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::wysiwyg::{self, MediaKind, MediaStop, VisualMap};
44
45/// Which view the body shows.
46#[derive(Clone, Copy, PartialEq, Eq, Debug)]
47pub enum View {
48 /// The raw document with a caret in source bytes.
49 Source,
50 /// Markup resolved to real styles, caret riding the rendered glyphs.
51 Wysiwyg,
52}
53
54/// How much of the source markup the WYSIWYG view exposes — a per-editor
55/// preference, orthogonal to [`View`]. Named for markup rather than for Markdown
56/// because leaf is grammar-agnostic: twig hands it Djot, HTML and XML on the same
57/// terms, and every rung below is about *delimiters*, whatever grammar spells
58/// them. The examples are Markdown only because that is what most documents are.
59///
60/// A single ladder over two underlying axes, because only three of their four
61/// combinations are coherent:
62///
63/// | | authoring off | authoring on |
64/// |---|---|---|
65/// | delimiters hidden | [`None`](Self::None) | [`Shortcuts`](Self::Shortcuts) |
66/// | caret line revealed | *incoherent* | [`Full`](Self::Full) |
67///
68/// The empty quadrant would show delimiters on the caret's line and then escape
69/// the ones you type — a surface that displays a syntax it refuses to accept.
70/// Someone who wants to read raw markup without authoring it has
71/// [`View::Source`], which is the better tool for it.
72///
73/// The two axes are read separately by the code that cares — see
74/// [`reveals_caret_line`](Self::reveals_caret_line) and
75/// [`authors`](Self::authors) — so neither behaviour has to know it's spelled
76/// as a ladder.
77#[derive(Clone, Copy, PartialEq, Eq, Debug, Default)]
78pub enum MarkupMode {
79 /// Delimiters stay hidden even on the caret's line, and typed syntax stays
80 /// literal — twig escapes anything that would open markup, so formatting
81 /// comes from commands (⌘b, the toolbar) instead of from spelling. The clean
82 /// reading surface for people who don't write markup by hand; the default,
83 /// and what Diaryx ships.
84 #[default]
85 None,
86 /// Delimiters stay hidden, but typing them authors real markup: `*x*`
87 /// becomes italic and the asterisks disappear into the styling
88 /// (Typora/Bear-shaped). For someone who knows the syntax but wants the
89 /// clean surface back once it has been applied.
90 Shortcuts,
91 /// The caret's line shows its raw markup while every other line renders
92 /// resolved (Obsidian live-preview-shaped), and typed syntax authors markup
93 /// — for people fluent in the document's grammar who want to see and edit
94 /// the delimiters they type.
95 Full,
96}
97
98impl MarkupMode {
99 /// Whether the rich view shows raw delimiters on the line holding the caret.
100 /// The rendering axis — read by [`Doc::reveal_line`] and threaded into the
101 /// WYSIWYG builder.
102 pub fn reveals_caret_line(self) -> bool {
103 matches!(self, MarkupMode::Full)
104 }
105
106 /// Whether typed markup characters author real formatting. The editing axis
107 /// — read by [`Doc::insert`], which escapes typed syntax when this is false.
108 pub fn authors(self) -> bool {
109 !matches!(self, MarkupMode::None)
110 }
111}
112
113/// How the WYSIWYG view treats a *soft break* — a bare newline inside a
114/// paragraph. An axis of its own, orthogonal to [`MarkupMode`] (which governs
115/// inline-markup delimiters) and to [`View`]: any reveal preference pairs with
116/// either flow. The renderer consults it when it lays a block's inline content
117/// into visual rows.
118#[derive(Clone, Copy, PartialEq, Eq, Debug, Default)]
119pub enum LineFlow {
120 /// A soft break folds into a space and the paragraph reflows to the
121 /// viewport width — flowing prose, where the source's line wrapping is
122 /// insignificant. The default, and what Diaryx ships.
123 #[default]
124 Fold,
125 /// A soft break renders as a line break exactly where it was written, so
126 /// the author's source line structure shows on screen unchanged — the mode
127 /// for people who lay out their prose deliberately (one sentence or clause
128 /// per line, semantic line breaks). The break is still a soft break in the
129 /// source; only its rendering changes.
