyg: the toggle swaps back");
6981 }
6982
6983 #[test]
6984 fn end_takes_the_line_the_view_is_showing() {
6985 // The line differs by view for the same document, and that is the point:
6986 // a bare newline inside a paragraph is a soft break, which WYSIWYG draws
6987 // as a space on one row and the source view as two lines.
6988 let mut d = doc_with("end_src", "one two\nthree\n");
6989 d.caret = 1;
6990 d.move_end(false);
6991 assert_eq!(d.caret, 7, "source: the end of the source line");
6992
6993 let mut d = wysiwyg_doc("end_wys", "one two\nthree\n");
6994 d.caret = 1;
6995 d.move_end(false);
6996 assert_eq!(
6997 d.caret, 13,
6998 "wysiwyg: the end of the row, soft break and all"
6999 );
7000 }
7001
7002 #[test]
7003 fn home_and_end_extend_the_selection_when_asked() {
7004 for (view, tag) in VIEWS {
7005 let mut d = doc_in(view, &format!("home_end_ext_{tag}"), "hello world");
7006 d.caret = 6;
7007 d.move_end(true);
7008 assert_eq!(d.selection(), Some((6, 11)), "{tag}: End extends");
7009 let mut d = doc_in(view, &format!("home_ext_{tag}"), "hello world");
7010 d.caret = 6;
7011 d.move_home(true);
7012 assert_eq!(d.selection(), Some((0, 6)), "{tag}: Home extends");
7013 }
7014 }
7015
7016 // ── kill to the line's start / end ───────────────────────────────────────
7017
7018 #[test]
7019 fn kill_to_the_line_start_and_end_in_both_views() {
7020 for (view, tag) in VIEWS {
7021 // The gap that reads as a paragraph break in each view: the source
7022 // view's lines are the renderer's rows only where the source says so.
7023 let gap = if view == View::Source { "\n" } else { "\n\n" };
7024 let mut d = doc_in(
7025 view,
7026 &format!("kill_end_{tag}"),
7027 &format!("one two{gap}three\n"),
7028 );
7029 d.caret = 3;
7030 d.delete_to_line_end();
7031 assert_eq!(
7032 d.source,
7033 format!("one{gap}three\n"),
7034 "{tag}: ^K to the line's end"
7035 );
7036 assert_eq!(d.caret, 3, "{tag}: the caret stays where it kills from");
7037
7038 let mut d = doc_in(
7039 view,
7040 &format!("kill_start_{tag}"),
7041 &format!("one two{gap}three\n"),
7042 );
7043 d.caret = 7; // the end of the first line
7044 d.delete_to_line_start();
7045 assert_eq!(
7046 d.source,
7047 format!("{gap}three\n"),
7048 "{tag}: ⌘⌫ to the line's start"
7049 );
7050 assert_eq!(d.caret, 0, "{tag}");
7051 }
7052 }
7053
7054 #[test]
7055 fn a_kill_at_the_line_s_edge_leaves_the_lines_joined() {
7056 // The decision: at the boundary both kills do nothing, rather than
7057 // eating the line break. "Line" is the view's own — in WYSIWYG it ends
7058 // at a soft wrap as often as at a newline, where there is nothing
7059 // written to delete — and a source newline is only half of the blank
7060 // line between two paragraphs, so taking it leaves a soft break rather
7061 // than the join it looks like. Backspace and Delete are the keys for it.
7062 for (view, tag) in VIEWS {
7063 let gap = if view == View::Source { "\n" } else { "\n\n" };
7064 let src = format!("one{gap}three\n");
7065 let mut d = doc_in(view, &format!("kill_edge_end_{tag}"), &src);
7066 d.caret = 3; // the end of "one"
7067 d.delete_to_line_end();
7068 assert_eq!(
7069 d.source, src,
7070 "{tag}: ^K at the line's end joined it to the next"
7071 );
7072
7073 let mut d = doc_in(view, &format!("kill_edge_start_{tag}"), &src);
7074 d.caret = 3 + gap.len(); // the start of "three"
7075 d.delete_to_line_start();
7076 assert_eq!(
7077 d.source, src,
7078 "{tag}: ⌘⌫ at the line's start joined it to the last"
7079 );
7080 }
7081 }
7082
7083 #[test]
7084 fn a_kill_takes_the_selection_when_there_is_one() {
7085 // What every other delete here does with one, so these two as well.
7086 for (view, tag) in VIEWS {
7087 for (name, kill) in [
7088 (
7089 "end",
7090 (|d: &mut Doc| d.delete_to_line_end()) as fn(&mut Doc),
7091 ),
7092 ("start", |d: &mut Doc| d.delete_to_line_start()),
7093 ] {
7094 let mut d = doc_in(view, &format!("kill_sel_{name}_{tag}"), "one two three\n");
7095 d.anchor = Some(4);
7096 d.caret = 7; // "two"
7097 kill(&mut d);
7098 assert_eq!(
7099 d.source, "one three\n",
7100 "{tag}: {name} ignored the selection"
7101 );
7102 assert_eq!(d.selection(), None, "{tag}: {name}");
7103 }
7104 }
7105 }
7106
7107 #[test]
7108 fn a_kill_takes_the_markup_it_empties_with_it() {
7109 // The same hazard a word-delete has: a WYSIWYG range covers what the
7110 // user can see, which for `**bold**` is the word and never the
7111 // delimiters, so a kill that stopped at the text would leave `a ****` —
7112 // markup wrapped around nothing.
7113 let mut d = wysiwyg_doc("kill_widen", "a **bold**\n");
7114 d.caret = d.source.find("bold").unwrap();
7115 d.delete_to_line_end();
7116 assert_eq!(d.source, "a \n");
7117 }
7118
7119 #[test]
7120 fn a_kill_is_undone_in_one_step() {
7121 for (view, tag) in VIEWS {
7122 let mut d = doc_in(view, &format!("kill_undo_{tag}"), "one two three\n");
7123 d.caret = 3;
7124 d.delete_to_line_end();
7125 assert_eq!(d.source, "one\n", "{tag}");
7126 d.undo();
7127 assert_eq!(d.source, "one two three\n", "{tag}: a kill takes one undo");
7128 }
7129 }
7130
7131 #[test]
7132 fn select_block_grabs_the_whole_paragraph_from_any_wrapped_row() {
7133 // Regression: triple-click used move_home/move_end over visual rows, so
7134 // it only worked on a paragraph's first row (a wrap-boundary offset maps
7135 // to the earlier row). select_block_at reads the AST, so every offset in
7136 // the paragraph selects the whole thing.
7137 let body = "one two three four five six seven eight\n";
7138 let mut d = doc_with("sel_block", body);
7139 d.view = View::Wysiwyg;
7140 d.build_visual(12); // force the paragraph to wrap into several rows
7141 assert!(d.vmap.num_rows() > 1, "test needs a wrapped paragraph");
7142 let para = (0, "one two three four five six seven eight".len());
7143 for off in [0usize, 8, 19, 28, 38] {
7144 d.caret = 0;
7145 d.anchor = None;
7146 d.select_block_at(off);
7147 assert_eq!(
7148 d.selection(),
7149 Some(para),
7150 "offset {off} should select the paragraph"
7151 );
7152 }
7153 }
7154
7155 #[test]
7156 fn select_block_uses_content_span_for_a_heading() {
7157 let mut d = doc_with("sel_head", "# Title\n\nbody\n");
7158 d.select_block_at(4); // inside "Title"
7159 // content_span excludes the "# " marker.
7160 assert_eq!(d.selected_text(), Some("Title"));
7161 d.select_block_at(10); // inside "body"
7162 assert_eq!(d.selected_text(), Some("body"));
7163 }
7164
7165 #[test]
7166 fn select_all_spans_the_document() {
7167 let mut d = doc_with("sel_all", "abc\n\ndef\n");
7168 d.select_all();
7169 assert_eq!(d.selection(), Some((0, d.source.len())));
7170 }
7171
7172 #[test]
7173 fn select_word_at_picks_the_surrounding_word() {
7174 let mut d = doc_with("sel_word", "hello world\n");
7175 d.select_word_at(8); // inside "world"
7176 assert_eq!(d.selection(), Some((6, 11)));
7177 // Double-clicking at end-of-word still grabs the word to its left.
7178 d.select_word_at(5); // the space between the words
7179 assert_eq!(d.selection(), Some((5, 6)));
7180 }
7181
7182 #[test]
7183 fn word_helpers_respect_utf8_boundaries() {
7184 // "café" is 5 bytes ('é' is two); motion must land on char boundaries.
7185 assert_eq!(
7186 golden("utf8", "|café ok", |d| d.move_word_right(false)),
7187 "café| ok"
7188 );
7189 assert_eq!(golden("utf8b", "café |ok", |d| d.delete_word_back()), "|ok");
7190 }
7191
7192 #[test]
7193 fn typing_inserts_at_the_caret_and_advances_it() {
7194 let mut d = doc_with("type", "hello\n");
7195 d.insert("Hi ");
7196 assert_eq!(d.source, "Hi hello\n");
7197 assert_eq!(d.caret, 3);
7198 assert!(d.dirty);
7199 }
7200
7201 #[test]
7202 fn backspace_deletes_the_char_before_the_caret() {
7203 let mut d = doc_with("bs", "hello\n");
7204 d.caret = 3; // after "hel"
7205 d.backspace();
7206 assert_eq!(d.source, "helo\n");
7207 assert_eq!(d.caret, 2);
7208 }
7209
7210 #[test]
7211 fn typing_replaces_the_selection() {
7212 let mut d = doc_with("replace", "a word b\n");
7213 d.anchor = Some(2);
7214 d.caret = 6; // "word" selected
7215 d.insert("X");
7216 assert_eq!(d.source, "a X b\n");
7217 assert_eq!(d.caret, 3);
7218 assert_eq!(d.anchor, None);
7219 }
7220
7221 #[test]
7222 fn toggle_bold_wraps_then_unwraps_the_selection() {
7223 let mut d = doc_with("bold", "a word b\n");
7224 d.anchor = Some(2);
7225 d.caret = 6;
7226 d.toggle(InlineKind::Strong);
7227 assert_eq!(d.source, "a **word** b\n");
7228 // The toggled region stays selected, so a second toggle reverses it.
7229 d.toggle(InlineKind::Strong);
7230 assert_eq!(d.source, "a word b\n");
7231 d.toggle(InlineKind::Strong);
7232 assert_eq!(d.source, "a **word** b\n");
7233 }
7234
7235 #[test]
7236 fn toggle_code_wraps_then_unwraps_the_selection() {
7237 let mut d = doc_with("code_rt", "a word b\n");
7238 d.anchor = Some(2);
7239 d.caret = 6;
7240 d.toggle(InlineKind::Verbatim);
7241 assert_eq!(d.source, "a `word` b\n");
7242 d.toggle(InlineKind::Verbatim);
7243 assert_eq!(d.source, "a word b\n");
7244 }
7245
7246 #[test]
7247 fn sticky_bold_with_no_selection_wraps_the_next_typed_text() {
7248 // ⌘b at a bare caret, then type: the text comes out bold with no
7249 // selection ever made — the word-processor "start bold here" gesture.
7250 let mut d = doc_with("sticky_wrap", "xy\n");
7251 d.caret = 1; // between x and y
7252 d.toggle(InlineKind::Strong);
7253 assert_eq!(d.source, "xy\n", "arming a mark must not edit the document");
7254 d.insert("A");
7255 assert_eq!(d.source, "x**A**y\n");
7256 }
7257
7258 #[test]
7259 fn sticky_bold_lights_the_toolbar_before_any_typing() {
7260 // The button must light the instant ⌘b is pressed, or the mode is
7261 // invisible until the first character lands.
7262 let mut d = doc_with("sticky_light", "xy\n");
7263 d.caret = 1;
7264 assert!(!d.active_inline_marks().contains(InlineKind::Strong));
7265 d.toggle(InlineKind::Strong);
7266 assert!(d.active_inline_marks().contains(InlineKind::Strong));
7267 }
7268
7269 #[test]
7270 fn sticky_bold_toggled_off_types_normally_again() {
7271 // ⌘b, type, ⌘b, type: the first run is bold, the second is not — all
7272 // in the flow of typing, the exact sequence the user described.
7273 let mut d = doc_with("sticky_off", "\n");
7274 d.caret = 0;
7275 d.toggle(InlineKind::Strong);
7276 d.insert("a");
7277 d.insert("b"); // continues inside the run, no re-arming
7278 assert_eq!(d.source, "**ab**\n");
7279 d.toggle(InlineKind::Strong); // ⌘b again — shed bold
7280 d.insert("c");
7281 assert_eq!(d.source, "**ab**c\n");
7282 }
7283
7284 #[test]
7285 fn continued_typing_after_a_sticky_run_stays_in_the_run() {
7286 // Once a mark is realised the caret sits inside the run, so plain typing
7287 // extends it rather than starting a second, adjacent bold span.
7288 let mut d = doc_with("sticky_cont", "\n");
7289 d.caret = 0;
7290 d.toggle(InlineKind::Emph);
7291 d.insert("h");
7292 d.insert("i");
7293 assert_eq!(d.source, "*hi*\n");
7294 }
7295
7296 #[test]
7297 fn moving_the_caret_disarms_a_sticky_mark() {
7298 // Arming a mark and then moving away must not style text elsewhere.
7299 let mut d = doc_with("sticky_disarm", "xy\n");
7300 d.caret = 0;
7301 d.toggle(InlineKind::Strong);
7302 d.move_right(false); // caret 0 → 1, disarms
7303 assert!(!d.active_inline_marks().contains(InlineKind::Strong));
7304 d.insert("A");
7305 assert_eq!(d.source, "xAy\n", "the mark must not follow the caret");
7306 }
7307
7308 #[test]
7309 fn stacked_sticky_marks_apply_together() {
7310 // ⌘b then ⌘i before typing: the text comes out both bold and italic.
7311 let mut d = doc_with("sticky_stack", "\n");
7312 d.caret = 0;
7313 d.toggle(InlineKind::Strong);
7314 d.toggle(InlineKind::Emph);
7315 d.insert("x");
7316 // Land the caret on the styled character and confirm both marks are live.
7317 d.anchor = Some(d.source.find('x').unwrap());
7318 d.caret = d.anchor.unwrap() + 1;
7319 let marks = d.active_inline_marks();
7320 assert!(marks.contains(InlineKind::Strong), "bold: {}", d.source);
7321 assert!(marks.contains(InlineKind::Emph), "italic: {}", d.source);
7322 }
7323
7324 // ── the mark-edge rule (see `Doc::splice`) ───────────────────────────────
7325
7326 #[test]
7327 fn a_space_typed_in_a_bold_run_never_leaves_the_delimiters_showing() {
7328 // The reported bug, keystroke for keystroke: ⌘b, "bold", space, "hey".
7329 // The space inside the run made `**bold **`, which is *not* bold — four
7330 // literal asterisks — so the rich view drew them, correctly and
7331 // uselessly, until the next character happened to close the run again.
7332 let mut d = wysiwyg_doc("edge_typing", "a \n");
7333 d.caret = 2;
7334 d.toggle(InlineKind::Strong);
7335 for c in "bold".chars() {
7336 d.insert(&c.to_string());
7337 }
7338 assert_eq!(d.source, "a **bold**\n");
7339 d.insert(" ");
7340 assert_eq!(
7341 d.source, "a **bold** \n",
7342 "the space belongs outside the run"
7343 );
7344 assert!(
7345 d.active_inline_marks().contains(InlineKind::Strong),
7346 "bold is still what's being typed, so the button stays lit"
7347 );
7348 // What the writer is looking at while all this happens: their words.
7349 d.build_visual(80);
7350 let drawn: String = d.vmap.rows[0].glyphs.iter().map(|g| g.ch).collect();
7351 assert_eq!(drawn, "a bold ", "no delimiter ever surfaces: {}", d.source);
7352 for c in "hey".chars() {
7353 d.insert(&c.to_string());
7354 }
7355 assert_eq!(
7356 d.source, "a **bold hey**\n",
7357 "one bold phrase, not two runs"
7358 );
7359 }
7360
7361 #[test]
7362 fn typing_past_a_space_can_still_leave_the_bold_behind() {
7363 // The other half: the marks stay armed across the space, so ⌘b turns
7364 // them off again there and the next word is plain — the run isn't
7365 // rejoined by a caret that was told not to.
7366 let mut d = wysiwyg_doc("edge_shed", "\n");
7367 d.caret = 0;
7368 d.toggle(InlineKind::Strong);
7369 for c in "bold ".chars() {
7370 d.insert(&c.to_string());
7371 }
7372 assert_eq!(d.source, "**bold** \n");
7373 d.toggle(InlineKind::Strong);
7374 assert!(!d.active_inline_marks().contains(InlineKind::Strong));
7375 d.insert("x");
7376 assert_eq!(d.source, "**bold** x\n");
7377 }
7378
7379 #[test]
7380 fn a_space_typed_first_of_all_still_leaves_the_mark_armed() {
7381 // ⌘b and then a space before any word: the space is not marked (nothing
7382 // is), and the word after it is.
7383 let mut d = wysiwyg_doc("edge_space_first", "a\n");
7384 d.caret = 1;
7385 d.toggle(InlineKind::Strong);
7386 d.insert(" ");
7387 assert_eq!(d.source, "a \n");
7388 assert!(d.active_inline_marks().contains(InlineKind::Strong));
7389 d.insert("b");
7390 assert_eq!(d.source, "a **b**\n");
7391 }
7392
7393 #[test]
7394 fn a_space_typed_at_either_edge_of_an_existing_mark_steps_outside_it() {
7395 let mut d = wysiwyg_doc("edge_tail", "x **bold**\n");
7396 d.caret = 8; // the caret's home at the end of the run's text
7397 d.insert(" ");
7398 assert_eq!(
7399 d.source, "x **bold** \n",
7400 "the space lands past the delimiters"
7401 );
7402 assert_eq!(d.caret, 11, "and the caret stands past it, outside the run");
7403
7404 let mut d = wysiwyg_doc("edge_head", "x **bold** y\n");
7405 d.caret = 4; // in front of the "b"
7406 d.insert(" ");
7407 assert_eq!(d.source, "x **bold** y\n");
7408 assert_eq!(d.caret, 3, "in front of the run, where the space was typed");
7409 }
7410
7411 #[test]
7412 fn a_delete_that_backs_a_space_onto_a_delimiter_moves_the_delimiter() {
7413 // Backspace over the last letter of a bold phrase.
7414 let mut d = wysiwyg_doc("edge_bksp", "a **bold h**\n");
7415 d.caret = 10; // past the "h"
7416 d.backspace();
7417 assert_eq!(d.source, "a **bold** \n");
7418 assert_eq!(d.caret, 11, "the caret keeps the place on screen it had");
7419 assert!(d.active_inline_marks().contains(InlineKind::Strong));
7420 d.insert("x");
7421 assert_eq!(d.source, "a **bold x**\n", "and typing rejoins the run");
7422 }
7423
7424 #[test]
7425 fn deleting_the_last_of_a_run_takes_its_delimiters_with_it() {
7426 // `**b**` with the `b` gone is `****`: two delimiters with nothing to
7427 // mark, which is only text. The marks live on in the caret instead.
7428 let mut d = wysiwyg_doc("edge_empty", "a **b** c\n");
7429 d.caret = 5;
7430 d.backspace();
7431 assert_eq!(d.source, "a c\n");
7432 assert!(d.active_inline_marks().contains(InlineKind::Strong));
7433 d.insert("x");
7434 assert_eq!(d.source, "a **x** c\n");
7435 }
7436
7437 #[test]
7438 fn typing_over_a_whole_bold_word_keeps_it_bold() {
7439 let mut d = wysiwyg_doc("edge_replace", "a **bold** c\n");
7440 d.anchor = Some(4);
7441 d.caret = 8; // the word, not its delimiters
7442 d.insert("x");
7443 assert_eq!(d.source, "a **x** c\n");
7444 }
7445
7446 #[test]
7447 fn a_code_span_keeps_the_space_it_is_given() {
7448 // Backticks are not whitespace-sensitive the way `**` is: `` `code ` ``
7449 // is still verbatim, so nothing is re-spelt. The repair asks the parser
7450 // rather than a table of kinds, and this is the answer it gets.
7451 let mut d = wysiwyg_doc("edge_code", "a `code` c\n");
7452 d.caret = 7;
7453 d.insert(" ");
7454 assert_eq!(d.source, "a `code ` c\n");
7455 }
7456
7457 #[test]
7458 fn a_delete_from_a_runs_outer_edge_reaches_into_the_run() {
7459 // A run's closing delimiter has a caret home on each side of it, one
7460 // column apart on screen — and a plain ← off the space after a bold word
7461 // lands on the outer one. The character drawn behind the caret there is
7462 // still the last letter of the phrase, so that is what Backspace takes;
7463 // the byte behind it is a `*` nobody can see.
7464 let mut d = wysiwyg_doc("edge_outer_close", "**bold** x\n");
7465 d.caret = 9;
7466 d.move_left(false);
7467 assert_eq!(d.caret, 8, "← rests past the delimiters, not inside them");
7468 d.backspace();
7469 assert_eq!(
7470 d.source, "**bol** x\n",
7471 "a letter of the phrase, not its `*`"
7472 );
7473 assert_eq!(d.caret, 5);
7474
7475 // And the mirror in front of the opening delimiter, where Delete's
7476 // character is the first letter of the run.
7477 let mut d = wysiwyg_doc("edge_outer_open", "x**bold**\n");
7478 d.caret = 1;
7479 d.delete_forward();
7480 assert_eq!(d.source, "x**old**\n");
7481 assert_eq!(d.caret, 3, "inside the run, in front of what is left of it");
7482 }
7483
7484 #[test]
7485 fn a_delete_at_a_run_edge_never_eats_a_delimiter() {
7486 // The byte beside the caret at either edge of a bold word is a `*` the
7487 // rich view draws nothing for. Taking it is not the character delete the
7488 // key was pressed for — it unspells the run and puts a literal asterisk
7489 // on screen (`a *bold** c`). The visible character is the one that goes.
7490 let mut d = wysiwyg_doc("edge_open_bksp", "a **bold** c\n");
7491 d.caret = 4; // in front of the "b"
7492 d.backspace();
7493 assert_eq!(d.source, "a**bold** c\n", "the space goes, the run stands");
7494
7495 let mut d = wysiwyg_doc("edge_close_del", "a **bold** c\n");
7496 d.caret = 8; // past the "d"
7497 d.delete_forward();
7498 assert_eq!(d.source, "a **bold**c\n");
7499 assert_eq!(d.caret, 8, "and the caret stays inside the run");
7500 d.insert("x");
7501 assert_eq!(d.source, "a **boldx**c\n");
7502
7503 // A code span's backticks are hidden the same way, so they are covered
7504 // by the same rule and not by a list of kinds.
7505 let mut d = wysiwyg_doc("edge_open_code", "a `code` c\n");
7506 d.caret = 3;
7507 d.backspace();
7508 assert_eq!(d.source, "a`code` c\n");
7509 }
7510
7511 #[test]
7512 fn the_source_view_deletes_the_delimiter_byte_it_is_shown() {
7513 // The asterisks are on the screen there and the caret can stand between
7514 // them, so a delete takes exactly the byte it is aimed at.
7515 let mut d = doc_with("edge_open_src", "a **bold** c\n");
7516 d.caret = 4;
7517 d.backspace();
7518 assert_eq!(d.source, "a *bold** c\n");
7519
7520 let mut d = doc_with("edge_close_src", "a **bold** c\n");
7521 d.caret = 8;
7522 d.delete_forward();
7523 assert_eq!(d.source, "a **bold* c\n");
7524 }
7525
7526 #[test]
7527 fn backspacing_the_space_out_of_a_bold_phrase_leaves_the_caret_in_it() {
7528 // The reported bug, keystroke for keystroke: ⌘b, "bold", space, Backspace.
7529 // The space had stepped outside the run (the mark-edge rule), taking the
7530 // caret with it, so the delete put it back down on the far side of the
7531 // closing `**` — one place on screen, and the wrong side of it. Typing
7532 // came out plain and the toolbar went dark, with nothing to see.
7533 let mut d = wysiwyg_doc("edge_bksp_space", "\n");
7534 d.caret = 0;
7535 d.toggle(InlineKind::Strong);
7536 for c in "bold".chars() {
7537 d.insert(&c.to_string());
7538 }
7539 d.insert(" ");
7540 assert_eq!(d.source, "**bold** \n");
7541 d.backspace();
7542 assert_eq!(
7543 d.source, "**bold**\n",
7544 "the space goes, the delimiters stay"
7545 );
7546 assert_eq!(d.caret, 6, "and the caret comes back inside the run");
7547 assert!(
7548 d.active_inline_marks().contains(InlineKind::Strong),
7549 "so the button is still lit"
7550 );
7551 d.insert("x");
7552 assert_eq!(
7553 d.source, "**boldx**\n",
7554 "and the next character is still bold"
7555 );
7556 }
7557
7558 #[test]
7559 fn a_second_backspace_there_deletes_a_letter_of_the_phrase() {
7560 // What the stranded caret did next: the byte behind it was the closing
7561 // `*`, so a second press took that instead of a letter — `**bold*`, the
7562 // styling gone and an asterisk on the screen where the word had been.
7563 let mut d = wysiwyg_doc("edge_bksp_twice", "\n");
7564 d.caret = 0;
7565 d.toggle(InlineKind::Strong);
7566 for c in "bold ".chars() {
7567 d.insert(&c.to_string());
7568 }
7569 assert_eq!(d.source, "**bold** \n");
7570 d.backspace();
7571 d.backspace();
7572 assert_eq!(d.source, "**bol**\n", "the delete lands inside the run");
7573 assert_eq!(d.caret, 5);
7574 }
7575
7576 #[test]
7577 fn a_delete_that_ends_at_a_nested_run_settles_inside_every_delimiter() {
7578 // `***both***` closes two runs with one stack of asterisks: the caret has
7579 // to walk in through all of them, or it lands between the emph and the
7580 // strong and types half-marked.
7581 let mut d = wysiwyg_doc("edge_bksp_nested", "***both*** \n");
7582 d.caret = 11;
7583 d.backspace();
7584 assert_eq!(d.source, "***both***\n");
7585 assert_eq!(d.caret, 7, "past the last letter, inside both runs");
7586 d.insert("x");
7587 assert_eq!(d.source, "***bothx***\n");
7588 }
7589
7590 #[test]
7591 fn a_delete_that_ends_mid_run_leaves_the_caret_where_it_fell() {
7592 // The settle only moves a caret a run actually closed over. Ordinary
7593 // deletes — inside a run, or in plain prose — are untouched.
7594 let mut d = wysiwyg_doc("edge_bksp_mid", "a **bold** c\n");
7595 d.caret = 8;
7596 d.backspace();
7597 assert_eq!(d.source, "a **bol** c\n");
7598 assert_eq!(d.caret, 7);
7599
7600 let mut d = wysiwyg_doc("edge_bksp_plain", "plain\n");
7601 d.caret = 5;
7602 d.backspace();
7603 assert_eq!(d.source, "plai\n");
7604 assert_eq!(d.caret, 4);
7605 }
7606
7607 #[test]
7608 fn the_source_view_leaves_a_delete_where_it_landed() {
7609 // The delimiters are on the screen there, so the offset past them is a
7610 // place the caret can be seen to be — nothing to settle.
7611 let mut d = doc_with("edge_bksp_src", "**bold** \n");
7612 d.caret = 9;
7613 d.backspace();
7614 assert_eq!(d.source, "**bold**\n");
7615 assert_eq!(d.caret, 8);
7616 }
7617
7618 #[test]
7619 fn the_mark_edge_rule_clears_every_delimiter_of_a_nested_run() {
7620 // `***both***` closes two runs with one stack of asterisks; a space that
7621 // clears only the inner one lands against the outer's and breaks that
7622 // instead.
7623 let mut d = wysiwyg_doc("edge_nested", "a ***both***\n");
7624 d.caret = 9;
7625 d.insert(" ");
7626 assert_eq!(d.source, "a ***both*** \n");
7627 assert_eq!(d.caret, 13);
7628 d.insert("x");
7629 assert_eq!(d.source, "a ***both x***\n");
7630 }
7631
7632 #[test]
7633 fn the_mark_edge_repair_undoes_with_the_keystroke_that_caused_it() {
7634 // The delimiter shuffle is not an edit the writer made, so it is not a
7635 // step they have to undo past.
7636 let mut d = wysiwyg_doc("edge_undo", "a **bold**\n");
7637 d.caret = 8;
7638 d.insert(" ");
7639 assert_eq!(d.source, "a **bold** \n");
7640 d.undo();
7641 assert_eq!(d.source, "a **bold**\n");
7642 }
7643
7644 #[test]
7645 fn the_source_view_types_the_space_where_it_was_asked_to() {
7646 // The rule is a rich-view courtesy. In the source view the delimiters are
7647 // on the screen and the user is editing the bytes they can see.
7648 let mut d = doc_with("edge_src", "a **bold** c\n");
7649 d.caret = 8;
7650 d.insert(" ");
7651 assert_eq!(d.source, "a **bold ** c\n");
7652 }
7653
7654 #[test]
7655 fn toggling_a_mark_over_a_selection_leaves_its_edge_whitespace_out() {
7656 // Double-clicking a word takes the space after it; bolding that must not
7657 // spell `**word **`, which is not bold at all.
7658 let mut d = wysiwyg_doc("edge_sel", "a word b\n");
7659 d.anchor = Some(2);
7660 d.caret = 7; // "word "
7661 d.toggle(InlineKind::Strong);
7662 assert_eq!(d.source, "a **word** b\n");
7663 d.toggle(InlineKind::Strong);
7664 assert_eq!(d.source, "a word b\n");
7665 d.toggle(InlineKind::Strong);
7666 assert_eq!(
7667 d.source, "a **word** b\n",
7668 "reapplying the mark must not wrap stale delimiter offsets"
7669 );
7670 // And a selection of nothing but whitespace has no word to mark.
7671 let mut d = wysiwyg_doc("edge_sel_ws", "a word b\n");
7672 d.anchor = Some(6);
7673 d.caret = 7;
7674 d.toggle(InlineKind::Strong);
7675 assert_eq!(d.source, "a word b\n");
7676 assert!(d.status.is_some());
7677 }
7678
7679 #[test]
7680 fn set_block_turns_a_paragraph_into_a_heading_at_the_caret() {
7681 let mut d = doc_with("head_set", "hello\n");
7682 d.caret = 2; // caret inside the paragraph, no selection
7683 d.set_block(BlockKind::Heading(1));
7684 assert_eq!(d.source, "# hello\n");
7685 }
7686
7687 #[test]
7688 fn set_block_heading_works_in_wysiwyg_view() {
7689 // The app defaults to WYSIWYG; the caret is a source offset either way.
7690 let mut d = wysiwyg_doc("head_wys", "hello\n");
7691 d.caret = 2;
7692 d.set_block(BlockKind::Heading(1));
7693 assert_eq!(d.source, "# hello\n");
7694 }
7695
7696 #[test]
7697 fn toggle_heading_applies_switches_and_reverts() {
7698 let mut d = doc_with("head_toggle", "hello\n");
7699 d.caret = 2;
7700 d.toggle_heading(1);
7701 assert_eq!(d.source, "# hello\n"); // paragraph → H1
7702 d.toggle_heading(2);
7703 assert_eq!(d.source, "## hello\n"); // H1 → H2 (different level switches)
7704 d.toggle_heading(2);
7705 assert_eq!(d.source, "hello\n"); // same level reverts to paragraph
7706 }
7707
7708 #[test]
7709 fn preserve_enter_at_a_line_end_lands_the_caret_on_the_new_blank_line() {
7710 // Regression: Enter at the end of a soft-break line (mid-paragraph) opened
7711 // the blank line but the caret rendered on the *next* line, because the
7712 // separator was a non-navigable decoration row. In Preserve flow that
7713 // blank line is a real caret home — the caret must resolve onto it, and
7714 // typing there makes the soft break that continues the paragraph.
7715 let src = "line one:\nsecond line\n";
7716 let mut d = wysiwyg_doc("pre_enter_lineend", src);
7717 d.set_line_flow(LineFlow::Preserve);
7718 d.build_visual_unwrapped(); // the GUI path (pixel-wrapped)
7719 d.caret = 9; // the visual end of row 0, at the soft-break '\n'
7720 d.newline();
7721 d.build_visual_unwrapped();
7722 assert_eq!(d.source, "line one:\n\nsecond line\n");
7723 assert_eq!(
7724 d.caret, 10,
7725 "caret sits on the new blank line, not the next line"
7726 );
7727 // The blank line is row 1, and the caret resolves onto it — not row 2.
7728 assert_eq!(
7729 d.vmap.pos_of_offset(10),
7730 (1, 0),
7731 "caret renders on the blank row"
7732 );
7733 assert!(
7734 !d.vmap.rows[1].decoration,
7735 "the blank line is navigable in Preserve"
7736 );
7737 // Typing there makes a soft break: one paragraph, three lines.
7738 d.insert("new clause,");
7739 assert_eq!(d.source, "line one:\nnew clause,\nsecond line\n");
7740 }
7741
7742 #[test]
7743 fn preserve_enter_makes_a_soft_break_not_a_paragraph() {
7744 // Mid-paragraph: Enter splits the line with a single `\n`, a soft break
7745 // that keeps it one paragraph — where Fold would open a second paragraph.
7746 let mut d = wysiwyg_doc("pre_enter_mid", "abcdef\n");
7747 d.set_line_flow(LineFlow::Preserve);
7748 d.caret = 3;
7749 d.newline();
7750 assert_eq!(d.source, "abc\ndef\n", "mid-line Enter is a soft break");
7751
7752 // End-of-paragraph: Enter then typing continues the same paragraph on a
7753 // new line (a soft break), not a fresh paragraph.
7754 let mut d = wysiwyg_doc("pre_enter_end", "abc\n");
7755 d.set_line_flow(LineFlow::Preserve);
7756 d.caret = 3;
7757 d.newline();
7758 d.insert("def");
7759 assert_eq!(
7760 d.source, "abc\ndef\n",
7761 "end-of-line Enter + typing is a soft break"
7762 );
7763 }
7764
7765 #[test]
7766 fn preserve_double_enter_still_makes_a_paragraph() {
7767 // Two Enters in a row promote to a real paragraph break: the second lands
7768 // on the blank line the first opened and takes the empty-line branch.
7769 let mut d = wysiwyg_doc("pre_enter_dbl", "abc\n");
7770 d.set_line_flow(LineFlow::Preserve);
7771 d.caret = 3;
7772 d.newline();
7773 d.newline();
7774 d.insert("def");
7775 assert_eq!(
7776 d.source, "abc\n\ndef\n",
7777 "double Enter is a paragraph break"
7778 );
7779 }
7780
7781 #[test]
7782 fn preserve_backspace_joins_across_a_soft_break() {
7783 // Backspace is the symmetric undo of a Preserve Enter: over the `\n` of a
7784 // soft break it deletes the single newline and joins the two lines.
7785 let mut d = wysiwyg_doc("pre_bs", "abc\ndef\n");
7786 d.set_line_flow(LineFlow::Preserve);
7787 d.build_visual(80);
7788 d.caret = 4; // start of "def", just past the soft break
7789 d.backspace();
7790 assert_eq!(
7791 d.source, "abcdef\n",
7792 "Backspace joins across the soft break"
7793 );
7794 assert_eq!(d.caret, 3, "caret lands where the lines meet");
7795 }
7796
7797 #[test]
7798 fn fold_enter_still_starts_a_new_paragraph() {
7799 // The default flow is unchanged: a lone `\n` would render as an invisible
7800 // space, so Enter keeps opening the paragraph break that actually shows.
7801 let mut d = wysiwyg_doc("fold_enter", "abcdef\n");
7802 d.caret = 3;
7803 d.newline();
7804 assert_eq!(
7805 d.source, "abc\n\ndef\n",
7806 "Fold mid-line Enter is a paragraph break"
7807 );
7808 }
7809
7810 #[test]
7811 fn wysiwyg_one_enter_starts_a_new_paragraph() {
7812 // Regression: one Enter left the caret between the two newlines, so typing
7813 // made a soft break (one paragraph) and you needed a second Enter.
7814 let mut d = wysiwyg_doc("wys_enter", "abc\n");
7815 d.caret = 3;
7816 d.newline();
7817 d.insert("def");
7818 assert_eq!(d.source, "abc\n\ndef\n"); // two paragraphs, not "abc\ndef\n"
7819 }
7820
7821 #[test]
7822 fn enter_at_the_end_of_a_bold_run_keeps_its_closing_delimiter_attached() {
7823 // Regression: Enter at the caret's natural End-of-line resting place
7824 // after a bold run with nothing following it (on screen: right after
7825 // "bold", before the hidden closing "**") spliced the paragraph break
7826 // at that very byte offset — which sits *before* the closing "**" in
7827 // the source, since the delimiter is hidden and emits no glyph of its
7828 // own for `push_row`'s "end of row" fallback to count. That severed the
7829 // mark: "**bold**\n" became "**bold\n\n**\n", stranding the closing
7830 // "**" alone on the new line instead of leaving "**bold**" intact with
7831 // a fresh empty paragraph after it.
7832 let mut d = wysiwyg_doc("bold_eol_enter", "**bold**\n");
7833 d.move_end(false); // the WYSIWYG End key, from caret 0
7834 assert_eq!(
7835 d.caret, 6,
7836 "caret rests right after \"bold\", before the hidden \"**\""
7837 );
7838 d.newline();
7839 assert!(
7840 d.source.starts_with("**bold**"),
7841 "the closing ** must stay attached to \"bold\": got {:?}",
7842 d.source
7843 );
7844 assert_eq!(
7845 d.source, "**bold**\n\n\n",
7846 "a fresh empty paragraph follows the still-intact bold run"
7847 );
7848 }
7849
7850 #[test]
7851 fn source_view_enter_is_a_single_newline() {
7852 let mut d = doc_with("src_enter", "abc\n");
7853 d.caret = 3;
7854 d.newline();
7855 assert_eq!(d.source, "abc\n\n");
7856 }
7857
7858 #[test]
7859 fn heading_applies_at_the_end_of_a_paragraph() {
7860 // The caret at a line end sits at the doc level; set_block must still find
7861 // the block on that line.
7862 let mut d = doc_with("head_end", "abc\n");
7863 d.caret = 3; // end of "abc"
7864 d.toggle_heading(1);
7865 assert_eq!(d.source, "# abc\n");
7866 }
7867
7868 #[test]
7869 fn heading_on_an_empty_new_paragraph_creates_one() {
7870 let mut d = wysiwyg_doc("head_empty", "abc\n");
7871 d.caret = 3;
7872 d.newline(); // caret now on a fresh, empty paragraph
7873 d.toggle_heading(1);
7874 d.insert("Title");
7875 assert!(d.source.contains("# Title"), "got {:?}", d.source);
7876 }
7877
7878 #[test]
7879 fn a_heading_typed_on_a_blank_line_keeps_the_caret_on_its_own_row() {
7880 // The reported bug, end to end: click a blank line with another one under
7881 // it, press H1, type. The text landed in the heading and the caret's
7882 // offset was right (the source view drew it there), but the rich view
7883 // drew it two rows lower, on the trailing blank line — the empty `# `
7884 // heading had left every row below it short by the marker's two bytes,
7885 // and the blank line ended up claiming the heading's own end offset.
7886 let mut d = wysiwyg_doc("head_blank", "one\n\ntwo\n\n\n\n");
7887 d.build_visual_unwrapped();
7888 d.caret = d.vmap.offset_of_pos(4, 0); // the first of the two blank lines
7889 d.toggle_heading(1);
7890 for c in "title".chars() {
7891 d.insert(&c.to_string());
7892 d.build_visual_unwrapped(); // as a frontend does, one frame per key
7893 }
7894 assert_eq!(d.source, "one\n\ntwo\n\n# title\n\n");
7895 assert_eq!(
7896 d.caret_pos(),
7897 (4, 5),
7898 "the caret draws at the end of the heading"
7899 );
7900 }
7901
7902 #[test]
7903 fn clicking_an_empty_heading_types_after_its_marker() {
7904 // The same anchor from the other side: the empty heading's row is its own
7905 // caret home, so a click on it must land past the hidden `# `. Landing in
7906 // front of the hashes made the first keystroke un-heading the line.
7907 let mut d = wysiwyg_doc("head_click", "# \n");
7908 d.build_visual_unwrapped();
7909 d.caret = d.vmap.offset_of_pos(0, 0);
7910 d.insert("x");
7911 assert_eq!(d.source, "# x\n");
7912 }
7913
7914 #[test]
7915 fn wysiwyg_enter_after_a_heading_makes_a_paragraph() {
7916 let mut d = wysiwyg_doc("head_enter", "# Title\n");
7917 d.caret = 7; // end of the heading
7918 d.newline();
7919 d.insert("body");
7920 assert_eq!(d.source, "# Title\n\nbody\n");
7921 }
7922
7923 #[test]
7924 fn wysiwyg_enter_continues_a_bullet_list() {
7925 let mut d = wysiwyg_doc("wys_bullet", "- item\n");
7926 d.caret = 6; // end of "item"
7927 d.newline();
7928 d.insert("two");
7929 assert_eq!(d.source, "- item\n- two\n");
7930 }
7931
7932 #[test]
7933 fn wysiwyg_enter_increments_an_ordered_list() {
7934 let mut d = wysiwyg_doc("wys_ol", "1. one\n");
7935 d.caret = 6; // end of "one"
7936 d.newline();
7937 d.insert("two");
7938 assert_eq!(d.source, "1. one\n2. two\n");
7939 }
7940
7941 #[test]
7942 fn wysiwyg_backspace_after_leaving_a_list_collapses_the_gap_cleanly() {
7943 // Regression for the "extra newline" left between a list and the paragraph
7944 // below it. Enter, Enter leaves the list on a fresh empty paragraph
7945 // (`- item\n\n\n\nnext`, a navigable blank between the two blocks); one
7946 // Backspace should then take the caret cleanly back to the end of the list
7947 // item, `- item\n\nnext`, not delete a single newline and strand it on the
7948 // odd `- item\n\n\nnext` — a blank line the eye reads as one separator but
7949 // no caret can land on. The map is rebuilt between keystrokes exactly as a
7950 // frontend does, since Backspace reads the stop table to place the delete.
7951 let mut d = wysiwyg_doc("wys_exit_bksp", "- item\n\nnext\n");
7952 d.caret = 6; // end of "item"
7953 d.newline();
7954 d.build_visual(80);
7955 d.newline(); // leave the list onto a fresh empty paragraph
7956 d.build_visual(80);
7957 assert_eq!(
7958 d.source, "- item\n\n\n\nnext\n",
7959 "double-Enter opens the empty paragraph"
7960 );
7961 d.backspace();
7962 assert_eq!(
7963 d.source, "- item\n\nnext\n",
7964 "one Backspace collapses the whole gap"
7965 );
7966 assert_eq!(
7967 d.caret, 6,
7968 "and lands the caret back at the end of the list item"
7969 );
7970 }
7971
7972 #[test]
7973 fn wysiwyg_backspace_on_stacked_blank_lines_still_removes_just_one() {
7974 // The stop-wise delete must not over-reach when there is no block boundary
7975 // to cross: two blank lines in a row are one caret stop apart, so pressing
7976 // Enter on an empty line and then Backspace removes exactly the one newline
7977 // it added — the lone-Enter / lone-Backspace symmetry, preserved.
7978 let mut d = wysiwyg_doc("wys_stack", "abc\n\n\n");
7979 d.caret = 5; // the empty paragraph the first Enter already opened
7980 d.build_visual(80);
7981 d.newline();
7982 d.build_visual(80);
7983 assert_eq!(
7984 d.source, "abc\n\n\n\n",
7985 "Enter on the blank line adds one newline"
7986 );
7987 d.backspace();
7988 assert_eq!(
7989 d.source, "abc\n\n\n",
7990 "Backspace takes back exactly that one newline"
7991 );
7992 }
7993
7994 #[test]
7995 fn wysiwyg_enter_on_an_empty_list_item_exits_the_list() {
7996 let mut d = wysiwyg_doc("wys_exit", "- a\n- \n");
7997 d.caret = 6; // end of the empty "- " item
7998 d.newline();
7999 d.insert("p");
8000 assert_eq!(d.source, "- a\n\np\n");
8001 }
8002
8003 #[test]
8004 fn wysiwyg_enter_does_not_mistake_a_setext_underline_for_a_list() {
8005 // `text\n- \n` is a setext heading — the `- ` is its underline, not a
8006 // list item, though it reads as a `- ` marker byte-for-byte. Enter must
8007 // not take the list-exit path (which would splice the `- ` away as if
8008 // leaving an empty item); the AST guard sends it to a normal break and
8009 // leaves the underline intact.
8010 let mut d = wysiwyg_doc("wys_setext", "text\n- \n");
8011 assert!(
8012 d.nodes().iter().any(|n| n.kind == Kind::Heading),
8013 "precondition: twig parses this as a heading, not a list",
8014 );
8015 d.caret = 7; // on the `- ` underline line
8016 d.newline();
8017 assert!(
8018 d.source.contains("- "),
8019 "the setext underline survives, not spliced away as a list item: {:?}",
8020 d.source,
8021 );
8022 }
8023
8024 #[test]
8025 fn wysiwyg_enter_in_a_code_block_is_a_literal_newline() {
8026 let mut d = wysiwyg_doc("wys_code", "```\nabc\n```\n");
8027 d.caret = 7; // end of "abc" inside the fence
8028 d.newline();
8029 d.insert("def");
8030 assert_eq!(d.source, "```\nabc\ndef\n```\n");
8031 }
8032
8033 #[test]
8034 fn wysiwyg_enter_continues_a_block_quote() {
8035 // Enter opens a new *paragraph* inside the quote, not a second line of
8036 // the same one. `> quote\n> more` is a soft break, which under
8037 // `LineFlow::Fold` renders as a space — the keystroke would look like it
8038 // did nothing. The quoted blank line is what makes the break visible, and
8039 // it's the same thing Enter does in running prose.
8040 let mut d = wysiwyg_doc("wys_quote", "> quote\n");
8041 d.caret = 7; // end of "quote"
8042 d.newline();
8043 d.insert("more");
8044 assert_eq!(d.source, "> quote\n>\n> more\n");
8045 // Still one quote, now holding two paragraphs — not a quote and a stray
8046 // line that fell out of it.
8047 let quotes = d
8048 .nodes()
8049 .iter()
8050 .filter(|n| n.kind == Kind::BlockQuote)
8051 .count();
8052 assert_eq!(quotes, 1);
8053 }
8054
8055 #[test]
8056 fn set_block_makes_a_heading_at_the_caret() {
8057 let mut d = doc_with("head", "Title\n\nbody\n");
8058 d.caret = 0;
8059 d.set_block(BlockKind::Heading(2));
8060 assert_eq!(d.source, "## Title\n\nbody\n");
8061 d.set_block(BlockKind::Paragraph);
8062 assert_eq!(d.source, "Title\n\nbody\n");
8063 }
8064
8065 // ── block containers (quote / list) ──────────────────────────────────────
8066
8067 #[test]
8068 fn toggle_blockquote_wraps_the_block_at_the_caret_and_reverses() {
8069 let g = |m, f: fn(&mut Doc)| golden("quote", m, f);
8070 assert_eq!(g("hel|lo\n", |d| d.toggle_blockquote()), "> hel|lo\n");
8071 assert_eq!(g("> hel|lo\n", |d| d.toggle_blockquote()), "hel|lo\n");
8072 // A caret at a line end sits at the doc level; the block is still found.
8073 assert_eq!(g("hello|\n", |d| d.toggle_blockquote()), "> hello|\n");
8074 }
8075
8076 #[test]
8077 fn toggle_blockquote_keeps_the_caret_in_a_hard_wrapped_paragraph() {
8078 // Every source line of the paragraph gets its own `> `, so a caret left
8079 // on its old byte offset falls one prefix per line above it too far
8080 // back — inside the markup it just asked for rather than in its word.
8081 assert_eq!(
8082 golden("quote_wrap", "aaa\nb|bb\nccc\n", |d| d.toggle_blockquote()),
8083 "> aaa\n> b|bb\n> ccc\n"
8084 );
8085 }
8086
8087 #[test]
8088 fn toggle_blockquote_works_in_wysiwyg_view() {
8089 let g = |n, m, f: fn(&mut Doc)| golden_in(View::Wysiwyg, n, m, f);
8090 assert_eq!(
8091 g("q_wys", "hel|lo\n", |d| d.toggle_blockquote()),
8092 "> hel|lo\n"
8093 );
8094 assert_eq!(
8095 g("q_wys2", "> hel|lo\n", |d| d.toggle_blockquote()),
8096 "hel|lo\n"
8097 );
8098 }
8099
8100 #[test]
8101 fn toggle_list_makes_a_list_and_converts_between_the_kinds() {
8102 let g = |m, f: fn(&mut Doc)| golden("list", m, f);
8103 assert_eq!(g("hel|lo\n", |d| d.toggle_list(false)), "- hel|lo\n");
8104 assert_eq!(g("hel|lo\n", |d| d.toggle_list(true)), "1. hel|lo\n");
8105 // The *other* kind converts in place instead of nesting, which is what
8106 // makes the two buttons one three-state control.
8107 assert_eq!(g("- hel|lo\n", |d| d.toggle_list(true)), "1. hel|lo\n");
8108 assert_eq!(g("1. hel|lo\n", |d| d.toggle_list(false)), "- hel|lo\n");
8109 // Its own kind, over the only item the list holds, takes it off.
8110 assert_eq!(g("- hel|lo\n", |d| d.toggle_list(false)), "hel|lo\n");
8111 }
8112
8113 #[test]
8114 fn toggle_list_works_in_wysiwyg_view() {
8115 let g = |n, m, f: fn(&mut Doc)| golden_in(View::Wysiwyg, n, m, f);
8116 assert_eq!(
8117 g("l_wys", "hel|lo\n", |d| d.toggle_list(true)),
8118 "1. hel|lo\n"
8119 );
8120 assert_eq!(
8121 g("l_wys2", "1. hel|lo\n", |d| d.toggle_list(false)),
8122 "- hel|lo\n"
8123 );
8124 assert_eq!(
8125 g("l_wys3", "- hel|lo\n", |d| d.toggle_list(false)),
8126 "hel|lo\n"
8127 );
8128 }
8129
8130 #[test]
8131 fn a_list_over_a_selection_numbers_each_block_and_stays_selected() {
8132 // The selection has to grow with the markup: twig takes a container off
8133 // only a range covering every block it holds, so the second press can
8134 // reverse the first only if the result is what's selected.
8135 let mut d = doc_with("list_sel", "abc\n\ndef\n");
8136 d.select_all();
8137 d.toggle_list(true);
8138 assert_eq!(d.source, "1. abc\n\n2. def\n");
8139 assert_eq!(d.selection(), Some((0, d.source.len())));
8140 d.toggle_list(true);
8141 assert_eq!(d.source, "abc\n\ndef\n");
8142 }
8143
8144 #[test]
8145 fn toggle_blockquote_nests_a_partly_covered_quote() {
8146 // twig's rule: covering only some of a container's blocks nests, because
8147 // taking the quote off would drag its uncovered siblings out with it.
8148 let mut d = doc_with("quote_nest", "> a\n>\n> b\n");
8149 d.caret = 2; // in the first quoted paragraph only
8150 d.toggle_blockquote();
8151 assert_eq!(d.source, "> > a\n>\n> b\n");
8152 }
8153
8154 #[test]
8155 fn a_container_toggle_opens_an_empty_one_on_a_blank_line() {
8156 // A blank line used to be no block for twig to wrap —
8157 // `toggle_block_container` answered `NotFound` — so Quote and the list
8158 // buttons did nothing on the very line the H1 button works on, and leaf
8159 // lent twig a scratch paragraph to wrap and took it back out again.
8160 // twig 3.2.0 opens an empty container there itself, so what is left here
8161 // is where the caret lands: inside the marker that was just written.
8162 let mut d = doc_with("quote_blank", "\nabc\n");
8163 d.caret = 0;
8164 d.toggle_blockquote();
8165 assert_eq!(d.source, "> \nabc\n");
8166 assert_eq!(
8167 d.caret, 2,
8168 "the caret belongs inside the quote it just opened"
8169 );
8170 assert!(d.status.is_none(), "{:?}", d.status);
8171 assert!(d.dirty);
8172
8173 // And the paragraph below is still its own block: an empty container one
8174 // soft break from `abc` would take that paragraph into the quote with it.
8175 let mut d = wysiwyg_doc("quote_blank_rows", "\nabc\n");
8176 d.caret = 0;
8177 d.toggle_blockquote();
8178 d.build_visual(80);
8179 assert_eq!(drawn_rows(&d), ["│ ", "", "abc"]);
8180
8181 // The same from the other side: a blank line directly under a paragraph
8182 // earns the blank line an empty block needs, rather than being read as a
8183 // soft break inside that paragraph.
8184 let mut d = doc_with("list_blank_below", "abc\n");
8185 d.caret = 4;
8186 d.toggle_list(false);
8187 assert_eq!(d.source, "abc\n\n- ");
8188 assert_eq!(d.caret, 7);
8189 }
8190
8191 #[test]
8192 fn enter_at_the_end_of_a_quote_stays_in_the_quote() {
8193 // The gesture the rendering fix is for. `newline` inside a quote already
8194 // wrote the right source — `> a\n` becomes `> a\n>\n> \n`, twig's own
8195 // spelling — but the two marker lines it adds belonged to no node until
8196 // twig 3.2.0, so the gutter stopped at `a` and the line the writer had
8197 // just made drew as plain prose under the quote.
8198 let mut d = wysiwyg_doc("quote_enter", "> a\n");
8199 d.caret = 3; // past `a`, at the end of the quoted line
8200 d.newline();
8201 assert_eq!(d.source, "> a\n>\n> \n");
8202 d.build_visual(80);
8203 assert_eq!(drawn_rows(&d), ["│ a", "│ ", "│ "]);
8204 // And the caret is on the new line, not stranded on the old one.
8205 assert_eq!(d.caret, 8);
8206 }
8207
8208 #[test]
8209 fn opening_a_container_on_a_blank_line_is_one_undo_step() {
8210 // It was three edits — scratch, wrap, unscratch — coalesced into one, and
8211 // now it is twig's single edit. Either way one ⌘z has to put the blank
8212 // line back rather than undoing into a half-built document.
8213 for open in [
8214 &(|d: &mut Doc| d.toggle_blockquote()) as &dyn Fn(&mut Doc),
8215 &|d: &mut Doc| d.toggle_list(false),
8216 &|d: &mut Doc| d.toggle_list(true),
8217 ] {
8218 let mut d = doc_with("container_blank_undo", "a\n\n\n\nb\n");
8219 d.caret = 3;
8220 open(&mut d);
8221 assert_ne!(d.source, "a\n\n\n\nb\n");
8222 d.undo();
8223 assert_eq!(d.source, "a\n\n\n\nb\n");
8224 }
8225 }
8226
8227 #[test]
8228 fn a_container_toggle_is_one_undo_step() {
8229 let mut d = doc_with("quote_undo", "hello\n");
8230 d.caret = 3;
8231 d.insert("X"); // a typing run the structural edit must not fold into
8232 d.toggle_blockquote();
8233 assert_eq!(d.source, "> helXlo\n");
8234 d.undo();
8235 assert_eq!(d.source, "helXlo\n");
8236 }
8237
8238 // ── links ────────────────────────────────────────────────────────────────
8239
8240 #[test]
8241 fn insert_link_wraps_the_selection_and_leaves_its_text_selected() {
8242 let mut d = doc_with("link_sel", "word here\n");
8243 d.anchor = Some(0);
8244 d.caret = 4;
8245 d.insert_link("http://x.dev");
8246 assert_eq!(d.source, "[word](http://x.dev) here\n");
8247 // The text, not the destination — so a second press re-points the link
8248 // the first one made rather than nesting one inside it.
8249 assert_eq!(d.selected_text(), Some("word"));
8250 d.insert_link("http://y.dev");
8251 assert_eq!(d.source, "[word](http://y.dev) here\n");
8252 assert_eq!(d.selected_text(), Some("word"));
8253 }
8254
8255 #[test]
8256 fn insert_image_at_the_caret_spells_the_markup_and_lands_past_it() {
8257 let mut d = doc_with("img_caret", "before after\n");
8258 d.caret = 7; // between "before " and "after"
8259 d.insert_image("cat.png", "a cat");
8260 assert_eq!(d.source, "before after\n");
8261 // The caret sits just past the inserted image, nothing selected.
8262 assert_eq!(d.selection(), None);
8263 assert_eq!(d.caret, 7 + "".len());
8264 }
8265
8266 /// The bug a real vault hit: a filename with spaces in it. Markdown ends a
8267 /// destination at the first space, so the `format!` this used to be wrote
8268 /// something that was not an image at all — and the reader saw the markup as
8269 /// text. twig owns the spelling now, and moves it into the angle form.
8270 #[test]
8271 fn insert_image_spells_a_destination_with_spaces_so_it_stays_an_image() {
8272 let mut d = doc_with("img_space", "x\n");
8273 d.caret = 0;
8274 d.insert_image("Jesus Commands the Apostles to Rest.jpg", "");
8275 assert_eq!(
8276 d.source,
8277 "x\n"
8278 );
8279 // And it reads back as an image pointing at the unescaped path — the angle
8280 // brackets are spelling, not part of the destination.
8281 d.caret = 2;
8282 assert_eq!(
8283 d.image_destination_at_caret(),
8284 Some("Jesus Commands the Apostles to Rest.jpg".to_string())
8285 );
8286 }
8287
8288 /// A `)` in a caption or a filename must not close the image early.
8289 #[test]
8290 fn insert_image_escapes_a_paren_in_either_half() {
8291 let mut d = doc_with("img_paren", "x\n");
8292 d.caret = 0;
8293 d.insert_image("a)b.png", "");
8294 assert_eq!(d.source, "b.png)x\n");
8295 d.caret = 2;
8296 assert_eq!(d.image_destination_at_caret(), Some("a)b.png".to_string()));
8297 }
8298
8299 #[test]
8300 fn insert_image_uses_the_selection_as_alt_text() {
8301 let mut d = doc_with("img_sel", "caption here\n");
8302 d.anchor = Some(0);
8303 d.caret = 7; // "caption"
8304 d.insert_image("p.png", "ignored fallback");
8305 assert_eq!(d.source, " here\n");
8306 }
8307
8308 #[test]
8309 fn insert_image_with_no_alt_leaves_empty_brackets() {
8310 let mut d = doc_with("img_noalt", "\n");
8311 d.caret = 0;
8312 d.insert_image("logo.svg", "");
8313 assert_eq!(d.source, "\n");
8314 }
8315
8316 #[test]
8317 fn insert_media_spells_a_video_as_html_and_reads_it_back_as_a_block() {
8318 // The round trip is the point: it's no use writing markup the reader
8319 // can't pick up again. This is the pair that only holds from twig 2.5.1
8320 // on — before it, the one-line form went in fine and came back as a
8321 // paragraph of raw tags, publishing no media at all.
8322 let mut d = doc_with("vid_rt", "\n");
8323 d.caret = 0;
8324 d.insert_media(MediaKind::Video, "clip.mp4", "a clip");
8325 assert_eq!(
8326 d.source,
8327 "<video src=\"clip.mp4\" controls>a clip</video>\n"
8328 );
8329
8330 d.build_visual(80);
8331 assert_eq!(d.vmap.media.len(), 1, "reads back as one block media");
8332 assert_eq!(d.vmap.media[0].kind, MediaKind::Video);
8333 assert_eq!(d.vmap.media[0].destination, "clip.mp4");
8334 assert_eq!(d.vmap.media[0].alt, "a clip");
8335 }
8336
8337 #[test]
8338 fn insert_media_spells_audio_with_its_own_tag() {
8339 let mut d = doc_with("aud_rt", "\n");
8340 d.caret = 0;
8341 d.insert_media(MediaKind::Audio, "take.mp3", "");
8342 assert_eq!(d.source, "<audio src=\"take.mp3\" controls></audio>\n");
8343 d.build_visual(80);
8344 assert_eq!(d.vmap.media[0].kind, MediaKind::Audio);
8345 }
8346
8347 #[test]
8348 fn insert_media_uses_the_selection_as_fallback_text() {
8349 // The same courtesy `insert_image` does with alt: select a caption,
8350 // insert, and the caption labels the thing rather than being replaced.
8351 let mut d = doc_with("vid_sel", "the talk here\n");
8352 d.anchor = Some(0);
8353 d.caret = 8; // "the talk"
8354 d.insert_media(MediaKind::Video, "talk.mp4", "ignored fallback");
8355 assert_eq!(
8356 d.source,
8357 "<video src=\"talk.mp4\" controls>the talk</video> here\n"
8358 );
8359 }
8360
8361 #[test]
8362 fn insert_media_with_an_image_kind_is_just_insert_image() {
8363 let mut d = doc_with("img_via_media", "\n");
8364 d.caret = 0;
8365 d.insert_media(MediaKind::Image, "logo.svg", "x");
8366 assert_eq!(d.source, "\n");
8367 }
8368
8369 // ── thematic breaks ─────────────────────────────────────────────────────
8370
8371 /// The node the source parses as at `caret` — what confirms an inserted
8372 /// `---` actually reads back as a rule, not stray text or a setext heading.
8373 ///
8374 /// The *narrowest* node covering the offset. Every ancestor covers it too,
8375 /// and since twig 2.8 that includes the `doc` root, which now carries a real
8376 /// span (it reported none before, so taking the first match used to land on
8377 /// the block by luck and now always answers `"doc"`).
8378 fn kind_at(d: &mut Doc, caret: usize) -> Option<Kind> {
8379 d.nodes()
8380 .into_iter()
8381 .filter(|n| n.span.start <= caret && caret < n.span.end)
8382 .min_by_key(|n| n.span.end - n.span.start)
8383 .map(|n| n.kind)
8384 }
8385
8386 #[test]
8387 fn a_task_box_toggles_at_the_caret_and_reads_back() {
8388 let mut d = doc_with("task_toggle", "- [ ] todo\n- [x] done\n");
8389 d.caret = 8; // inside "todo"
8390 assert_eq!(d.task_checked_at_caret(), Some(false));
8391 d.toggle_task_checked();
8392 assert_eq!(d.source, "- [x] todo\n- [x] done\n");
8393 assert_eq!(d.task_checked_at_caret(), Some(true));
8394 d.toggle_task_checked();
8395 assert_eq!(d.source, "- [ ] todo\n- [x] done\n");
8396 }
8397
8398 #[test]
8399 fn a_click_toggles_a_box_without_taking_the_caret_with_it() {
8400 // The whole reason `toggle_task_at` exists apart from the caret form:
8401 // ticking a box elsewhere must not move the cursor out of what's being
8402 // typed.
8403 let mut d = doc_with("task_click", "- [ ] first\n- [ ] second\n");
8404 d.caret = 8; // inside "first"
8405 let second = d.source.find("second").unwrap();
8406 d.toggle_task_at(second);
8407 assert_eq!(d.source, "- [ ] first\n- [x] second\n");
8408 assert_eq!(d.caret, 8, "the caret stayed in the first item");
8409 }
8410
8411 #[test]
8412 fn a_plain_item_gains_and_loses_a_box() {
8413 let mut d = doc_with("task_mint", "- plain\n");
8414 d.caret = 4;
8415 assert_eq!(d.task_checked_at_caret(), None);
8416 d.toggle_task_item();
8417 assert_eq!(d.source, "- [ ] plain\n");
8418 assert_eq!(
8419 d.task_checked_at_caret(),
8420 Some(false),
8421 "a new box arrives unticked"
8422 );
8423 d.toggle_task_item();
8424 assert_eq!(d.source, "- plain\n");
8425 }
8426
8427 #[test]
8428 fn ticking_a_box_that_isnt_there_reports_rather_than_minting_one() {
8429 // `set checked` must not silently convert a bullet into a task — that is
8430 // `toggle_task_item`'s job, and twig refuses it here.
8431 let mut d = doc_with("task_none", "- plain\n");
8432 d.caret = 4;
8433 d.toggle_task_checked();
8434 assert_eq!(d.source, "- plain\n", "nothing written");
8435 assert!(
8436 d.status.is_some(),
8437 "the refusal should reach the status line"
8438 );
8439 }
8440
8441 #[test]
8442 fn a_task_item_in_a_quote_is_found_past_the_quote_marker() {
8443 let mut d = doc_with("task_quote", "> - [ ] nested\n");
8444 d.caret = d.source.find("nested").unwrap();
8445 assert_eq!(d.task_checked_at_caret(), Some(false));
8446 d.toggle_task_checked();
8447 assert_eq!(d.source, "> - [x] nested\n");
8448 }
8449
8450 #[test]
8451 fn insert_thematic_break_parts_the_paragraph_around_the_caret() {
8452 // A rule is a block, so twig's `insert_thematic_break` alone lands it
8453 // after the whole paragraph. `split_block` parts the paragraph first and
8454 // the rule is aimed at the *first* half, which is what a rule button is
8455 // understood to do — and what leaf spelled by hand until twig grew both
8456 // halves of the gesture.
8457 let mut d = doc_with("hr_mid", "before after\n");
8458 d.caret = 7; // between "before " and "after"
8459 d.insert_thematic_break();
8460 assert_eq!(d.source, "before \n\n---\n\nafter\n");
8461 assert_eq!(d.selection(), None);
8462 assert_eq!(
8463 kind_at(&mut d, "before \n\n".len()),
8464 Some(Kind::ThematicBreak)
8465 );
8466 }
8467
8468 #[test]
8469 fn insert_thematic_break_at_a_paragraph_s_end_splits_nothing() {
8470 // At the end there is nothing to part, and a split there writes the
8471 // separator anyway — a blank line and the empty slot the next paragraph
8472 // would fill — which the rule then landed above: `para\n\n* * *\n\n\n`,
8473 // two blank lines nothing fills. Now the rule lands after the paragraph,
8474 // where the split-and-aim was sending it regardless. Both formats, and
8475 // both shapes of a last line — terminated, and still being typed —
8476 // because the two reach the split through different doors: Markdown's
8477 // paragraph span stops before its newline, so `para\n` at 4 never split
8478 // there, but `para` at 4 did.
8479 for (fmt, rule) in [(Format::Markdown, "---"), (Format::Djot, "* * *")] {
8480 for src in ["para\n", "para"] {
8481 let mut d = Doc::from_source(src.into(), fmt).unwrap();
8482 d.caret = 4;
8483 d.insert_thematic_break();
8484 assert_eq!(d.source, format!("para\n\n{rule}\n"), "{fmt:?} {src:?}");
8485 assert_eq!(d.caret, d.source.len());
8486 }
8487 // Mid-document the slot sat between the rule and the next block.
8488 let mut d = Doc::from_source("para\n\nnext\n".into(), fmt).unwrap();
8489 d.caret = 4;
8490 d.insert_thematic_break();
8491 assert_eq!(d.source, format!("para\n\n{rule}\n\nnext\n"), "{fmt:?}");
8492 // Trailing whitespace is nothing to part either.
8493 let mut d = Doc::from_source("para \n".into(), fmt).unwrap();
8494 d.caret = 4;
8495 d.insert_thematic_break();
8496 assert_eq!(d.source, format!("para \n\n{rule}\n"), "{fmt:?}");
8497 }
8498 }
8499
8500 #[test]
8501 fn insert_thematic_break_at_a_paragraph_s_start_lands_before_it() {
8502 // The split at the start parts nothing, but it is kept on purpose:
8503 // `|para` becomes `\npara` with the caret on a blank line, and twig
8504 // (3.5.2) writes a rule aimed at a blank line ON that line — the only
8505 // way "before the paragraph" is reachable through a gesture that only
8506 // places after. Before 3.5.2 this came out as `\n\n---\n\npara`.
8507 for (fmt, rule) in [(Format::Markdown, "---"), (Format::Djot, "* * *")] {
8508 let mut d = Doc::from_source("para\n".into(), fmt).unwrap();
8509 d.caret = 0;
8510 d.insert_thematic_break();
8511 assert_eq!(d.source, format!("{rule}\n\npara\n"), "{fmt:?}");
8512 let mut d = Doc::from_source("prev\n\npara\n".into(), fmt).unwrap();
8513 d.caret = 6;
8514 d.insert_thematic_break();
8515 assert_eq!(d.source, format!("prev\n\n{rule}\n\npara\n"), "{fmt:?}");
8516 }
8517 }
8518
8519 #[test]
8520 fn insert_thematic_break_on_a_blank_line_takes_that_line() {
8521 // The gap between two blocks is where a click lands the caret; the
8522 // rule goes on the blank, one blank each side.
8523 let mut d = doc_with("hr_gap", "a\n\nb\n");
8524 d.caret = 2;
8525 d.insert_thematic_break();
8526 assert_eq!(d.source, "a\n\n---\n\nb\n");
8527 }
8528
8529 #[test]
8530 fn insert_table_at_a_paragraph_s_end_splits_nothing() {
8531 // The same door as the rule's, through the placement they share.
8532 let mut d = Doc::from_source("para\n".into(), Format::Djot).unwrap();
8533 d.caret = 4;
8534 d.insert_table(1, 1);
8535 assert_eq!(d.source, "para\n\n| |\n|---|\n| |\n");
8536 let mut d = doc_with("table_end_typed", "para");
8537 d.caret = 4;
8538 d.insert_table(1, 1);
8539 assert_eq!(d.source, "para\n\n| |\n| --- |\n| |\n");
8540 assert!(d.caret_in_table());
8541 }
8542
8543 #[test]
8544 fn insert_thematic_break_spells_the_rule_the_format_s_own_way() {
8545 // The whole point of delegating: `---` is Markdown's, `* * *` is djot's,
8546 // and leaf wrote the first into both until twig started spelling it.
8547 let mut md = doc_with("hr_md", "para\n");
8548 md.caret = 2;
8549 md.insert_thematic_break();
8550 assert_eq!(md.source, "pa\n\n---\n\nra\n");
8551
8552 let mut dj = Doc::from_source("para\n".into(), Format::Djot).unwrap();
8553 dj.caret = 2;
8554 dj.insert_thematic_break();
8555 assert_eq!(dj.source, "pa\n\n* * *\n\nra\n");
8556 }
8557
8558 #[test]
8559 fn insert_table_parts_the_paragraph_and_lands_in_the_first_header_cell() {
8560 // The table goes *at* the caret the way the rule does: the paragraph is
8561 // parted first, and twig writes the grid after its first half. The
8562 // caret then sits in the first header cell — selected, as Tab would
8563 // leave it — so the next keystroke is the heading.
8564 let mut d = doc_with("table_mid", "before after\n");
8565 d.caret = 7;
8566 d.insert_table(2, 3);
8567 assert_eq!(
8568 d.source,
8569 "before \n\n| | | |\n| --- | --- | --- |\n| | | |\n| | | |\n\nafter\n"
8570 );
8571 assert!(d.caret_in_table());
8572 let first_bar = d.source.find('|').unwrap();
8573 assert!(
8574 d.caret > first_bar && d.caret < d.source.find("| ---").unwrap(),
8575 "caret {} is not in the header row",
8576 d.caret
8577 );
8578 d.insert("Name");
8579 assert!(d.source.starts_with("before \n\n| Name | | |\n"));
8580 // And the grid the table was written into is one the table keys walk
8581 // (over the map a frontend rebuilds after every edit).
8582 d.build_visual(80);
8583 assert!(d.cell_tab(true));
8584 d.insert("Qty");
8585 assert!(d.source.starts_with("before \n\n| Name | Qty | |\n"));
8586 }
8587
8588 #[test]
8589 fn insert_table_spells_the_grid_the_format_s_own_way() {
8590 // Djot's delimiter row is unpadded, and leaf never has to know that.
8591 let mut dj = Doc::from_source("para\n".into(), Format::Djot).unwrap();
8592 dj.caret = 2;
8593 dj.insert_table(1, 2);
8594 assert_eq!(dj.source, "pa\n\n| | |\n|---|---|\n| | |\n\nra\n");
8595 assert!(dj.caret_in_table());
8596 }
8597
8598 #[test]
8599 fn insert_table_refuses_where_the_format_spells_no_table() {
8600 let mut d = Doc::from_source("<p>ab</p>\n".into(), Format::Html).unwrap();
8601 d.caret = 4;
8602 d.insert_table(1, 1);
8603 assert_eq!(d.source, "<p>ab</p>\n");
8604 assert!(d.status.as_deref().unwrap_or("").contains("not supported"));
8605 assert!(!d.capabilities().table);
8606 }
8607
8608 #[test]
8609 fn insert_table_reports_a_zero_shape_and_writes_nothing() {
8610 let mut d = doc_with("table_zero", "para\n");
8611 d.caret = 2;
8612 d.insert_table(0, 2);
8613 assert_eq!(d.source, "para\n");
8614 assert!(d.status.as_deref().unwrap_or("").starts_with("table:"));
8615 }
8616
8617 #[test]
8618 fn clicking_below_a_final_thematic_break_can_type_after_it() {
8619 let mut d = wysiwyg_doc("hr_final_click", "---\n");
8620 d.build_visual(80);
8621 d.click(d.vmap.num_rows() + 2, 0, false);
8622 assert_eq!(d.caret, d.source.len(), "the caret belongs after the rule");
8623 d.insert("after");
8624 assert_eq!(d.source, "---\nafter");
8625 }
8626
8627 #[test]
8628 fn enter_in_a_nested_list_item_keeps_the_new_item_nested() {
8629 // The same bytes are two documents. In Markdown ` - b` is a nested item
8630 // and the next one belongs beside it, at its indent. In Djot a list
8631 // marker can't interrupt a paragraph, so those bytes are literal text in
8632 // item `a` and there is only one item — writing ` - ` under it would add
8633 // no item at all, just more text, and the new sibling has to go to
8634 // column zero. Both spellings come out of the *enclosing item's* line.
8635 let mut md = wysiwyg_doc("enter_nested_md", "- a\n - b\n");
8636 md.caret = "- a\n - b".len();
8637 md.newline();
8638 assert_eq!(md.source, "- a\n - b\n - \n");
8639 assert_eq!(list_items(&mut md), 3);
8640
8641 let mut dj = Doc::from_source("- a\n - b\n".into(), Format::Djot).unwrap();
8642 dj.view = View::Wysiwyg;
8643 dj.build_visual(80);
8644 dj.caret = "- a\n - b".len();
8645 dj.newline();
8646 assert_eq!(dj.source, "- a\n - b\n- \n");
8647 assert_eq!(list_items(&mut dj), 2);
8648
8649 // Where Djot's nesting is real — opened by a blank line — the indent is
8650 // reproduced there too, and the two formats agree again.
8651 let mut dj = Doc::from_source("- a\n\n - b\n".into(), Format::Djot).unwrap();
8652 dj.view = View::Wysiwyg;
8653 dj.build_visual(80);
8654 dj.caret = "- a\n\n - b".len();
8655 dj.newline();
8656 assert_eq!(dj.source, "- a\n\n - b\n - \n");
8657 assert_eq!(list_items(&mut dj), 3);
8658 }
8659
8660 #[test]
8661 fn tab_nests_an_item_at_the_column_its_own_marker_asks_for() {
8662 // Tab replaces the line's whole prefix with the one twig spells, so the
8663 // quote markers, the parent's indent and an ordered marker's extra
8664 // column are all its answer rather than leaf's arithmetic.
8665 for (name, body, caret, want) in [
8666 ("bullet", "- a\n- b\n", 6, "- a\n - b\n"),
8667 ("ordered", "1. a\n2. b\n", 8, "1. a\n 1. b\n"),
8668 ("quoted", "> - a\n> - b\n", 10, "> - a\n> - b\n"),
8669 // A checkbox is markup the item's own text wraps past, but a nested
8670 // list may only open at the *list* marker's column — four in from
8671 // there is a paragraph continuation, and `- [ ] a\n - [ ] b`
8672 // parses as one item, not two.
8673 ("task", "- [ ] a\n- [ ] b\n", 14, "- [ ] a\n - [ ] b\n"),
8674 (
8675 "quoted task",
8676 "> - [ ] a\n> - [ ] b\n",
8677 18,
8678 "> - [ ] a\n> - [ ] b\n",
8679 ),
8680 ] {
8681 let mut doc = wysiwyg_doc(name, body);
8682 doc.caret = caret;
8683 doc.indent();
8684 assert_eq!(doc.source, want, "{name}");
8685 // The nesting is real, not just indented text.
8686 assert_eq!(list_items(&mut doc), 2, "{name}");
8687 }
8688 }
8689
8690 #[test]
8691 fn backspace_only_outdents_where_the_format_says_there_is_an_item() {
8692 // The same bytes, the two formats disagreeing, and a gesture that used
8693 // to read the bytes. ` - b` is a nested item in Markdown, so Backspace
8694 // at its marker outdents. In Djot a marker can't interrupt a paragraph,
8695 // so those bytes are literal text inside item `a` — there is nothing to
8696 // outdent, and treating them as a marker turned one item into two, a
8697 // structural edit from a keystroke that should delete one character.
8698 //
8699 // twig's `line_prefix` is what tells them apart: it reports the marker
8700 // on the Markdown line and nothing on the Djot one, which is a
8701 // continuation. No byte scan can reach that answer.
8702 let src = "- a\n - b\n";
8703 let at = "- a\n - ".len();
8704
8705 let mut md = Doc::from_source(src.into(), Format::Markdown).unwrap();
8706 md.view = View::Wysiwyg;
8707 md.build_visual(80);
8708 md.caret = at;
8709 md.backspace();
8710 assert_eq!(md.source, "- a\n- b\n");
8711 assert_eq!(list_items(&mut md), 2);
8712
8713 let mut dj = Doc::from_source(src.into(), Format::Djot).unwrap();
8714 dj.view = View::Wysiwyg;
8715 dj.build_visual(80);
8716 dj.caret = at;
8717 dj.backspace();
8718 assert_eq!(dj.source, "- a\n -b\n"); // an ordinary character delete
8719 assert_eq!(list_items(&mut dj), 1); // and the structure is untouched
8720 }
8721
8722 #[test]
8723 fn enter_in_a_checklist_item_starts_another_unchecked_one() {
8724 // Leaf used to spell the next item from the marker bytes it scanned, and
8725 // its scanner stopped at the bullet — so Enter in a checklist wrote `- `
8726 // and dropped out of the checklist. twig reproduces the whole
8727 // continuation, and a fresh item is always unticked however the one above
8728 // it stands.
8729 for (name, body, want) in [
8730 ("unchecked", "- [ ] a\n", "- [ ] a\n- [ ] \n"),
8731 ("checked", "- [x] a\n", "- [x] a\n- [ ] \n"),
8732 ] {
8733 let mut doc = wysiwyg_doc(name, body);
8734 doc.caret = body.trim_end_matches('\n').len();
8735 doc.newline();
8736 assert_eq!(doc.source, want, "{name}");
8737 // Both items are checklist items — the new one is a box, not the
8738 // plain bullet the old marker scan left behind — and it is unticked
8739 // whichever way the one above it faces.
8740 let boxes: Vec<Option<bool>> = doc
8741 .nodes()
8742 .iter()
8743 .filter(|n| n.kind == Kind::TaskListItem)
8744 .map(|n| n.checked)
8745 .collect();
8746 assert_eq!(boxes.len(), 2, "{name}");
8747 assert_eq!(boxes[1], Some(false), "{name}");
8748 }
8749 }
8750
8751 #[test]
8752 fn a_split_takes_the_space_the_caret_was_in_front_of() {
8753 // Splicing a break at the caret strands the space the words were parted
8754 // at on the head of the second block, where it reads as an indent nobody
8755 // typed. twig's split consumes it.
8756 for (name, body, caret, want) in [
8757 ("para", "one two\n", 3, "one\n\ntwo\n"),
8758 ("item", "- one two\n", 5, "- one\n- two\n"),
8759 ("quote", "> one two\n", 5, "> one\n>\n> two\n"),
8760 // A heading takes leaf's own path, which has to match.
8761 ("heading", "# one two\n", 5, "# one\n\ntwo\n"),
8762 ] {
8763 let mut doc = wysiwyg_doc(name, body);
8764 doc.caret = caret;
8765 doc.newline();
8766 assert_eq!(doc.source, want, "{name}");
8767 }
8768 }
8769
8770 #[test]
8771 fn enter_at_the_end_of_a_heading_opens_a_paragraph() {
8772 // The one place leaf keeps its own break: `split_block` repeats the `#`,
8773 // and Enter after a title is how the body under it is asked for.
8774 let mut doc = wysiwyg_doc("head_enter", "# Title\n");
8775 doc.caret = "# Title".len();
8776 doc.newline();
8777 doc.insert("body");
8778 assert_eq!(doc.source, "# Title\n\nbody\n");
8779 assert_eq!(
8780 doc.nodes()
8781 .iter()
8782 .filter(|n| n.kind == Kind::Heading)
8783 .count(),
8784 1
8785 );
8786 }
8787
8788 #[test]
8789 fn enter_in_a_quoted_list_item_starts_the_next_quoted_item() {
8790 // A quoted item's marker doesn't open its line, so a scan that starts at
8791 // column zero finds a `>` where it wanted a bullet, calls the line "not a
8792 // list" and hands Enter to the plain-quote branch — which writes `> ` and
8793 // drops the list. The next item has to carry the whole prefix.
8794 for (name, body, want) in [
8795 ("flat", "> - a\n", "> - a\n> - \n"),
8796 ("sibling", "> - a\n> - b\n", "> - a\n> - b\n> - \n"),
8797 ("nested", "> - a\n> - b\n", "> - a\n> - b\n> - \n"),
8798 ("ordered", "> 1. a\n> 2. b\n", "> 1. a\n> 2. b\n> 3. \n"),
8799 ("twice quoted", "> > - a\n", "> > - a\n> > - \n"),
8800 ] {
8801 let mut doc = wysiwyg_doc(name, body);
8802 doc.caret = body.trim_end_matches('\n').len();
8803 doc.newline();
8804 assert_eq!(doc.source, want, "{name}");
8805 // The marker isn't just spelled right, it parses as an item.
8806 assert_eq!(list_items(&mut doc), body.lines().count() + 1, "{name}");
8807 }
8808 }
8809
8810 #[test]
8811 fn an_empty_quoted_item_leaves_the_list_and_stays_in_the_quote() {
8812 // Double-Enter exits the list. Unquoted that means a blank line, but a
8813 // *bare* blank line would end the quote too and drop the caret out of it,
8814 // so the separator keeps its `>` and the caret's line keeps its `> `.
8815 let mut doc = wysiwyg_doc("quoted_exit", "> - a\n> - \n");
8816 doc.caret = "> - a\n> - ".len();
8817 doc.newline();
8818 assert_eq!(doc.source, "> - a\n>\n> \n");
8819 assert_eq!(list_items(&mut doc), 1);
8820 // What "still in the quote" means for the next keystroke: the caret sits
8821 // behind the prefix, and what's typed there lands inside the quote as a
8822 // paragraph of its own — not as more of item `a`.