130 Preserve,
131}
132
133/// What the file behind a document looks like right now, against the bytes leaf
134/// last read from it or wrote to it — the question a frontend asks before it
135/// saves (a `Changed` file plus a `dirty` document is an overwrite about to
136/// happen) or when its window regains focus. See [`Doc::disk_state`].
137#[derive(Clone, Copy, Debug, PartialEq, Eq)]
138pub enum DiskState {
139 /// The file holds exactly the bytes leaf last read or wrote.
140 Unchanged,
141 /// Someone else wrote the file since. Saving overwrites their work; see
142 /// [`Doc::reload`] for the other direction.
143 Changed,
144 /// The file is gone — deleted or renamed away. A save recreates it.
145 Missing,
146 /// There is a path, but the file couldn't be read (permissions, a directory
147 /// in the way): leaf can't tell, and won't guess.
148 Unreadable,
149 /// No file behind this document yet — see [`Doc::blank`]. Nothing can have
150 /// changed under a document that was never on disk.
151 Untitled,
152}
153
154/// The inline marks in force at a point in the document — what a toolbar
155/// lights up. A `Copy` bitset rather than a `HashSet`, because
156/// [`Doc::active_inline_marks`] is called on every frame that draws a toolbar
157/// and a set that allocates to answer "is Bold on?" is a set that shouldn't.
158#[derive(Clone, Copy, Debug, Default, PartialEq, Eq)]
159pub struct InlineMarks(u8);
160
161impl InlineMarks {
162 /// Every kind, in the order [`InlineMarks::iter`] yields them.
163 const ALL: [InlineKind; 8] = [
164 InlineKind::Strong,
165 InlineKind::Emph,
166 InlineKind::Verbatim,
167 InlineKind::Mark,
168 InlineKind::Superscript,
169 InlineKind::Subscript,
170 InlineKind::Insert,
171 InlineKind::Delete,
172 ];
173
174 pub const fn empty() -> Self {
175 InlineMarks(0)
176 }
177
178 /// Private: the set is an *answer*, and adding a mark to it doesn't mark
179 /// anything ([`Doc::toggle`] does that). `FromIterator` is the way in.
180 fn insert(&mut self, kind: InlineKind) {
181 self.0 |= Self::bit(kind);
182 }
183
184 /// Flip `kind` in the set — the sticky-marks toggle at a collapsed caret.
185 fn flip(&mut self, kind: InlineKind) {
186 self.0 ^= Self::bit(kind);
187 }
188
189 /// The symmetric difference: which marks differ between the two sets. Used
190 /// to resolve the marks already in force at the caret against the pending
191 /// delta — a bit set in the delta flips the base mark for the next keystroke.
192 fn xor(self, other: InlineMarks) -> InlineMarks {
193 InlineMarks(self.0 ^ other.0)
194 }
195
196 /// Whether `kind` is in force — the toolbar's "is Bold active?".
197 pub fn contains(self, kind: InlineKind) -> bool {
198 self.0 & Self::bit(kind) != 0
199 }
200
201 pub fn is_empty(self) -> bool {
202 self.0 == 0
203 }
204
205 /// The marks in force, for a frontend that renders whatever is on rather
206 /// than asking after a fixed list.
207 pub fn iter(self) -> impl Iterator<Item = InlineKind> {
208 Self::ALL.into_iter().filter(move |&k| self.contains(k))
209 }
210
211 fn bit(kind: InlineKind) -> u8 {
212 1 << match kind {
213 InlineKind::Strong => 0,
214 InlineKind::Emph => 1,
215 InlineKind::Verbatim => 2,
216 InlineKind::Mark => 3,
217 InlineKind::Superscript => 4,
218 InlineKind::Subscript => 5,
219 InlineKind::Insert => 6,
220 InlineKind::Delete => 7,
221 }
222 }
223}
224
225impl FromIterator<InlineKind> for InlineMarks {
226 fn from_iter<I: IntoIterator<Item = InlineKind>>(iter: I) -> Self {
227 let mut m = InlineMarks::empty();
228 for k in iter {
229 m.insert(k);
230 }
231 m
232 }
233}
234
235/// What kind of edit produced an undo group. Same-kind edits in a row coalesce
236/// into one undo step (a run of typed characters undoes together); `Other` never
237/// coalesces, so a paste, format toggle, or block change is always its own step.