8823 doc.insert("x");
8824 assert_eq!(doc.source, "> - a\n>\n> x\n");
8825 assert!(
8826 doc.editor
8827 .ancestors_at(doc.caret - 1)
8828 .is_ok_and(|c| c.into_iter().any(|m| m.kind == Kind::BlockQuote))
8829 );
8830 }
8831
8832 #[test]
8833 fn backspace_at_a_quoted_marker_takes_the_marker_and_leaves_the_quote() {
8834 // The marker is hidden block markup, so Backspace over it is structural —
8835 // but only the marker is the list's. Splicing from the line start would
8836 // take the `>` with it and silently unquote the line.
8837 let mut doc = wysiwyg_doc("quoted_bksp", "> - a\n");
8838 doc.caret = "> - ".len();
8839 doc.backspace();
8840 assert_eq!(doc.source, "> a\n");
8841 assert_eq!(list_items(&mut doc), 0);
8842
8843 // A nested one outdents instead, moving the bullet within the quote
8844 // rather than moving the quote.
8845 let mut doc = wysiwyg_doc("quoted_outdent", "> - a\n> - b\n");
8846 doc.caret = "> - a\n> - ".len();
8847 doc.backspace();
8848 assert_eq!(doc.source, "> - a\n> - b\n");
8849 assert_eq!(list_items(&mut doc), 2);
8850 }
8851
8852 #[test]
8853 fn only_a_bare_paragraph_is_parted_around_the_caret() {
8854 // The split is deliberately narrow. Parting a fenced block would leave
8855 // two fences with a rule between them, and parting a list item would
8856 // mint an item nobody asked for on the way to a rule that lands after
8857 // the list either way — so both keep the whole block intact and take the
8858 // rule after it. A caret in a quote is likewise left alone.
8859 for (name, body, caret, want) in [
8860 (
8861 "code",
8862 "```\nfn x() {}\n```\n",
8863 8,
8864 "```\nfn x() {}\n```\n\n---\n",
8865 ),
8866 ("list", "- one two\n", 6, "- one two\n\n---\n"),
8867 ("quote", "> one two\n", 6, "> one two\n>\n> ---\n"),
8868 ] {
8869 let mut d = doc_with(&format!("hr_narrow_{name}"), body);
8870 d.caret = caret;
8871 d.insert_thematic_break();
8872 assert_eq!(d.source, want, "{name}: the block should stay whole");
8873 }
8874 }
8875
8876 #[test]
8877 fn insert_thematic_break_replaces_the_selection() {
8878 // Now that the rule lands *at* the caret again, replacing the selection
8879 // is coherent once more: the text goes, and the rule takes its place.
8880 // The space the deletion left leading the second half is consumed by the
8881 // split rather than opening the new paragraph with it.
8882 let mut d = doc_with("hr_sel", "one two three\n");
8883 d.anchor = Some(4);
8884 d.caret = 7; // "two"
8885 d.insert_thematic_break();
8886 assert_eq!(d.source, "one \n\n---\n\nthree\n");
8887 assert_eq!(d.selection(), None);
8888 }
8889
8890 #[test]
8891 fn insert_thematic_break_clears_a_code_block_and_a_table_rather_than_refusing() {
8892 // Both are blocks the rule lands *after*. Leaf used to refuse a fence,
8893 // because writing `---` into one is code, not a rule — twig now walks out
8894 // to the block that owns the caret's line, so there is nothing to refuse.
8895 let mut code = doc_with("hr_code", "```\nfn x() {}\n```\n");
8896 code.caret = 5; // inside the fenced code
8897 code.insert_thematic_break();
8898 assert_eq!(code.source, "```\nfn x() {}\n```\n\n---\n");
8899 assert_eq!(code.status, None, "no refusal to report any more");
8900
8901 let mut table = doc_with("hr_table", "| a | b |\n|---|---|\n| 1 | 2 |\n");
8902 table.caret = 3; // in the header row
8903 table.insert_thematic_break();
8904 assert_eq!(table.source, "| a | b |\n|---|---|\n| 1 | 2 |\n\n---\n");
8905 }
8906
8907 #[test]
8908 fn insert_thematic_break_in_a_list_item_ends_the_list() {
8909 // The un-indented rule cannot continue the list, so it closes the list
8910 // and lands at the top level rather than nested inside it.
8911 let mut d = doc_with("hr_list", "- one\n- two\n");
8912 d.caret = "- one\n- tw".len(); // mid "two"
8913 d.insert_thematic_break();
8914 d.build_visual(80);
8915 let rule_at = d.source.find("---").unwrap();
8916 assert_eq!(kind_at(&mut d, rule_at), Some(Kind::ThematicBreak));
8917 assert!(
8918 !d.nodes().iter().any(|n| n.kind == Kind::BulletList
8919 && n.span.start <= rule_at
8920 && rule_at < n.span.end),
8921 "the rule must not be nested inside the list"
8922 );
8923 }
8924
8925 #[test]
8926 fn insert_thematic_break_in_a_blockquote_stays_in_the_quote() {
8927 // Leaf used to end the quote. twig gives the rule the quote's own prefix,
8928 // which is the document the gesture was actually asked for.
8929 let mut d = doc_with("hr_quote", "> hello\n");
8930 d.caret = 4; // inside the quoted text
8931 d.insert_thematic_break();
8932 assert_eq!(d.source, "> hello\n>\n> ---\n");
8933 d.build_visual(80);
8934 let rule_at = d.source.find("---").unwrap();
8935 assert_eq!(kind_at(&mut d, rule_at), Some(Kind::ThematicBreak));
8936 assert!(
8937 d.nodes().iter().any(|n| n.kind == Kind::BlockQuote
8938 && n.span.start <= rule_at
8939 && rule_at < n.span.end),
8940 "the rule belongs to the quote it was asked for"
8941 );
8942 }
8943
8944 // ── typing against a block picture ────────────────────────────────────────
8945
8946 /// A rendered-view document with the caret parked on one of the picture's two
8947 /// stops, and the map already built — the state a frontend is in between
8948 /// drawing a frame and the next keystroke.
8949 fn doc_at_picture(name: &str, src: &str, side: MediaStop) -> Doc {
8950 let mut d = doc_in(View::Wysiwyg, name, src);
8951 d.build_visual_unwrapped();
8952 let start = src.find("".len(),
8956 };
8957 d
8958 }
8959
8960 /// The block media the map publishes, after rebuilding it — "is this still a
8961 /// picture, or has it become a line of text with an image in it?"
8962 fn media_count(d: &mut Doc) -> usize {
8963 d.build_visual_unwrapped();
8964 d.vmap.media.len()
8965 }
8966
8967 #[test]
8968 fn typing_past_a_block_picture_opens_a_paragraph_under_it() {
8969 // The accident this prevents: tap the blank page under a photo (which
8970 // lands on the picture's trailing stop), type, and `xy` is a
8971 // paragraph with an *inline* image — the photo stops being drawn.
8972 let mut d = doc_at_picture("pic_after", "hi\n\n\n", MediaStop::After);
8973 d.insert("xy");
8974 assert_eq!(d.source, "hi\n\n\n\nxy\n");
8975 assert_eq!(media_count(&mut d), 1, "still a picture");
8976 }
8977
8978 #[test]
8979 fn typing_in_front_of_a_block_picture_opens_a_paragraph_above_it() {
8980 let mut d = doc_at_picture("pic_before", "hi\n\n\n", MediaStop::Before);
8981 d.insert("xy");
8982 assert_eq!(d.source, "hi\n\nxy\n\n\n");
8983 assert_eq!(media_count(&mut d), 1);
8984 }
8985
8986 #[test]
8987 fn a_picture_that_opens_the_document_still_takes_a_paragraph_above_it() {
8988 let mut d = doc_at_picture("pic_first", "\n", MediaStop::Before);
8989 d.insert("x");
8990 assert_eq!(d.source, "x\n\n\n");
8991 assert_eq!(media_count(&mut d), 1);
8992 }
8993
8994 #[test]
8995 fn one_undo_puts_the_picture_back_the_way_it_was_found() {
8996 // The opened paragraph is part of the keystroke, not an edit the writer
8997 // made — so it undoes with the character, not a step later.
8998 let mut d = doc_at_picture("pic_undo", "hi\n\n\n", MediaStop::After);
8999 d.insert("x");
9000 assert_eq!(d.source, "hi\n\n\n\nx\n");
9001 d.undo();
9002 assert_eq!(d.source, "hi\n\n\n");
9003 }
9004
9005 #[test]
9006 fn pasting_against_a_block_picture_opens_a_paragraph_too() {
9007 // ⌘V dissolves the picture exactly as a keystroke does.
9008 let mut d = doc_at_picture("pic_paste", "hi\n\n\n", MediaStop::After);
9009 d.paste("pasted");
9010 assert_eq!(d.source, "hi\n\n\n\npasted\n");
9011 assert_eq!(media_count(&mut d), 1);
9012 }
9013
9014 #[test]
9015 fn typing_beside_an_inline_image_is_ordinary_editing() {
9016 // An inline image has no placeholder row and no stops of its own. Opening
9017 // a paragraph mid-sentence would be the bug, not the fix.
9018 let mut d = doc_in(View::Wysiwyg, "pic_inline", "see  here\n");
9019 d.build_visual_unwrapped();
9020 d.caret = "see ".len();
9021 d.insert("!");
9022 assert_eq!(d.source, "see ! here\n");
9023 }
9024
9025 #[test]
9026 fn source_view_types_raw_markup_against_an_image_untouched() {
9027 // Source view is for writing the markup itself; a break inserted behind
9028 // the writer's back there would be the editor arguing with them.
9029 let mut d = doc_in(View::Source, "pic_src", "\n");
9030 d.caret = "".len();
9031 d.insert("x");
9032 assert_eq!(d.source, "x\n");
9033 }
9034
9035 #[test]
9036 fn typing_over_a_selection_that_starts_at_a_picture_stop_replaces_it() {
9037 // A selection is replaced, not joined into, so there is nothing to
9038 // protect: the range takes the picture with it.
9039 let mut d = doc_at_picture("pic_sel", "hi\n\n\n", MediaStop::Before);
9040 d.anchor = Some(d.caret);
9041 d.caret = d.source.find("".len();
9042 d.insert("x");
9043 assert_eq!(d.source, "hi\n\nx\n");
9044 }
9045
9046 #[test]
9047 fn backspace_past_a_block_picture_deletes_the_picture_not_its_last_byte() {
9048 // What this actually cost: a real vault's photo, to one stray Backspace.
9049 // The caret past `` was deleting the closing paren — invisible
9050 // in the rendered view — and the photo became the text `\n", MediaStop::After);
9052 d.backspace();
9053 assert_eq!(d.source, "hi\n");
9054 assert_eq!(media_count(&mut d), 0, "the picture went, in one piece");
9055 d.undo();
9056 assert_eq!(
9057 d.source, "hi\n\n\n",
9058 "and comes back in one piece"
9059 );
9060 }
9061
9062 #[test]
9063 fn backspace_in_front_of_a_block_picture_steps_out_instead_of_merging_it() {
9064 // Deleting the break here would join the picture to the paragraph above,
9065 // where it is an *inline* image and stops being drawn. Step over the
9066 // boundary; the next press deletes in the paragraph the caret reached.
9067 let mut d = doc_at_picture("pic_bs_before", "hi\n\n\n", MediaStop::Before);
9068 d.backspace();
9069 assert_eq!(d.source, "hi\n\n\n", "nothing deleted");
9070 assert_eq!(d.caret, 2, "the caret stepped up to the end of `hi`");
9071 d.backspace();
9072 assert_eq!(d.source, "h\n\n\n", "and now it deletes there");
9073 assert_eq!(media_count(&mut d), 1, "the picture was never at risk");
9074 }
9075
9076 #[test]
9077 fn forward_delete_in_front_of_a_block_picture_deletes_the_picture() {
9078 // The mirror. A byte-step here eats the `!` and leaves a link.
9079 let mut d = doc_at_picture("pic_del", "hi\n\n\n\nbye\n", MediaStop::Before);
9080 d.delete_forward();
9081 assert_eq!(d.source, "hi\n\nbye\n");
9082 assert_eq!(media_count(&mut d), 0);
9083 }
9084
9085 #[test]
9086 fn forward_delete_past_a_block_picture_steps_over_the_boundary() {
9087 let mut d = doc_at_picture(
9088 "pic_del_after",
9089 "hi\n\n\n\nbye\n",
9090 MediaStop::After,
9091 );
9092 d.delete_forward();
9093 assert_eq!(d.source, "hi\n\n\n\nbye\n", "nothing deleted");
9094 assert_eq!(
9095 d.caret,
9096 d.source.find("bye").unwrap(),
9097 "the caret stepped down to `bye`"
9098 );
9099 }
9100
9101 #[test]
9102 fn a_picture_that_is_the_whole_document_still_deletes_cleanly() {
9103 let mut d = doc_at_picture("pic_only", "\n", MediaStop::After);
9104 d.backspace();
9105 assert_eq!(d.source, "\n");
9106 assert_eq!(media_count(&mut d), 0);
9107 }
9108
9109 #[test]
9110 fn a_word_delete_takes_the_picture_whole_or_steps_out_of_it() {
9111 // ⌥⌫ past a picture would otherwise eat a "word" of its markup.
9112 let mut d = doc_at_picture("pic_wordbs", "hi there\n\n\n", MediaStop::After);
9113 d.delete_word_back();
9114 assert_eq!(d.source, "hi there\n");
9115
9116 // And in front of one it runs *through* the paragraph break into the
9117 // prose above, which merges the picture inline — so it steps out first,
9118 // and the second press deletes the word it was aimed at.
9119 let mut d = doc_at_picture("pic_wordbs2", "hi there\n\n\n", MediaStop::Before);
9120 d.delete_word_back();
9121 assert_eq!(d.source, "hi there\n\n\n");
9122 d.delete_word_back();
9123 assert_eq!(
9124 d.source, "hi \n\n\n",
9125 "the word above went, the picture stayed"
9126 );
9127 assert_eq!(media_count(&mut d), 1);
9128 }
9129
9130 #[test]
9131 fn source_view_deletes_raw_markup_against_an_image_untouched() {
9132 let mut d = doc_in(View::Source, "pic_src_del", "\n");
9133 d.caret = "".len();
9134 d.backspace();
9135 assert_eq!(d.source, ";
9136 }
9137
9138 #[test]
9139 fn image_destination_at_caret_reads_the_image_under_the_caret() {
9140 let mut d = doc_with("img_read", "\n");
9141 d.caret = 3; // inside the image markup
9142 assert_eq!(d.image_destination_at_caret(), Some("cat.png".to_string()));
9143 // Past the image, the caret is in no image.
9144 d.caret = "".len();
9145 assert_eq!(d.image_destination_at_caret(), None);
9146 }
9147
9148 #[test]
9149 fn set_media_rows_reserves_blank_filler_rows_the_frontend_paints_over() {
9150 // The image is one placeholder row by default, and `set_media_rows` grows
9151 // it to the height the frontend measured: the label row plus blank
9152 // `decoration` fillers that hold the vertical space a raster is drawn into.
9153 let mut d = wysiwyg_doc("img_rows", "intro\n\n\n\nend\n");
9154 assert_eq!(d.vmap.media.len(), 1);
9155 let img_row = d.vmap.media[0].rows_span.start;
9156 assert_eq!(
9157 d.vmap.media[0].rows_span,
9158 img_row..img_row + 1,
9159 "default is one row"
9160 );
9161
9162 d.set_media_rows(HashMap::from([("cat.png".to_string(), 4)]));
9163 d.build_visual(80);
9164 assert_eq!(d.vmap.media.len(), 1, "still one image, now taller");
9165 let span = d.vmap.media[0].rows_span.clone();
9166 assert_eq!(span.end - span.start, 4, "reserves the four rows asked for");
9167 // The label row carries the mark and its glyphs; the three below are blank
9168 // decoration — drawn, but no caret and no text.
9169 assert!(
9170 d.vmap.rows[span.start].media.is_some(),
9171 "mark rides the first row"
9172 );
9173 for r in (span.start + 1)..span.end {
9174 assert!(d.vmap.rows[r].decoration, "filler row {r} is decoration");
9175 assert!(d.vmap.rows[r].glyphs.is_empty(), "filler row {r} is blank");
9176 assert!(
9177 d.vmap.rows[r].media.is_none(),
9178 "only the first row is marked"
9179 );
9180 }
9181 }
9182
9183 #[test]
9184 fn a_taller_image_adds_no_caret_stops_and_motion_steps_over_its_fillers() {
9185 // The extra rows are pure spacers: the caret's only homes stay the stop in
9186 // front of the image and the one just past it, so walking the document top
9187 // to bottom visits the same offsets whether the image is 1 row or 5.
9188 let body = "ab\n\n\n\ncd\n";
9189 let stops_at = |rows: usize| -> Vec<usize> {
9190 let mut d = wysiwyg_doc("img_stops", body);
9191 if rows > 1 {
9192 d.set_media_rows(HashMap::from([("p.png".to_string(), rows)]));
9193 d.build_visual(80);
9194 }
9195 d.caret = 0;
9196 let mut seen = vec![d.caret];
9197 loop {
9198 d.move_right(false);
9199 if *seen.last().unwrap() == d.caret {
9200 break;
9201 }
9202 seen.push(d.caret);
9203 }
9204 seen
9205 };
9206 assert_eq!(
9207 stops_at(1),
9208 stops_at(5),
9209 "reserving rows must not add stops"
9210 );
9211 }
9212
9213 #[test]
9214 fn insert_link_repoints_the_link_at_a_bare_caret() {
9215 let mut d = doc_with("link_repoint", "[word](http://x.dev)\n");
9216 d.caret = 3; // in the link's text, nothing selected
9217 d.insert_link("http://y.dev");
9218 assert_eq!(d.source, "[word](http://y.dev)\n");
9219 assert_eq!(d.selected_text(), Some("word"));
9220 }
9221
9222 #[test]
9223 fn insert_link_on_an_empty_range_autolinks_a_url() {
9224 // A link with no text of its own is an autolink, and twig spells it —
9225 // `<…>` is the canonical form and needs no text typed into it, so the
9226 // caret lands after it rather than selecting a finished link.
9227 let mut d = doc_with("link_empty", "\n");
9228 d.caret = 0;
9229 d.insert_link("http://x.dev");
9230 assert_eq!(d.source, "<http://x.dev>\n");
9231 assert_eq!(d.selection(), None);
9232 assert_eq!(d.caret, 14);
9233 }
9234
9235 #[test]
9236 fn insert_link_on_an_empty_range_falls_back_for_a_non_url() {
9237 // `<./notes.md>` is literal text in both formats and `<foo>` is raw HTML
9238 // in Markdown, so a destination that can't autolink doubles as the text
9239 // instead — which is then selected, ready to be typed over.
9240 let mut d = doc_with("link_rel", "\n");
9241 d.caret = 0;
9242 d.insert_link("./notes.md");
9243 assert_eq!(d.source, "[./notes.md](./notes.md)\n");
9244 assert_eq!(d.selection(), Some((1, 11)));
9245 d.insert("Notes");
9246 assert_eq!(d.source, "[Notes](./notes.md)\n");
9247 }
9248
9249 #[test]
9250 fn insert_link_repoints_the_autolink_the_caret_stands_in() {
9251 // The autolink's text is its URL, so re-pointing replaces the whole
9252 // node — the caret must not splice a second link inside the first.
9253 let mut d = doc_with("link_repoint_auto", "see <https://x.dev> ok\n");
9254 d.caret = 10;
9255 d.insert_link("https://y.dev");
9256 assert_eq!(d.source, "see <https://y.dev> ok\n");
9257 }
9258
9259 #[test]
9260 fn code_language_reads_and_edits_through_the_fence() {
9261 let mut d = doc_with("code_lang", "```rust\nlet x = 1;\n```\n");
9262 d.caret = 10; // inside the code body
9263 assert_eq!(d.code_language_at_caret().as_deref(), Some("rust"));
9264 assert!(d.caret_in_fenced_code());
9265
9266 d.set_code_language("python");
9267 assert!(
9268 d.source.starts_with("```python\n"),
9269 "source: {:?}",
9270 d.source
9271 );
9272 assert_eq!(d.code_language_at_caret().as_deref(), Some("python"));
9273
9274 // Clearing it leaves a bare fence and no label.
9275 d.set_code_language("");
9276 assert!(d.source.starts_with("```\n"), "source: {:?}", d.source);
9277 assert_eq!(d.code_language_at_caret(), None);
9278
9279 // A caret outside any code block edits nothing.
9280 let mut p = doc_with("code_lang_none", "just prose\n");
9281 assert!(!p.caret_in_fenced_code());
9282 p.set_code_language("rust");
9283 assert_eq!(p.source, "just prose\n");
9284 }
9285
9286 #[test]
9287 fn a_language_the_fence_cannot_carry_is_refused_not_written() {
9288 // Markdown's info string ends at whitespace, so `two words` would write
9289 // a fence that reads back with a different language than the one asked
9290 // for. twig refuses it; leaf reports that and leaves the source alone.
9291 // The old splice trimmed the ends and wrote whatever was left.
9292 let mut d = doc_with("code_lang_bad", "```rust\nx\n```\n");
9293 d.caret = 10;
9294 d.set_code_language("two words");
9295 assert_eq!(d.source, "```rust\nx\n```\n", "source should be untouched");
9296 assert!(d.status.is_some(), "the refusal should be reported");
9297 assert_eq!(d.code_language_at_caret().as_deref(), Some("rust"));
9298 }
9299
9300 #[test]
9301 fn link_destination_at_caret_reads_both_spellings() {
9302 let mut d = doc_with("link_dest", "see [t](https://x.dev) ok\n");
9303 d.caret = 5;
9304 assert_eq!(
9305 d.link_destination_at_caret().as_deref(),
9306 Some("https://x.dev")
9307 );
9308 d.caret = 0;
9309 assert_eq!(d.link_destination_at_caret(), None);
9310
9311 // An autolink has no `destination`; its text is the URL.
9312 let mut a = doc_with("link_dest_auto", "see <https://x.dev> ok\n");
9313 a.caret = 10;
9314 assert_eq!(
9315 a.link_destination_at_caret().as_deref(),
9316 Some("https://x.dev")
9317 );
9318 a.caret = 21;
9319 assert_eq!(a.link_destination_at_caret(), None);
9320 }
9321
9322 #[test]
9323 fn locate_finds_the_block_a_declared_id_names() {
9324 // The Book of Mormon shape: one document per chapter, one `{#v…}` per
9325 // verse. The locator has to land on the *verse*, which is the whole
9326 // reason a link carries one.
9327 let src = "{#v1}\nI, Nephi, having been born of goodly parents.\n\n\
9328 {#v2}\nYea, I make a record in the language of my father.\n";
9329 let mut d = Doc::from_source(src.to_string(), Format::Djot).unwrap();
9330 let v2 = d.locate("v2").expect("the document declares `{#v2}`");
9331 assert_eq!(
9332 d.source[v2.start..v2.end].trim_end(),
9333 "Yea, I make a record in the language of my father."
9334 );
9335 // The attribute line is not part of it: `start` is a place to put a
9336 // caret, and `{#v2}` is markup the caret has no business landing in.
9337 assert!(d.source[..v2.start].ends_with("{#v2}\n"));
9338 assert_eq!(d.locate("v99"), None);
9339 }
9340
9341 #[test]
9342 fn locate_reads_a_heading_by_its_words_when_the_format_mints_no_ids() {
9343 // Markdown has no ids at all — twig mints none, and `{#custom}` in a
9344 // Markdown heading is literal text. So `#the-second-part` can only be
9345 // the heading's own words, which is the rule every Markdown renderer
9346 // already follows and therefore the one a link was authored against.
9347 let src = "# Title\n\nintro\n\n## The Second Part\n\nbody\n\n## Third\n\nmore\n";
9348 let mut d = doc_with("locate_md", src);
9349 let hit = d.locate("the-second-part").expect("the heading's slug");
9350 assert!(d.source[hit.start..].starts_with("## The Second Part"));
9351 // Bounded by the next heading that isn't under it, so a peek shows the
9352 // section rather than only its title.
9353 assert_eq!(
9354 &d.source[hit.start..hit.end],
9355 "## The Second Part\n\nbody\n\n"
9356 );
9357
9358 // A subsection does not end its parent: `# Title` runs to `## Third`'s
9359 // sibling only because there is no other `#`, so it covers the lot.
9360 let title = d.locate("title").expect("the top heading");
9361 assert_eq!(title.end, d.source.len());
9362 }
9363
9364 #[test]
9365 fn locate_reads_a_djot_auto_id_however_the_link_spelled_it() {
9366 // djot mints `Some-Heading-Here`; a link to it is written
9367 // `#some-heading-here` by nearly everything that writes links. Both
9368 // spellings are one question.
9369 let src = "## Some Heading Here\n\nbody\n";
9370 let mut d = Doc::from_source(src.to_string(), Format::Djot).unwrap();
9371 let exact = d.locate("Some-Heading-Here").expect("djot's own spelling");
9372 let slugged = d.locate("some-heading-here").expect("the link's spelling");
9373 assert_eq!(exact, slugged);
9374 // The section, not the heading line — there is more to show than a title.
9375 assert_eq!(&d.source[exact.start..exact.end], src);
9376 }
9377
9378 #[test]
9379 fn locate_ignores_an_empty_locator_and_one_that_slugs_to_nothing() {
9380 let mut d = doc_with("locate_empty", "# Title\n\nbody\n");
9381 assert_eq!(d.locate(""), None);
9382 assert_eq!(d.locate(" "), None);
9383 // All punctuation: it names nothing, and must not be read as "match the
9384 // first heading whose slug is also empty".
9385 assert_eq!(d.locate("!!!"), None);
9386 }
9387
9388 #[test]
9389 fn locate_gives_a_duplicated_id_to_the_first_block_that_claims_it() {
9390 // The document's mistake, and the answer every other anchor
9391 // implementation gives — the alternative is for a link to mean whichever
9392 // of the two a walk happened to reach first.
9393 let src = "{#dup}\nfirst.\n\n{#dup}\nsecond.\n";
9394 let mut d = Doc::from_source(src.to_string(), Format::Djot).unwrap();
9395 let hit = d.locate("dup").expect("the first `{#dup}`");
9396 assert_eq!(d.source[hit.start..hit.end].trim_end(), "first.");
9397 }
9398
9399 #[test]
9400 fn insert_footnote_writes_both_halves_and_lands_the_caret_in_the_note() {
9401 // The button's whole job: a reference where the caret was, a definition
9402 // to give it meaning, and the caret waiting in the empty note so the
9403 // next keystroke is the note's first word.
9404 let mut d = doc_with("fn_insert", "A claim and more.\n");
9405 d.caret = 7; // just past "A claim"
9406 d.insert_footnote();
9407 assert!(
9408 d.source.starts_with("A claim[^1] and more."),
9409 "{:?}",
9410 d.source
9411 );
9412 assert!(
9413 d.source.contains("[^1]:"),
9414 "the definition too: {:?}",
9415 d.source
9416 );
9417 assert_eq!(d.status, None);
9418
9419 let reference = d.source.find("[^1]").unwrap();
9420 let note = d
9421 .footnote_at(reference + 2)
9422 .expect("the reference just written");
9423 assert_eq!(note.label, "1");
9424 assert_eq!(note.text.as_deref(), Some(""), "the note starts empty");
9425 assert_eq!(Some(d.caret), note.offset, "the caret waits in the note");
9426 // …and typing there is typing into the note, not near it.
9427 d.insert("the note");
9428 assert_eq!(
9429 d.footnote_at(reference + 2).and_then(|f| f.text),
9430 Some("the note".to_string())
9431 );
9432 }
9433
9434 #[test]
9435 fn insert_footnote_numbers_past_the_notes_already_written() {
9436 // A second press must not hand back a label somebody else is using: twig
9437 // reuses a defined label rather than appending a rival definition, so a
9438 // repeat of `1` would quietly point the new reference at the old note.
9439 let mut d = doc_with("fn_insert_number", "One[^1] two.\n\n[^1]: first\n");
9440 d.caret = 7; // past `[^1]`, before " two."
9441 d.insert_footnote();
9442 assert!(d.source.starts_with("One[^1][^2] two."), "{:?}", d.source);
9443 assert_eq!(d.source.matches("[^2]:").count(), 1);
9444 }
9445
9446 #[test]
9447 fn insert_footnote_counts_a_dangling_reference_and_ignores_a_named_one() {
9448 // `[^2]` with no definition is still a 2 that means something to whoever
9449 // wrote it — stepping over it would mint a note for their reference. A
9450 // word label takes no number, so it blocks none.
9451 let mut d = doc_with("fn_insert_dangling", "a[^2] b[^why] c\n\n[^why]: named\n");
9452 d.caret = d.source.find(" c").unwrap();
9453 d.insert_footnote();
9454 assert!(d.source.contains("[^1]:"), "1 is free: {:?}", d.source);
9455 assert!(
9456 d.source.starts_with("a[^2] b[^why][^1] c"),
9457 "{:?}",
9458 d.source
9459 );
9460 }
9461
9462 #[test]
9463 fn insert_footnote_marks_the_selection_rather_than_replacing_it() {
9464 // A reference annotates the words before it. Consuming the selection —
9465 // which is what an insert normally does — would delete the very claim
9466 // the author selected in order to footnote.
9467 let mut d = doc_with("fn_insert_sel", "A claim and more.\n");
9468 d.anchor = Some(2);
9469 d.caret = 7; // "claim" selected
9470 d.insert_footnote();
9471 assert!(
9472 d.source.starts_with("A claim[^1] and more."),
9473 "{:?}",
9474 d.source
9475 );
9476 }
9477
9478 #[test]
9479 fn a_note_just_written_still_knows_where_its_reference_is() {
9480 // The authoring loop in one test: press the button, type the note, ask to
9481 // go back. The caret ends at the note's last byte — which is the *end* of
9482 // the definition's span, the one offset the query used to exclude — so
9483 // this is where the round trip either works or doesn't.
9484 let mut d = doc_with("fn_insert_return", "A claim and more.\n");
9485 d.caret = 7;
9486 d.insert_footnote();
9487 d.insert("the note");
9488 assert_eq!(d.source, "A claim[^1] and more.\n\n[^1]: the note\n");
9489 let back = d
9490 .footnote_definition_at_caret()
9491 .expect("still in the note we just typed");
9492 assert_eq!(back.label, "1");
9493 // …and following it lands on the reference's label, where a reader's
9494 // return leg lands.
9495 assert_eq!(back.offset, Some(9));
9496 assert_eq!(&d.source[9..10], "1");
9497 }
9498
9499 #[test]
9500 fn insert_footnote_takes_one_undo_for_both_halves() {
9501 // twig writes the pair as a single edit; the point of that is here.
9502 let before = "A claim and more.\n";
9503 let mut d = doc_with("fn_insert_undo", before);
9504 d.caret = 7;
9505 d.insert_footnote();
9506 assert_ne!(d.source, before);
9507 d.undo();
9508 assert_eq!(d.source, before, "one undo takes back both halves");
9509 }
9510
9511 #[test]
9512 fn insert_footnote_refuses_a_format_that_cannot_spell_one() {
9513 // HTML is authorable — it spells the inline marks — and has no footnote.
9514 // The refusal says so rather than writing brackets that would render as
9515 // brackets.
9516 let src = "<p>A claim.</p>\n";
9517 let mut d = Doc::from_source(src.to_string(), Format::Html).unwrap();
9518 assert!(!Capabilities::of(Format::Html).footnote);
9519 d.caret = 5;
9520 d.insert_footnote();
9521 assert_eq!(d.source, src, "nothing written");
9522 assert!(d.status.is_some_and(|s| s.starts_with("footnote:")));
9523 }
9524
9525 #[test]
9526 fn insert_footnote_leaves_the_caret_on_a_real_stop_in_the_rich_view() {
9527 // The empty body is the one place this could go wrong: the definition
9528 // renders as a `[1] ` marker the caret cannot occupy, so a caret aimed a
9529 // byte early would draw up in the paragraph above the note it belongs to.
9530 let mut d = doc_in(View::Wysiwyg, "fn_insert_stop", "A claim and more.\n");
9531 d.place_caret(7, false);
9532 d.insert_footnote();
9533 d.build_visual(80); // the frame a frontend draws after the edit
9534 assert_eq!(
9535 d.vmap.snap_to_stop(d.caret),
9536 d.caret,
9537 "the caret sits on a stop"
9538 );
9539 let (row, _) = d.caret_pos();
9540 assert!(
9541 drawn_rows(&d)[row].contains("[1]"),
9542 "the caret is on the note's row, not above it: {:?}",
9543 drawn_rows(&d)
9544 );
9545 }
9546
9547 #[test]
9548 fn footnote_at_caret_resolves_a_reference_to_its_note() {
9549 // `[^1]` spans 7..11; its label byte is at 9. The definition follows a
9550 // blank line, as one has to.
9551 let mut d = doc_with("fn_at_caret", "A claim[^1] and more.\n\n[^1]: the note\n");
9552 d.caret = 9;
9553 let f = d
9554 .footnote_at_caret()
9555 .expect("the caret stands in a reference");
9556 assert_eq!(f.label, "1");
9557 assert_eq!(f.text.as_deref(), Some("the note"));
9558 // The offset points at the note's first word, not at the definition's
9559 // `[` — the marker is decoration with no caret stop on it.
9560 assert_eq!(f.offset, Some(29));
9561 assert_eq!(&d.source[29..37], "the note");
9562 // …and `end` closes the range, so a frontend can ask which rendered rows
9563 // the note occupies rather than re-deriving them from the text.
9564 assert_eq!(f.end, Some(37));
9565 assert_eq!(&d.source[f.offset.unwrap()..f.end.unwrap()], "the note");
9566 }
9567
9568 /// Two definitions in a row: each is its own note, and neither reaches into
9569 /// the other.
9570 ///
9571 /// A djot definition's span used to run past the blank line into the first
9572 /// byte of whatever followed, so this answered `"first note.\n\n["` — and the
9573 /// offsets named the *next* note's rows too, showing a reader two footnotes
9574 /// when they had asked about one. twig 3.1 ends the span after the block's
9575 /// own last line; the test outlives the workaround leaf carried for it.
9576 #[test]
9577 fn footnote_at_stops_a_note_at_the_definition_after_it() {
9578 let src = "Claim[^2a] and [^2b].\n\n[^2a]: first note.\n\n[^2b]: second note.\n";
9579 for format in [Format::Markdown, Format::Djot] {
9580 let mut d = Doc::from_source(src.to_string(), format).unwrap();
9581 d.caret = 7;
9582 let f = d.footnote_at_caret().expect("a reference");
9583 assert_eq!(f.text.as_deref(), Some("first note."), "in {format:?}");
9584 assert_eq!(
9585 &src[f.offset.unwrap()..f.end.unwrap()],
9586 "first note.",
9587 "in {format:?}"
9588 );
9589 }
9590 }
9591
9592 /// The other side of that boundary: a blank line *inside* a definition is
9593 /// interior to it, and the note keeps its second paragraph.
9594 ///
9595 /// This is what the old body scan cost. It stopped at the first line not
9596 /// indented under the note — a blank line is not — so a two-paragraph note
9597 /// came back as its first paragraph, and "go to note" framed half of it.
9598 /// Reading the span twig gives is both simpler and right.
9599 #[test]
9600 fn footnote_at_keeps_a_notes_second_paragraph() {
9601 let src = "Claim[^1].\n\n[^1]: first para.\n\n second para.\n\nAfter.\n";
9602 let mut d = Doc::from_source(src.to_string(), Format::Djot).unwrap();
9603 d.caret = 7;
9604 let f = d.footnote_at_caret().expect("a reference");
9605 assert_eq!(f.text.as_deref(), Some("first para.\n\n second para."));
9606 // And it stops there — `After.` is the next block, not more note.
9607 assert_eq!(
9608 &src[f.offset.unwrap()..f.end.unwrap()],
9609 f.text.as_deref().unwrap()
9610 );
9611 assert!(!f.text.as_deref().unwrap().contains("After"));
9612 }
9613
9614 #[test]
9615 fn footnote_at_bounds_a_note_whose_body_is_empty() {
9616 // `[^1]:` with nothing after it. The range is empty rather than
9617 // inverted, and still points inside the definition — which is what keeps
9618 // a frontend's row lookup from walking off into the block above.
9619 let src = "A claim[^1].\n\n[^1]:\n";
9620 let mut d = doc_with("fn_empty_body", src);
9621 d.caret = 9;
9622 let f = d.footnote_at_caret().expect("a reference");
9623 assert_eq!(f.text.as_deref(), Some(""));
9624 assert_eq!(f.offset, f.end, "an empty note is an empty range");
9625 assert!(f.offset.unwrap() >= src.find("[^1]:").unwrap());
9626 }
9627
9628 #[test]
9629 fn footnote_at_caret_ignores_a_caret_that_stands_in_no_reference() {
9630 let mut d = doc_with(
9631 "fn_at_caret_none",
9632 "A claim[^1] and more.\n\n[^1]: the note\n",
9633 );
9634 d.caret = 2; // in the prose
9635 assert_eq!(d.footnote_at_caret(), None);
9636 }
9637
9638 #[test]
9639 fn footnote_at_caret_is_not_a_link_query_and_vice_versa() {
9640 // The two are deliberately separate: a reference names a note in this
9641 // document, a link names somewhere to leave for, and answering one with
9642 // the other is what made a reference click do nothing at all.
9643 let mut d = doc_with("fn_vs_link", "a[^1] b [t](https://x.dev)\n\n[^1]: note\n");
9644 d.caret = 3; // the `1` of `[^1]`
9645 assert!(d.footnote_at_caret().is_some());
9646 assert_eq!(
9647 d.link_destination_at_caret(),
9648 None,
9649 "a reference is not a link"
9650 );
9651
9652 d.caret = 10; // inside the link's label
9653 assert_eq!(d.footnote_at_caret(), None, "a link is not a reference");
9654 assert_eq!(
9655 d.link_destination_at_caret().as_deref(),
9656 Some("https://x.dev")
9657 );
9658 }
9659
9660 #[test]
9661 fn footnote_at_caret_reports_an_undefined_reference_rather_than_nothing() {
9662 // A `[^99]` the document never defines is a real state — a note deleted
9663 // out from under its reference — and the label is what lets a frontend
9664 // say so. `None` here would be indistinguishable from "not on a
9665 // reference", which is the wrong thing to tell a reader.
9666 let mut d = doc_with("fn_undefined", "A claim[^99] and more.\n");
9667 d.caret = 9;
9668 let f = d
9669 .footnote_at_caret()
9670 .expect("the reference is still a reference");
9671 assert_eq!(f.label, "99");
9672 assert_eq!(f.text, None);
9673 assert_eq!(f.offset, None);
9674 }
9675
9676 #[test]
9677 fn footnote_at_caret_reads_a_word_label_and_a_multiline_note() {
9678 // Labels are not always numbers, and a note's body runs past its first
9679 // line — the indented continuation belongs to the note, so it comes back
9680 // with it (source bytes, verbatim, as documented).
9681 let src = "see[^note] here\n\n[^note]: first line\n second line\n";
9682 let mut d = doc_with("fn_word_label", src);
9683 d.caret = 6;
9684 let f = d
9685 .footnote_at_caret()
9686 .expect("the caret stands in a reference");
9687 assert_eq!(f.label, "note");
9688 assert_eq!(f.text.as_deref(), Some("first line\n second line"));
9689 }
9690
9691 #[test]
9692 fn footnote_at_answers_for_an_offset_the_caret_is_nowhere_near() {
9693 // The point of the offset form: a pointer hovering a reference asks what
9694 // note it names, and must not drag the caret along to ask.
9695 let mut d = doc_with("fn_at_off", "A claim[^1] and more.\n\n[^1]: the note\n");
9696 d.caret = 0;
9697 let f = d.footnote_at(9).expect("offset 9 stands in the reference");
9698 assert_eq!(f.label, "1");
9699 assert_eq!(f.text.as_deref(), Some("the note"));
9700 assert_eq!(d.caret, 0, "asking must not move the caret");
9701 assert_eq!(d.footnote_at(2), None, "offset 2 is prose");
9702 }
9703
9704 #[test]
9705 fn footnote_definition_at_caret_points_back_at_the_reference() {
9706 // The return leg. `[^1]` spans 7..11, so its label — the only byte of it
9707 // the caret can rest on — is at 9.
9708 let mut d = doc_with("fn_def", "A claim[^1] and more.\n\n[^1]: the note\n");
9709 d.caret = 30; // inside the note's body
9710 let f = d
9711 .footnote_definition_at_caret()
9712 .expect("the caret stands in a definition");
9713 assert_eq!(f.label, "1");
9714 assert_eq!(f.offset, Some(9));
9715 assert_eq!(&d.source[7..11], "[^1]");
9716 }
9717
9718 #[test]
9719 fn footnote_definition_at_covers_where_a_go_to_note_actually_lands() {
9720 // The two legs have to meet: wherever `footnote_at` sends the caret, the
9721 // definition query must answer for — otherwise arriving at a note leaves
9722 // the reader somewhere the way back isn't offered.
9723 let src = "A claim[^1] and more.\n\n[^1]: the note\n";
9724 let mut d = doc_with("fn_def_marker", src);
9725 let landed = d.footnote_at(9).unwrap().offset.unwrap();
9726 assert_eq!(
9727 d.footnote_definition_at(landed).and_then(|f| f.offset),
9728 Some(9),
9729 "the note a reference sends you to offers the way back"
9730 );
9731 }
9732
9733 #[test]
9734 fn footnote_definition_at_caret_ignores_prose_and_the_reference_itself() {
9735 // The two queries answer for disjoint places, which is what lets one
9736 // gesture mean "down to the note" in one and "back up" in the other
9737 // without either having to remember which way the reader is going.
9738 let mut d = doc_with("fn_def_none", "A claim[^1] and more.\n\n[^1]: the note\n");
9739 d.caret = 2; // prose
9740 assert_eq!(d.footnote_definition_at_caret(), None);
9741 d.caret = 9; // the reference
9742 assert_eq!(d.footnote_definition_at_caret(), None);
9743 assert!(
9744 d.footnote_at_caret().is_some(),
9745 "which is the reference's own query"
9746 );
9747 }
9748
9749 #[test]
9750 fn footnote_definition_at_caret_reports_an_orphan_note_rather_than_nothing() {
9751 // Nothing cites `[^2]`. Answering `None` would say "you are not in a
9752 // note", which is false and leaves a frontend unable to explain why the
9753 // way back is missing.
9754 let src = "A claim[^1].\n\n[^1]: cited\n\n[^2]: orphan\n";
9755 let mut d = doc_with("fn_def_orphan", src);
9756 d.caret = src.find("orphan").unwrap();
9757 let f = d
9758 .footnote_definition_at_caret()
9759 .expect("an orphan is still a definition");
9760 assert_eq!(f.label, "2");
9761 assert_eq!(f.offset, None);
9762 }
9763
9764 #[test]
9765 fn footnote_definition_at_caret_returns_to_the_first_of_repeated_references() {
9766 // One label, cited twice. The first is where the reader most likely came
9767 // from, and the only answer that doesn't depend on how they got here.
9768 let src = "One[^a] and two[^a].\n\n[^a]: the note\n";
9769 let mut d = doc_with("fn_def_repeat", src);
9770 d.caret = src.find("the note").unwrap();
9771 let f = d.footnote_definition_at_caret().expect("a definition");
9772 assert_eq!(
9773 f.offset,
9774 Some(5),
9775 "the first `[^a]`'s label, not the second's"
9776 );
9777 assert_eq!(&src[3..7], "[^a]");
9778 }
9779
9780 #[test]
9781 fn footnote_navigation_is_a_round_trip_through_placed_carets() {
9782 // Down and back up, each leg found from the document rather than from a
9783 // memory of the other — so it still works for a reader who scrolled to
9784 // the notes instead of jumping there.
9785 //
9786 // `place_caret` rather than assigning `caret`, because that is what a
9787 // frontend calls: it snaps to a real caret stop, and a jump that lands
9788 // on a byte the caret can't rest on would arrive somewhere the return
9789 // leg no longer answers for. `build_map` first, since snapping is a
9790 // no-op until the map exists — which is exactly how this went unnoticed
9791 // when the offsets pointed at the `[^` markers.
9792 let mut d = doc_with("fn_round", "A claim[^1] and more.\n\n[^1]: the note\n");
9793 d.build_map(None);
9794 d.place_caret(9, false);
9795 let down = d
9796 .footnote_at_caret()
9797 .expect("a reference")
9798 .offset
9799 .expect("a note");
9800 d.place_caret(down, false);
9801 let up = d
9802 .footnote_definition_at_caret()
9803 .expect("a definition")
9804 .offset
9805 .expect("a reference");
9806 d.place_caret(up, false);
9807 assert_eq!(d.caret, up, "the way back is a stop the caret can occupy");
9808 assert_eq!(
9809 d.footnote_at_caret().expect("back on the reference").label,
9810 "1"
9811 );
9812 }
9813
9814 #[test]
9815 fn insert_link_hands_the_destination_to_twig_raw() {
9816 // Escaping is twig's, and format-specific: Markdown ends a destination
9817 // at the first space and needs the `<…>` form, where djot would read
9818 // those angle brackets as part of the URL.
9819 let mut d = doc_with("link_space", "word\n");
9820 d.anchor = Some(0);
9821 d.caret = 4;
9822 d.insert_link("a b");
9823 assert_eq!(d.source, "[word](<a b>)\n");
9824 }
9825
9826 #[test]
9827 fn insert_link_reports_a_destination_no_format_can_carry() {
9828 let mut d = doc_with("link_bad", "word\n");
9829 d.anchor = Some(0);
9830 d.caret = 4;
9831 d.insert_link("a\nb");
9832 assert_eq!(d.source, "word\n"); // untouched, not quietly rewritten
9833 assert!(
9834 d.status.is_some(),
9835 "InvalidArgument should reach the status line"
9836 );
9837 assert!(!d.dirty);
9838 }
9839
9840 #[test]
9841 fn insert_link_works_in_wysiwyg_view() {
9842 let mut d = wysiwyg_doc("link_wys", "word here\n");
9843 d.anchor = Some(0);
9844 d.caret = 4;
9845 d.insert_link("http://x.dev");
9846 assert_eq!(d.source, "[word](http://x.dev) here\n");
9847 assert_eq!(d.selected_text(), Some("word"));
9848 // The map the caret has to keep riding is rebuilt each frame; motion
9849 // over the fresh one must still land on a real stop (the debug_assert).
9850 d.build_visual(80);
9851 d.move_right(false);
9852 d.move_left(false);
9853 }
9854
9855 #[test]
9856 fn click_maps_a_row_col_to_a_byte_offset() {
9857 let mut d = doc_with("click", "ab\ncd\n");
9858 d.click(1, 1, false); // row 1 ("cd"), col 1 -> the 'd'
9859 assert_eq!(d.caret, 4);
9860 }
9861
9862 // A pixel-hit-test placement (the GUI's `place_caret`) must land on a caret
9863 // stop just as the `(row, col)` click path does, so the caret can never come
9864 // to rest in the blank gap between two paragraphs — where it would draw in one
9865 // place and type in another.
9866 #[test]
9867 fn place_caret_snaps_out_of_the_blank_gap_between_paragraphs() {
9868 // "A\n\nB": offset 2 is the gap the paragraph break is drawn with, not a
9869 // caret stop (stops are 0,1,3,4).
9870 let mut d = wysiwyg_doc("place_gap", "A\n\nB");
9871 assert!(!d.vmap.is_stop(2), "offset 2 should be an unreachable gap");
9872 d.place_caret(2, false);
9873 assert!(d.vmap.is_stop(d.caret), "caret {} is not a stop", d.caret);
9874 assert_eq!(d.caret, 1, "should snap to the end of the paragraph above");
9875 }
9876
9877 #[test]
9878 fn place_caret_dragging_through_the_gap_keeps_selection_on_stops() {
9879 let mut d = wysiwyg_doc("place_gap_drag", "A\n\nB");
9880 d.place_caret(0, false); // anchor at the start of "A"
9881 d.place_caret(2, true); // drag into the gap
9882 assert!(d.vmap.is_stop(d.caret), "caret {} is not a stop", d.caret);
9883 let (s, e) = d.selection().expect("a selection");
9884 assert!(
9885 d.vmap.is_stop(s) && d.vmap.is_stop(e),
9886 "selection {s}..{e} off a stop"
9887 );
9888 }
9889
9890 #[test]
9891 fn place_caret_on_a_real_stop_is_left_untouched() {
9892 let mut d = wysiwyg_doc("place_stop", "A\n\nB");
9893 d.place_caret(3, false); // the start of "B" — a genuine stop
9894 assert_eq!(d.caret, 3);
9895 }
9896
9897 // An *empty paragraph* (two blank lines, an intentional blank line the user
9898 // opened) is a real caret stop, unlike the gap — a click into it must stay.
9899 #[test]
9900 fn place_caret_rests_in_an_empty_paragraph() {
9901 let mut d = wysiwyg_doc("place_empty_para", "A\n\n\n\nB");
9902 let empty = 3; // the navigable empty row's offset (stops: 0,1,3,5,6)
9903 assert!(d.vmap.is_stop(empty));
9904 d.place_caret(empty, false);
9905 assert_eq!(d.caret, empty);
9906 }
9907
9908 // The content end of a hidden mark is a home too (`VisualMap::mark_ends`):
9909 // a drag over the word `bold` ends there, and a caret placed there stays.
9910 #[test]
9911 fn place_caret_rests_at_the_end_of_a_hidden_marks_content() {
9912 let src = "| A | B |\n| --- | --- |\n| **bold** | other |\n";
9913 let mut d = wysiwyg_doc("place_mark_end", src);
9914 let start = src.find("bold").unwrap();
9915 d.place_caret(start, false);
9916 d.place_caret(start + 4, true);
9917 assert_eq!(d.selection(), Some((start, start + 4)), "the whole word");
9918 d.toggle(InlineKind::Strong);
9919 assert_eq!(d.source, src.replace("**bold**", "bold"));
9920 }
9921
9922 #[test]
9923 fn right_steps_onto_the_end_of_a_mark_and_then_past_its_delimiter() {
9924 let mut d = wysiwyg_doc("right_mark_end", "a **bold** b");
9925 d.caret = 7; // before the `d`
9926 d.move_right(false);
9927 assert_eq!(d.caret, 8, "onto the end of the bold");
9928 assert!(d.active_inline_marks().contains(InlineKind::Strong));
9929 d.move_right(false);
9930 assert_eq!(d.caret, 10, "past the closing `**`");
9931 assert!(!d.active_inline_marks().contains(InlineKind::Strong));
9932 d.move_left(false);
9933 assert_eq!(d.caret, 8);
9934 d.move_left(false);
9935 assert_eq!(d.caret, 7);
9936 // Typing at the inner home extends the bold.
9937 d.caret = 8;
9938 d.insert("!");
9939 assert_eq!(d.source, "a **bold!** b");
9940 }
9941
9942 #[test]
9943 fn a_marks_end_home_follows_an_edit_through_the_incremental_map() {
9944 // The splice path shifts the home with the block it is in, and the
9945 // re-rendered block finds its own again.
9946 let mut d = wysiwyg_doc("mark_end_splice", "x\n\na **bold** b\n\ny\n");
9947 d.build_visual_unwrapped();
9948 d.edit(0, 0, "zz");
9949 d.build_visual_unwrapped();
9950 wysiwyg::assert_maps_eq(&d.vmap, &reference_map(&d.source), "after a shift");
9951 assert!(d.vmap.is_stop(d.source.find("bold").unwrap() + 4));
9952 let at = d.source.find("bold").unwrap();
9953 d.edit(at, at, "very ");
9954 d.build_visual_unwrapped();
9955 wysiwyg::assert_maps_eq(&d.vmap, &reference_map(&d.source), "after a re-render");
9956 assert!(d.vmap.is_stop(d.source.find("bold").unwrap() + 4));
9957 }
9958
9959 fn wysiwyg_doc(name: &str, body: &str) -> Doc {
9960 doc_in(View::Wysiwyg, name, body)
9961 }
9962
9963 /// How many list items the source actually parses into — the check that a
9964 /// marker Leaf wrote is a marker the format agrees is one.
9965 fn list_items(doc: &mut Doc) -> usize {
9966 doc.editor
9967 .nodes()
9968 .unwrap()
9969 .iter()
9970 .filter(|n| n.kind == Kind::ListItem || n.kind == Kind::TaskListItem)
9971 .count()
9972 }
9973
9974 /// A from-scratch, cache-free WYSIWYG map for `source` — the ground truth the
9975 /// incremental (`build_spliced` / `build_cached`) path must always match.
9976 fn reference_map(source: &str) -> crate::wysiwyg::VisualMap {
9977 reference_map_revealing(source, None)
9978 }
9979
9980 /// [`reference_map`] with a reveal line — the ground truth for the
9981 /// `MarkupMode::Full` builds, where the map is a function of the caret's
9982 /// line as well as the text.
9983 fn reference_map_revealing(
9984 source: &str,
9985 reveal: Option<Range<usize>>,
9986 ) -> crate::wysiwyg::VisualMap {
9987 // The same parse `Doc` uses. With twig's plain defaults instead, the two
9988 // sides disagree on what the *document* is before the renderer is even
9989 // reached — a bare `:word` is a text directive to one and prose to the
9990 // other — and the mismatch reads as a splice bug that isn't one.
9991 let mut ed =
9992 twig::Editor::new_ext(source.as_bytes(), Format::Markdown, parse_extensions()).unwrap();
9993 let nodes = ed.nodes().unwrap();
9994 crate::wysiwyg::build(
9995 &nodes,
9996 source,
9997 None,
9998 false,
9999 &std::collections::HashMap::new(),
10000 reveal,
10001 )
10002 }
10003
10004 fn maps_differ(a: &crate::wysiwyg::VisualMap, b: &crate::wysiwyg::VisualMap) -> bool {
10005 if a.rows.len() != b.rows.len() {
10006 return true;
10007 }
10008 for (ra, rb) in a.rows.iter().zip(&b.rows) {
10009 if ra.end_src != rb.end_src || ra.glyphs.len() != rb.glyphs.len() {
10010 return true;
10011 }
10012 for (ga, gb) in ra.glyphs.iter().zip(&rb.glyphs) {
10013 if ga.ch != gb.ch || ga.src != gb.src {
10014 return true;
10015 }
10016 }
10017 }
10018 false
10019 }
10020
10021 #[test]
10022 fn incremental_build_matches_a_fresh_build_across_edits() {
10023 // Every `Doc` edit rebuilds through `build_spliced` (the single-block
10024 // fast path, gated on twig's `dirty_range`) or falls back to
10025 // `build_cached`. After each edit the map must be byte-identical to a
10026 // from-scratch build — this is the correctness net under the splice.
10027 let docs = [
10028 "# Title\n\nThe quick brown fox jumps.\n\nAnother paragraph here.\n\n- a\n- b\n",
10029 "para one\n\n> quote **bold** text\n> continued line\n\ntail paragraph\n",
10030 "alpha\n\nbeta\n\ngamma\n\ndelta\n\nepsilon\n\nzeta\n",
10031 // A footnote definition is a root beside `doc`, merged back into the
10032 // top-level list by `wysiwyg::top_blocks`. The random edits below
10033 // make and unmake definitions as they go (a deleted `:` turns one
10034 // back into a paragraph, and vice versa), which is exactly the
10035 // structural churn the splice path has to notice and bail out of.
10036 "text[^1] here\n\n[^1]: the note\n\nmore text[^b]\n\n[^b]: second\n",
10037 // A comment is a top-level block that draws no rows — a layout entry
10038 // at zero rows either side of blocks that do. The edits below type
10039 // into the blocks around it (a splice past a hidden block), and
10040 // break the comment open into prose and back (a structural change).
10041 "intro\n\n<!-- exec -->\n```\ncode\n```\n\nafter the comment\n\n<!-- trail -->\n",
10042 // Link reference definitions: a hidden block that an edit can turn
10043 // into a paragraph (a deleted `:`) and back, and whose own bytes an
10044 // edit can land in.
10045 "see [a] and [b]\n\n[a]: /a\n\nmid text\n\n[b]: /b\n",
10046 ];
10047 // A deterministic mix: mostly single characters (which stay inside one
10048 // block → splice), plus edits that reshape structure (a paragraph break,
10049 // a heading marker, a code fence → fallback), so both paths are exercised.
10050 let inserts = ["x", "y", "\n\n", "#", "`", " ", "z"];
10051 for src in docs {
10052 let mut d = wysiwyg_doc("diff", src);
10053 d.build_visual_unwrapped();
10054 wysiwyg::assert_maps_eq(&d.vmap, &reference_map(&d.source), "initial");
10055
10056 for step in 0..60usize {
10057 let len = d.source.len();
10058 let raw = (step * 13 + 5) % (len + 1);
10059 let pos = (raw..=len).find(|&i| d.source.is_char_boundary(i)).unwrap();
10060 let pre = d.source.clone();
10061 let action;
10062 if step % 3 == 0 && pos < len {
10063 let end = (pos + 1..=len)
10064 .find(|&i| d.source.is_char_boundary(i))
10065 .unwrap();
10066 action = format!("delete [{pos},{end})");
10067 d.edit(pos, end, "");
10068 } else {
10069 let ins = inserts[step % inserts.len()];
10070 action = format!("insert {ins:?} @ {pos}");
10071 d.edit(pos, pos, ins);
10072 }
10073 d.build_visual_unwrapped();
10074 if maps_differ(&d.vmap, &reference_map(&d.source)) {
10075 panic!(
10076 "FIRST MISMATCH at step {step}: {action}\n pre = {pre:?}\n post = {:?}",
10077 d.source
10078 );
10079 }
10080 }
10081 }
10082 }
10083
10084 /// A frontend is handed [`Doc::vmap`] and may present it differently:
10085 /// leaf-ratatui splices blank filler rows under an oversized heading so the
10086 /// raster it paints there has somewhere to stand, and leaves them in the map
10087 /// because the caret and the mouse both read it between frames. The splice
10088 /// path addresses that map by *row index*, against the block layout the last
10089 /// build recorded — so handed a map with rows in it that no block owns, it
10090 /// laid the re-rendered block over one of the fillers and carried the rows
10091 /// the block really occupied into the suffix. One stranded copy of the
10092 /// edited line, and everything below it a row further down, per keystroke.
10093 ///
10094 /// A map that isn't the one the layout describes is a map this path can't
10095 /// patch, whoever changed it and for whatever reason. It rebuilds instead.
10096 #[test]
10097 fn an_edit_over_a_map_a_frontend_reshaped_rebuilds_it_whole() {
10098 let mut d = wysiwyg_doc("reshaped", "# Title\n\nThe quick brown fox jumps.\n");
10099 d.build_visual_unwrapped();
10100
10101 // Stand in for the heading filler rows: two blank rows past the heading
10102 // that no block accounts for. Cloning a real row keeps every field
10103 // plausible — it is the row *count* the splice can't survive.
10104 let filler = d.vmap.rows[0].clone();
10105 d.vmap.rows.insert(1, filler.clone());
10106 d.vmap.rows.insert(1, filler);
10107
10108 // An edit inside the last block: the single-block case the splice path
10109 // is for, and the one the frontend hits on every keystroke.
10110 let at = d.source.len() - 1;
10111 d.edit(at, at, "!");
10112 d.build_visual_unwrapped();
10113
10114 wysiwyg::assert_maps_eq(&d.vmap, &reference_map(&d.source), "after the edit");
10115 }
10116
10117 #[test]
10118 fn incremental_build_matches_a_fresh_build_under_full_reveal() {
10119 // The same correctness net as `incremental_build_matches_a_fresh_build_
10120 // across_edits`, under `MarkupMode::Full` — where the map depends on
10121 // the caret's *line* as well as the text, so the two caches have a new
10122 // way to be wrong. Both are exercised: the block cache can hand back
10123 // rows built for a line that is no longer the revealed one, and the
10124 // splice path can reuse a suffix that still has yesterday's line raw.
10125 //
10126 // Caret motion is interleaved with the edits deliberately, because a
10127 // caret that only ever moved with the edit would never cross a line
10128 // without also dirtying it — the case where a stale reveal survives.
10129 let docs = [
10130 "# Title\n\n*one* and **two**\n\n[lk](http://x) and `code`\n\n- a *b*\n",
10131 "para *em* one\n\n> quote **bold** text\n\ntail ~~del~~ paragraph\n",
10132 ];
10133 let inserts = ["x", "*", "\n\n", "#", "`", " ", "_"];
10134 for src in docs {
10135 let mut d = wysiwyg_doc("reveal_diff", src);
10136 d.set_markup_mode(MarkupMode::Full);
10137
10138 for step in 0..60usize {
10139 let len = d.source.len();
10140 let raw = (step * 13 + 5) % (len + 1);
10141 let pos = (raw..=len).find(|&i| d.source.is_char_boundary(i)).unwrap();
10142 let pre = d.source.clone();
10143 let action;
10144 if step % 3 == 0 && pos < len {
10145 let end = (pos + 1..=len)
10146 .find(|&i| d.source.is_char_boundary(i))
10147 .unwrap();
10148 action = format!("delete [{pos},{end})");
10149 d.edit(pos, end, "");
10150 } else {
10151 let ins = inserts[step % inserts.len()];
10152 action = format!("insert {ins:?} @ {pos}");
10153 d.edit(pos, pos, ins);
10154 }
10155 // Walk the caret somewhere else in the document, independently
10156 // of where the edit landed.
10157 let want = (step * 29 + 11) % (d.source.len() + 1);
10158 d.caret = (want..=d.source.len())
10159 .find(|&i| d.source.is_char_boundary(i))
10160 .unwrap();
10161 d.build_visual_unwrapped();
10162
10163 let want = reference_map_revealing(&d.source, d.reveal_line());
10164 if maps_differ(&d.vmap, &want) {
10165 panic!(
10166 "FIRST MISMATCH at step {step}: {action}, caret {}\n pre = {pre:?}\n post = {:?}",
10167 d.caret, d.source
10168 );
10169 }
10170 }
10171 }
10172 }
10173
10174 #[test]
10175 fn caret_motion_across_lines_rebuilds_only_under_full() {
10176 // The cache-key change has to earn its keep in both directions: `Full`
10177 // must rebuild when the caret changes line (or the reveal would never
10178 // move), and the hidden modes must *not* (or every arrow key would pay
10179 // for a feature they don't use). The existing `cache_motion` test pins
10180 // the second for the default mode; this pins the pair against a mode
10181 // change alone.
10182 let body = "*one* here\n\n*two* there\n";
10183
10184 let mut full = doc_in(View::Wysiwyg, "motion_full", body);
10185 full.set_markup_mode(MarkupMode::Full);
10186 caret_at(&mut full, "one");
10187 let before = full.revision();
10188 caret_at(&mut full, "two");
10189 assert_eq!(full.revision(), before, "motion is not an edit");
10190 assert!(
10191 drawn_rows(&full).iter().any(|r| r == "*two* there"),
10192 "the map followed the caret: {:?}",
10193 drawn_rows(&full)
10194 );
10195
10196 let mut hidden = doc_in(View::Wysiwyg, "motion_hidden", body);
10197 caret_at(&mut hidden, "one");
10198 let key = hidden.vmap_key.clone();
10199 caret_at(&mut hidden, "two");
10200 assert_eq!(
10201 hidden.vmap_key, key,
10202 "a hidden mode rebuilds nothing on motion"
10203 );
10204 }
10205
10206 #[test]
10207 fn wysiwyg_down_crosses_a_paragraph_boundary() {
10208 // Regression: the blank separator row used to share the previous
10209 // paragraph's end offset, so Down got pinned at the boundary (while Up
10210 // still crossed). Both directions must step through it symmetrically.
10211 //
10212 // It's now stepped *over* rather than onto: the blank line between two
10213 // paragraphs is the boundary being drawn, not a line of the document, so
10214 // one press of Down crosses it. The goal column survives the crossing —
10215 // col 3 at the end of "abc" is col 3 at the end of "def".
10216 let mut d = wysiwyg_doc("wys_down", "abc\n\ndef\n");
10217 d.caret = 3; // end of "abc" (row 0)
10218 d.move_down(false);
10219 assert_eq!(d.caret_pos().0, 2, "Down should reach the second paragraph");
10220 assert_eq!(d.caret, 8); // end of "def", col 3 kept
10221 d.move_up(false);
10222 assert_eq!(d.caret_pos().0, 0, "Up should come back symmetrically");
10223 assert_eq!(d.caret, 3);
10224 }
10225
10226 #[test]
10227 fn wysiwyg_up_and_down_are_inverse_across_paragraphs() {
10228 // The second Up and the second Down here run off the ends of the
10229 // document, which is no longer a place a press is swallowed: they carry
10230 // the caret to the start and the end of the text. The claim in the
10231 // middle — that a Down retraces the Up that crossed the paragraph gap —
10232 // is the one this test is for, and it is asserted where it is made.
10233 let mut d = wysiwyg_doc("wys_updown", "abc\n\ndef\n");
10234 d.caret = 5; // start of "def"
10235 let start = d.caret_pos();
10236 d.move_up(false);
10237 assert_eq!(d.caret_pos().0, 0, "Up reaches the first paragraph");
10238 d.move_up(false);
10239 assert_eq!(d.caret, 0, "a second Up runs on to the document's start");
10240 d.move_down(false);
10241 assert_eq!(d.caret_pos(), start, "Down retraces Up exactly");
10242 d.move_down(false);
10243 assert_eq!(d.caret, 8, "a second Down runs on to the document's end");
10244 }
10245
10246 #[test]
10247 fn wysiwyg_new_paragraph_shows_before_typing() {
10248 // Regression: two Enters at the end of a paragraph produced trailing
10249 // newlines with no AST node, so the caret appeared stuck on the old line
10250 // until a character was typed. It must ride down onto the new line now.
10251 let mut d = doc_with("wys_newpara", "abc\n");
10252 d.view = View::Wysiwyg;
10253 d.caret = 3;
10254 d.insert("\n");
10255 d.insert("\n"); // source is now "abc\n\n\n", caret at 5
10256 assert_eq!(d.source, "abc\n\n\n");
10257 d.build_visual(80);
10258 let (row, _) = d.caret_pos();
10259 assert!(
10260 row >= 2,
10261 "caret should have moved down to the new line, got row {row}"
10262 );
10263 assert!(
10264 d.vmap.num_rows() >= 3,
10265 "the blank lines should render as rows"
10266 );
10267 }
10268
10269 #[test]
10270 fn wysiwyg_enter_between_paragraphs_lands_on_an_empty_line() {
10271 // The reported bug: Enter at the end of a paragraph that has another
10272 // paragraph below put the caret at the *start of the next paragraph* —
10273 // the empty paragraph it opened had no row, so the caret snapped onto
10274 // "World". It must now sit on its own empty line, with a blank spacer
10275 // above it (the paragraph gap).
10276 let mut d = wysiwyg_doc("wys_gap_mid", "Hello\n\nWorld\n");
10277 d.caret = 5; // end of "Hello"
10278 d.newline();
10279 d.build_visual(80);
10280 let (row, col) = d.caret_pos();
10281 assert_eq!(col, 0, "caret should start an empty line, not sit in text");
10282 assert_eq!(
10283 d.vmap.row_width(row),
10284 0,
10285 "caret's row must be empty, not 'World'"
10286 );
10287 assert!(
10288 row >= 2,
10289 "a blank spacer row should sit above the caret, got row {row}"
10290 );
10291 // The row above the caret is a real (empty) gap, and "Hello" stays put.
10292 assert_eq!(
10293 d.vmap.row_width(row - 1),
10294 0,
10295 "the row above the caret is a gap"
10296 );
10297 let row0: String = d.vmap.rows[0].glyphs.iter().map(|g| g.ch).collect();
10298 assert_eq!(row0, "Hello", "the paragraph above the caret must not move");
10299 }
10300
10301 #[test]
10302 fn wysiwyg_enter_at_eof_shows_a_gap_before_typing() {
10303 // At the document end a single Enter must also show the paragraph gap —
10304 // a blank spacer row above the caret — so the layout already matches how
10305 // it will look once the new paragraph has text.
10306 let mut d = wysiwyg_doc("wys_gap_eof", "Hello");
10307 d.caret = 5; // end of "Hello", no trailing newline
10308 d.newline(); // source becomes "Hello\n\n"
10309 d.build_visual(80);
10310 let (row, col) = d.caret_pos();
10311 assert_eq!(col, 0);
10312 assert!(
10313 row >= 2,
10314 "caret should sit below a blank spacer, got row {row}"
10315 );
10316 assert_eq!(
10317 d.vmap.row_width(row - 1),
10318 0,
10319 "the row above the caret is a gap"
10320 );
10321 }
10322
10323 #[test]
10324 fn wysiwyg_typing_after_enter_does_not_shift_the_caret_row() {
10325 // The spacer is view-only: typing the new paragraph must not reflow the
10326 // caret onto a different row — the transient view already matched the
10327 // settled one.
10328 let mut d = wysiwyg_doc("wys_no_reflow", "Hello\n\nWorld\n");
10329 d.caret = 5;
10330 d.newline();
10331 d.build_visual(80);
10332 let before = d.caret_pos();
10333 d.insert("New");
10334 d.build_visual(80);
10335 let after = d.caret_pos();
10336 assert_eq!(
10337 after.0, before.0,
10338 "typing must not move the caret to another row ({before:?} -> {after:?})"
10339 );
10340 }
10341
10342 #[test]
10343 fn wysiwyg_return_on_the_last_code_line_keeps_the_caret_in_the_block() {
10344 // Return at the end of the block's last line writes an empty line the
10345 // map used to drop, so the caret landed on `after` and the next
10346 // keystroke went into the paragraph below instead of into the code.
10347 let mut d = wysiwyg_doc("code_return", "prose\n\n```\nalpha\nbeta\n```\n\nafter\n");
10348 d.caret = d.source.find("beta").unwrap() + "beta".len();
10349 d.build_visual(80);
10350 let before = d.caret_pos().0;
10351
10352 d.newline();
10353 d.build_visual(80);
10354 assert_eq!(d.source, "prose\n\n```\nalpha\nbeta\n\n```\n\nafter\n");
10355
10356 let (row, col) = d.caret_pos();
10357 assert_eq!(row, before + 1, "the caret moves down one row");
10358 assert_eq!(col, 0, "onto the head of the empty line");
10359 let span = d.vmap.code_blocks[0].rows_span.clone();
10360 assert!(
10361 span.contains(&row),
10362 "caret row {row} is outside the block's rows {span:?}"
10363 );
10364
10365 // The whole point: what is typed next is code.
10366 d.insert("gamma");
10367 assert_eq!(d.source, "prose\n\n```\nalpha\nbeta\ngamma\n```\n\nafter\n");
10368 }
10369
10370 #[test]
10371 fn wysiwyg_hides_frontmatter_from_the_caret_and_copy() {
10372 let fm = "---\ntitle: hi\n---\n";
10373 let body = format!("{fm}# leaf\n\nbody\n");
10374 let mut d = wysiwyg_doc("wys_fm", &body);
10375 // Opening lifts the caret out of the now-hidden frontmatter.
10376 assert_eq!(
10377 d.caret,
10378 fm.len(),
10379 "caret should start at the first real block"
10380 );
10381 // Left at the content start can't step back into frontmatter.
10382 d.move_left(false);
10383 assert_eq!(d.caret, fm.len(), "left must not enter frontmatter");
10384 // Doc-start lands on the content floor, not offset 0.
10385 d.move_doc_start(false);
10386 assert_eq!(d.caret, fm.len());
10387 // Select-all + copy never include the frontmatter bytes.
10388 d.select_all();
10389 let sel = d.selected_text().unwrap().to_string();
10390 assert!(!sel.contains("title"), "copy leaked frontmatter: {sel:?}");
10391 assert!(
10392 sel.starts_with("# leaf"),
10393 "selection should begin at content: {sel:?}"
10394 );
10395 }
10396
10397 #[test]
10398 fn typing_in_a_frontmatter_only_document_lands_after_the_frontmatter() {
10399 // A fresh note is frontmatter and nothing else. With no rendered block
10400 // to floor the caret it opened at offset 0 — before the opening `---` —
10401 // so the first keystroke wrote itself in front of the metadata and the
10402 // file came out as `This---\ntitle: …`.
10403 let fm = "---\ntitle: 2026-08-29\nid: f8s32cd\n---\n";
10404 let mut d = wysiwyg_doc("wys_fm_only", fm);
10405 assert_eq!(d.caret, fm.len(), "caret must open past the frontmatter");
10406 // Nothing is rendered, so the caret draws at the origin of an empty view
10407 // — the same place an empty document puts it.
10408 assert_eq!(d.caret_pos(), (0, 0));
10409 d.insert("This");
10410 assert_eq!(d.source, format!("{fm}This"));
10411 }
10412
10413 /// `select_range` is the verb for a range a host already knows the bytes of,
10414 /// so it must not snap — and must still hold every invariant `place_caret`
10415 /// holds, the frontmatter floor above all.
10416 #[test]
10417 fn select_range_takes_the_range_as_given_but_still_floors_it() {
10418 let fm = "---\ntitle: foo\n---\n\n";
10419 let body = format!("{fm}body foo here\n");
10420 let mut d = wysiwyg_doc("wys_select_range", &body);
10421
10422 // The `foo` in the body: taken exactly, not snapped to a caret stop.
10423 let at = body.rfind("foo").unwrap();
10424 d.select_range(at, at + 3);
10425 assert_eq!(d.selection(), Some((at, at + 3)));
10426 assert_eq!(d.selected_text(), Some("foo"));
10427
10428 // The `foo` in the hidden frontmatter: below the floor, so both ends
10429 // come up to it rather than parking the caret in the metadata, where a
10430 // later keystroke would rewrite `title:`.
10431 let hidden = body.find("foo").unwrap();
10432 assert!(hidden < d.vmap.content_start);
10433 d.select_range(hidden, hidden + 3);
10434 assert!(
10435 d.caret >= d.vmap.content_start && d.anchor.unwrap() >= d.vmap.content_start,
10436 "a range under the floor must not leave the caret in the frontmatter"
10437 );
10438
10439 // Past the end, and mid-character, are both brought back to something
10440 // sliceable rather than panicking the next reader of the range.
10441 let multi = wysiwyg_doc("wys_select_range_utf8", "héllo\n");
10442 let mut d = multi;
10443 d.select_range(2, 9_999);
10444 assert_eq!(d.caret, d.source.len());
10445 assert!(d.source.is_char_boundary(d.anchor.unwrap()));
10446 assert!(d.source.is_char_boundary(d.caret));
10447 }
10448
10449 /// The bug `select_range` exists for: a match butting up against a hidden
10450 /// delimiter. `place_caret` snaps to the nearest *visible* stop, which is
10451 /// the one before the `**`.
10452 #[test]
10453 fn select_range_does_not_snap_off_a_hidden_delimiter() {
10454 let mut d = wysiwyg_doc("wys_select_range_bold", "a **needle** in it\n");
10455 let at = d.source.find("needle").unwrap();
10456 d.select_range(at, at + 6);
10457 assert_eq!(d.selected_text(), Some("needle"), "not \"needl\"");
10458 }
10459
10460 #[test]
10461 fn wysiwyg_backspace_at_content_start_leaves_frontmatter_intact() {
10462 // Backspace deletes `prev_boundary..caret` directly; at the first real
10463 // block that boundary is inside the hidden frontmatter, so it must be a
10464 // no-op rather than eating the closing `---`.
10465 let fm = "---\ntitle: hi\n---\n";
10466 let body = format!("{fm}leaf\n");
10467 let mut d = wysiwyg_doc("wys_fm_bs", &body);
10468 assert_eq!(d.caret, fm.len());
10469 d.backspace();
10470 assert_eq!(d.source, body, "backspace must not touch frontmatter");
10471 d.delete_word_back();
10472 assert_eq!(
10473 d.source, body,
10474 "word-delete must not touch frontmatter either"
10475 );
10476 }
10477
10478 #[test]
10479 fn wysiwyg_edits_inside_a_vis_directive_block_without_disturbing_its_fences() {
10480 // diaryx's `:::vis{.audience}` visibility block — any `:::name{.class}`
10481 // fenced div, really, since core parses these on for every document
10482 // now (`parse_extensions`). The container is a `directive` node, an
10483 // `is_block_container` kind like `block_quote`, so the caret works
10484 // inside its child paragraph exactly as it would inside a quote: typing
10485 // edits the paragraph, and the `:::vis{...}` / `:::` fences round-trip
10486 // untouched.
10487 let body = ":::vis{.public .family}\nhello\n:::\nafter\n";
10488 let mut d = wysiwyg_doc("wys_vis", body);
10489 d.caret = body.find("hello").unwrap() + "hello".len();
10490 d.insert("!");
10491 assert_eq!(
10492 d.source, ":::vis{.public .family}\nhello!\n:::\nafter\n",
10493 "typing inside the block edits its content in place"
10494 );
10495 assert!(
10496 d.source.contains(":::vis{.public .family}"),
10497 "opening fence survives"
10498 );
10499 assert!(d.source.contains(":::\nafter"), "closing fence survives");
10500 }
10501
10502 #[test]
10503 fn source_view_still_reaches_frontmatter() {
10504 // The metadata is only *hidden*, never lost: the source view edits and
10505 // selects it in full, and it's always preserved on save.
10506 let fm = "---\ntitle: hi\n---\n";
10507 let body = format!("{fm}# leaf\n");
10508 let mut d = doc_with("src_fm", &body);
10509 d.select_all();
10510 let sel = d.selected_text().unwrap();
10511 assert!(
10512 sel.contains("title"),
10513 "source view should select everything"
10514 );
10515 d.move_doc_start(false);
10516 assert_eq!(d.caret, 0, "source view can reach offset 0");
10517 }
10518
10519 const TABLE: &str = "| Name | Qty |\n|:-----|----:|\n| Pear | 3 |\n| Fig | 12 |\n";
10520
10521 #[test]
10522 fn wysiwyg_right_crosses_a_cell_border_without_stalling() {
10523 // The border and padding between two cells all share one source offset,
10524 // so a column-stepping caret would sit on `│` and then stall there
10525 // forever. Right must step: end of "Name" -> start of "Qty".
10526 let mut d = wysiwyg_doc("tbl_right", TABLE);
10527 d.caret = TABLE.find("Name").unwrap() + 4; // just after "Name"
10528 d.move_right(false);
10529 assert_eq!(
10530 d.caret,
10531 TABLE.find("Qty").unwrap(),
10532 "should land in the next cell"
10533 );
10534 let (r, c) = d.caret_pos();
10535 assert_eq!(d.vmap.rows[r].glyphs[c].ch, 'Q');
10536 }
10537
10538 #[test]
10539 fn wysiwyg_left_crosses_back_to_the_previous_cell() {
10540 let mut d = wysiwyg_doc("tbl_left", TABLE);
10541 d.caret = TABLE.find("Qty").unwrap();
10542 d.move_left(false);
10543 assert_eq!(
10544 d.caret,
10545 TABLE.find("Name").unwrap() + 4,
10546 "end of the previous cell"
10547 );
10548 }
10549
10550 #[test]
10551 fn wysiwyg_down_steps_over_a_table_rule() {
10552 // Between the header and the first body row sits a `├───┼───┤` rule.
10553 // It's drawn but holds no caret, so one Down must reach "Pear".
10554 let mut d = wysiwyg_doc("tbl_down", TABLE);
10555 d.caret = TABLE.find("Name").unwrap();
10556 d.move_down(false);
10557 assert_eq!(
10558 d.caret,
10559 TABLE.find("Pear").unwrap(),
10560 "one Down reaches the body row"
10561 );
10562 d.move_down(false);
10563 assert_eq!(d.caret, TABLE.find("Fig").unwrap());
10564 }
10565
10566 #[test]
10567 fn wysiwyg_tab_walks_the_cells_and_shift_tab_walks_back() {
10568 let mut d = wysiwyg_doc("tbl_tab", TABLE);
10569 d.caret = TABLE.find("Name").unwrap();
10570 // A hop lands with the destination cell's whole content selected, the
10571 // caret at its end — so typing replaces the cell like a form field.
10572 assert!(d.cell_hop(true));
10573 assert_eq!(
10574 d.selected_text(),
10575 Some("Qty"),
10576 "the target cell comes up selected"
10577 );
10578 assert_eq!(d.caret, TABLE.find("Qty").unwrap() + "Qty".len());
10579 assert!(d.cell_hop(true), "Tab wraps onto the next row's first cell");
10580 assert_eq!(d.selected_text(), Some("Pear"));
10581 assert!(d.cell_hop(false));
10582 assert_eq!(d.selected_text(), Some("Qty"));
10583 }
10584
10585 #[test]
10586 fn tab_outside_a_table_is_not_a_cell_hop() {
10587 // `cell_hop` reports false so the frontend can indent as usual.
10588 let mut d = wysiwyg_doc("tbl_none", "just a paragraph\n");
10589 d.caret = 4;
10590 assert!(!d.cell_hop(true));
10591 assert_eq!(d.caret, 4, "a refused hop leaves the caret alone");
10592 }
10593
10594 #[test]
10595 fn tab_at_the_last_cell_declines_rather_than_leaving_the_table() {
10596 let mut d = wysiwyg_doc("tbl_edge", TABLE);
10597 d.caret = TABLE.rfind("12").unwrap(); // the final cell
10598 assert!(!d.cell_hop(true), "no cell after the last one");
10599 d.caret = TABLE.find("Name").unwrap();
10600 assert!(!d.cell_hop(false), "no cell before the first one");
10601 }
10602
10603 #[test]
10604 fn wysiwyg_vertical_cell_motion_holds_the_column() {
10605 // Down/Up step to the cell above/below in the *same column*, not back to
10606 // the top-left the way a naive row/col motion over the picture would.
10607 let mut d = wysiwyg_doc("tbl_vert", TABLE);
10608 d.caret = TABLE.find("Qty").unwrap();
10609 // Each vertical hop selects the destination cell, holding the column.
10610 assert!(d.cell_move_vertical(true));
10611 assert_eq!(d.selected_text(), Some("3"), "Down holds column 1");
10612 assert!(d.cell_move_vertical(true));
10613 assert_eq!(d.selected_text(), Some("12"), "Down again, still column 1");
10614 assert!(!d.cell_move_vertical(true), "no row below the last");
10615 assert!(d.cell_move_vertical(false));
10616 assert_eq!(d.selected_text(), Some("3"), "Up holds column 1");
10617 assert!(d.cell_move_vertical(false));
10618 assert_eq!(d.selected_text(), Some("Qty"), "Up onto the header");
10619 assert!(!d.cell_move_vertical(false), "no row above the header");
10620 }
10621
10622 #[test]
10623 fn tab_off_the_last_cell_grows_a_row_and_enters_it() {
10624 let mut d = wysiwyg_doc("tbl_grow", TABLE);
10625 d.caret = TABLE.rfind("12").unwrap();
10626 let rows_before = d.source.matches('\n').count();
10627 assert!(d.cell_tab(true), "acts as a table key");
10628 assert_eq!(
10629 d.source.matches('\n').count(),
10630 rows_before + 1,
10631 "a fresh row was appended"
10632 );
10633 assert!(d.caret_in_table(), "the caret entered the new row");
10634 // The caret sits in the new row's first cell — past the old last cell.