238#[derive(Clone, Copy, PartialEq, Eq)]
239enum EditKind {
240 Insert,
241 Delete,
242 /// One step of an IME composition — see [`Doc::edit_composing`]. Its own kind
243 /// rather than `Insert`'s because a composition is not typing: each step
244 /// *replaces* the last (`か` → `かん` → `感`), so the run has to coalesce even
245 /// though no two steps insert the same bytes, and it must not fold into the
246 /// typed characters on either side of it.
247 Compose,
248 Other,
249}
250
251/// Which side of the caret a delete looks for an in-cell `<br>` break to swallow
252/// whole — see [`Doc::cell_break_at`]. `Backward` is Backspace (a break ending at
253/// the caret), `Forward` is Delete (one starting at it).
254#[derive(Clone, Copy)]
255enum BreakEdge {
256 Backward,
257 Forward,
258}
259
260/// A re-spelling of one inline mark run, held ready in case the edit about to
261/// happen breaks it — see [`Doc::mark_edge_fix`] and [`Doc::repair_mark_edges`].
262/// Every offset in it is in the coordinates the document will have *after* the
263/// plain edit, since that is when it may be applied.
264struct MarkEdgeFix {
265 /// The run's kind, and an offset inside what was its content: together they
266 /// answer "did the plain edit actually break this mark?" — the question that
267 /// decides whether any of this is applied at all.
268 kind: InlineKind,
269 probe: usize,
270 /// The byte range to re-spell (the run's delimiters included) and its new
271 /// spelling, with the edge whitespace moved outside the delimiters.
272 start: usize,
273 end: usize,
274 text: String,
275 /// Where the caret belongs afterwards — the same place on screen it would
276 /// have had, which is now on the other side of a delimiter.
277 caret: usize,
278 /// The marks in force for text typed at that caret. The caret can land
279 /// outside a run it was inside, and the marks have to survive the move or
280 /// the toolbar goes dark mid-word.
281 want: InlineMarks,
282}
283
284/// The caret and selection at one moment — the part of a history step twig's
285/// `Change` cannot carry, because the caret is leaf's state and twig only knows
286/// about bytes. leaf serializes it into the opaque per-state blob twig now
287/// stores in its own undo history (see `record_caret`), so undo and redo hand
288/// back the caret that matches the source they restore.
289#[derive(Clone, Copy)]
290struct CaretState {
291 caret: usize,
292 anchor: Option<usize>,
293}
294
295impl CaretState {
296 /// Pack into the fixed 17-byte blob leaf hands twig: the caret as a u64,
297 /// then an anchor-present flag and the anchor. twig copies these bytes and
298 /// never reads them.
299 fn to_blob(self) -> [u8; 17] {
300 let mut b = [0u8; 17];
301 b[..8].copy_from_slice(&(self.caret as u64).to_le_bytes());
302 if let Some(a) = self.anchor {
303 b[8] = 1;
304 b[9..].copy_from_slice(&(a as u64).to_le_bytes());
305 }
306 b
307 }
308
309 /// Recover a state from twig's blob, or `None` when it is empty or the wrong
310 /// length — a state twig restored that never had a caret set on it, which
311 /// leaves the caller to fall back to the edit site.
312 fn from_blob(b: &[u8]) -> Option<Self> {
313 let b: &[u8; 17] = b.try_into().ok()?;
314 let caret = u64::from_le_bytes(b[..8].try_into().unwrap()) as usize;
315 let anchor = (b[8] != 0).then(|| u64::from_le_bytes(b[9..].try_into().unwrap()) as usize);
316 Some(CaretState { caret, anchor })
317 }
318}
319
320/// A footnote reference and the note it names — the answer to
321/// [`Doc::footnote_at`].
322///
323/// The two `Option`s move together: a reference whose definition is missing has
324/// neither a body to show nor a place to jump to, and one that resolved has
325/// both.
326#[derive(Clone, PartialEq, Eq, Debug)]
327pub struct FootnoteRef {
328 /// The reference's label — the `1` of `[^1]`, with neither the `^` that
329 /// spells it a footnote nor the brackets around it.
330 pub label: String,
331 /// The note's body as source bytes (see
332 /// [`wysiwyg::footnote_body_span`](crate::wysiwyg)), or `None` when the
333 /// document defines no `[^label]:` to read one from.
334 pub text: Option<String>,
335 /// Where the note's *body* starts, for a "go to note" that moves the caret
336 /// there. `None` alongside a `None` `text`.