10635 assert!(d.caret > TABLE.rfind("12").unwrap());
10636 }
10637
10638 #[test]
10639 fn return_in_a_table_drops_a_cell_and_grows_a_row_at_the_bottom() {
10640 let mut d = wysiwyg_doc("tbl_ret", TABLE);
10641 d.caret = TABLE.find("Name").unwrap();
10642 assert!(d.cell_return(), "acts as a table key");
10643 assert_eq!(
10644 d.selected_text(),
10645 Some("Pear"),
10646 "Return drops one cell, selecting it"
10647 );
10648 // From the last row, Return appends a row and enters it.
10649 d.caret = TABLE.rfind("Fig").unwrap();
10650 let rows_before = d.source.matches('\n').count();
10651 assert!(d.cell_return());
10652 assert_eq!(d.source.matches('\n').count(), rows_before + 1);
10653 assert!(d.caret_in_table());
10654 }
10655
10656 #[test]
10657 fn return_and_tab_outside_a_table_decline() {
10658 let mut d = wysiwyg_doc("tbl_decline", "just a paragraph\n");
10659 d.caret = 4;
10660 assert!(!d.cell_return(), "no table: the frontend inserts a newline");
10661 assert!(!d.cell_tab(true), "no table: the frontend indents");
10662 assert!(
10663 !d.cell_line_break(),
10664 "no table: the frontend breaks the line"
10665 );
10666 }
10667
10668 #[test]
10669 fn shift_return_inserts_an_in_cell_break_the_renderer_reads_as_a_line() {
10670 let mut d = wysiwyg_doc("tbl_break", TABLE);
10671 d.caret = TABLE.find("Pear").unwrap() + 4; // just after "Pear"
10672 assert!(d.cell_line_break(), "acts as a table key");
10673 assert!(
10674 d.source.contains("Pear<br>"),
10675 "spelled as an inline <br>: {}",
10676 d.source
10677 );
10678 assert!(d.caret_in_table(), "still in the cell, past the break");
10679 // The break renders as a real line: the "Pear" cell now draws two lines,
10680 // so the table's picture is one row taller than a single-line table.
10681 d.build_visual(80);
10682 let table = &d.vmap.tables[0];
10683 let cell = &table.grid[1].cells[0]; // first body row, first column
10684 assert!(
10685 cell.glyphs.iter().any(|g| g.ch == '\n'),
10686 "the cell carries the break as a newline glyph for the frontend to split"
10687 );
10688 }
10689
10690 #[test]
10691 fn shift_return_in_a_markdown_cell_leaves_a_semantic_hard_break_not_raw_html() {
10692 // twig promotes the in-cell `<br>` to a `hard_break`, so the break reads
10693 // back as structure — the whole point of routing through insert_line_break
10694 // instead of splicing raw `<br>` bytes.
10695 let mut d = wysiwyg_doc("tbl_break_semantic", TABLE);
10696 d.caret = TABLE.find("Pear").unwrap() + 4;
10697 assert!(d.cell_line_break());
10698 let kinds: Vec<Kind> = d
10699 .editor
10700 .nodes()
10701 .unwrap()
10702 .iter()
10703 .map(|n| n.kind.clone())
10704 .collect();
10705 assert!(kinds.contains(&Kind::HardBreak), "got {kinds:?}");
10706 assert!(
10707 !kinds.contains(&Kind::RawInline),
10708 "still raw HTML: {kinds:?}"
10709 );
10710 }
10711
10712 #[test]
10713 fn backspace_over_an_in_cell_break_deletes_the_whole_br_not_a_byte() {
10714 // The `<br>` draws as one newline glyph, so Backspace over it must take
10715 // all four bytes — a one-byte delete would strand a visible `<br` in the
10716 // cell (the reported bug).
10717 let mut d = wysiwyg_doc("tbl_break_bs", TABLE);
10718 d.caret = TABLE.find("Pear").unwrap() + 4;
10719 assert!(d.cell_line_break());
10720 assert!(d.source.contains("Pear<br>"), "precondition: {}", d.source);
10721 d.backspace(); // caret sits just past the break
10722 assert!(
10723 !d.source.contains("<br"),
10724 "no half-deleted <br left: {}",
10725 d.source
10726 );
10727 assert!(
10728 d.source.contains("| Pear |"),
10729 "the cell is back to one line: {}",
10730 d.source
10731 );
10732 }
10733
10734 #[test]
10735 fn delete_forward_over_an_in_cell_break_deletes_the_whole_br() {
10736 let mut d = wysiwyg_doc("tbl_break_del", TABLE);
10737 d.caret = TABLE.find("Pear").unwrap() + 4;
10738 assert!(d.cell_line_break());
10739 d.caret = TABLE.find("Pear").unwrap() + 4; // back onto the break's start
10740 d.delete_forward();
10741 assert!(
10742 !d.source.contains("<br"),
10743 "no half-deleted <br: {}",
10744 d.source
10745 );
10746 assert!(
10747 d.source.contains("| Pear |"),
10748 "cell back to one line: {}",
10749 d.source
10750 );
10751 }
10752
10753 #[test]
10754 fn shift_return_in_a_djot_cell_is_swallowed_and_leaves_the_row_intact() {
10755 // Djot has no idiomatic in-cell break, so twig refuses it. The gesture is
10756 // still consumed (a real newline would split the one-line row), but the
10757 // cell must be left exactly as it was — no non-idiomatic `<br>` spliced in.
10758 let src = "| Name | Qty |\n|:-----|----:|\n| Pear | 3 |\n";
10759 let mut d = Doc::from_source(src.to_string(), Format::Djot).unwrap();
10760 d.caret = src.find("Pear").unwrap() + 4;
10761 assert!(d.caret_in_table(), "caret should be inside the djot table");
10762 assert!(
10763 d.cell_line_break(),
10764 "the key is consumed, not passed to the frontend"
10765 );
10766 assert_eq!(d.source, src, "the djot cell is left untouched");
10767 assert!(
10768 !d.source.contains("<br>"),
10769 "no non-idiomatic <br> spliced into djot"
10770 );
10771 assert!(
10772 d.status.is_some(),
10773 "the refusal is surfaced on the status line"
10774 );
10775 }
10776
10777 #[test]
10778 fn typing_in_a_cell_edits_that_cell() {
10779 // Editing comes free once offsets map correctly: the caret is a source
10780 // offset, so a normal splice lands inside the pipe table.
10781 let mut d = wysiwyg_doc("tbl_type", TABLE);
10782 d.caret = TABLE.find("Pear").unwrap() + 4;
10783 d.insert("s");
10784 assert!(d.source.contains("| Pears | 3 |"), "got {:?}", d.source);
10785 }
10786
10787 #[test]
10788 fn motion_and_delete_treat_an_emoji_as_one_character() {
10789 // 👨👩👧 is a single grapheme built from three emoji joined by ZWJ — 18
10790 // bytes, several codepoints. Right-arrow must clear it in one step, and
10791 // backspace must remove the whole cluster, not a stray joiner.
10792 let family = "👨👩👧";
10793 let mut d = doc_with("emoji", &format!("a{family}b\n"));
10794 d.caret = 1; // just after 'a', before the emoji
10795 d.move_right(false);
10796 assert_eq!(
10797 d.caret,
10798 1 + family.len(),
10799 "one step clears the whole cluster"
10800 );
10801 assert_eq!(&d.source[d.caret..d.caret + 1], "b");
10802
10803 d.backspace(); // delete the emoji as a unit
10804 assert_eq!(d.source, "ab\n");
10805 assert_eq!(d.caret, 1);
10806 }
10807
10808 #[test]
10809 fn motion_handles_a_combining_accent_as_one_character() {
10810 // "e" + U+0301 (combining acute) renders as one é.
10811 let mut d = doc_with("combining", "e\u{0301}x\n");
10812 d.caret = 0;
10813 d.move_right(false);
10814 assert_eq!(
10815 d.caret,
10816 "e\u{0301}".len(),
10817 "steps past base + combining mark"
10818 );
10819 }
10820
10821 #[test]
10822 fn undo_then_redo_round_trips_an_edit() {
10823 let mut d = doc_with("undo", "hello\n");
10824 d.caret = 5;
10825 d.insert("!");
10826 assert_eq!(d.source, "hello!\n");
10827 d.undo();
10828 assert_eq!(d.source, "hello\n");
10829 assert_eq!(d.caret, 5, "undo restores the caret");
10830 d.redo();
10831 assert_eq!(d.source, "hello!\n");
10832 }
10833
10834 #[test]
10835 fn a_run_of_typing_undoes_as_one_step() {
10836 let mut d = doc_with("coalesce", "\n");
10837 d.caret = 0;
10838 d.insert("a");
10839 d.insert("b");
10840 d.insert("c");
10841 assert_eq!(d.source, "abc\n");
10842 d.undo(); // the whole typed run, not just "c"
10843 assert_eq!(d.source, "\n");
10844 d.undo(); // nothing left — the run was one step
10845 assert_eq!(d.source, "\n");
10846 assert_eq!(d.status.as_deref(), Some("nothing to undo"));
10847 }
10848
10849 // ── IME composition ──────────────────────────────────────────────────────
10850
10851 #[test]
10852 fn a_composition_run_undoes_as_one_step() {
10853 let mut d = doc_with("compose", "\n");
10854 d.caret = 0;
10855 // What an IME does: each step replaces the last one's provisional bytes.
10856 d.edit_composing(0, 0, "k");
10857 d.edit_composing(0, 1, "か");
10858 d.edit_composing(0, 3, "かん");
10859 d.edit_composing(0, 6, "感"); // the commit
10860 d.end_composition();
10861 assert_eq!(d.source, "感\n");
10862 d.undo(); // the whole composition, not its last keystroke
10863 assert_eq!(d.source, "\n");
10864 assert_eq!(d.status.as_deref(), None, "the run was a single step");
10865 }
10866
10867 #[test]
10868 fn two_compositions_are_two_undo_steps() {
10869 let mut d = doc_with("compose_two", "\n");
10870 d.caret = 0;
10871 d.edit_composing(0, 0, "か");
10872 d.edit_composing(0, 3, "蚊");
10873 d.end_composition();
10874 d.edit_composing(3, 3, "き");
10875 d.edit_composing(3, 6, "木");
10876 d.end_composition();
10877 assert_eq!(d.source, "蚊木\n");
10878 d.undo();
10879 assert_eq!(d.source, "蚊\n", "only the second composition");
10880 d.undo();
10881 assert_eq!(d.source, "\n");
10882 }
10883
10884 #[test]
10885 fn a_composition_does_not_fold_into_the_typing_around_it() {
10886 let mut d = doc_with("compose_typing", "\n");
10887 d.caret = 0;
10888 d.insert("a");
10889 d.insert("b");
10890 d.edit_composing(2, 2, "か");
10891 d.edit_composing(2, 5, "蚊");
10892 d.end_composition();
10893 d.insert("c");
10894 assert_eq!(d.source, "ab蚊c\n");
10895 d.undo();
10896 assert_eq!(d.source, "ab蚊\n");
10897 d.undo();
10898 assert_eq!(d.source, "ab\n");
10899 d.undo();
10900 assert_eq!(d.source, "\n");
10901 }
10902
10903 #[test]
10904 fn ending_a_composition_that_never_began_leaves_a_typing_run_alone() {
10905 let mut d = doc_with("compose_spurious", "\n");
10906 d.caret = 0;
10907 d.insert("a");
10908 d.end_composition(); // an IME unmarking unprompted
10909 d.insert("b");
10910 assert_eq!(d.source, "ab\n");
10911 d.undo();
10912 assert_eq!(d.source, "\n", "still one typed run");
10913 }
10914
10915 // ── the clipboard's rich flavor ──────────────────────────────────────────
10916
10917 #[test]
10918 fn an_inline_selection_publishes_html_without_a_paragraph_wrapper() {
10919 let mut d = doc_with("sel_inline", "a **bold** c\n");
10920 d.anchor = Some(2);
10921 d.caret = 10; // `**bold**`, inside the paragraph
10922 assert_eq!(d.selection_html().as_deref(), Some("<strong>bold</strong>"));
10923 }
10924
10925 #[test]
10926 fn a_whole_block_selection_keeps_its_paragraph() {
10927 let mut d = doc_with("sel_block", "a **bold** c\n");
10928 d.anchor = Some(0);
10929 d.caret = 12; // the entire paragraph
10930 assert_eq!(
10931 d.selection_html().as_deref(),
10932 Some("<p>a <strong>bold</strong> c</p>")
10933 );
10934 }
10935
10936 #[test]
10937 fn a_multi_block_selection_keeps_its_structure() {
10938 let mut d = doc_with("sel_multi", "para\n\n- one\n- two\n");
10939 d.select_all();
10940 let html = d.selection_html().expect("renders");
10941 assert!(html.contains("<p>para</p>"), "{html:?}");
10942 assert!(html.contains("<li>one</li>"), "{html:?}");
10943 }
10944
10945 #[test]
10946 fn a_word_inside_a_heading_publishes_as_text_not_a_heading() {
10947 // The fragment `Head` is a paragraph standalone; the *document* says it
10948 // sits inside one block, so the wrapper is an artifact either way.
10949 let mut d = doc_with("sel_heading", "# Head line\n");
10950 d.anchor = Some(2);
10951 d.caret = 6;
10952 assert_eq!(d.selection_html().as_deref(), Some("Head"));
10953 }
10954
10955 #[test]
10956 fn no_selection_publishes_no_html() {
10957 let mut d = doc_with("sel_none", "a b\n");
10958 d.caret = 1;
10959 assert_eq!(d.selection_html(), None);
10960 }
10961
10962 #[test]
10963 fn pasting_html_converts_it_and_is_one_undo_step() {
10964 let mut d = doc_with("paste_html", "x\n");
10965 d.caret = 1;
10966 assert!(d.paste_html("<p>a <strong>b</strong> c</p>"));
10967 assert_eq!(d.source, "xa **b** c\n");
10968 d.undo();
10969 assert_eq!(d.source, "x\n", "the whole paste, in one step");
10970 }
10971
10972 #[test]
10973 fn pasting_html_replaces_the_selection() {
10974 let mut d = doc_with("paste_html_sel", "keep drop\n");
10975 d.anchor = Some(5);
10976 d.caret = 9;
10977 assert!(d.paste_html("<em>new</em>"));
10978 assert_eq!(d.source, "keep *new*\n");
10979 }
10980
10981 #[test]
10982 fn html_that_would_paste_garbage_declines_so_the_caller_falls_back() {
10983 let mut d = doc_with("paste_html_bad", "x\n");
10984 d.caret = 1;
10985 // twig builds no table from HTML; raw `<table>` in prose is worse than
10986 // the plain flavor the caller still holds.
10987 assert!(!d.paste_html("<table><tr><td>a</td></tr></table>"));
10988 assert_eq!(d.source, "x\n", "declined edits nothing");
10989 }
10990
10991 #[test]
10992 fn copy_then_paste_round_trips_through_the_html_flavor() {
10993 let mut d = doc_with("clip_round", "a **b** and [l](https://x.dev)\n");
10994 d.select_all();
10995 let html = d.selection_html().expect("renders");
10996 let mut into = doc_with("clip_round_dst", "\n");
10997 into.caret = 0;
10998 assert!(into.paste_html(&html));
10999 assert_eq!(into.source, "a **b** and [l](https://x.dev)\n");
11000 }
11001
11002 #[test]
11003 fn moving_the_caret_starts_a_new_undo_group() {
11004 let mut d = doc_with("break", "\n");
11005 d.caret = 0;
11006 d.insert("a");
11007 d.insert("b"); // "ab\n", caret at 2
11008 d.move_left(false); // breaks the run
11009 d.insert("X"); // "aXb\n"
11010 assert_eq!(d.source, "aXb\n");
11011 d.undo();
11012 assert_eq!(
11013 d.source, "ab\n",
11014 "first undo removes only the post-move insert"
11015 );
11016 d.undo();
11017 assert_eq!(d.source, "\n", "second undo removes the earlier run");
11018 }
11019
11020 #[test]
11021 fn undo_reverses_a_format_toggle() {
11022 let mut d = doc_with("fmt_undo", "a word b\n");
11023 d.anchor = Some(2);
11024 d.caret = 6;
11025 d.toggle(InlineKind::Strong);
11026 assert_eq!(d.source, "a **word** b\n");
11027 d.undo();
11028 assert_eq!(d.source, "a word b\n");
11029 }
11030
11031 #[test]
11032 fn undo_back_to_the_saved_state_clears_dirty() {
11033 let mut d = doc_with("dirty_undo", "hello\n");
11034 assert!(!d.dirty);
11035 d.caret = 5;
11036 d.insert("!");
11037 assert!(d.dirty);
11038 d.undo();
11039 assert!(
11040 !d.dirty,
11041 "undoing to the saved source is not a modification"
11042 );
11043 }
11044
11045 #[test]
11046 fn a_new_edit_invalidates_redo() {
11047 let mut d = doc_with("redo_inv", "\n");
11048 d.caret = 0;
11049 d.insert("a");
11050 d.undo();
11051 d.insert("b"); // diverges — the redo of "a" is now gone
11052 d.redo();
11053 assert_eq!(d.source, "b\n");
11054 }
11055
11056 #[test]
11057 fn can_undo_and_can_redo_follow_the_history_a_menu_would_enable_by() {
11058 let mut d = doc_with("can_undo", "hello\n");
11059 assert!(
11060 !d.can_undo() && !d.can_redo(),
11061 "a fresh document has no history"
11062 );
11063 d.caret = 5;
11064 d.insert("!");
11065 assert!(
11066 d.can_undo() && !d.can_redo(),
11067 "an edit is a step to take back"
11068 );
11069 d.undo();
11070 assert!(!d.can_undo() && d.can_redo(), "undone: only redo remains");
11071 d.redo();
11072 assert!(d.can_undo() && !d.can_redo(), "redone: back to undoable");
11073 d.undo();
11074 d.insert("?");
11075 assert!(
11076 d.can_undo() && !d.can_redo(),
11077 "a fresh edit ends the redo chain"
11078 );
11079 // A coalesced run over-counts steps — the bound is what a menu needs,
11080 // and it reconciles the moment twig reports the history empty.
11081 d.insert("a");
11082 d.insert("b");
11083 while d.can_undo() {
11084 d.undo();
11085 }
11086 assert_eq!(d.source, "hello\n");
11087 assert!(!d.can_undo());
11088 // A reading surface has nothing to undo, whatever the history holds.
11089 d.redo();
11090 d.set_read_only(true);
11091 assert!(!d.can_undo() && !d.can_redo());
11092 }
11093
11094 #[test]
11095 fn undo_on_empty_history_is_a_no_op() {
11096 let mut d = doc_with("undo_empty", "hi\n");
11097 d.undo();
11098 assert_eq!(d.source, "hi\n");
11099 assert_eq!(d.status.as_deref(), Some("nothing to undo"));
11100 }
11101
11102 #[test]
11103 fn a_one_character_paste_is_its_own_undo_step() {
11104 for view in [View::Source, View::Wysiwyg] {
11105 let mut d = doc_in(view, "paste_step", "ab\n");
11106 d.caret = 0;
11107 d.insert("x");
11108 d.insert("y"); // a run of typing
11109 d.paste("z"); // one character, but pasted — not part of that run
11110 assert_eq!(d.source, "xyzab\n");
11111 d.undo();
11112 assert_eq!(d.source, "xyab\n", "the paste undoes on its own");
11113 assert_eq!(d.caret, 2, "and hands back the caret it found");
11114 d.undo();
11115 assert_eq!(d.source, "ab\n", "the typed run is still one step under it");
11116 }
11117 }
11118
11119 #[test]
11120 fn the_same_character_typed_still_joins_the_run() {
11121 // The other half of the pair: `z` is a keystroke here and a paste above,
11122 // and the two undo differently. Nothing about the *string* says which —
11123 // which is why provenance has to come from the door the caller uses.
11124 for view in [View::Source, View::Wysiwyg] {
11125 let mut d = doc_in(view, "typed_run", "ab\n");
11126 d.caret = 0;
11127 d.insert("x");
11128 d.insert("y");
11129 d.insert("z");
11130 d.undo();
11131 assert_eq!(d.source, "ab\n", "one run, one step");
11132 }
11133 }
11134
11135 #[test]
11136 fn undo_restores_the_caret_to_where_it_was_not_to_the_edit_site() {
11137 for view in [View::Source, View::Wysiwyg] {
11138 let mut d = doc_in(view, "undo_caret", "hello world\n");
11139 d.caret = 11; // standing at the end of "world", away from the edit
11140 d.edit(0, 5, "goodbye");
11141 assert_eq!(d.source, "goodbye world\n");
11142 d.undo();
11143 assert_eq!(d.source, "hello world\n");
11144 // The undone edit ends at offset 5; the user was at 11.
11145 assert_eq!(d.caret, 11, "the caret comes back with the bytes");
11146 }
11147 }
11148
11149 #[test]
11150 fn undo_restores_the_selection_the_edit_replaced() {
11151 for view in [View::Source, View::Wysiwyg] {
11152 let mut d = doc_in(view, "undo_sel", "a word b\n");
11153 d.anchor = Some(2);
11154 d.caret = 6; // "word" selected
11155 d.insert("X");
11156 assert_eq!(d.source, "a X b\n");
11157 d.undo();
11158 assert_eq!(d.source, "a word b\n");
11159 assert_eq!(d.selection(), Some((2, 6)), "the selection comes back too");
11160 }
11161 }
11162
11163 #[test]
11164 fn redo_restores_the_caret_the_edit_left_behind() {
11165 for view in [View::Source, View::Wysiwyg] {
11166 let mut d = doc_in(view, "redo_caret", "hello world\n");
11167 d.caret = 11;
11168 d.edit(0, 5, "goodbye");
11169 assert_eq!(d.caret, 7, "the edit left the caret after its new text");
11170 d.undo();
11171 d.redo();
11172 assert_eq!(d.source, "goodbye world\n");
11173 assert_eq!(d.caret, 7, "redo puts it back where the edit had it");
11174 }
11175 }
11176
11177 #[test]
11178 fn undoing_a_typed_run_restores_the_caret_from_before_the_whole_run() {
11179 for view in [View::Source, View::Wysiwyg] {
11180 let mut d = doc_in(view, "run_caret", "hi\n");
11181 d.caret = 2;
11182 d.insert("a");
11183 d.insert("b");
11184 d.insert("c");
11185 assert_eq!(d.source, "hiabc\n");
11186 d.undo();
11187 assert_eq!(d.source, "hi\n");
11188 assert_eq!(d.caret, 2, "before the run, not before its last keystroke");
11189 d.redo();
11190 assert_eq!(d.caret, 5, "and redo restores the end of the whole run");
11191 }
11192 }
11193
11194 #[test]
11195 fn undo_restores_the_caret_across_a_format_toggle() {
11196 // A toggle reaches twig without going through `splice`, so it has to
11197 // record its own step — miss it and every stack depth below it is off by
11198 // one, and undo starts handing back another edit's caret.
11199 for view in [View::Source, View::Wysiwyg] {
11200 let mut d = doc_in(view, "fmt_caret", "a word b\n");
11201 d.caret = 8;
11202 d.anchor = Some(2);
11203 d.caret = 6;
11204 d.toggle(InlineKind::Strong);
11205 assert_eq!(d.source, "a **word** b\n");
11206 d.undo();
11207 assert_eq!(d.source, "a word b\n");
11208 assert_eq!(
11209 d.selection(),
11210 Some((2, 6)),
11211 "the toggled selection comes back"
11212 );
11213 }
11214 }
11215
11216 #[test]
11217 fn an_edit_after_an_undo_truncates_the_caret_history_with_twigs() {
11218 // The drift that would never announce itself: twig drops its redo stack
11219 // on any fresh edit, so a leaf redo entry that outlives it would restore
11220 // a caret from the timeline that edit abandoned.
11221 for view in [View::Source, View::Wysiwyg] {
11222 let mut d = doc_in(view, "redo_trunc", "hello world\n");
11223 d.caret = 11;
11224 d.edit(0, 5, "goodbye"); // step A, caret 11 → 7
11225 d.undo();
11226 assert_eq!(d.caret, 11);
11227 d.caret = 0;
11228 d.insert("X"); // diverges: A's redo is gone from twig
11229 assert_eq!(d.source, "Xhello world\n");
11230
11231 d.redo();
11232 assert_eq!(d.source, "Xhello world\n", "nothing to redo onto");
11233 assert_eq!(d.status.as_deref(), Some("nothing to redo"));
11234 d.undo();
11235 assert_eq!(d.source, "hello world\n");
11236 assert_eq!(
11237 d.caret, 0,
11238 "the surviving step's caret, not the dropped one"
11239 );
11240 }
11241 }
11242
11243 #[test]
11244 fn indent_and_outdent_move_the_caret_line_with_its_text() {
11245 for view in [View::Source, View::Wysiwyg] {
11246 let g = |m, f: fn(&mut Doc)| golden_in(view, "indent_line", m, f);
11247 assert_eq!(g("he|llo\n", |d| d.indent()), " he|llo\n");
11248 assert_eq!(g(" he|llo\n", |d| d.outdent()), "he|llo\n");
11249 // Indentation the caret is standing *in* collapses to the line start
11250 // rather than dragging the caret into the text.
11251 assert_eq!(g("| hello\n", |d| d.outdent()), "|hello\n");
11252 // A line with none to give back is left exactly as it was.
11253 assert_eq!(g("he|llo\n", |d| d.outdent()), "he|llo\n");
11254 // Less than a full level gives back what it has.
11255 assert_eq!(g(" he|llo\n", |d| d.outdent()), "he|llo\n");
11256 // A tab is one level however many spaces it isn't.
11257 assert_eq!(g("\the|llo\n", |d| d.outdent()), "he|llo\n");
11258 }
11259 }
11260
11261 #[test]
11262 fn one_indent_level_leaves_a_paragraph_a_paragraph() {
11263 // Why the level is two spaces and not the four both frontends type
11264 // today. Four is markdown's indented-code-block marker, so a Tab on a
11265 // paragraph would silently restyle it as code — a width that changes
11266 // what the document *means* isn't an indent. Pinned because the number
11267 // is the kind of thing a later list-aware pass would reach for.
11268 let mut d = doc_with("indent_kind", "hello\n");
11269 d.caret = 2;
11270 d.indent();
11271 assert_eq!(d.source, " hello\n");
11272 assert!(
11273 d.nodes().iter().any(|n| n.kind == Kind::Para),
11274 "still prose after a Tab"
11275 );
11276 assert!(!d.nodes().iter().any(|n| n.kind == Kind::CodeBlock));
11277
11278 // The four-space level this replaces, for contrast: same text, and twig
11279 // reparses the paragraph into a code block.
11280 let mut wide = doc_with("indent_kind_4", " hello\n");
11281 wide.build_visual(80);
11282 assert!(
11283 wide.nodes().iter().any(|n| n.kind == Kind::CodeBlock),
11284 "four spaces is a code block, not an indented paragraph"
11285 );
11286 }
11287
11288 #[test]
11289 fn indent_nests_a_list_item_under_its_parent() {
11290 // Tab indents a list item by its own marker width, landing its marker at
11291 // the parent's content column so twig reparses it as a nested list.
11292 for view in [View::Source, View::Wysiwyg] {
11293 let mut d = doc_in(view, "indent_nest", "- a\n- b\n");
11294 d.caret = 6; // on the second item
11295 d.indent();
11296 assert_eq!(d.source, "- a\n - b\n");
11297 let lists = d
11298 .nodes()
11299 .iter()
11300 .filter(|n| n.kind == Kind::BulletList)
11301 .count();
11302 assert_eq!(lists, 2, "the indented item is a nested list");
11303 }
11304 }
11305
11306 #[test]
11307 fn indent_nests_an_ordered_item_at_its_marker_width() {
11308 // An ordered marker `1. ` is three columns wide, so a two-space step
11309 // (which nests a bullet) leaves it flat. Regression: Tab must use the
11310 // marker width, three, so the item actually nests — and the source
11311 // renumbers so the sub-list restarts at 1 and the outer list resumes.
11312 for view in [View::Source, View::Wysiwyg] {
11313 let mut d = doc_in(view, "indent_ord", "1. a\n2. b\n3. c\n");
11314 d.caret = d.source.find('b').unwrap();
11315 d.indent();
11316 assert_eq!(d.source, "1. a\n 1. b\n2. c\n");
11317 let lists = d
11318 .nodes()
11319 .iter()
11320 .filter(|n| n.kind == Kind::OrderedList)
11321 .count();
11322 assert_eq!(lists, 2, "the indented item is a nested ordered list");
11323 }
11324 }
11325
11326 #[test]
11327 fn indent_leaves_a_lists_first_item_put() {
11328 // The first item of a list has no sibling above it to nest under, so Tab
11329 // is a no-op there — the marker stays at column zero rather than being
11330 // shoved into indentation twig can't read as a sub-list.
11331 for view in [View::Source, View::Wysiwyg] {
11332 let mut d = doc_in(view, "indent_first", "- a\n- b\n");
11333 d.caret = 1; // on the FIRST item
11334 d.indent();
11335 assert_eq!(d.source, "- a\n- b\n", "the first item doesn't nest");
11336 // The sibling below still nests, proving the guard is per-item.
11337 d.caret = d.source.find('b').unwrap();
11338 d.indent();
11339 assert_eq!(d.source, "- a\n - b\n");
11340 }
11341 }
11342
11343 #[test]
11344 fn hidden_mode_keeps_typed_markup_literal() {
11345 // The Diaryx default: typing `*hi*` gives the characters, not emphasis —
11346 // twig escapes what would open markup, so the source is `\*hi\*` and the
11347 // AST is a plain string. Formatting is the commands' job in this mode.
11348 let mut d = doc_in(View::Wysiwyg, "hidden_literal", "");
11349 d.insert("*hi*");
11350 assert_eq!(d.source, "\\*hi\\*");
11351 assert!(
11352 d.nodes()
11353 .iter()
11354 .all(|n| n.kind != Kind::Emph && n.kind != Kind::Strong)
11355 );
11356 }
11357
11358 #[test]
11359 fn hidden_mode_escapes_a_line_start_block_marker() {
11360 // A `#`/`-`/`>` at a line start would open a block, so Hidden mode keeps
11361 // it literal too — a Diaryx user's "# 1 idea" stays prose, not a heading.
11362 let mut d = doc_in(View::Wysiwyg, "hidden_block", "");
11363 d.insert("# hi");
11364 assert_eq!(d.source, "\\# hi");
11365 assert!(d.nodes().iter().all(|n| n.kind != Kind::Heading));
11366 }
11367
11368 #[test]
11369 fn authoring_modes_keep_typed_markup_live() {
11370 // Both authoring rungs of the ladder: typing `*hi*` really is emphasis
11371 // (no escape), the same as source view — escaping is `None`'s alone, and
11372 // it's the axis, not the reveal, that decides.
11373 for (view, mode) in [
11374 (View::Wysiwyg, MarkupMode::Shortcuts),
11375 (View::Wysiwyg, MarkupMode::Full),
11376 (View::Source, MarkupMode::None),
11377 ] {
11378 let mut d = doc_in(view, "live_markup", "");
11379 d.set_markup_mode(mode);
11380 d.insert("*hi*");
11381 assert_eq!(d.source, "*hi*", "{mode:?} in {view:?} types raw markup");
11382 }
11383 }
11384
11385 #[test]
11386 fn hidden_mode_overwrite_undoes_in_one_step() {
11387 // Typing over a selection escapes the replacement *and* stays a single
11388 // undo — the selection-delete and the literal insert fold together, so
11389 // one undo brings the whole selection back, like a plain overwrite.
11390 let mut d = doc_in(View::Wysiwyg, "hidden_overwrite", "a word b\n");
11391 d.anchor = Some(2);
11392 d.caret = 6; // "word"
11393 d.insert("*");
11394 assert_eq!(d.source, "a \\* b\n", "the replacement is escaped");
11395 d.undo();
11396 assert_eq!(d.source, "a word b\n");
11397 assert_eq!(d.selection(), Some((2, 6)), "one undo, selection restored");
11398 }
11399
11400 #[test]
11401 fn backspace_over_an_escaped_char_takes_the_hidden_backslash_too() {
11402 // Type `*` in Hidden mode → `\*` (drawn as one `*`); one Backspace clears
11403 // the whole visual character, never stranding the hidden `\`.
11404 let mut d = doc_in(View::Wysiwyg, "bsp_escape", "");
11405 d.insert("*");
11406 assert_eq!(d.source, "\\*");
11407 d.backspace();
11408 assert_eq!(d.source, "", "the escape backslash went with the *");
11409 // A *literal* backslash (source view, no escape) is an ordinary char.
11410 let mut s = doc_in(View::Source, "bsp_lit", "a\\b\n");
11411 s.caret = 3; // after `b`
11412 s.backspace();
11413 assert_eq!(s.source, "a\\\n", "only the b is deleted, the \\ stays");
11414 }
11415
11416 #[test]
11417 fn hidden_mode_leaves_structural_markup_alone() {
11418 // Enter continues a bullet list by writing a real `- ` marker (an
11419 // `insert_raw`, not the typing path), so Hidden mode's escaping never
11420 // touches it — the list keeps working.
11421 let mut d = doc_in(View::Wysiwyg, "hidden_struct", "- item\n");
11422 d.caret = 6;
11423 d.newline();
11424 d.insert("two");
11425 assert_eq!(d.source, "- item\n- two\n");
11426 }
11427
11428 #[test]
11429 fn markup_mode_defaults_to_none_and_round_trips() {
11430 // Diaryx's default is the clean `None` surface; a markup-fluent
11431 // frontend can climb the ladder, and the choice sticks.
11432 let mut d = doc_in(View::Wysiwyg, "markup_mode", "hi\n");
11433 assert_eq!(d.markup_mode(), MarkupMode::None, "None by default");
11434 for mode in [MarkupMode::Shortcuts, MarkupMode::Full, MarkupMode::None] {
11435 d.set_markup_mode(mode);
11436 assert_eq!(d.markup_mode(), mode);
11437 }
11438 }
11439
11440 #[test]
11441 fn full_mode_reveals_only_the_caret_line() {
11442 // The mode's whole claim: the caret's line shows its raw delimiters and
11443 // every other line stays resolved. Two paragraphs with identical markup
11444 // so the only difference between the rows is where the caret is.
11445 let mut d = doc_in(
11446 View::Wysiwyg,
11447 "reveal_caret_line",
11448 "*one* here\n\n*two* there\n",
11449 );
11450 d.set_markup_mode(MarkupMode::Full);
11451
11452 caret_at(&mut d, "one");
11453 let rows = drawn_rows(&d);
11454 assert!(
11455 rows.iter().any(|r| r == "*one* here"),
11456 "caret's line raw: {rows:?}"
11457 );
11458 assert!(
11459 rows.iter().any(|r| r == "two there"),
11460 "other line resolved: {rows:?}"
11461 );
11462
11463 // Move to the other paragraph: the reveal follows, and the line just
11464 // left goes back to being resolved.
11465 caret_at(&mut d, "two");
11466 let rows = drawn_rows(&d);
11467 assert!(
11468 rows.iter().any(|r| r == "*two* there"),
11469 "caret's line raw: {rows:?}"
11470 );
11471 assert!(
11472 rows.iter().any(|r| r == "one here"),
11473 "left line resolved: {rows:?}"
11474 );
11475 }
11476
11477 #[test]
11478 fn revealing_a_coloured_highlight_shows_the_emoji_that_spelled_it() {
11479 // The emoji is a delimiter, not content — so `MarkupMode::Full` owes it
11480 // the same treatment as an emphasis's `*`: hidden while the caret is
11481 // elsewhere, shown in full where the caret lands. That falls out of
11482 // `delims` reading the bytes between the mark's span and its content
11483 // span, which is exactly `==🔴 ` and `==`, rather than from a table
11484 // of spellings — so the no-space form `==🟢green==` reveals right too.
11485 let mut d = doc_in(
11486 View::Wysiwyg,
11487 "reveal_coloured_mark",
11488 "a ==🔴 red== one\n\nb ==plain== two\n",
11489 );
11490 d.set_markup_mode(MarkupMode::Full);
11491
11492 caret_at(&mut d, "red");
11493 let rows = drawn_rows(&d);
11494 assert!(
11495 rows.iter().any(|r| r == "a ==🔴 red== one"),
11496 "the caret's line shows the colour it was written with: {rows:?}"
11497 );
11498 assert!(
11499 rows.iter().any(|r| r == "b plain two"),
11500 "and every other line stays resolved: {rows:?}"
11501 );
11502
11503 // Away from it, the emoji goes back to being markup — the reader sees
11504 // the words and the wash.
11505 caret_at(&mut d, "two");
11506 let rows = drawn_rows(&d);
11507 assert!(
11508 rows.iter().any(|r| r == "a red one"),
11509 "resolved again: {rows:?}"
11510 );
11511 }
11512
11513 #[test]
11514 fn hidden_modes_never_reveal_wherever_the_caret_is() {
11515 // The two rungs below `Full` share a rendering: delimiters stay hidden
11516 // even under the caret. `Shortcuts` differing from `None` only in what
11517 // typing does is exactly the point of splitting the axes.
11518 for mode in [MarkupMode::None, MarkupMode::Shortcuts] {
11519 let mut d = doc_in(View::Wysiwyg, "reveal_hidden", "*one* here\n");
11520 d.set_markup_mode(mode);
11521 caret_at(&mut d, "one");
11522 let rows = drawn_rows(&d);
11523 assert!(
11524 rows.iter().any(|r| r == "one here"),
11525 "{mode:?} hides: {rows:?}"
11526 );
11527 assert!(
11528 !rows.iter().any(|r| r.contains('*')),
11529 "{mode:?} shows no `*`: {rows:?}"
11530 );
11531 }
11532 }
11533
11534 #[test]
11535 fn revealed_delimiters_are_the_authors_own_spelling() {
11536 // Delimiters are re-read from the source rather than synthesized per
11537 // kind, so a line comes back spelled the way it was written: `_em_` does
11538 // not turn into `*em*`, and a two-backtick fence keeps both backticks.
11539 let body = "_em_ and __st__ and ``lit ` tick`` and [lk](http://x) and ~~del~~\n";
11540 let mut d = doc_in(View::Wysiwyg, "reveal_spelling", body);
11541 d.set_markup_mode(MarkupMode::Full);
11542 caret_at(&mut d, "em");
11543 let rows = drawn_rows(&d);
11544 assert!(
11545 rows.iter().any(|r| r == body.trim_end()),
11546 "the revealed line is its own source: {rows:?}"
11547 );
11548 }
11549
11550 #[test]
11551 fn revealed_heading_shows_its_hashes() {
11552 // The `# ` marker is a block-level prefix, not an inline delimiter, so
11553 // it takes its own path — but it reveals on the same rule.
11554 let mut d = doc_in(View::Wysiwyg, "reveal_heading", "# Title\n\nbody\n");
11555 d.set_markup_mode(MarkupMode::Full);
11556
11557 caret_at(&mut d, "Title");
11558 assert!(
11559 drawn_rows(&d).iter().any(|r| r == "# Title"),
11560 "{:?}",
11561 drawn_rows(&d)
11562 );
11563
11564 caret_at(&mut d, "body");
11565 let rows = drawn_rows(&d);
11566 assert!(
11567 rows.iter().any(|r| r == "Title"),
11568 "hashes hidden again: {rows:?}"
11569 );
11570 }
11571
11572 #[test]
11573 fn revealed_delimiters_are_caret_stops() {
11574 // A delimiter that is drawn but can't be reached is worse than one
11575 // that's hidden: the mode exists so the markup can be *edited*. Every
11576 // revealed byte must be somewhere the caret can stand.
11577 let mut d = doc_in(View::Wysiwyg, "reveal_stops", "*em* x\n");
11578 d.set_markup_mode(MarkupMode::Full);
11579 caret_at(&mut d, "em");
11580 let opener = d.source.find('*').unwrap();
11581 assert!(d.vmap.is_stop(opener), "the opening `*` is a caret stop");
11582 assert!(
11583 d.vmap.is_stop(opener + 3),
11584 "the closing `*` is a caret stop"
11585 );
11586 }
11587
11588 #[test]
11589 fn setext_heading_reveals_nothing_across_its_newline() {
11590 // A setext heading's underline is on another line, so it is not the
11591 // caret line's to reveal — and emitting it would inject a `\n` glyph
11592 // that splits the row where the author wrote no break.
11593 let mut d = doc_in(View::Wysiwyg, "reveal_setext", "Title\n=====\n\nbody\n");
11594 d.set_markup_mode(MarkupMode::Full);
11595 caret_at(&mut d, "Title");
11596 let rows = drawn_rows(&d);
11597 assert!(
11598 rows.iter().any(|r| r == "Title"),
11599 "title renders alone: {rows:?}"
11600 );
11601 assert!(
11602 !rows.iter().any(|r| r.contains('=')),
11603 "no underline leaks in: {rows:?}"
11604 );
11605 }
11606
11607 #[test]
11608 fn markup_mode_axes_split_the_ladder() {
11609 // The two behaviours the ladder spells: `Shortcuts` is the middle rung
11610 // that authors markup but still hides it, and it's the only rung where
11611 // the two axes disagree.
11612 assert!(!MarkupMode::None.authors());
11613 assert!(!MarkupMode::None.reveals_caret_line());
11614 assert!(MarkupMode::Shortcuts.authors());
11615 assert!(!MarkupMode::Shortcuts.reveals_caret_line());
11616 assert!(MarkupMode::Full.authors());
11617 assert!(MarkupMode::Full.reveals_caret_line());
11618 }
11619
11620 #[test]
11621 fn indenting_an_empty_dash_item_under_text_dodges_the_setext_collapse() {
11622 // Tabbing an empty `- ` under a text line would spell `- hello\n - `,
11623 // which twig (correctly, per CommonMark — pandoc agrees) reparses as a
11624 // setext H2. leaf swaps the dash for a `*` so the item stays an empty
11625 // nested bullet and `hello` stays prose: the file round-trips instead of
11626 // hiding a heading the user never asked for.
11627 for view in [View::Source, View::Wysiwyg] {
11628 let mut d = doc_in(view, "setext_guard", "- hello\n- \n");
11629 d.caret = d.source.find("- \n").unwrap() + 2; // after the empty marker
11630 d.indent();
11631 assert_eq!(d.source, "- hello\n * \n");
11632 assert!(
11633 d.nodes().iter().all(|n| n.kind != Kind::Heading),
11634 "no heading"
11635 );
11636 // And it's genuinely a nested list, not a flat one.
11637 assert_eq!(
11638 d.nodes()
11639 .iter()
11640 .filter(|n| n.kind == Kind::BulletList)
11641 .count(),
11642 2
11643 );
11644 }
11645 }
11646
11647 #[test]
11648 fn indenting_a_dash_item_with_content_keeps_its_dash() {
11649 // With content, `- x` can't be a setext underline, so there's nothing to
11650 // dodge: the marker stays a dash and nests as an ordinary sub-bullet.
11651 let mut d = doc_in(View::Wysiwyg, "setext_ok", "- hello\n- x\n");
11652 d.caret = d.source.find('x').unwrap();
11653 d.indent();
11654 assert_eq!(d.source, "- hello\n - x\n");
11655 }
11656
11657 #[test]
11658 fn the_setext_swap_undoes_as_one_step_with_the_indent() {
11659 // The dash→`*` repair coalesces into the Tab, so a single undo restores
11660 // the whole pre-Tab state rather than stranding a half-collapsed doc.
11661 let mut d = doc_in(View::Wysiwyg, "setext_undo", "- hello\n- \n");
11662 d.caret = d.source.find("- \n").unwrap() + 2;
11663 d.indent();
11664 assert_eq!(d.source, "- hello\n * \n");
11665 d.undo();
11666 assert_eq!(d.source, "- hello\n- \n", "one undo, not two");
11667 }
11668
11669 #[test]
11670 fn indent_leaves_a_nested_lists_first_item_put_too() {
11671 // The guard is about siblings, not depth: the first item of an *inner*
11672 // list (already nested under `a`) still has nothing before it at its own
11673 // level, so Tab can't take it deeper.
11674 let mut d = doc_in(View::Wysiwyg, "indent_first_nested", "- a\n - b\n - c\n");
11675 d.caret = d.source.find('b').unwrap();
11676 d.indent();
11677 assert_eq!(d.source, "- a\n - b\n - c\n", "inner first item holds");
11678 // But `c` (a sibling of `b`) nests under `b`.
11679 d.caret = d.source.find('c').unwrap();
11680 d.indent();
11681 assert_eq!(d.source, "- a\n - b\n - c\n");
11682 }
11683
11684 #[test]
11685 fn backspace_at_a_nested_item_start_outdents_it() {
11686 // Backspace with the caret right after a nested item's marker gives back
11687 // one level of nesting, the mirror of Tab — and renumbers the flattened
11688 // ordered list back to a clean run.
11689 let mut d = doc_in(View::Wysiwyg, "bsp_outdent", "1. a\n 1. b\n2. c\n");
11690 d.caret = d.source.find('b').unwrap(); // start of the nested item's content
11691 d.backspace();
11692 assert_eq!(d.source, "1. a\n2. b\n3. c\n");
11693 }
11694
11695 #[test]
11696 fn backspace_at_a_top_level_item_start_strips_the_marker() {
11697 // At the outermost level there's no nesting left to give back, so the same
11698 // keystroke drops the bullet and leaves a plain paragraph.
11699 let mut d = doc_in(View::Wysiwyg, "bsp_strip", "- a\n- b\n");
11700 d.caret = d.source.find('b').unwrap(); // right after `- `
11701 d.backspace();
11702 assert_eq!(d.source, "- a\nb\n", "the marker is gone, the text stays");
11703 }
11704
11705 #[test]
11706 fn backspace_mid_item_still_deletes_a_character() {
11707 // The list behaviour is armed only at the item's content start; anywhere
11708 // else Backspace is the ordinary character delete.
11709 let mut d = doc_in(View::Wysiwyg, "bsp_mid", "- ab\n");
11710 d.caret = d.source.find('b').unwrap(); // between `a` and `b`
11711 d.backspace();
11712 assert_eq!(d.source, "- b\n");
11713 }
11714
11715 #[test]
11716 fn backspace_at_a_heading_start_strips_the_marker() {
11717 // The `# ` is markup the rich view hides, so Backspace over it takes the
11718 // whole marker and leaves a paragraph. Deleting a byte of it instead left
11719 // `#Title` — no longer a heading, with the hash now literal text the user
11720 // never typed and has to delete again.
11721 let mut d = doc_in(View::Wysiwyg, "bsp_head", "## Title\n");
11722 d.caret = d.source.find('T').unwrap(); // right after `## `
11723 d.backspace();
11724 assert_eq!(d.source, "Title\n");
11725 assert_eq!(
11726 d.caret, 0,
11727 "the caret stays with the text it was in front of"
11728 );
11729 }
11730
11731 #[test]
11732 fn backspace_at_a_heading_start_keeps_the_block_around_it() {
11733 // Only the heading's own marker goes — the quote (or list) it sits in is
11734 // untouched, exactly as un-heading it should be.
11735 let mut d = doc_in(View::Wysiwyg, "bsp_head_quote", "> # Title\n");
11736 d.caret = d.source.find('T').unwrap();
11737 d.backspace();
11738 assert_eq!(d.source, "> Title\n");
11739 }
11740
11741 #[test]
11742 fn backspace_at_a_heading_start_takes_its_closing_sequence_too() {
11743 // `# Title #`'s trailing hashes are hidden at the other end; leaving them
11744 // behind would surface the same stray hash the marker delete just avoided.
11745 let mut d = doc_in(View::Wysiwyg, "bsp_head_closed", "# Title #\n");
11746 d.caret = d.source.find('T').unwrap();
11747 d.backspace();
11748 assert_eq!(d.source, "Title\n");
11749 // And it's one edit: a single undo puts the whole heading back.
11750 d.undo();
11751 assert_eq!(d.source, "# Title #\n");
11752 }
11753
11754 #[test]
11755 fn backspace_mid_heading_still_deletes_a_character() {
11756 // The heading behaviour is armed only at the content's start; anywhere
11757 // else Backspace is the ordinary character delete.
11758 let mut d = doc_in(View::Wysiwyg, "bsp_head_mid", "# ab\n");
11759 d.caret = d.source.find('b').unwrap();
11760 d.backspace();
11761 assert_eq!(d.source, "# b\n");
11762 }
11763
11764 #[test]
11765 fn source_view_backspace_still_edits_the_heading_marker_literally() {
11766 // In source view the `# ` is text on the screen the user is deleting a
11767 // byte of, so it keeps its literal meaning — the same split the list
11768 // ladder and Enter draw between the two views.
11769 let mut d = doc_with("bsp_head_src", "# Title\n");
11770 d.caret = d.source.find('T').unwrap();
11771 d.backspace();
11772 assert_eq!(d.source, "#Title\n");
11773 }
11774
11775 #[test]
11776 fn outdent_unnests_an_ordered_item_in_one_press() {
11777 // Shift+Tab gives back exactly the marker width the indent added, so a
11778 // nested ordered item unnests in a single press, and the flattened list
11779 // renumbers back to a clean 1, 2, 3.
11780 let mut d = doc_with("outdent_ord", "1. a\n 2. b\n3. c\n");
11781 d.caret = d.source.find('b').unwrap();
11782 d.outdent();
11783 assert_eq!(d.source, "1. a\n2. b\n3. c\n");
11784 let lists = d
11785 .nodes()
11786 .iter()
11787 .filter(|n| n.kind == Kind::OrderedList)
11788 .count();
11789 assert_eq!(lists, 1, "back to one flat list");
11790 }
11791
11792 #[test]
11793 fn table_insert_row_adds_a_row_below_the_caret() {
11794 let mut d = doc_with("tbl_ins_row", "| a | b |\n| --- | --- |\n| 1 | 2 |\n");
11795 d.caret = d.source.find('1').unwrap(); // in the body row
11796 d.table_insert_row(true);
11797 assert_eq!(d.source, "| a | b |\n| --- | --- |\n| 1 | 2 |\n| | |\n");
11798 }
11799
11800 #[test]
11801 fn table_insert_and_delete_column_at_the_caret() {
11802 let mut d = doc_with("tbl_col", "| a | b |\n| --- | --- |\n| 1 | 2 |\n");
11803 d.caret = d.source.find('a').unwrap(); // column 0
11804 d.table_insert_column(true); // add a column to the right of `a`
11805 assert_eq!(
11806 d.source,
11807 "| a | | b |\n| --- | --- | --- |\n| 1 | | 2 |\n"
11808 );
11809 d.caret = d.source.find('b').unwrap(); // now the third column
11810 d.table_delete_column();
11811 assert_eq!(d.source, "| a | |\n| --- | --- |\n| 1 | |\n");
11812 }
11813
11814 // ── ragged formats ───────────────────────────────────────────────────────
11815 // No format spells every gesture. HTML writes the inline marks as a tag pair
11816 // and no heading, list, quote or link; Markdown spells five of the eight
11817 // marks — the highlight only because leaf parses with `highlight`, which is
11818 // why the question is asked with the extensions; djot spells all eight and
11819 // no in-cell break. leaf asks twig per
11820 // gesture (`Doc::supports`) and refuses at the door, rather than letting each
11821 // op discover the fact on its own — one of them didn't.
11822
11823 /// An HTML document in the rich view, ready for a gesture.
11824 fn html_doc(body: &str) -> Doc {
11825 let mut d = Doc::from_source(body.to_string(), Format::Html).unwrap();
11826 d.view = View::Wysiwyg;
11827 d.build_visual(80);
11828 d
11829 }
11830
11831 #[test]
11832 fn a_table_gesture_leaves_an_html_table_alone() {
11833 // The regression this guard exists for. twig's table editor consults no
11834 // `Syntax` table — it spells a grid, not a delimiter — so it rebuilt an
11835 // HTML `<table>` as a *pipe table* and reported success: the whole
11836 // element replaced by `| a | b |`, silently, on one press of a toolbar
11837 // button. Every grid op went the same way.
11838 let src = "<table><tr><td>a</td><td>b</td></tr><tr><td>c</td><td>d</td></tr></table>\n";
11839 // A table of named operations, which is what it looks like.
11840 #[allow(clippy::type_complexity)]
11841 let ops: [(&str, &dyn Fn(&mut Doc)); 7] = [
11842 ("insert row", &|d: &mut Doc| d.table_insert_row(true)),
11843 ("delete row", &|d: &mut Doc| d.table_delete_row()),
11844 ("insert column", &|d: &mut Doc| d.table_insert_column(true)),
11845 ("delete column", &|d: &mut Doc| d.table_delete_column()),
11846 ("align", &|d: &mut Doc| {
11847 d.table_set_alignment(Alignment::Right)
11848 }),
11849 ("move row", &|d: &mut Doc| d.table_move_row(true)),
11850 ("move column", &|d: &mut Doc| d.table_move_column(true)),
11851 ];
11852 for (name, op) in ops {
11853 let mut d = html_doc(src);
11854 d.caret = d.source.find('a').unwrap();
11855 assert!(d.caret_in_table(), "{name}: the caret really is in a table");
11856 op(&mut d);
11857 assert_eq!(d.source, src, "{name} rewrote an HTML table");
11858 assert!(
11859 !d.dirty,
11860 "{name} marked the document dirty without editing it"
11861 );
11862 assert!(d.status.is_some(), "{name} refused without saying why");
11863 }
11864 }
11865
11866 #[test]
11867 fn the_block_gestures_html_cannot_spell_are_refused_with_a_reason() {
11868 // A task box is a form control in HTML and a footnote has no native
11869 // spelling at all — the two gestures twig 3.5 still spells nothing
11870 // for, now that a quote, a list, a link and an image print through
11871 // its renderer (see the test below).
11872 let src = "<h1>Title</h1>\n<p>Hello world</p>\n<ul><li>one</li></ul>\n";
11873 // A table of named operations, which is what it looks like.
11874 #[allow(clippy::type_complexity)]
11875 let ops: [(&str, &dyn Fn(&mut Doc)); 3] = [
11876 ("task item", &|d: &mut Doc| d.toggle_task_item()),
11877 ("task tick", &|d: &mut Doc| d.toggle_task_checked()),
11878 ("footnote", &|d: &mut Doc| d.insert_footnote()),
11879 ];
11880 for (name, op) in ops {
11881 let mut d = html_doc(src);
11882 let at = d.source.find("Hello").unwrap();
11883 d.caret = at;
11884 d.anchor = Some(at + 5); // a selection, for the ops that want one
11885 op(&mut d);
11886 assert_eq!(d.source, src, "{name} edited an HTML document");
11887 assert!(
11888 !d.dirty,
11889 "{name} marked the document dirty without editing it"
11890 );
11891 let status = d.status.as_deref().unwrap_or("");
11892 assert!(
11893 status.contains("html"),
11894 "{name}: the refusal should name the format, got {status:?}"
11895 );
11896 }
11897 }
11898
11899 #[test]
11900 fn html_spells_a_quote_a_list_a_link_and_an_image_through_the_renderer() {
11901 // twig 3.5: where HTML has no marker alphabet it prints the fresh
11902 // node — a `<blockquote>` around the paragraph, a `<ul>`/`<ol>` with
11903 // the paragraph as its item, an `<a>` or `<img>` over the selection.
11904 // Until then every one of these was a refusal; now each is a real
11905 // edit, which is what the toolbar's capability flags say too.
11906 let src = "<h1>Title</h1>\n<p>Hello world</p>\n<ul><li>one</li></ul>\n";
11907 #[allow(clippy::type_complexity)]
11908 let ops: [(&str, &dyn Fn(&mut Doc), &str); 5] = [
11909 (
11910 "quote",
11911 &|d: &mut Doc| d.toggle_blockquote(),
11912 "<blockquote>",
11913 ),
11914 ("list", &|d: &mut Doc| d.toggle_list(false), "<ul>\n<li>"),
11915 (
11916 "ordered list",
11917 &|d: &mut Doc| d.toggle_list(true),
11918 "<ol>\n<li>",
11919 ),
11920 (
11921 "link",
11922 &|d: &mut Doc| d.insert_link("https://example.dev"),
11923 "<a href=\"https://example.dev\">Hello</a>",
11924 ),
11925 (
11926 "image",
11927 &|d: &mut Doc| d.insert_image("pic.png", "alt"),
11928 "<img alt=\"Hello\" src=\"pic.png\">",
11929 ),
11930 ];
11931 for (name, op, expect) in ops {
11932 let mut d = html_doc(src);
11933 let at = d.source.find("Hello").unwrap();
11934 d.caret = at;
11935 d.anchor = Some(at + 5);
11936 op(&mut d);
11937 assert!(d.source.contains(expect), "{name}: got {:?}", d.source);
11938 assert!(d.dirty, "{name}: a real edit");
11939 assert_eq!(
11940 d.status, None,
11941 "{name}: a supported gesture reports nothing"
11942 );
11943 }
11944 }
11945
11946 #[test]
11947 fn html_spells_a_heading_as_its_tag_pair() {
11948 // twig 3.4 rebuilds a heading or paragraph as its tag pair, attributes
11949 // along — the one block gesture whose HTML shape it can write. So ⌘2
11950 // in an HTML document is a real edit, and ⌘0 takes it back.
11951 let src = "<h1>Title</h1>\n<p>Hello world</p>\n";
11952 let mut d = html_doc(src);
11953 d.caret = d.source.find("Hello").unwrap();
11954 d.toggle_heading(2);
11955 assert_eq!(d.source, "<h1>Title</h1>\n<h2>Hello world</h2>\n");
11956 assert!(d.dirty);
11957 assert_eq!(d.status, None, "a supported gesture reports nothing");
11958 d.toggle_heading(2);
11959 assert_eq!(d.source, src, "the same level again is back to a paragraph");
11960 }
11961
11962 #[test]
11963 fn html_spells_the_inline_marks_and_the_rule() {
11964 // The other half, and why one per-document flag stopped being enough:
11965 // ⌘B in an HTML document writes `<strong>` — the tag the serializer
11966 // already emits and the parser reads straight back as the same mark —
11967 // and the rule button writes an `<hr>`. Refusing these on the old
11968 // "HTML is parse-only" reading would now be leaf's own limitation.
11969 let mut d = html_doc("<p>Hello world</p>\n");
11970 let at = d.source.find("world").unwrap();
11971 d.caret = at;
11972 d.anchor = Some(at + 5);
11973 d.toggle(InlineKind::Strong);
11974 assert_eq!(d.source, "<p>Hello <strong>world</strong></p>\n");
11975 assert!(d.dirty);
11976 assert_eq!(d.status, None, "a supported gesture reports nothing");
11977
11978 // And off again — the toggle reverses, which is the property that makes
11979 // authoring in HTML worth offering rather than a one-way trip.
11980 d.toggle(InlineKind::Strong);
11981 assert_eq!(d.source, "<p>Hello world</p>\n");
11982
11983 let mut d = html_doc("<p>Hello world</p>\n");
11984 d.caret = d.source.find("world").unwrap();
11985 d.insert_thematic_break();
11986 assert!(d.source.contains("<hr>"), "got {:?}", d.source);
11987 }
11988
11989 #[test]
11990 fn a_mark_the_format_cannot_spell_arms_nothing() {
11991 // `toggle` with a collapsed caret doesn't reach twig at all — it arms a
11992 // sticky mark for the next text typed. Guarding only the twig call
11993 // leaves that path live, promising a mark the gesture will not write and
11994 // then swallowing the error inside `insert`.
11995 //
11996 // Markdown carries this, on the superscript now rather than on the
11997 // highlight: `^x^` is text there in any configuration, whereas twig
11998 // 3.3.1 authors `==x==` for an editor holding the `highlight` extension,
11999 // which every leaf document does.
12000 let mut d = doc_with("mark", "Hello world\n");
12001 d.view = View::Wysiwyg;
12002 d.build_visual(80);
12003 d.caret = d.source.find("world").unwrap();
12004 d.toggle(InlineKind::Superscript);
12005 assert!(d.pending_marks.is_empty(), "no mark should be armed");
12006 assert!(d.status.as_deref().unwrap_or("").contains("markdown"));
12007 d.insert("X");
12008 assert_eq!(d.source, "Hello Xworld\n");
12009 }
12010
12011 #[test]
12012 fn markdown_authors_a_highlight_and_a_strikethrough() {
12013 // twig 3.3.1: the two marks Markdown reads and, until it, refused to
12014 // write. `==x==` is authorable because leaf's own `parse_extensions`
12015 // turns `highlight` on — twig will only mint bytes this editor's reparse
12016 // reads back — and `~~x~~` because GFM strikethrough is parsed by
12017 // default, so the refusal there was never right for any leaf document.
12018 for (kind, marked) in [
12019 (InlineKind::Mark, "a ==word== b\n"),
12020 (InlineKind::Delete, "a ~~word~~ b\n"),
12021 ] {
12022 let mut d = doc_with("author_mark", "a word b\n");
12023 d.anchor = Some(2);
12024 d.caret = 6;
12025 d.toggle(kind);
12026 assert_eq!(d.source, marked, "{kind:?}");
12027 assert_eq!(d.status, None, "{kind:?}: a supported gesture is silent");
12028 assert!(d.dirty, "{kind:?}");
12029 // The region stays selected, so the second press reverses it — the
12030 // property that separates authoring from a one-way trip.
12031 d.toggle(kind);
12032 assert_eq!(d.source, "a word b\n", "{kind:?}");
12033 }
12034 }
12035
12036 #[test]
12037 fn an_authored_highlight_reads_back_as_a_mark() {
12038 // The round trip the extension gate exists to protect: what the toggle
12039 // writes, the reparse must read back as a `mark` rather than as two
12040 // literal `=` pairs. A `Role::Mark` glyph is that answer, taken from the
12041 // rebuilt map rather than from the source text.
12042 let mut d = doc_with("mark_roundtrip", "a word b\n");
12043 d.view = View::Wysiwyg;
12044 d.build_visual(80);
12045 d.anchor = Some(2);
12046 d.caret = 6;
12047 d.toggle(InlineKind::Mark);
12048 assert_eq!(d.source, "a ==word== b\n");
12049 d.build_visual(80);
12050 let w = d
12051 .vmap
12052 .rows
12053 .iter()
12054 .flat_map(|r| r.glyphs.iter())
12055 .find(|g| g.ch == 'w')
12056 .expect("the highlighted word");
12057 assert_eq!(w.style.role, crate::Role::Mark(None));
12058 }
12059
12060 #[test]
12061 fn a_highlight_takes_a_colour_changes_it_and_gives_it_back() {
12062 // The three states of one gesture, in the order a palette is pressed:
12063 // an uncoloured highlight takes the prefix, a coloured one has it
12064 // replaced, and `None` takes it away with the space that was part of the
12065 // spelling.
12066 let mut d = doc_with("mark_colour", "a ==word== b\n");
12067 d.caret = d.source.find("word").unwrap();
12068 d.set_mark_color(Some(MarkColor::Red));
12069 assert_eq!(d.source, "a ==🔴 word== b\n");
12070 assert_eq!(d.status, None);
12071 assert!(d.dirty);
12072
12073 d.set_mark_color(Some(MarkColor::Blue));
12074 assert_eq!(d.source, "a ==🔵 word== b\n");
12075
12076 d.set_mark_color(None);
12077 assert_eq!(d.source, "a ==word== b\n");
12078 }
12079
12080 #[test]
12081 fn the_caret_keeps_its_place_in_the_text_across_a_colour() {
12082 // The prefix is written *before* the word, so an offset in the word has
12083 // to ride its width — a caret that stayed put would be a caret that
12084 // walked backwards through the text it was standing in.
12085 let mut d = doc_with("mark_colour_caret", "a ==word== b\n");
12086 let word = d.source.find("word").unwrap();
12087 d.caret = word + 2; // between `wo` and `rd`
12088 d.set_mark_color(Some(MarkColor::Red));
12089 assert_eq!(&d.source[d.caret..d.caret + 2], "rd", "still before `rd`");
12090
12091 // And back the other way when the prefix goes.
12092 d.set_mark_color(None);
12093 assert_eq!(&d.source[d.caret..d.caret + 2], "rd");
12094 }
12095
12096 #[test]
12097 fn the_colour_at_the_caret_is_what_the_palette_lights() {
12098 let mut d = doc_with("mark_colour_read", "a ==🔴 red== and ==plain== b\n");
12099 d.caret = d.source.find("red").unwrap();
12100 assert!(d.caret_in_mark());
12101 assert_eq!(d.mark_color_at_caret(), Some(MarkColor::Red));
12102
12103 d.caret = d.source.find("plain").unwrap();
12104 assert!(d.caret_in_mark(), "a highlight with no colour is still one");
12105 assert_eq!(d.mark_color_at_caret(), None);
12106
12107 d.caret = d.source.find(" and ").unwrap() + 2;
12108 assert!(!d.caret_in_mark());
12109 assert_eq!(d.mark_color_at_caret(), None);
12110 }
12111
12112 #[test]
12113 fn a_colour_without_a_highlight_says_so_and_writes_nothing() {
12114 // The gesture colours a highlight that exists; it does not make one.
12115 // Two presses is the price of a coloured highlight from bare text, and
12116 // the reason is undo — one press that spliced twice would take two
12117 // presses to take back.
12118 let mut d = doc_with("mark_colour_none", "a word b\n");
12119 d.caret = d.source.find("word").unwrap();
12120 d.set_mark_color(Some(MarkColor::Red));
12121 assert_eq!(d.source, "a word b\n");
12122 assert!(d.status.is_some(), "it should say why");
12123 assert!(!d.dirty);
12124
12125 // Clearing where there is nothing to clear is the same refusal, not a
12126 // quiet success — the caret is in no highlight either way.
12127 d.status = None;
12128 d.set_mark_color(None);
12129 assert_eq!(d.source, "a word b\n");
12130 assert!(d.status.is_some());
12131 }
12132
12133 #[test]
12134 fn clearing_an_uncoloured_highlight_is_a_quiet_no_op() {
12135 // twig answers this one *successfully* with a `Change` describing some
12136 // earlier edit, so a caller that trusted the change would jump the caret
12137 // to wherever that was. Core answers it before asking.
12138 let mut d = doc_with("mark_colour_noop", "a ==word== b\n");
12139 d.toggle(InlineKind::Strong); // an earlier edit for a stale change to name
12140 d.caret = d.source.find("word").unwrap();
12141 let (source, caret) = (d.source.clone(), d.caret);
12142 d.set_mark_color(None);
12143 assert_eq!(d.source, source);
12144 assert_eq!(
12145 d.caret, caret,
12146 "the caret must not ride a change that isn't one"
12147 );
12148 assert_eq!(d.status, None, "and it is not an error either");
12149 }
12150
12151 #[test]
12152 fn djot_spells_the_highlight_and_not_its_colour() {
12153 // The reason the palette is its own capability rather than the Highlight
12154 // button's: `{=word=}` is a highlight djot writes happily, and there is
12155 // no djot spelling for a colour on it.
12156 assert!(Capabilities::of(Format::Djot).mark);
12157 assert!(!Capabilities::of(Format::Djot).mark_color);
12158 assert!(Capabilities::of(Format::Markdown).mark_color);
12159
12160 let mut d = Doc::from_source("a {=word=} b\n".into(), Format::Djot).unwrap();
12161 d.caret = d.source.find("word").unwrap();
12162 assert!(
12163 d.caret_in_mark(),
12164 "the caret is in a highlight all the same"
12165 );
12166 d.set_mark_color(Some(MarkColor::Red));
12167 assert_eq!(d.source, "a {=word=} b\n");
12168 assert!(
12169 d.status.as_deref().unwrap_or("").contains("djot"),
12170 "and the refusal names the document's format: {:?}",
12171 d.status
12172 );
12173 }
12174
12175 #[test]
12176 fn a_coloured_highlight_is_one_undo_step_and_reads_back_as_its_colour() {
12177 // The round trip that matters for a palette: the bytes twig writes are
12178 // bytes its own reparse reads back as a colour, so the swatch that was
12179 // pressed is the swatch that lights afterwards.
12180 let mut d = doc_with("mark_colour_undo", "a word b\n");
12181 d.anchor = Some(2);
12182 d.caret = 6;
12183 d.toggle(InlineKind::Mark);
12184 d.caret = d.source.find("word").unwrap();
12185 d.set_mark_color(Some(MarkColor::Green));
12186 assert_eq!(d.source, "a ==🟢 word== b\n");
12187 assert_eq!(d.mark_color_at_caret(), Some(MarkColor::Green));
12188
12189 // One splice, one step: the colour comes off and the highlight stays.
12190 d.undo();
12191 assert_eq!(d.source, "a ==word== b\n");
12192 d.undo();
12193 assert_eq!(d.source, "a word b\n");
12194 }
12195
12196 #[test]
12197 fn every_colour_leaf_names_is_one_twig_writes() {
12198 // The two enums are one vocabulary, and this is what says so: each of
12199 // leaf's colours writes an emoji twig's reparse reads back as *that*
12200 // colour, so `twig_mark_color`'s table cannot quietly pair red with
12201 // orange.
12202 for color in MarkColor::ALL {
12203 let mut d = doc_with("mark_colour_all", "a ==word== b\n");
12204 d.caret = d.source.find("word").unwrap();
12205 d.set_mark_color(Some(color));
12206 assert_eq!(d.status, None, "{color:?}");
12207 assert_eq!(d.mark_color_at_caret(), Some(color), "{color:?}");
12208 }
12209 }
12210
12211 #[test]
12212 fn a_fresh_highlight_takes_a_colour_without_moving_the_caret_first() {
12213 // The two presses a coloured highlight is made of, in the state the
12214 // first one leaves: `toggle` selects the whole `==word==` and puts the
12215 // caret one past the closing `==`, which is *not* in the mark. Asking at
12216 // the caret alone would refuse to colour the highlight just written —
12217 // the selection's start is what answers.
12218 let mut d = doc_with("mark_colour_fresh", "a word b\n");
12219 d.anchor = Some(2);
12220 d.caret = 6;
12221 d.toggle(InlineKind::Mark);
12222 assert_eq!(d.source, "a ==word== b\n");
12223 assert_eq!(d.caret, 10, "the caret twig leaves, past the closing `==`");
12224
12225 assert!(d.caret_in_mark(), "the selected highlight is the one meant");
12226 d.set_mark_color(Some(MarkColor::Yellow));
12227 assert_eq!(d.source, "a ==🟡 word== b\n");
12228 assert_eq!(d.status, None);
12229 }
12230
12231 #[test]
12232 fn one_press_highlights_a_selection_and_colours_it() {
12233 // What a toolbar swatch means over a plain selection, and the undo it
12234 // has to have: one press, one step. Two steps would leave an uncoloured
12235 // highlight behind on the way back, which is a state the author never
12236 // asked for and never saw.
12237 let mut d = doc_with("highlight_one", "a word b\n");
12238 d.anchor = Some(2);
12239 d.caret = 6;
12240 d.highlight(Some(MarkColor::Purple));
12241 assert_eq!(d.source, "a ==\u{1F7E3} word== b\n");
12242 assert_eq!(d.status, None);
12243
12244 d.undo();
12245 assert_eq!(d.source, "a word b\n", "one press, one undo");
12246 }
12247
12248 #[test]
12249 fn one_press_on_an_existing_highlight_only_recolours_it() {
12250 // The other half: inside a highlight there is nothing to make, so the
12251 // compound is the plain gesture and the text is untouched.
12252 let mut d = doc_with("highlight_recolour", "a ==\u{1F534} word== b\n");
12253 d.caret = d.source.find("word").unwrap();
12254 d.highlight(Some(MarkColor::Blue));
12255 assert_eq!(d.source, "a ==\u{1F535} word== b\n");
12256 d.undo();
12257 assert_eq!(d.source, "a ==\u{1F534} word== b\n", "the highlight stays");
12258 }
12259
12260 #[test]
12261 fn one_press_with_no_colour_over_a_selection_just_highlights_it() {
12262 // `None` means "no colour", and over bare text that is the Highlight
12263 // button's own job. The fold must not happen here — there is no second
12264 // splice, and folding would take the *previous* edit into this one.
12265 let mut d = doc_with("highlight_none", "a word b and more\n");
12266 d.caret = d.source.find("more").unwrap() + 4; // after "more"
12267 d.insert("!"); // an earlier edit for a wrong fold to swallow
12268 d.anchor = Some(2);
12269 d.caret = 6;
12270 d.highlight(None);
12271 assert_eq!(d.source, "a ==word== b and more!\n");
12272
12273 d.undo();
12274 assert_eq!(
12275 d.source, "a word b and more!\n",
12276 "only the highlight came off"
12277 );
12278 d.undo();
12279 assert_eq!(
12280 d.source, "a word b and more\n",
12281 "and the edit before it survived"
12282 );
12283 }
12284
12285 #[test]
12286 fn one_press_at_a_bare_caret_in_no_highlight_writes_nothing() {
12287 // `toggle` at a collapsed caret arms a mark for text not yet typed, and
12288 // a colour cannot be armed with it — so the compound declines rather
12289 // than leaving half a promise.
12290 let mut d = doc_with("highlight_bare", "a word b\n");
12291 d.caret = 4;
12292 d.highlight(Some(MarkColor::Red));
12293 assert_eq!(d.source, "a word b\n");
12294 assert!(d.pending_marks.is_empty(), "and nothing armed");
12295 assert!(d.status.is_some());
12296 }
12297
12298 #[test]
12299 fn a_read_only_document_takes_no_colour() {
12300 let mut d = doc_with("mark_colour_ro", "a ==word== b\n");
12301 d.caret = d.source.find("word").unwrap();
12302 d.set_read_only(true);
12303 d.set_mark_color(Some(MarkColor::Red));
12304 assert_eq!(d.source, "a ==word== b\n");
12305 }
12306
12307 #[test]
12308 fn a_sticky_highlight_wraps_the_next_typed_text_in_markdown() {
12309 // The other door into `toggle`: no selection, so nothing reaches twig
12310 // until `insert` realises the armed mark. It is armed now — the guard
12311 // above asks `Doc::supports`, which asks with the extensions — and what
12312 // it writes is the same `==…==`.
12313 let mut d = doc_with("sticky_mark", "xy\n");
12314 d.caret = 1;
12315 d.toggle(InlineKind::Mark);
12316 assert!(d.pending_marks.contains(InlineKind::Mark));
12317 d.insert("Z");
12318 assert_eq!(d.source, "x==Z==y\n");
12319 }
12320
12321 #[test]
12322 fn html_documents_still_take_typed_text() {
12323 // The guard covers *markup* gestures and must not touch plain editing:
12324 // twig's splicer is language-neutral, and typing into an HTML document
12325 // is the thing that does work today.
12326 let mut d = html_doc("<p>Hello world</p>\n");
12327 d.caret = d.source.find("world").unwrap();
12328 d.insert("big ");
12329 assert_eq!(d.source, "<p>Hello big world</p>\n");
12330 assert!(d.dirty);
12331 d.backspace();
12332 assert_eq!(d.source, "<p>Hello bigworld</p>\n");
12333 d.undo();
12334 d.undo();
12335 assert_eq!(d.source, "<p>Hello world</p>\n");
12336 }
12337
12338 #[test]
12339 fn authorable_is_the_coarse_question_and_capabilities_the_useful_one() {
12340 // `authorable` only separates "there is a door in" from "there is not",
12341 // and HTML is on the near side of that line — which is exactly why a
12342 // toolbar must not be built from it.
12343 let html = Doc::from_source("<p>x</p>\n".into(), Format::Html).unwrap();
12344 assert!(html.authorable());
12345 assert!(
12346 !Doc::from_source("<r>x</r>".into(), Format::Xml)
12347 .unwrap()
12348 .authorable()
12349 );
12350
12351 let caps = html.capabilities();
12352 assert!(caps.bold && caps.italic && caps.code && caps.mark);
12353 assert!(caps.thematic_break && caps.cell_line_break);
12354 // A heading is a tag pair twig rebuilds (3.4), and since 3.5 so are a
12355 // quote, a list, a code block's language, a link and an image — each
12356 // printed as a fresh node where HTML has no marker to rewrite. A task
12357 // box is a form control and a footnote has no spelling, so those two
12358 // are what keeps the record ragged.
12359 assert!(caps.heading && caps.blockquote && caps.bullet_list);
12360 assert!(caps.link && caps.image && caps.code_language);
12361 assert!(!caps.task && !caps.footnote);
12362 // The one flag that isn't twig's answer: an HTML `<table>` is a grid
12363 // twig's table editor would happily re-emit as `| a | b |`.
12364 assert!(!caps.table);
12365
12366 // The two lightweight formats spell everything leaf offers — and still
12367 // differ from each other, which is the other half of why one boolean
12368 // can't serve.
12369 for fmt in [Format::Markdown, Format::Djot] {
12370 let caps = Capabilities::of(fmt);
12371 assert!(
12372 caps.heading && caps.blockquote && caps.ordered_list,
12373 "{fmt:?}"
12374 );
12375 assert!(
12376 caps.task && caps.link && caps.image && caps.table,
12377 "{fmt:?}"
12378 );
12379 }
12380 // Both spell the highlight and the strikethrough: djot natively, and
12381 // Markdown because `Capabilities` asks with `parse_extensions` rather
12382 // than with twig's defaults — `==x==` is text under those, and a mark
12383 // under the `highlight` leaf always parses with.
12384 for fmt in [Format::Markdown, Format::Djot] {
12385 let caps = Capabilities::of(fmt);
12386 assert!(caps.mark && caps.strike, "{fmt:?}");
12387 }
12388 // What still separates them, now that the highlight doesn't: djot has
12389 // no in-cell break, and Markdown spells neither of the scripts.
12390 assert!(Capabilities::of(Format::Djot).superscript);
12391 assert!(!Capabilities::of(Format::Markdown).superscript);
12392 assert!(Capabilities::of(Format::Markdown).cell_line_break);
12393 assert!(!Capabilities::of(Format::Djot).cell_line_break);
12394
12395 // A parse-only format answers no to every one of them, so the coarse
12396 // predicate and the record agree there.
12397 let caps = Capabilities::of(Format::Xml);
12398 assert!(!caps.bold && !caps.heading && !caps.table && !caps.thematic_break);
12399 }
12400
12401 #[test]
12402 fn a_refused_gesture_says_so_where_twig_would_have_said_it() {
12403 // The guard exists to name the *document's* format rather than twig's
12404 // internals, so the message has to survive being one leaf writes itself.
12405 // Checked against a gesture twig also refuses, since that is the pair
12406 // most at risk of drifting apart — the task box, once the code
12407 // language stopped being one (twig 3.5).
12408 let mut d = html_doc("<p>Hello</p>\n");
12409 d.caret = d.source.find("Hello").unwrap();
12410 d.toggle_task_item();
12411 assert_eq!(d.status.as_deref(), Some("task: not supported in html"));
12412 assert!(!d.dirty);
12413 }
12414
12415 #[test]
12416 fn table_set_alignment_respells_the_delimiter() {
12417 let mut d = doc_with("tbl_align", "| a | b |\n| --- | --- |\n| 1 | 2 |\n");
12418 d.caret = d.source.find('b').unwrap();
12419 d.table_set_alignment(Alignment::Right);
12420 assert_eq!(d.source, "| a | b |\n| --- | ---: |\n| 1 | 2 |\n");
12421 }
12422
12423 #[test]
12424 fn each_empty_table_cell_has_its_own_editable_home() {
12425 // Regression: an empty cell has no twig content_span, so both cells of a
12426 // `| | |` row collapsed onto the row's start (before the first `│`).
12427 // Typing there inserted *before* the table (`hello| | |`); nav couldn't
12428 // tell the cells apart. Each empty cell must now have a distinct home
12429 // inside it.
12430 let mut d = wysiwyg_doc("tbl_empty", "| a | b |\n| --- | --- |\n| | |\n");
12431 let (c0, c1) = {
12432 let cells = &d.vmap.tables[0].grid[1].cells;
12433 (cells[0].start, cells[1].start)
12434 };
12435 assert!(
12436 c0 < c1,
12437 "the two empty cells have distinct homes: {c0} < {c1}"
12438 );
12439 d.caret = c0;
12440 d.insert("x");
12441 assert_eq!(
12442 d.source, "| a | b |\n| --- | --- |\n| x | |\n",
12443 "typed inside the cell"
12444 );
12445 }
12446
12447 #[test]
12448 fn arrows_step_into_each_empty_table_cell() {
12449 let mut d = wysiwyg_doc("tbl_empty_nav", "| a | b |\n| --- | --- |\n| | |\n");
12450 let (c0, c1) = {
12451 let cells = &d.vmap.tables[0].grid[1].cells;
12452 (cells[0].start, cells[1].start)
12453 };
12454 d.caret = d.source.find('b').unwrap(); // in the header's second cell
12455 let mut seen = std::collections::HashSet::new();
12456 for _ in 0..6 {
12457 d.move_right(false);
12458 seen.insert(d.caret);
12459 }
12460 assert!(
12461 seen.contains(&c0),
12462 "right arrow reaches the first empty cell"
12463 );
12464 assert!(
12465 seen.contains(&c1),
12466 "right arrow reaches the second empty cell"
12467 );
12468 }
12469
12470 #[test]
12471 fn table_op_off_a_table_is_a_no_op_with_a_status() {
12472 let mut d = doc_with("tbl_none", "just text\n");
12473 d.caret = 3;
12474 d.table_insert_row(true);
12475 assert_eq!(d.source, "just text\n", "nothing changed");
12476 assert!(d.status.is_some(), "a status explains why");
12477 assert!(!d.caret_in_table());
12478 }
12479
12480 #[test]
12481 fn enter_in_an_ordered_list_renumbers_the_following_items() {
12482 // Inserting an item mid-list left the source markers stale (`1. 2. 2. 3.`);
12483 // the renumber pass keeps them sequential, matching what the view draws.
12484 let mut d = wysiwyg_doc("enter_renumber", "1. a\n2. b\n3. c\n");
12485 d.caret = d.source.find('a').unwrap() + 1; // end of item a
12486 d.newline();
12487 d.insert("x");
12488 assert_eq!(d.source, "1. a\n2. x\n3. b\n4. c\n");
12489 }
12490
12491 #[test]
12492 fn outdent_with_nothing_to_give_back_records_no_undo_step() {
12493 for view in [View::Source, View::Wysiwyg] {
12494 let mut d = doc_in(view, "outdent_noop", "hello\n");
12495 d.caret = 2;
12496 d.outdent();
12497 assert_eq!(d.source, "hello\n");
12498 assert!(!d.dirty, "a no-op is not a modification");
12499 d.undo();
12500 assert_eq!(
12501 d.status.as_deref(),
12502 Some("nothing to undo"),
12503 "spends no undo step"
12504 );
12505 assert_eq!(d.source, "hello\n");
12506 }
12507 }
12508
12509 #[test]
12510 fn indent_shifts_every_selected_line_and_keeps_them_selected() {
12511 for view in [View::Source, View::Wysiwyg] {
12512 let mut d = doc_in(view, "indent_sel", "one\n\ntwo\n");
12513 d.anchor = Some(0);
12514 d.caret = 7; // through "two"
12515 d.indent();
12516 assert_eq!(
12517 d.source, " one\n\n two\n",
12518 "the blank line keeps no trailing pad"
12519 );
12520 // Selected, so a second Tab lands on the same lines rather than on
12521 // whatever the shifted offsets now cover.
12522 assert_eq!(d.selection(), Some((0, 12)));
12523 d.indent();
12524 assert_eq!(d.source, " one\n\n two\n");
12525 }
12526 }
12527
12528 #[test]
12529 fn outdent_takes_what_each_line_has_and_leaves_the_rest_alone() {
12530 for view in [View::Source, View::Wysiwyg] {
12531 let mut d = doc_in(view, "outdent_sel", " two\n one\nnone\n");
12532 d.anchor = Some(0);
12533 d.caret = 15;
12534 d.outdent();
12535 assert_eq!(d.source, "two\none\nnone\n");
12536 }
12537 }
12538
12539 #[test]
12540 fn a_tab_undoes_as_one_step_however_many_lines_it_moved() {
12541 for view in [View::Source, View::Wysiwyg] {
12542 let mut d = doc_in(view, "indent_undo", "one\n\ntwo\n");
12543 d.anchor = Some(0);
12544 d.caret = 7;
12545 d.indent();
12546 assert_eq!(d.source, " one\n\n two\n");
12547 d.undo();
12548 assert_eq!(d.source, "one\n\ntwo\n", "one step, not one per line");
12549 assert_eq!(
12550 d.selection(),
12551 Some((0, 7)),
12552 "with the selection it was aimed at"
12553 );
12554 d.redo();
12555 assert_eq!(d.source, " one\n\n two\n");
12556 assert_eq!(
12557 d.selection(),
12558 Some((0, 12)),
12559 "redo replays the caret the indent placed, not the one splice left"
12560 );
12561 }
12562 }
12563
12564 #[test]
12565 fn vertical_motion_keeps_the_column() {
12566 let mut d = doc_with("move", "abcd\nef\n");
12567 d.caret = 3; // "abc|d" on row 0, col 3
12568 d.move_down(false); // row 1 "ef" only has cols 0..2 -> clamps to end
12569 assert_eq!(d.caret, 7); // just after "ef"
12570 }
12571
12572 // ── goal column ──────────────────────────────────────────────────────────
12573
12574 #[test]
12575 fn vertical_motion_goal_column_survives_a_short_line() {
12576 // Regression: re-deriving the column from the clamped position on
12577 // every step permanently forgets it once a short line clamps it.
12578 // Down through "xy" (2 cols) and into "ghijkl" must return to col 4.
12579 let g = |m, f: fn(&mut Doc)| golden("goalcol", m, f);
12580 assert_eq!(
12581 g("abcd|ef\nxy\nghijkl\n", |d| {
12582 d.move_down(false); // clamps to end of "xy"
12583 d.move_down(false); // restores col 4 on the long line
12584 }),
12585 "abcdef\nxy\nghij|kl\n"
12586 );
12587 }
12588
12589 #[test]
12590 fn goal_column_state_is_set_by_vertical_motion_and_cleared_by_horizontal() {
12591 let mut d = doc_with("goalcol_state", "abcdef\nxy\nghijkl\n");
12592 assert_eq!(d.goal_col, None);
12593 d.caret = 4; // row 0, col 4
12594 d.move_down(false); // clamps into "xy"; goal stays the original col
12595 assert_eq!(d.goal_col, Some(4));
12596 assert_eq!(d.caret_pos(), (1, 2));
12597
12598 // A horizontal motion drops the goal column...
12599 d.move_left(false);
12600 assert_eq!(d.goal_col, None);
12601
12602 // ...so the next vertical motion picks up the *new* column (1), not
12603 // the stale one (4).
12604 d.move_down(false);
12605 assert_eq!(d.goal_col, Some(1));
12606 assert_eq!(d.caret_pos(), (2, 1));
12607 }
12608
12609 #[test]
12610 fn editing_clears_the_goal_column() {
12611 let mut d = doc_with("goalcol_edit", "abcdef\nxy\nghijkl\n");
12612 d.caret = 4;
12613 d.move_down(false);
12614 assert_eq!(d.goal_col, Some(4));
12615 d.insert("Z");
12616 assert_eq!(d.goal_col, None);
12617 }
12618
12619 #[test]
12620 fn vertical_motion_on_an_empty_document_is_a_no_op() {
12621 let mut d = doc_with("empty_vert", "");
12622 d.move_down(false);
12623 assert_eq!(d.caret, 0);
12624 d.move_up(false);
12625 assert_eq!(d.caret, 0);
12626 }
12627
12628 // ── the document's edges ─────────────────────────────────────────────────
12629
12630 #[test]
12631 fn vertical_motion_at_the_document_edges_runs_to_them_in_both_views() {
12632 // The reproduction, and the disagreement: Down on the last line ran to
12633 // the end of the document in the source view — by accident, an
12634 // out-of-range row clamping to the end of the string — and did nothing
12635 // whatever in the view leaf opens in. One rule now, in both.
12636 for (view, tag) in VIEWS {
12637 let mut d = doc_in(view, &format!("edge_{tag}"), "abc");
12638 d.caret = 1;
12639 d.move_down(false);
12640 assert_eq!(d.caret, 3, "{tag}: Down on the last line runs to the end");
12641 d.move_up(false);
12642 assert_eq!(d.caret, 0, "{tag}: Up on the first line runs to the start");
12643 }
12644 }
12645
12646 #[test]
12647 fn vertical_motion_at_the_edges_carries_the_column_across_the_lines_between() {
12648 // Down off the bottom is a motion like any other, so it latches a goal
12649 // column — and Up comes back to the column the caret left, not to the
12650 // one the document's end happened to be in.
12651 for (view, tag) in VIEWS {
12652 let gap = if view == View::Source { "\n" } else { "\n\n" };
12653 let src = format!("abcdef{gap}ghijkl");
12654 let mut d = doc_in(view, &format!("edge_goal_{tag}"), &src);
12655 d.caret = 2; // row 0, col 2
12656 d.move_down(false);
12657 assert_eq!(d.caret_pos().1, 2, "{tag}: Down keeps the column");
12658 d.move_down(false);
12659 assert_eq!(
12660 d.caret,
12661 src.len(),
12662 "{tag}: Down off the bottom reaches the end"
12663 );
12664 d.move_up(false);
12665 assert_eq!(
12666 d.caret_pos().1,
12667 2,
12668 "{tag}: Up returns to the column Down left"
12669 );
12670 }
12671 }
12672
12673 #[test]
12674 fn vertical_motion_with_nowhere_to_go_latches_no_goal_column() {
12675 // `goal_col.get_or_insert` ran *before* the early return at row 0, so an
12676 // Up that did nothing still armed a goal column, and the next Down aimed
12677 // at a column the caret had never been in.
12678 for (view, tag) in VIEWS {
12679 let mut d = doc_in(view, &format!("noop_goal_{tag}"), "abc\n\ndef");
12680 d.caret = 0;
12681 d.move_up(false);
12682 assert_eq!(d.caret, 0, "{tag}: already at the start");
12683 assert_eq!(d.goal_col, None, "{tag}: a no-op Up latched a goal column");
12684
12685 d.caret = d.source.len();
12686 d.move_down(false);
12687 assert_eq!(d.caret, d.source.len(), "{tag}: already at the end");
12688 assert_eq!(
12689 d.goal_col, None,
12690 "{tag}: a no-op Down latched a goal column"
12691 );
12692 }
12693 }
12694
12695 // ── soft wrap ────────────────────────────────────────────────────────────
12696 // Every other test here builds the map at 80 columns, where no fixture is
12697 // long enough to fold. A wrap is where one offset belongs to two rows at
12698 // once, and it broke everything that asks the caret what row it is on.
12699
12700 /// The wrapped fixture these cases share, folded at 12 columns into
12701 /// `one two ` / `three four ` / `five six ` / `seven eight`.
12702 fn wrapped_doc(name: &str) -> Doc {
12703 let mut d = wysiwyg_doc(name, "one two three four five six seven eight");
12704 d.build_visual(12);
12705 d
12706 }
12707
12708 #[test]
12709 fn home_and_end_work_from_a_wrapped_row() {
12710 // The reproduction: offset 19 is the `f` of "five", the first character
12711 // of the third row — and also the offset the second row ends at. It
12712 // resolved to the *second* row, so End aimed at a place the caret was
12713 // already in and did nothing, while Home walked backwards onto a row the
12714 // caret had left.
12715 let mut d = wrapped_doc("wrap_home_end");
12716 d.caret = 19;
12717 assert_eq!(
12718 d.caret_pos(),
12719 (2, 0),
12720 "the wrap boundary opens the third row"
12721 );
12722 d.move_end(false);
12723 assert_eq!(d.caret, 27, "End stalled at the wrap boundary");
12724 d.move_home(false);
12725 assert_eq!(d.caret, 19, "Home left the row the caret was on");
12726 }
12727
12728 #[test]
12729 fn end_of_a_wrapped_row_stays_put_when_pressed_again() {
12730 // The row's end is the last offset that is only ever its own: the offset
12731 // past it opens the row below, and aiming there would send a second
12732 // press on to *that* row's end, and a third to the next — End walking
12733 // down the paragraph rather than sitting where it landed.
12734 let mut d = wrapped_doc("wrap_end_twice");
12735 d.caret = 12; // inside "three", on the second row
12736 d.move_end(false);
12737 assert_eq!(
12738 d.caret, 18,
12739 "the end of `three four`, before the space the wrap ate"
12740 );
12741 assert_eq!(d.caret_pos(), (1, 10), "drawn on the row it is the end of");
12742 d.move_end(false);
12743 assert_eq!(d.caret, 18, "a second End moved the caret");
12744 d.move_home(false);
12745 assert_eq!(d.caret, 8, "Home takes the row's own start");
12746 }
12747
12748 #[test]
12749 fn vertical_motion_crosses_a_soft_wrap() {
12750 // Down aimed at the row below's column 0, an offset that resolved *up*
12751 // to the row above's end — so it landed on the offset it already had and
12752 // the caret could never leave a paragraph's first row.
12753 let mut d = wrapped_doc("wrap_down");
12754 d.caret = 0;
12755 for (want, row) in [(8, 1), (19, 2), (28, 3), (39, 3)] {
12756 d.move_down(false);
12757 assert_eq!(d.caret, want, "Down stalled");
12758 assert_eq!(d.caret_pos().0, row, "Down landed on the wrong row");
12759 }
12760 d.move_down(false);
12761 assert_eq!(d.caret, 39, "the last row's Down runs to the end and stops");
12762
12763 // ...and back up, one row per press. The goal column is the end of the
12764 // last row, past every other row's width, so each press clamps to the
12765 // row's own last offset rather than to the one that opens the next.
12766 let mut d = wrapped_doc("wrap_up");
12767 d.caret = 39;
12768 for (want, pos) in [(27, (2, 8)), (18, (1, 10)), (7, (0, 7)), (0, (0, 0))] {
12769 d.move_up(false);
12770 assert_eq!(d.caret, want, "Up stalled");
12771 assert_eq!(d.caret_pos(), pos, "Up landed on the wrong row");
12772 }
12773 }
12774
12775 #[test]
12776 fn a_kill_on_a_wrapped_row_stops_at_the_row() {
12777 // The kills take the same line Home and End do, so in WYSIWYG they take
12778 // the visual row — and a soft wrap has no newline in it to delete, so
12779 // nothing is joined by reaching the end of one.
12780 let mut d = wrapped_doc("wrap_kill");
12781 d.caret = 19; // the `f` of "five", opening the third row
12782 d.delete_to_line_end();
12783 // The space the wrap ate goes with the row it was drawn on: sparing it
12784 // would leave "four seven", two spaces where the row had been.
12785 assert_eq!(d.source, "one two three four seven eight");
12786
12787 // Backwards from the row's last caret position — which is *before* that
12788 // space, so this one survives, being on the far side of the caret.
12789 let mut d = wrapped_doc("wrap_kill_back");
12790 d.caret = 27;
12791 d.delete_to_line_start();
12792 assert_eq!(d.source, "one two three four seven eight");
12793 }
12794
12795 // ── document start / end ────────────────────────────────────────────────
12796
12797 #[test]
12798 fn move_doc_start_and_end_jump_to_the_edges() {
12799 let g = |m, f: fn(&mut Doc)| golden("doc_edges", m, f);
12800 assert_eq!(
12801 g("hello\nwor|ld\n", |d| d.move_doc_start(false)),
12802 "|hello\nworld\n"
12803 );
12804 assert_eq!(
12805 g("hel|lo\nworld\n", |d| d.move_doc_end(false)),
12806 "hello\nworld\n|"
12807 );
12808 // Already at the edge: a no-op.
12809 assert_eq!(g("|hello\n", |d| d.move_doc_start(false)), "|hello\n");
12810 assert_eq!(g("hello|\n", |d| d.move_doc_end(false)), "hello\n|");
12811 }
12812
12813 #[test]
12814 fn move_doc_start_and_end_extend_the_selection() {
12815 assert_eq!(
12816 golden("doc_edges_ext_end", "hello wor|ld\n", |d| d
12817 .move_doc_end(true)),
12818 "hello wor[ld\n|]"
12819 );
12820 assert_eq!(
12821 golden("doc_edges_ext_start", "hello wor|ld\n", |d| d
12822 .move_doc_start(true)),
12823 "[|hello wor]ld\n"
12824 );
12825 }
12826
12827 #[test]
12828 fn move_doc_start_and_end_on_an_empty_document_are_a_no_op() {
12829 let mut d = doc_with("empty_edges", "");
12830 d.move_doc_end(false);
12831 assert_eq!(d.caret, 0);
12832 d.move_doc_start(false);
12833 assert_eq!(d.caret, 0);
12834 }
12835
12836 // ── arrow collapses an active selection ─────────────────────────────────
12837
12838 #[test]
12839 fn arrow_collapses_selection_to_its_near_edge() {
12840 let mut d = doc_with("collapse", "hello world\n");
12841
12842 // Forward selection (anchor before caret): Right -> end, Left -> start.
12843 d.anchor = Some(2);
12844 d.caret = 7;
12845 d.move_right(false);
12846 assert_eq!((d.caret, d.anchor), (7, None));
12847
12848 d.anchor = Some(2);
12849 d.caret = 7;
12850 d.move_left(false);
12851 assert_eq!((d.caret, d.anchor), (2, None));
12852
12853 // Backward selection (anchor after caret): edges are the same
12854 // regardless of which end the caret started on.
12855 d.anchor = Some(7);
12856 d.caret = 2;
12857 d.move_right(false);
12858 assert_eq!((d.caret, d.anchor), (7, None));
12859
12860 d.anchor = Some(7);
12861 d.caret = 2;
12862 d.move_left(false);
12863 assert_eq!((d.caret, d.anchor), (2, None));
12864 }
12865
12866 #[test]
12867 fn arrow_with_extend_keeps_growing_the_selection() {
12868 let mut d = doc_with("collapse_extend", "hello world\n");
12869 d.anchor = Some(2);
12870 d.caret = 7;
12871 d.move_right(true); // extend: no collapse, caret steps one further
12872 assert_eq!((d.caret, d.anchor), (8, Some(2)));
12873 }
12874
12875 #[test]
12876 fn arrow_without_a_selection_moves_one_character_as_before() {
12877 let mut d = doc_with("no_collapse", "hello\n");
12878 d.caret = 2;
12879 d.move_right(false);
12880 assert_eq!(d.caret, 3);
12881 d.move_left(false);
12882 assert_eq!(d.caret, 2);
12883 }
12884
12885 /// Press Right until it stops, collecting the offsets walked through. Every
12886 /// caret bug in the WYSIWYG view shows up here as a walk that ends early:
12887 /// two stops sharing one source offset can't be moved between, so the caret
12888 /// stalls on the first of them and the walk never reaches the rest.
12889 fn walk_right(d: &mut Doc) -> Vec<usize> {
12890 let mut seen = vec![d.caret];
12891 for _ in 0..2000 {
12892 let before = d.caret;
12893 d.move_right(false);
12894 if d.caret == before {
12895 break;
12896 }
12897 seen.push(d.caret);
12898 }
12899 seen
12900 }
12901
12902 #[test]
12903 fn the_caret_crosses_a_soft_break() {
12904 // A newline inside a paragraph is a `soft_break`, which twig gives no
12905 // span of its own — the space it renders as used to borrow the offset of
12906 // the character before it, and a caret can't move without changing
12907 // offset. Right must walk clean off the end of the first line.
12908 let mut d = wysiwyg_doc("soft_break_walk", "one two\nthree four\n");
12909 d.caret = 0;
12910 let seen = walk_right(&mut d);
12911 assert_eq!(seen, (0..=18).collect::<Vec<_>>(), "walk stalled: {seen:?}");
12912 }
12913
12914 #[test]
12915 fn line_flow_preserve_resplits_the_map_and_defaults_to_fold() {
12916 // The paragraph holds one soft break. Folded (the default) it lays out as
12917 // a single reflowed row; Preserve re-lays it as a row per source line.
12918 // The setter must invalidate the cached map for the change to show, and
12919 // again on the way back — so a round trip returns to the folded layout.
12920 let mut d = wysiwyg_doc("line_flow", "one two\nthree four\n");
12921 assert_eq!(d.line_flow(), LineFlow::Fold, "fold is the default");
12922 d.build_visual(80);
12923 assert_eq!(d.vmap.num_rows(), 1, "fold: one flowing row");
12924
12925 d.set_line_flow(LineFlow::Preserve);
12926 d.build_visual(80);
12927 assert_eq!(d.vmap.num_rows(), 2, "preserve: a row per source line");
12928
12929 d.set_line_flow(LineFlow::Fold);
12930 d.build_visual(80);
12931 assert_eq!(d.vmap.num_rows(), 1, "fold again: back to one row");
12932 }
12933
12934 #[test]
12935 fn the_caret_still_crosses_a_preserved_soft_break() {
12936 // Preserve renders the soft break as a row boundary rather than a space,
12937 // but the caret must still reach every offset — the break's own offset is
12938 // the first row's end stop, so Right walks clean off the end of line one
12939 // onto line two, exactly as it does when the break is folded.
12940 let mut d = wysiwyg_doc("preserve_walk", "one two\nthree four\n");
12941 d.set_line_flow(LineFlow::Preserve);
12942 d.build_visual(80);
12943 d.caret = 0;
12944 let seen = walk_right(&mut d);
12945 assert_eq!(seen, (0..=18).collect::<Vec<_>>(), "walk stalled: {seen:?}");
12946 }
12947
12948 #[test]
12949 fn the_caret_walks_a_code_block() {
12950 // Every glyph of a code block used to map to the block's start, so the
12951 // whole block was a single offset and the caret couldn't move inside it.
12952 let src = "```rust\nlet x = 1;\nfn f() {}\n```\n";
12953 let mut d = wysiwyg_doc("code_walk", src);
12954 d.caret = 0;
12955 let seen = walk_right(&mut d);
12956 // The fences are markup: hidden, and no caret stop. The code between
12957 // them is reached a character at a time.
12958 let code = src.find("let").unwrap()..src.find("\n```").unwrap();
12959 for off in code.clone() {
12960 assert!(seen.contains(&off), "offset {off} unreachable: {seen:?}");
12961 }
12962 assert!(seen.contains(&code.end), "no stop after the last line");
12963 }
12964
12965 #[test]
12966 fn the_caret_walks_an_indented_code_block() {
12967 // An indented block's text has the four-space indent stripped, so it
12968 // isn't a verbatim slice and its lines have to be re-found. The caret
12969 // lands on the code, never in the indent.
12970 let src = " indented\n code\n";
12971 let mut d = wysiwyg_doc("indent_code_walk", src);
12972 d.caret = 0;
12973 let seen = walk_right(&mut d);
12974 assert!(seen.contains(&src.find("indented").unwrap()));
12975 assert!(seen.contains(&src.find("code").unwrap()));
12976 assert!(
12977 !seen.contains(&0) || seen[0] == 0,
12978 "the caret starts where it was put"
12979 );
12980 // Nothing in the stripped indent is a stop.
12981 for off in [1, 2, 3] {
12982 assert!(!seen.contains(&off), "landed in the indent at {off}");
12983 }
12984 }
12985
12986 #[test]
12987 fn the_caret_leaves_a_tight_heading() {
12988 // "# H" with text directly under it: the heading row's end and the
12989 // separator row's end are the same offset. Right used to find the
12990 // separator's copy, set the caret to where it already was, and stop.
12991 let mut d = wysiwyg_doc("tight_heading_walk", "# H\ntext\n");
12992 d.caret = 2; // the "H"
12993 let seen = walk_right(&mut d);
12994 assert!(
12995 seen.len() > 2,
12996 "Right stalled at the heading's end: {seen:?}"
12997 );
12998 assert!(
12999 seen.contains(&8),
13000 "never reached the end of \"text\": {seen:?}"
13001 );
13002 }
13003
13004 #[test]
13005 fn the_caret_skips_the_gap_between_two_paragraphs() {
13006 // The blank line between two paragraphs is the boundary itself. The
13007 // caret used to be able to sit on it, and typing there landed in the
13008 // previous paragraph — "A\n\nB" became "A\nx\nB", one paragraph with a
13009 // soft break, so the text visibly snapped back up.
13010 let mut d = wysiwyg_doc("gap_skip", "A\n\nB\n");
13011 d.caret = 1; // the end of "A"
13012 d.move_right(false);
13013 assert_eq!(d.caret, 3, "Right stopped in the gap");
13014 d.insert("x");
13015 assert_eq!(d.source, "A\n\nxB\n", "typing landed outside B");
13016 }
13017
13018 #[test]
13019 fn down_from_a_paragraph_lands_on_the_next_one() {
13020 let mut d = wysiwyg_doc("gap_down", "A\n\nB\n");
13021 d.caret = 0;
13022 d.move_down(false);
13023 assert_eq!(d.caret, 3, "Down stopped in the gap");
13024 }
13025
13026 #[test]
13027 fn clicking_the_gap_lands_on_real_text() {
13028 // A click can still *reach* the gap — it's drawn, so it's clickable.
13029 // It has to resolve to somewhere the caret can be.
13030 let mut d = wysiwyg_doc("gap_click", "A\n\nB\n");
13031 d.click(1, 0, false); // the gap row
13032 assert!(
13033 d.caret == 1 || d.caret == 3,
13034 "click left the caret in the gap at {}",
13035 d.caret
13036 );
13037 d.insert("x");
13038 // Either edge of the boundary is a fair place to land; inside it isn't.
13039 assert!(
13040 d.source == "Ax\n\nB\n" || d.source == "A\n\nxB\n",
13041 "click in the gap typed into the boundary: {:?}",
13042 d.source
13043 );
13044 }
13045
13046 #[test]
13047 fn enter_opens_an_empty_paragraph_the_caret_can_type_into() {
13048 // Enter inserts a paragraph break, which leaves a blank line spare on
13049 // either side of a new one. That middle line is a real empty paragraph:
13050 // the caret lands there, and typing makes a paragraph rather than
13051 // extending a neighbour.
13052 let mut d = wysiwyg_doc("gap_enter", "A\n\nB\n");
13053 d.caret = 1;
13054 d.newline();
13055 assert_eq!(d.source, "A\n\n\n\nB\n");
13056 d.build_visual(80);
13057 let (row, _) = d.caret_pos();
13058 assert!(
13059 d.vmap.row_is_navigable(row),
13060 "the caret landed on a gap row"
13061 );
13062 d.insert("x");
13063 assert_eq!(
13064 d.source, "A\n\nx\n\nB\n",
13065 "the new paragraph merged into a neighbour"
13066 );
13067 }
13068
13069 #[test]
13070 fn enter_at_the_end_of_the_document_opens_a_paragraph_too() {
13071 let mut d = wysiwyg_doc("gap_eof", "A\n");
13072 d.caret = 1;
13073 d.newline();
13074 d.build_visual(80);
13075 let (row, _) = d.caret_pos();
13076 assert!(
13077 d.vmap.row_is_navigable(row),
13078 "the caret landed on a gap row"
13079 );
13080 d.insert("x");
13081 assert!(
13082 d.source.starts_with("A\n\n") && d.source.contains('x'),
13083 "typing at the end merged into A: {:?}",
13084 d.source
13085 );
13086 }
13087
13088 #[test]
13089 fn triple_click_selects_a_paragraph_across_its_soft_breaks() {
13090 // A paragraph broken over two source lines is one paragraph. Selecting
13091 // it must not stop at the newline inside it — that newline is markup the
13092 // rich-text view exists to hide.
13093 let src = "one two\nthree four\n\nnext\n";
13094 let mut d = wysiwyg_doc("triple_para", src);
13095 d.select_block_at(2);
13096 assert_eq!(
13097 d.selected_text(),
13098 Some("one two\nthree four"),
13099 "stopped at the soft break"
13100 );
13101 }
13102
13103 #[test]
13104 fn the_wheel_can_scroll_away_from_a_caret_that_stays_put() {
13105 // The reader scrolls down past the caret's row. Nothing moved the
13106 // caret, so the view must stay where it was put — the old code revealed
13107 // the caret every frame, which dragged the view straight back and made
13108 // the document unscrollable past the caret.
13109 let mut d = wysiwyg_doc("scroll_free", "a\n\nb\n\nc\n\nd\n\ne\n");
13110 d.caret = 0;
13111 d.follow_caret(0, 3, 9); // first frame: the caret is at the top
13112 d.scroll = 4; // the wheel
13113 d.follow_caret(0, 3, 9);
13114 assert_eq!(
13115 d.scroll, 4,
13116 "the wheel was overruled by a caret that never moved"
13117 );
13118 }
13119
13120 #[test]
13121 fn moving_the_caret_brings_the_view_back_to_it() {
13122 let mut d = wysiwyg_doc("scroll_follow", "a\n\nb\n\nc\n\nd\n\ne\n");
13123 d.caret = 0;
13124 d.follow_caret(0, 3, 9);
13125 d.scroll = 6; // scrolled away
13126 d.move_right(false); // ...and now the caret moves
13127 let (row, _) = d.caret_pos();
13128 d.follow_caret(row, 3, 9);
13129 assert!(
13130 d.scroll <= row && row < d.scroll + 3,
13131 "caret row {row} off screen at scroll {}",
13132 d.scroll
13133 );
13134 }
13135
13136 #[test]
13137 fn scrolling_stops_at_the_last_row() {
13138 let mut d = wysiwyg_doc("scroll_clamp", "a\n\nb\n");
13139 d.caret = 0;
13140 d.follow_caret(0, 3, 3); // a first frame, so the caret isn't "new"
13141 d.scroll = 999; // the wheel, spun hard
13142 d.follow_caret(0, 3, 3);
13143 assert_eq!(d.scroll, 2, "scrolled into the void past the document");
13144 }
13145
13146 #[test]
13147 fn every_cell_of_a_wide_table_is_reachable() {
13148 // A table whose cells are far wider than the surface: the columns are
13149 // cut to fit and the text wraps inside them, so no cell hangs off the
13150 // right edge where the caret can never go.
13151 let src = "| Ingredient | Notes |\n|---|---|\n\
13152 | flour milled coarse | sift it twice before folding it in |\n";
13153 let mut d = wysiwyg_doc("wide_table_walk", src);
13154 d.build_visual(30);
13155 d.caret = 0;
13156 let seen = walk_right(&mut d);
13157 for word in ["Ingredient", "Notes", "coarse", "folding"] {
13158 let at = src.find(word).unwrap();
13159 assert!(seen.contains(&at), "{word:?} at {at} unreachable: {seen:?}");
13160 }
13161 }
13162
13163 // ── view parity ──────────────────────────────────────────────────────────
13164 // `doc_with` pins the source view, so everything above tests a view users
13165 // never start in — `Doc::open` opens in WYSIWYG. These run the motion and
13166 // deletion golden cases through *both*, plus the WYSIWYG cases the two
13167 // can't share: where the source carries markup the rendered text is a
13168 // different string, and the views agreeing would itself be the bug.
13169
13170 const VIEWS: [(View, &str); 2] = [(View::Source, "source"), (View::Wysiwyg, "wysiwyg")];
13171
13172 /// Run `action` in both views on one `|`-marked fixture and assert they
13173 /// agree. Plain prose only: with no markup to hide, WYSIWYG renders the
13174 /// source verbatim, so the two views are looking at the same text and any
13175 /// disagreement is one of them having lost the plot.
13176 fn both_views(name: &str, marked: &str, action: fn(&mut Doc)) -> String {
13177 let (src, caret) = parse_caret(marked);
13178 let run = |view: View, tag: &str| {
13179 let mut d = doc_in(view, &format!("{name}_{tag}"), &src);
13180 d.caret = caret;
13181 action(&mut d);
13182 render_caret(&d)
13183 };
13184 let source = run(VIEWS[0].0, VIEWS[0].1);
13185 let wysiwyg = run(VIEWS[1].0, VIEWS[1].1);
13186 assert_eq!(source, wysiwyg, "the views disagree on {marked:?}");
13187 source
13188 }
13189
13190 #[test]
13191 fn word_motion_agrees_across_the_views_on_plain_prose() {
13192 let g = both_views;
13193 assert_eq!(
13194 g("par_wl", "hello wor|ld", |d| d.move_word_left(false)),
13195 "hello |world"
13196 );
13197 assert_eq!(
13198 g("par_wl2", "hello| world", |d| d.move_word_left(false)),
13199 "|hello world"
13200 );
13201 assert_eq!(
13202 g("par_wr", "hel|lo world", |d| d.move_word_right(false)),
13203 "hello| world"
13204 );
13205 assert_eq!(
13206 g("par_wr2", "hello| world", |d| d.move_word_right(false)),
13207 "hello world|"
13208 );
13209 assert_eq!(
13210 g("par_punct", "|foo.bar", |d| d.move_word_right(false)),
13211 "foo|.bar"
13212 );
13213 assert_eq!(
13214 g("par_ext", "hello |world", |d| d.move_word_right(true)),
13215 "hello [world|]"
13216 );
13217 }
13218
13219 #[test]
13220 fn word_deletion_agrees_across_the_views_on_plain_prose() {
13221 let g = both_views;
13222 assert_eq!(
13223 g("par_db", "hello world|", |d| d.delete_word_back()),
13224 "hello |"
13225 );
13226 assert_eq!(
13227 g("par_df", "hello |world", |d| d.delete_word_forward()),
13228 "hello |"
13229 );
13230 assert_eq!(
13231 g("par_db2", "foo |bar baz", |d| d.delete_word_back()),
13232 "|bar baz"
13233 );
13234 assert_eq!(g("par_utf8", "café |ok", |d| d.delete_word_back()), "|ok");
13235 }
13236
13237 #[test]
13238 fn character_motion_and_deletion_agree_across_the_views_on_plain_prose() {
13239 let g = both_views;
13240 assert_eq!(g("par_r", "he|llo", |d| d.move_right(false)), "hel|lo");
13241 assert_eq!(g("par_l", "he|llo", |d| d.move_left(false)), "h|ello");
13242 assert_eq!(g("par_bs", "hel|lo", |d| d.backspace()), "he|lo");
13243 assert_eq!(g("par_del", "hel|lo", |d| d.delete_forward()), "hel|o");
13244 }
13245
13246 #[test]
13247 fn wysiwyg_motion_steps_a_grapheme_cluster_the_way_the_source_view_does() {
13248 // The reproduction: the stop table was built one stop per `char`, so
13249 // Right parked the caret 4 bytes into a ZWJ sequence — a place the
13250 // source view, which steps by grapheme, can't reach and backspace can't
13251 // survive. The two views must land on the same offset.
13252 let family = "👨👩👧"; // three emoji strung together with joiners: one cluster
13253 for (view, tag) in VIEWS {
13254 let mut d = doc_in(view, &format!("cluster_{tag}"), &format!("a{family}b\n"));
13255 d.caret = 1;
13256 d.move_right(false);
13257 assert_eq!(d.caret, 1 + family.len(), "{tag} parked inside the cluster");
13258
13259 // ...and the edit that used to sever a joiner off the front of it.
13260 d.backspace();
13261 assert_eq!(d.source, "ab\n", "{tag} split the cluster");
13262 assert_eq!(d.caret, 1);
13263 }
13264 }
13265
13266 #[test]
13267 fn wysiwyg_motion_treats_a_combining_accent_as_one_character() {
13268 for (view, tag) in VIEWS {
13269 let mut d = doc_in(view, &format!("combining_{tag}"), "e\u{0301}x\n");
13270 d.caret = 0;
13271 d.move_right(false);
13272 assert_eq!(
13273 d.caret,
13274 "e\u{0301}".len(),
13275 "{tag} stopped on the combining mark"
13276 );
13277 }
13278 }
13279
13280 #[test]
13281 fn no_wysiwyg_motion_can_park_the_caret_inside_a_cluster() {
13282 // The general form: whatever route the caret takes through a document
13283 // full of clusters, it never lands between the codepoints of one — so no
13284 // motion-then-backspace sequence can leave a dangling joiner behind.
13285 use unicode_segmentation::UnicodeSegmentation;
13286
13287 let src = "a👨👩👧b e\u{0301}mo👨👩👧ji\n\nnext 👩🚀 line\n";
13288 let mut d = wysiwyg_doc("cluster_walk", src);
13289 d.caret = 0;
13290 let boundaries: Vec<usize> = src
13291 .grapheme_indices(true)
13292 .map(|(i, _)| i)
13293 .chain(std::iter::once(src.len()))
13294 .collect();
13295 for off in walk_right(&mut d) {
13296 assert!(
13297 boundaries.contains(&off),
13298 "Right stopped at {off}, inside a grapheme cluster"
13299 );
13300 }
13301 }
13302
13303 #[test]
13304 fn wysiwyg_word_motion_stays_out_of_hidden_delimiters() {
13305 // The reproduction: ⌥→ from inside the opening `**` computed its
13306 // boundary over the raw source and landed on byte 8 — inside the
13307 // *closing* `**`, which `caret_pos` draws at column 6, immediately after
13308 // "bold". The caret drew past the bold word and sat inside it.
13309 let mut d = wysiwyg_doc("wys_word_delim", "a **bold** c\n");
13310 d.caret = 2;
13311 d.move_word_right(false);
13312 assert!(
13313 d.vmap.is_stop(d.caret),
13314 "landed at {}, not a caret stop",
13315 d.caret
13316 );
13317 assert_eq!(d.caret, 10, "should land on the space after \"bold\"");
13318 // The rendered row is "a bold c": column 6 is the space just past "bold",
13319 // and now the caret is really there rather than only drawn there.
13320 assert_eq!(d.caret_pos(), (0, 6));
13321
13322 // ...and back again: ⌥← returns to the "b", not into the opening `**`.
13323 d.move_word_left(false);
13324 assert_eq!(d.caret, 4);
13325 assert_eq!(d.caret_pos(), (0, 2));
13326 }
13327
13328 #[test]
13329 fn wysiwyg_word_delete_takes_the_markup_with_the_word() {
13330 // The reproduction: ⌥⌫ from after "bold" walked the raw source, stopped
13331 // inside the closing `**`, and left "a ** c\n" — delimiters with no
13332 // opener. Glyph space covers the word alone, which would leave
13333 // "a **** c": markup wrapped around nothing. The word and the styling
13334 // that was only ever the word's go together.
13335 let mut d = wysiwyg_doc("wys_word_del_back", "a **bold** c\n");
13336 d.caret = 10;
13337 d.delete_word_back();
13338 assert_eq!(d.source, "a c\n");
13339 assert_eq!(d.caret, 2);
13340
13341 let mut d = wysiwyg_doc("wys_word_del_fwd", "a **bold** c\n");
13342 d.caret = 4; // the "b"
13343 d.delete_word_forward();
13344 assert_eq!(d.source, "a c\n");
13345 }
13346
13347 #[test]
13348 fn wysiwyg_word_delete_empties_a_nested_mark_and_a_code_span_too() {
13349 let src = "a ***bold*** c\n";
13350 let mut d = wysiwyg_doc("wys_word_del_nest", src);
13351 d.caret = src.find(" c").unwrap();
13352 d.delete_word_back();
13353 assert_eq!(
13354 d.source, "a c\n",
13355 "the emph inside the strong empties it too"
13356 );
13357
13358 let src = "a `code` c\n";
13359 let mut d = wysiwyg_doc("wys_word_del_code", src);
13360 d.caret = src.find(" c").unwrap();
13361 d.delete_word_back();
13362 assert_eq!(d.source, "a c\n");
13363 }
13364
13365 #[test]
13366 fn wysiwyg_word_delete_keeps_a_mark_that_still_has_text() {
13367 // Only an *emptied* node goes. Take one word of two and the `**` still
13368 // has a job to do — over the word that's left, with the space the delete
13369 // pushed against the opening delimiter moved out in front of it, or the
13370 // run would be no run at all (`** words**` is literal asterisks — see
13371 // the mark-edge rule on `splice`).
13372 let src = "a **two words** c\n";
13373 let mut d = wysiwyg_doc("wys_word_del_partial", src);
13374 d.caret = src.find(" words").unwrap();
13375 d.delete_word_back();
13376 assert_eq!(d.source, "a **words** c\n");
13377 }
13378
13379 #[test]
13380 fn source_view_word_motion_still_walks_the_markup() {
13381 // The other half of the decision: in the source view the `**` are
13382 // characters like any other — they're on the screen, so word motion has
13383 // to stop at them and a word-delete has to leave them behind. Only
13384 // WYSIWYG hides them, so only WYSIWYG steps over them.
13385 let g = |n, m, f: fn(&mut Doc)| golden(n, m, f);
13386 assert_eq!(
13387 g("src_word_motion", "a |**bold** c\n", |d| d
13388 .move_word_right(false)),
13389 "a **bold|** c\n"
13390 );
13391 // The same caret as the WYSIWYG reproduction, and the opposite outcome:
13392 // here "a ** c\n" is right, because `bold**` is what's to the left of it.
13393 assert_eq!(
13394 g("src_word_del", "a **bold**| c\n", |d| d.delete_word_back()),
13395 "a **| c\n"
13396 );
13397 }
13398
13399 #[test]
13400 fn every_wysiwyg_motion_lands_on_a_caret_stop() {
13401 // The single invariant both bugs violated: the caret draws and edits at
13402 // the same place only when it's on a stop. `debug_assert_on_a_stop`
13403 // makes the same claim in-place; this pins it from the outside, over a
13404 // document with every kind of thing the map has to be careful about.
13405 // At two widths: the wide one every other test builds at, where no
13406 // fixture folds, and one narrow enough that they all do. A soft wrap is
13407 // where an offset stops being on exactly one row, and testing only the
13408 // width that never wraps is how the caret came to be pinned at the first
13409 // one Down reached.
13410 let src = "# Title\n\na **bold** e\u{0301}mo👨👩👧ji `x` c\n\n\
13411 - item one\n\n| A | B |\n|---|---|\n| x | y |\n";
13412 // A table of named operations, which is what it looks like.
13413 #[allow(clippy::type_complexity)]
13414 let motions: [(&str, fn(&mut Doc)); 8] = [
13415 ("right", |d| d.move_right(false)),
13416 ("left", |d| d.move_left(false)),
13417 ("word_right", |d| d.move_word_right(false)),
13418 ("word_left", |d| d.move_word_left(false)),
13419 ("down", |d| d.move_down(false)),
13420 ("up", |d| d.move_up(false)),
13421 ("home", |d| d.move_home(false)),
13422 ("end", |d| d.move_end(false)),
13423 ];
13424 for width in [80, 12] {
13425 let mut d = wysiwyg_doc("stop_invariant", src);
13426 d.build_visual(width);
13427 let stops: Vec<usize> = (0..=src.len()).filter(|&o| d.vmap.is_stop(o)).collect();
13428 assert!(stops.len() > 20, "fixture should have plenty of stops");
13429 for start in stops {
13430 for (name, motion) in &motions {
13431 d.caret = start;
13432 d.anchor = None;
13433 motion(&mut d);
13434 assert!(
13435 d.vmap.is_stop(d.caret),
13436 "{name} from {start} at width {width} landed at {} — not a caret stop",
13437 d.caret
13438 );
13439 }
13440 }
13441 }
13442 }
13443
13444 #[test]
13445 fn no_wysiwyg_motion_is_a_dead_end() {
13446 // Down held to the bottom of a document reaches the bottom, and Up held
13447 // to the top reaches the top — from anywhere, at a width that wraps. The
13448 // invariant above says a motion lands somewhere legal; this one says it
13449 // gets somewhere at all, which is what a caret pinned at a wrap boundary
13450 // was quietly failing to do while every assertion around it held.
13451 let src = "# Title\n\none two three four five six seven eight nine ten\n\n\
13452 - item one two three four five\n\nlast\n";
13453 for width in [80, 12] {
13454 let mut d = wysiwyg_doc("no_dead_end", src);
13455 d.build_visual(width);
13456 let stops: Vec<usize> = (0..=src.len()).filter(|&o| d.vmap.is_stop(o)).collect();
13457 let (first, last) = (stops[0], stops[stops.len() - 1]);
13458 for &start in &stops {
13459 for (name, motion, want) in [
13460 (
13461 "down",
13462 (|d: &mut Doc| d.move_down(false)) as fn(&mut Doc),
13463 last,
13464 ),
13465 ("up", |d: &mut Doc| d.move_up(false), first),
13466 ] {
13467 d.caret = start;
13468 d.anchor = None;
13469 d.goal_col = None;
13470 // Every row, plus the presses the edges take, plus slack.
13471 for _ in 0..d.vmap.num_rows() + 4 {
13472 motion(&mut d);
13473 }
13474 assert_eq!(
13475 d.caret, want,
13476 "{name} held from {start} at width {width} never arrived"
13477 );
13478 }
13479 }
13480 }
13481 }
13482 // ── display columns ──────────────────────────────────────────────────────
13483 // A `col` is a terminal cell, not a character. The two are the same number
13484 // for the ASCII the fixtures above are written in, which is how they came
13485 // apart in the first place: `你` is one character drawn in two cells, so a
13486 // column counted in characters names a cell the text isn't in — one earlier
13487 // for every wide character to its left.
13488
13489 #[test]
13490 fn a_wide_character_is_two_columns_wide() {
13491 // The reproduction: `你` is one char and two cells, so the caret just
13492 // past it drew at column 1 — inside the character it had already left.
13493 for (view, tag) in VIEWS {
13494 let mut d = doc_in(view, &format!("wide_col_{tag}"), "你好\n");
13495 d.caret = "你".len();
13496 assert_eq!(d.caret_pos(), (0, 2), "{tag}: caret drew inside 你");
13497 d.caret = "你好".len();
13498 assert_eq!(d.caret_pos(), (0, 4), "{tag}");
13499 }
13500 }
13501
13502 #[test]
13503 fn a_cluster_is_as_wide_as_it_is_drawn_not_as_its_codepoints_measure() {
13504 // `👨👩👧` is five codepoints — two-cell, joiner, two-cell, joiner,
13505 // two-cell — measuring six cells one at a time, but the character they
13506 // spell is drawn in two. Width belongs to the cluster, not the glyph,
13507 // and the frontends measure it the same way.
13508 let family = "👨👩👧";
13509 for (view, tag) in VIEWS {
13510 let src = format!("a{family}b\n");
13511 let mut d = doc_in(view, &format!("wide_cluster_{tag}"), &src);
13512 d.caret = 1 + family.len();
13513 assert_eq!(
13514 d.caret_pos(),
13515 (0, 3),
13516 "{tag}: 'a' is one cell, the family two"
13517 );
13518 }
13519 }
13520
13521 #[test]
13522 fn both_cells_of_a_wide_character_mean_the_character() {
13523 // Clicking the far half of `好` is still clicking `好`: half a character
13524 // is not a place the caret can be, so it comes to rest at the
13525 // character's start — the column it would have been drawn at anyway.
13526 for (view, tag) in VIEWS {
13527 let mut d = doc_in(view, &format!("wide_click_{tag}"), "你好\n");
13528 for col in [2, 3] {
13529 d.caret = 0;
13530 d.click(0, col, false);
13531 assert_eq!(d.caret, "你".len(), "{tag}: click at col {col}");
13532 assert_eq!(d.caret_pos(), (0, 2), "{tag}: click at col {col}");
13533 }
13534 // Past the last cell is the line's end, as it is for ASCII.
13535 d.click(0, 9, false);
13536 assert_eq!(d.caret, "你好".len(), "{tag}: click past the end");
13537 }
13538 }
13539
13540 #[test]
13541 fn every_offset_survives_the_trip_out_to_a_column_and_back() {
13542 // The mapping is only a mapping if it inverts: the cell the caret is
13543 // drawn in has to be the cell that brings it back to the same offset.
13544 // Over a fixture where a character may be one cell or two, and one
13545 // codepoint or five.
13546 use unicode_segmentation::UnicodeSegmentation;
13547
13548 let src = "ab 你好 c\n\n👨👩👧 e\u{0301}x 漢字\n\nplain ascii\n";
13549
13550 let mut d = doc_in(View::Source, "roundtrip_source", src);
13551 // Every offset the source view's caret can occupy: it steps by grapheme
13552 // cluster, so those are its boundaries.
13553 for (off, _) in src
13554 .grapheme_indices(true)
13555 .chain(std::iter::once((src.len(), "")))
13556 {
13557 d.caret = off;
13558 let (row, col) = d.caret_pos();
13559 d.click(row, col, false);
13560 assert_eq!(d.caret, off, "source: {off} → ({row}, {col}) → {}", d.caret);
13561 }
13562
13563 // And in WYSIWYG, where the offsets the caret can occupy are the map's
13564 // stops rather than every boundary.
13565 let mut d = doc_in(View::Wysiwyg, "roundtrip_wysiwyg", src);
13566 let stops: Vec<usize> = (0..=src.len()).filter(|&o| d.vmap.is_stop(o)).collect();
13567 assert!(stops.len() > 20, "fixture should have plenty of stops");
13568 for off in stops {
13569 d.caret = off;
13570 let (row, col) = d.caret_pos();
13571 d.click(row, col, false);
13572 assert_eq!(
13573 d.caret, off,
13574 "wysiwyg: {off} → ({row}, {col}) → {}",
13575 d.caret
13576 );
13577 }
13578 }
13579
13580 #[test]
13581 fn vertical_motion_aims_at_a_column_the_reader_can_see() {
13582 // Down from under `世` lands under the glyph in that cell, not two
13583 // characters further along the line. The goal is a column, so a line of
13584 // wide characters and a line of ASCII line up the way they're drawn.
13585 //
13586 // The gap differs by view: a bare newline inside a paragraph is a soft
13587 // break, which WYSIWYG draws as a space on a single row. The views share
13588 // a grid only where the source's lines are the renderer's rows too.
13589 for (view, tag) in VIEWS {
13590 let gap = if view == View::Source { "\n" } else { "\n\n" };
13591 let src = format!("你好世{gap}abcdef\n");
13592 let mut d = doc_in(view, &format!("goal_wide_{tag}"), &src);
13593 d.caret = "你好".len();
13594 assert_eq!(d.caret_pos().1, 4, "{tag}: `世` is drawn at column 4");
13595 d.move_down(false);
13596 assert_eq!(d.caret_pos().1, 4, "{tag}: goal column lost");
13597 assert!(
13598 d.source[d.caret..].starts_with('e'),
13599 "{tag}: landed on the wrong glyph"
13600 );
13601 }
13602 }
13603
13604 #[test]
13605 fn a_goal_column_landing_inside_a_wide_character_lands_on_it() {
13606 // Down from column 3 onto `你好`, whose characters start at columns 0
13607 // and 2: column 3 is the *second* cell of `好`. There is nowhere to be
13608 // between the cells of one character, so the caret rests on it — and on
13609 // its start, which is the only offset there that is a caret stop.
13610 for (view, tag) in VIEWS {
13611 let gap = if view == View::Source { "\n" } else { "\n\n" };
13612 let src = format!("abcdef{gap}你好\n");
13613 let mut d = doc_in(view, &format!("goal_inside_{tag}"), &src);
13614 let line = src.find('你').unwrap();
13615 d.caret = 3;
13616 d.move_down(false);
13617 assert_eq!(d.caret, line + "你".len(), "{tag}: landed off `好`'s start");
13618 assert_eq!(d.caret_pos().1, 2, "{tag}: drew between `好`'s cells");
13619 }
13620 }
13621
13622 #[test]
13623 fn a_caret_in_a_table_cell_of_wide_text_draws_where_the_text_is() {
13624 // The column the cell's text is laid out in is measured in cells, so the
13625 // caret walking that text has to be too — the two agreeing is the whole
13626 // point of the grid staying square.
13627 let mut d = wysiwyg_doc("table_wide", "| A | B |\n|---|---|\n| 你好 | y |\n");
13628 let at = d.source.find("你").unwrap();
13629 d.caret = at;
13630 let (row, col) = d.caret_pos();
13631 // `│ ` opens the row, so the cell's text starts at column 2; `好` is two
13632 // cells further along.
13633 assert_eq!(col, 2, "the cell's first character");
13634 d.move_right(false);
13635 assert_eq!(
13636 d.caret_pos(),
13637 (row, 4),
13638 "`好` is drawn past `你`'s two cells"
13639 );
13640 assert_eq!(d.caret, at + "你".len());
13641 }
13642
13643 // ── active inline marks ───────────────────────────────────────────────────
13644
13645 /// The marks at a `|`-marked fixture's caret, in `InlineMarks::iter` order.
13646 fn marks(view: View, name: &str, marked: &str) -> Vec<InlineKind> {
13647 let (src, caret) = parse_caret(marked);
13648 let mut d = doc_in(view, name, &src);
13649 d.caret = caret;
13650 d.active_inline_marks().iter().collect()
13651 }
13652
13653 /// The marks over the selection `[start, end)`.
13654 fn marks_over(view: View, name: &str, src: &str, start: usize, end: usize) -> Vec<InlineKind> {
13655 let mut d = doc_in(view, name, src);
13656 d.anchor = Some(start);
13657 d.caret = end;
13658 d.active_inline_marks().iter().collect()
13659 }
13660
13661 #[test]
13662 fn a_caret_in_a_mark_reports_it() {
13663 for (view, tag) in VIEWS {
13664 let m = |marked| marks(view, &format!("marks_in_{tag}"), marked);
13665 assert_eq!(m("a **bo|ld** b"), [InlineKind::Strong], "{tag}");
13666 assert_eq!(m("a *it|alic* b"), [InlineKind::Emph], "{tag}");
13667 assert_eq!(m("a `co|de` b"), [InlineKind::Verbatim], "{tag}");
13668 // Plain text under no mark lights nothing — the toolbar's resting state.
13669 assert_eq!(m("a| **bold** b"), [], "{tag}");
13670 assert!(m("plain t|ext").is_empty(), "{tag}");
13671 }
13672 }
13673
13674 #[test]
13675 fn nested_marks_all_report() {
13676 // Bold *and* italic: a toolbar lights both buttons, so the set has both —
13677 // the ancestor chain is a chain, and every mark on it is in force.
13678 for (view, tag) in VIEWS {
13679 assert_eq!(
13680 marks(
13681 view,
13682 &format!("marks_nested_{tag}"),
13683 "**bold and *bo|th*** end"
13684 ),
13685 [InlineKind::Strong, InlineKind::Emph],
13686 "{tag}"
13687 );
13688 }
13689 }
13690
13691 #[test]
13692 fn the_caret_at_a_marks_edge_reports_it_where_typing_would_extend_it() {
13693 // The offsets a WYSIWYG caret actually reaches at a bold run's edges are
13694 // the first byte of its text and the byte after its last — both inside
13695 // the mark's span, both places typing lands inside the bold. The offset
13696 // past the closing delimiter is the next text, and reports nothing.
13697 let src = "a **bold** b";
13698 let inner_start = src.find("bold").unwrap(); // 4
13699 let inner_end = inner_start + "bold".len(); // 8, on the closing `**`
13700 for (view, tag) in VIEWS {
13701 let mut d = doc_in(view, &format!("marks_edge_{tag}"), src);
13702 for off in [2, 3, inner_start, inner_end, 9] {
13703 d.caret = off;
13704 assert!(
13705 d.active_inline_marks().contains(InlineKind::Strong),
13706 "{tag}: offset {off} is inside the strong span"
13707 );
13708 }
13709 for off in [0, 1, 10, 11, 12] {
13710 d.caret = off;
13711 assert!(
13712 !d.active_inline_marks().contains(InlineKind::Strong),
13713 "{tag}: offset {off} is outside the strong run"
13714 );
13715 }
13716 }
13717 }
13718
13719 #[test]
13720 fn a_mark_ends_the_same_way_at_the_end_of_the_buffer_as_in_the_middle() {
13721 // Regression: twig resolves an offset that is one node's end and the
13722 // next one's start to the node that *starts* there, so `**bold**|\n`
13723 // isn't bold. With nothing following there's no tie to break and the
13724 // chain still ended at the mark, which made a trailing `\n` — not the
13725 // text — decide whether the caret after a bold word reported bold. It's
13726 // the offset past the mark either way, and typing there is plain either
13727 // way. A blank document typed into is exactly this shape.
13728 for (view, tag) in VIEWS {
13729 let m = |name: String, marked| marks(view, &name, marked);
13730 assert_eq!(
13731 m(format!("marks_eob_{tag}"), "**bold**|"),
13732 [],
13733 "{tag}: no trailing newline"
13734 );
13735 assert_eq!(
13736 m(format!("marks_eol_{tag}"), "**bold**|\n"),
13737 [],
13738 "{tag}: with one"
13739 );
13740 // And the last offset that *is* in the mark still is.
13741 assert_eq!(
13742 m(format!("marks_eob_in_{tag}"), "**bold*|*"),
13743 [InlineKind::Strong],
13744 "{tag}"
13745 );
13746 }
13747 }
13748
13749 #[test]
13750 fn a_selection_reports_a_mark_only_when_it_covers_the_whole_thing() {
13751 let src = "a **bold** b";
13752 let (b, d_) = (src.find("bold").unwrap(), src.find("bold").unwrap() + 4);
13753 for (view, tag) in VIEWS {
13754 let m = |s, e| marks_over(view, &format!("marks_sel_{tag}"), src, s, e);
13755 // The whole bold word, and a slice of it.
13756 assert_eq!(m(b, d_), [InlineKind::Strong], "{tag}: the whole word");
13757 assert_eq!(m(b + 1, d_ - 1), [InlineKind::Strong], "{tag}: a slice");
13758 // Ending exactly at the closing delimiter's start is still all-bold:
13759 // an exclusive end sits *past* the last selected character, so the
13760 // question is asked of the character, not the boundary.
13761 assert_eq!(
13762 m(b, d_ + 2),
13763 [InlineKind::Strong],
13764 "{tag}: through the close"
13765 );
13766 // Half in, half out: Bold lit here would claim a press turns it off.
13767 assert_eq!(m(0, d_), [], "{tag}: leading plain text");
13768 assert_eq!(m(b, src.len()), [], "{tag}: trailing plain text");
13769 }
13770 }
13771
13772 #[test]
13773 fn a_selection_across_two_runs_of_the_same_mark_reports_nothing() {
13774 // Both ends are bold, but the space between them isn't — two runs are two
13775 // nodes, which is exactly what the node id catches and a kind-only
13776 // comparison would not.
13777 let src = "**one** **two**";
13778 for (view, tag) in VIEWS {
13779 let m = marks_over(view, &format!("marks_runs_{tag}"), src, 2, 13);
13780 assert_eq!(m, [], "{tag}: `one** **two` is not all bold");
13781 }
13782 }
13783
13784 #[test]
13785 fn marks_read_the_document_as_it_is_edited() {
13786 // The point of asking twig every frame instead of caching: the answer has
13787 // to follow the toggle that changed it.
13788 let mut d = wysiwyg_doc("marks_live", "one two\n");
13789 d.anchor = Some(0);
13790 d.caret = 3;
13791 assert!(d.active_inline_marks().is_empty(), "plain to start");
13792 d.toggle(InlineKind::Strong);
13793 assert_eq!(d.source, "**one** two\n");
13794 // `toggle` leaves the bolded text selected, so the button it lit stays lit.
13795 assert!(d.active_inline_marks().contains(InlineKind::Strong));
13796 d.toggle(InlineKind::Strong);
13797 assert!(d.active_inline_marks().is_empty(), "and off again");
13798 }
13799
13800 #[test]
13801 fn a_link_is_not_an_inline_mark() {
13802 // `link`/`str` are inline nodes, but nothing on the inline toolbar
13803 // toggles them — a set with a "link mark" in it would have no button.
13804 for (view, tag) in VIEWS {
13805 assert_eq!(
13806 marks(view, &format!("marks_link_{tag}"), "a [te|xt](u) b"),
13807 [],
13808 "{tag}"
13809 );
13810 }
13811 }
13812
13813 // ── blank documents ───────────────────────────────────────────────────────
13814
13815 #[test]
13816 fn a_blank_document_is_untitled_empty_and_markdown() {
13817 let mut d = Doc::blank().unwrap();
13818 assert!(d.is_untitled());
13819 assert_eq!(d.path, PathBuf::new());
13820 assert_eq!(
13821 d.file_name(),
13822 "untitled",
13823 "the header has to show something"
13824 );
13825 assert_eq!(d.format_name(), "markdown");
13826 assert_eq!(d.source, "");
13827 assert!(!d.dirty, "nothing typed yet is nothing to lose");
13828 assert_eq!(d.disk_state(), DiskState::Untitled);
13829 // And it's a document you can be in: the default view renders it.
13830 d.build_visual(80);
13831 assert_eq!(d.caret, 0);
13832 }
13833
13834 #[test]
13835 fn saving_an_untitled_document_asks_for_a_name_instead_of_writing() {
13836 let mut d = Doc::blank().unwrap();
13837 d.insert("hello");
13838 assert!(d.dirty);
13839 d.save();
13840 assert_eq!(d.status.as_deref(), Some("untitled — save as…"));
13841 assert!(d.dirty, "it must not come away believing it saved");
13842 assert!(d.is_untitled(), "and it still has no file");
13843 }
13844
13845 #[test]
13846 fn a_blank_document_becomes_a_real_one_at_the_first_save_as() {
13847 let p = temp_path("blank_save_as");
13848 let mut d = Doc::blank().unwrap();
13849 // Plain text — a blank doc opens in Hidden mode, where a typed `#` would
13850 // be kept literal (`\#`); this test is about save-as, not escaping (which
13851 // has its own test), so it types nothing that escaping would touch.
13852 d.insert("hi");
13853 d.save_as(p.clone());
13854 assert_eq!(std::fs::read_to_string(&p).unwrap(), "hi");
13855 assert!(!d.is_untitled());
13856 assert!(!d.dirty);
13857 assert_eq!(d.file_name(), p.file_name().unwrap().to_string_lossy());
13858 assert_eq!(
13859 d.disk_state(),
13860 DiskState::Unchanged,
13861 "the watermark is stamped"
13862 );
13863 // And ⌘S is a plain save from here on.
13864 d.insert("!");
13865 d.save();
13866 assert_eq!(std::fs::read_to_string(&p).unwrap(), "hi!");
13867 let _ = std::fs::remove_file(&p);
13868 }
13869
13870 // ── a file that isn't there yet ───────────────────────────────────────────
13871
13872 /// A unique path in the temp dir with the given extension, guaranteed not to
13873 /// exist — what `leaf notes.md` is handed when the file has never been made.
13874 fn missing_path(name: &str, ext: &str) -> PathBuf {
13875 static SEQ: std::sync::atomic::AtomicUsize = std::sync::atomic::AtomicUsize::new(0);
13876 let seq = SEQ.fetch_add(1, std::sync::atomic::Ordering::Relaxed);
13877 let mut p = std::env::temp_dir();
13878 p.push(format!("leaf_test_new_{name}_{seq}.{ext}"));
13879 let _ = std::fs::remove_file(&p);
13880 p
13881 }
13882
13883 #[test]
13884 fn a_file_that_doesnt_exist_opens_as_an_empty_named_document() {
13885 let p = missing_path("named", "md");
13886 let mut d = Doc::open_or_create(p.clone()).unwrap();
13887
13888 assert_eq!(d.source, "", "nothing was read, so there's nothing in it");
13889 assert!(!d.dirty, "an untouched new buffer has nothing to lose");
13890 assert!(
13891 !d.is_untitled(),
13892 "it has the name the user asked for — ^S must not detour to Save As"
13893 );
13894 assert_eq!(d.file_name(), p.file_name().unwrap().to_str().unwrap());
13895 assert!(d.path.is_absolute(), "the same absolute path `open` stores");
13896 assert!(!p.exists(), "and opening it wrote nothing");
13897 // And it's a document you can be in.
13898 d.build_visual(80);
13899 assert_eq!(d.caret, 0);
13900 }
13901
13902 #[test]
13903 fn a_new_file_is_created_by_its_first_save() {
13904 let p = missing_path("first_save", "md");
13905 let mut d = Doc::open_or_create(p.clone()).unwrap();
13906 d.insert("hello\n");
13907 assert!(d.dirty);
13908 d.save();
13909
13910 assert_eq!(
13911 std::fs::read_to_string(&p).unwrap(),
13912 "hello\n",
13913 "a plain ^S wrote it — no Save As, no name to invent"
13914 );
13915 assert!(!d.dirty);
13916 assert_eq!(d.disk_state(), DiskState::Unchanged);
13917 let _ = std::fs::remove_file(&p);
13918 }
13919
13920 #[test]
13921 fn a_new_file_takes_its_format_from_the_extension() {
13922 // The one thing `blank` can't do: with no name it has to assume Markdown,
13923 // and typing djot into a Markdown parse is the wrong buffer.
13924 let dj = missing_path("format", "dj");
13925 assert_eq!(Doc::open_or_create(dj).unwrap().format_name(), "djot");
13926 let md = missing_path("format", "md");
13927 assert_eq!(Doc::open_or_create(md).unwrap().format_name(), "markdown");
13928 }
13929
13930 #[test]
13931 fn a_new_file_reports_itself_missing_until_it_is_saved() {
13932 // Not `Untitled` — that's the answer for a document with no path, and it
13933 // would tell a frontend there is nothing a save could collide with. Here
13934 // there is a path, and the file simply isn't at it yet.
13935 let p = missing_path("disk_state", "md");
13936 let mut d = Doc::open_or_create(p.clone()).unwrap();
13937 assert_eq!(d.disk_state(), DiskState::Missing);
13938
13939 // Somebody else creates it while the buffer is open: that's an overwrite
13940 // the frontend has to be able to prompt about, exactly as for an opened
13941 // file. Their bytes, not ours, so `Changed`.
13942 std::fs::write(&p, "theirs\n").unwrap();
13943 assert_eq!(d.disk_state(), DiskState::Changed);
13944
13945 // Saving makes the file ours and re-stamps the watermark.
13946 d.insert("ours\n");
13947 d.save();
13948 assert_eq!(d.disk_state(), DiskState::Unchanged);
13949 assert_eq!(std::fs::read_to_string(&p).unwrap(), "ours\n");
13950 let _ = std::fs::remove_file(&p);
13951 }
13952
13953 #[test]
13954 fn open_or_create_still_opens_a_file_that_is_there() {
13955 let d = doc_with("open_or_create_existing", "body\n");
13956 let reopened = Doc::open_or_create(d.path.clone()).unwrap();
13957 assert_eq!(reopened.source, "body\n");
13958 assert_eq!(reopened.disk_state(), DiskState::Unchanged);
13959 }
13960
13961 #[test]
13962 fn a_missing_file_with_no_readable_extension_is_still_an_error() {
13963 // A mistyped flag or a stray argument must not become a buffer promising
13964 // to save somewhere — the same refusal `open` gives a real file.
13965 let mut p = std::env::temp_dir();
13966 p.push("leaf_test_new_bad_ext.wat");
13967 assert!(Doc::open_or_create(p).is_err());
13968 let mut none = std::env::temp_dir();
13969 none.push("leaf_test_new_no_ext");
13970 assert!(Doc::open_or_create(none).is_err());
13971 }
13972
13973 #[test]
13974 fn a_new_file_in_a_directory_that_doesnt_exist_opens_but_wont_save() {
13975 // Opening reads nothing, so there is nothing to fail on yet; the write is
13976 // where it fails, and it says so rather than claiming a save.
13977 let p = std::env::temp_dir().join("leaf_test_no_such_dir_c41/doc.md");
13978 let mut d = Doc::open_or_create(p).unwrap();
13979 d.insert("x");
13980 d.save();
13981 assert!(
13982 d.status.as_deref().unwrap().starts_with("save failed:"),
13983 "got {:?}",
13984 d.status
13985 );
13986 assert!(d.dirty, "it must not come away believing it saved");
13987 }
13988
13989 // ── save as ───────────────────────────────────────────────────────────────
13990
13991 /// A unique path in the temp dir that no fixture wrote — a Save As target.
13992 fn temp_path(name: &str) -> PathBuf {
13993 static SEQ: std::sync::atomic::AtomicUsize = std::sync::atomic::AtomicUsize::new(0);
13994 let seq = SEQ.fetch_add(1, std::sync::atomic::Ordering::Relaxed);
13995 let mut p = std::env::temp_dir();
13996 p.push(format!("leaf_test_target_{name}_{seq}.md"));
13997 let _ = std::fs::remove_file(&p);
13998 p
13999 }
14000
14001 #[test]
14002 fn save_as_moves_the_document_and_leaves_the_old_file_alone() {
14003 let mut d = doc_with("save_as_move", "original\n");
14004 let old = d.path.clone();
14005 let new = temp_path("save_as_move");
14006 d.insert("edited: ");
14007 d.save_as(new.clone());
14008
14009 assert_eq!(std::fs::read_to_string(&new).unwrap(), "edited: original\n");
14010 assert_eq!(
14011 std::fs::read_to_string(&old).unwrap(),
14012 "original\n",
14013 "Save As doesn't touch the file it came from"
14014 );
14015 assert_eq!(d.path, new, "the document moved");
14016 assert!(!d.dirty);
14017 assert_eq!(
14018 d.status.as_deref(),
14019 Some(&*format!("saved {}", d.file_name()))
14020 );
14021
14022 // Every later save follows it, which is the whole difference from a copy.
14023 d.caret = 0;
14024 d.insert("re-");
14025 d.save();
14026 assert_eq!(
14027 std::fs::read_to_string(&new).unwrap(),
14028 "re-edited: original\n"
14029 );
14030 assert_eq!(std::fs::read_to_string(&old).unwrap(), "original\n");
14031 let _ = std::fs::remove_file(&new);
14032 }
14033
14034 #[test]
14035 fn save_as_overwrites_an_existing_target() {
14036 // The picker already asked; asking again down here is the same question
14037 // twice, and the second one has no way to be answered.
14038 let new = temp_path("save_as_over");
14039 std::fs::write(&new, "theirs\n").unwrap();
14040 let mut d = doc_with("save_as_over", "ours\n");
14041 d.save_as(new.clone());
14042 assert_eq!(std::fs::read_to_string(&new).unwrap(), "ours\n");
14043 let _ = std::fs::remove_file(&new);
14044 }
14045
14046 #[test]
14047 fn a_save_as_that_fails_leaves_the_document_where_it_was() {
14048 let mut d = doc_with("save_as_fail", "body\n");
14049 let old = d.path.clone();
14050 d.insert("x");
14051 // A directory that doesn't exist: the write can't land.
14052 let bad = std::env::temp_dir().join("leaf_test_no_such_dir_9f2/doc.md");
14053 d.save_as(bad);
14054
14055 assert_eq!(
14056 d.path, old,
14057 "the document must not move to a file that isn't there"
14058 );
14059 assert!(d.dirty, "and must not believe it saved");
14060 assert!(
14061 d.status.as_deref().unwrap().starts_with("save failed:"),
14062 "the same failure a plain save reports, got {:?}",
14063 d.status
14064 );
14065 // The original is still the document's file, and still saveable.
14066 d.save();
14067 assert_eq!(std::fs::read_to_string(&old).unwrap(), "xbody\n");
14068 assert!(!d.dirty);
14069 }
14070
14071 #[test]
14072 fn save_as_renames_without_reparsing_the_format() {
14073 // `.dj` on the name doesn't make the buffer djot: it was parsed as
14074 // Markdown and still is, and saying otherwise would be a conversion the
14075 // user never asked for (and an undo history thrown away to do it).
14076 let mut d = doc_with("save_as_format", "**b**\n");
14077 let mut new = temp_path("save_as_format");
14078 new.set_extension("dj");
14079 d.save_as(new.clone());
14080 assert_eq!(d.format_name(), "markdown");
14081 let _ = std::fs::remove_file(&new);
14082 }
14083
14084 // ── external change / reload ──────────────────────────────────────────────
14085
14086 #[test]
14087 fn an_untouched_file_reports_unchanged() {
14088 let mut d = doc_with("disk_clean", "body\n");
14089 assert_eq!(d.disk_state(), DiskState::Unchanged);
14090 // Editing the buffer is not editing the file.
14091 d.insert("x");
14092 assert_eq!(d.disk_state(), DiskState::Unchanged);
14093 assert!(d.dirty);
14094 // Saving re-stamps the watermark rather than reporting our own bytes back.
14095 d.save();
14096 assert_eq!(d.disk_state(), DiskState::Unchanged);
14097 }
14098
14099 #[test]
14100 fn a_file_written_underneath_reports_changed() {
14101 let mut d = doc_with("disk_changed", "body\n");
14102 std::fs::write(&d.path, "someone else\n").unwrap();
14103 assert_eq!(d.disk_state(), DiskState::Changed);
14104 // Dirty *and* changed is the clobber: both halves are readable, and
14105 // leaf-core takes neither side.
14106 d.insert("x");
14107 assert!(d.dirty && d.disk_state() == DiskState::Changed);
14108 // Saving anyway is allowed — the frontend asked, or chose not to.
14109 d.save();
14110 assert_eq!(std::fs::read_to_string(&d.path).unwrap(), "xbody\n");
14111 assert_eq!(d.disk_state(), DiskState::Unchanged);
14112 }
14113
14114 #[test]
14115 fn a_file_rewritten_with_the_same_bytes_is_unchanged() {
14116 // The hash is what makes this honest: the file was written (a fresh
14117 // mtime), and nothing about the document is stale.
14118 let d = doc_with("disk_same_bytes", "body\n");
14119 std::fs::write(&d.path, "body\n").unwrap();
14120 assert_eq!(d.disk_state(), DiskState::Unchanged);
14121 }
14122
14123 #[test]
14124 fn a_deleted_file_reports_missing() {
14125 let mut d = doc_with("disk_missing", "body\n");
14126 std::fs::remove_file(&d.path).unwrap();
14127 assert_eq!(d.disk_state(), DiskState::Missing);
14128 // A save recreates it, and the document is whole again.
14129 d.save();
14130 assert_eq!(d.disk_state(), DiskState::Unchanged);
14131 assert_eq!(std::fs::read_to_string(&d.path).unwrap(), "body\n");
14132 }
14133
14134 #[test]
14135 fn reload_replaces_the_document_with_the_file() {
14136 for (view, tag) in VIEWS {
14137 let mut d = doc_in(view, &format!("reload_{tag}"), "one\n\ntwo\n");
14138 d.insert("edited ");
14139 assert!(d.dirty);
14140 std::fs::write(&d.path, "one\n\ntwo\n\nthree\n").unwrap();
14141 d.reload();
14142
14143 assert_eq!(d.source, "one\n\ntwo\n\nthree\n", "{tag}");
14144 assert!(!d.dirty, "{tag}: the file is what we have");
14145 assert_eq!(d.disk_state(), DiskState::Unchanged, "{tag}");
14146 assert_eq!(
14147 d.status.as_deref(),
14148 Some(&*format!("reloaded {}", d.file_name()))
14149 );
14150 // The reloaded tree is live, not the old parse.
14151 d.caret = d.source.find("three").unwrap();
14152 assert_eq!(d.breadcrumb(), "doc › para › str", "{tag}");
14153 }
14154 }
14155
14156 #[test]
14157 fn reload_clamps_the_caret_and_drops_the_selection() {
14158 let mut d = doc_with("reload_caret", "a long first line\n");
14159 d.caret = 12;
14160 d.anchor = Some(4);
14161 std::fs::write(&d.path, "short\n").unwrap();
14162 d.reload();
14163 assert_eq!(d.caret, d.source.len(), "clamped into the shorter file");
14164 assert_eq!(
14165 d.anchor, None,
14166 "a selection over bytes that changed is a lie"
14167 );
14168 assert!(d.selection().is_none());
14169
14170 // A caret the file still has room for stays put.
14171 let mut d = doc_with("reload_caret_keep", "one\n\ntwo\n");
14172 d.caret = 2;
14173 std::fs::write(&d.path, "one\n\ntwo\n\nthree\n").unwrap();
14174 d.reload();
14175 assert_eq!(d.caret, 2);
14176 }
14177
14178 /// A silent reload is something that happened *to* a reader — a formatter,
14179 /// a `git checkout` — so it has to be undoable like anything else that
14180 /// changes the document, and undoable as one step rather than as however
14181 /// many the file happens to differ by.
14182 #[test]
14183 fn reload_is_one_undo_step_and_keeps_the_history_under_it() {
14184 let mut d = doc_with("reload_undo", "body\n");
14185 d.insert("x");
14186 assert_eq!(d.source, "xbody\n");
14187 std::fs::write(&d.path, "replaced\n").unwrap();
14188 d.reload();
14189 assert_eq!(d.source, "replaced\n");
14190 assert!(!d.dirty, "a reload lands clean");
14191
14192 // One ^Z takes the whole swap off, and hands back the unsaved work it
14193 // replaced — which is unsaved again, because the file no longer says it.
14194 d.undo();
14195 assert_eq!(d.source, "xbody\n", "the reload comes off in one step");
14196 assert!(d.dirty, "and what it comes back to is unsaved");
14197 // …and the history under it is still there.
14198 d.undo();
14199 assert_eq!(
14200 d.source, "body\n",
14201 "the typing before the reload undoes too"
14202 );
14203 // Redo walks back up through the reload.
14204 d.redo();
14205 d.redo();
14206 assert_eq!(d.source, "replaced\n");
14207 }
14208
14209 /// A file rewritten with the bytes it already had is not an edit, so it
14210 /// must not leave an undo step behind for something nobody did.
14211 #[test]
14212 fn reloading_identical_bytes_pushes_no_undo_step() {
14213 let mut d = doc_with("reload_same", "body\n");
14214 d.insert("x");
14215 std::fs::write(&d.path, "xbody\n").unwrap();
14216 d.reload();
14217 assert_eq!(d.source, "xbody\n");
14218 assert!(!d.dirty, "the file now says what the buffer does");
14219 d.undo();
14220 assert_eq!(
14221 d.source, "body\n",
14222 "one step back is the typing, not a no-op"
14223 );
14224 }
14225
14226 #[test]
14227 fn a_reload_that_cant_read_leaves_the_document_alone() {
14228 let mut d = doc_with("reload_gone", "body\n");
14229 d.insert("x");
14230 std::fs::remove_file(&d.path).unwrap();
14231 d.reload();
14232 assert_eq!(d.source, "xbody\n", "the unsaved work is still here");
14233 assert!(d.dirty);
14234 assert!(
14235 d.status.as_deref().unwrap().starts_with("reload failed:"),
14236 "{:?}",
14237 d.status
14238 );
14239
14240 // And an untitled document has nothing to reload from.
14241 let mut d = Doc::blank().unwrap();
14242 d.insert("typed");
14243 d.reload();
14244 assert_eq!(d.source, "typed");
14245 assert_eq!(d.status.as_deref(), Some("no file to reload"));
14246 }
14247
14248 #[test]
14249 fn a_read_only_document_refuses_every_door() {
14250 let mut d = doc_with("readonly", "one two three\n");
14251 d.insert("x");
14252 assert!(d.dirty, "writable first, so the undo step exists");
14253 d.set_read_only(true);
14254 let before = d.source.clone();
14255 d.insert("y");
14256 d.backspace();
14257 d.undo();
14258 d.redo();
14259 assert_eq!(d.source, before, "no door moved a byte");
14260 d.set_read_only(false);
14261 d.undo();
14262 assert_ne!(d.source, before, "off again, the same doors work");
14263 }
14264
14265 /// The doors that go to twig's own verbs rather than through the splice.
14266 /// Typed text in the rendered view under the default markup mode is the
14267 /// everyday one — it is what a keystroke in leaf-web or the Apple views
14268 /// becomes — and it walked straight past the gate.
14269 #[test]
14270 fn a_read_only_document_refuses_the_doors_around_the_splice() {
14271 let mut d = wysiwyg_doc(
14272 "readonly-doors",
14273 "one two three\n\n| a | b |\n|---|---|\n| c | d |\n",
14274 );
14275 d.set_markup_mode(MarkupMode::None);
14276 d.set_read_only(true);
14277 let before = d.source.clone();
14278 d.place_caret(3, false);
14279 d.insert("y");
14280 d.insert_link("https://example.com");
14281 d.insert_image("a.png", "alt");
14282 d.insert_thematic_break();
14283 d.insert_footnote();
14284 d.place_caret(0, false);
14285 d.place_caret(3, true);
14286 d.toggle(InlineKind::Strong);
14287 d.toggle_heading(2);
14288 d.set_block(BlockKind::Paragraph);
14289 d.toggle_list(false);
14290 d.toggle_blockquote();
14291 d.toggle_task_item();
14292 d.newline();
14293 d.indent();
14294 d.set_code_language("rust");
14295 let in_cell = d.source.find("| c").unwrap() + 2;
14296 d.place_caret(in_cell, false);
14297 assert!(d.caret_in_table(), "the caret is in the grid");
14298 assert!(!d.cell_line_break(), "the cell break reports the refusal");
14299 assert_eq!(d.source, before, "no door moved a byte");
14300 assert!(!d.dirty, "nothing to save");
14301 d.set_read_only(false);
14302 d.place_caret(3, false);
14303 d.insert("y");
14304 assert_ne!(d.source, before, "off again, the same doors work");
14305 }
14306
14307 #[test]
14308 fn a_selection_quote_carries_its_context_on_char_boundaries() {
14309 let mut d = doc_with("quote", "before 你好 exact 世界 after\n");
14310 let start = d.source.find("exact").unwrap();
14311 d.place_caret(start, false);
14312 d.place_caret(start + "exact".len(), true);
14313 let q = d.selection_quote(3).unwrap();
14314 assert_eq!(q.exact, "exact");
14315 assert_eq!(
14316 q.prefix, "你好 ",
14317 "chars, not bytes — the multibyte pair counts as two"
14318 );
14319 assert_eq!(q.suffix, " 世界");
14320 assert_eq!(&d.source[q.start..q.end], "exact");
14321 // At the edges the context clips rather than erring.
14322 d.place_caret(0, false);
14323 d.place_caret(6, true);
14324 let q = d.selection_quote(40).unwrap();
14325 assert_eq!(q.prefix, "");
14326 assert_eq!(q.exact, "before");
14327 // No selection is no quote.
14328 d.place_caret(0, false);
14329 assert!(d.selection_quote(3).is_none());
14330 }
14331
14332 #[test]
14333 fn highlights_are_kept_sorted_and_answer_point_queries() {
14334 let mut d = doc_with("hl", "one two three\n");
14335 d.set_highlights(vec![
14336 Highlight {
14337 start: 8,
14338 end: 13,
14339 id: "b".into(),
14340 color: None,
14341 marker: None,
14342 },
14343 Highlight {
14344 start: 0,
14345 end: 3,
14346 id: "a".into(),
14347 color: Some("#ffe066".into()),
14348 marker: None,
14349 },
14350 Highlight {
14351 start: 5,
14352 end: 5,
14353 id: "empty".into(),
14354 color: None,
14355 marker: None,
14356 },
14357 ]);
14358 assert_eq!(
14359 d.highlights()
14360 .iter()
14361 .map(|h| h.id.as_str())
14362 .collect::<Vec<_>>(),
14363 ["a", "b"],
14364 "sorted by start, the empty range dropped"
14365 );
14366 assert_eq!(d.highlight_at(1).map(|h| h.id.as_str()), Some("a"));
14367 assert_eq!(d.highlight_at(3), None, "end is exclusive");
14368 assert_eq!(d.highlight_at(8).map(|h| h.id.as_str()), Some("b"));
14369 d.set_highlights(Vec::new());
14370 assert!(d.highlights().is_empty(), "a replace is a replace");
14371 }
14372
14373 /// `Highlight::covering` and the cursor over it are what both painters ask
14374 /// per glyph, so they have to answer the same as the scan they replaced —
14375 /// including in the gaps, which is where most glyphs are.
14376 #[test]
14377 fn covering_answers_from_a_sorted_list_without_scanning_all_of_it() {
14378 let hl = |start: usize, end: usize, id: &str| Highlight {
14379 start,
14380 end,
14381 id: id.into(),
14382 color: None,
14383 marker: None,
14384 };
14385 // Disjoint, as search hits are: in a range, in a gap, and past the end.
14386 let hits: Vec<Highlight> = (0..20).map(|i| hl(i * 10, i * 10 + 3, "hit")).collect();
14387 assert_eq!(Highlight::covering(&hits, 0).map(|h| h.start), Some(0));
14388 assert_eq!(Highlight::covering(&hits, 102).map(|h| h.start), Some(100));
14389 assert_eq!(
14390 Highlight::covering(&hits, 105),
14391 None,
14392 "a gap covers nothing"
14393 );
14394 assert_eq!(Highlight::covering(&hits, 103), None, "end is exclusive");
14395 assert_eq!(Highlight::covering(&hits, 9_999), None);
14396 assert_eq!(Highlight::covering(&[], 0), None);
14397
14398 // Nested: first by start, so a hit inside an annotation still resolves
14399 // to the annotation — and the range that stops short doesn't mask it.
14400 let nested = vec![hl(0, 20, "outer"), hl(5, 10, "inner")];
14401 assert_eq!(
14402 Highlight::covering(&nested, 7).map(|h| h.id.as_str()),
14403 Some("outer")
14404 );
14405 assert_eq!(
14406 Highlight::covering(&nested, 15).map(|h| h.id.as_str()),
14407 Some("outer")
14408 );
14409 }
14410
14411 /// The cursor is an optimisation, so the only thing worth asserting is that
14412 /// it is not also a change of answer — at every offset, over a list with a
14413 /// nest in it, walked forwards and then backwards.
14414 #[test]
14415 fn the_highlight_cursor_answers_exactly_what_a_fresh_scan_would() {
14416 let hl = |start: usize, end: usize, id: &str| Highlight {
14417 start,
14418 end,
14419 id: id.into(),
14420 color: None,
14421 marker: None,
14422 };
14423 let mut list = vec![
14424 hl(0, 20, "outer"),
14425 hl(5, 10, "inner"),
14426 hl(30, 33, "hit"),
14427 hl(40, 43, "hit"),
14428 ];
14429 list.sort_by_key(|h| (h.start, h.end));
14430
14431 let mut cursor = HighlightCursor::new(&list);
14432 for offset in 0..50 {
14433 assert_eq!(
14434 cursor.at(offset).map(|h| h.id.as_str()),
14435 Highlight::covering(&list, offset).map(|h| h.id.as_str()),
14436 "cursor disagrees at {offset}"
14437 );
14438 }
14439 // Backwards: the cursor re-seats rather than answering from where it
14440 // had got to, so a painter that revisits a row is still told the truth.
14441 for offset in (0..50).rev() {
14442 assert_eq!(
14443 cursor.at(offset).map(|h| h.id.as_str()),
14444 Highlight::covering(&list, offset).map(|h| h.id.as_str()),
14445 "cursor disagrees walking back at {offset}"
14446 );
14447 }
14448 }
14449}