337 ///
338 /// The body rather than the definition, because this is an offset to put a
339 /// caret on and the `[^1]:` marker is decoration the caret can't occupy —
340 /// aiming at the definition's first byte snaps to the nearest real stop,
341 /// which is up in the paragraph above the note. It is also simply where a
342 /// reader following a reference wants to land: at the note's first word,
343 /// ready to read or amend it.
344 pub offset: Option<usize>,
345 /// Where the note's body ends, exclusive — so a frontend can ask which
346 /// *rendered rows* the note occupies and draw those instead of [`text`](Self::text).
347 ///
348 /// The rows are the note with its markup resolved: `see *later*` reaches a
349 /// frontend as an italic run, not as asterisks. `text` is the source bytes
350 /// and stays the honest answer for anything that wants the note as written
351 /// (a search index, a copy); this pair of offsets is for anything that wants
352 /// it as *read*. `None` alongside a `None` `offset`.
353 pub end: Option<usize>,
354}
355
356/// A footnote definition and the reference that sends a reader to it — the
357/// answer to [`Doc::footnote_definition_at`], and the other half of the round
358/// trip [`FootnoteRef`] starts.
359///
360/// A note is a place a reader *arrives*, so the useful thing to know while
361/// standing in one is the way back. Without this the jump to a note is a
362/// one-way door: the definitions sit at the foot of the document, so returning
363/// by hand means scrolling back up and finding the sentence again.
364#[derive(Clone, PartialEq, Eq, Debug)]
365pub struct FootnoteDef {
366 /// The definition's label — the `1` of `[^1]: …`, marker and colon stripped,
367 /// spelled exactly as [`FootnoteRef::label`] spells the same footnote's.
368 pub label: String,
369 /// Where the reference's *label* is, for a "back to reference" that moves
370 /// the caret there. `None` for a note nothing refers to — an orphan, which
371 /// is worth being able to say rather than silently doing nothing.
372 ///
373 /// The label rather than the reference's first byte, for
374 /// [`FootnoteRef::offset`]'s reason: a reference's brackets are decoration
375 /// and its label is the only part of it the caret can rest on.
376 ///
377 /// The *first* reference, when a label is cited more than once: a repeated
378 /// citation has no one true home, and the first is both the one a reader
379 /// most likely came from and the only choice that doesn't depend on how
380 /// they got here.
381 pub offset: Option<usize>,
382}
383
384/// Where a locator lands — the answer to [`Doc::locate`].
385///
386/// A locator (the `v2` of a `chapter.dj#v2`) names a *place* rather than a
387/// document, and a place is a span rather than a point: a reader following one
388/// wants the caret at its first byte, and a reader merely *peeking* at one wants
389/// the block it covers drawn. Both are served by carrying the whole span, and
390/// only one of the two can be recovered from an offset alone.
391#[derive(Clone, PartialEq, Eq, Debug)]
392pub struct Landing {
393 /// The first byte of the block the locator names — where a caret goes.
394 pub start: usize,
395 /// One past its last byte, so a frontend can map the pair through
396 /// [`VisualMap::row_range_for`](crate::wysiwyg::VisualMap::row_range_for) to the rendered rows the block occupies and draw
397 /// those, the way a footnote peek draws a note ([`FootnoteRef::end`]).
398 pub end: usize,
399}
400
401/// A selection cited out of the source: the text itself, up to a requested
402/// number of characters either side, and the byte range it came from. See
403/// [`Doc::selection_quote`].
404///
405/// The prefix and suffix are what make the quote *re-findable*: the same text
406/// can occur twice, and a little of what surrounded it is how a later reader —
407/// or the same document after an edit — tells the occurrences apart. The Web
408/// Annotation model calls this a `TextQuoteSelector`; the shape is older than
409/// the name.
410#[derive(Debug, Clone, PartialEq, Eq)]
411pub struct Quote {
412 /// The selected source, verbatim.
413 pub exact: String,
414 /// What immediately preceded it — possibly empty, at the document's start.
415 pub prefix: String,
416 /// What immediately followed it — possibly empty, at the document's end.
417 pub suffix: String,
418 /// Byte offset in the source where the selection begins.
419 pub start: usize,
420 /// Byte offset where it ends (exclusive).
421 pub end: usize,
422}
423
424/// A host-painted range of the source — an annotation's footprint, a search
425/// hit, a reviewer's mark. Leaf renders it (a background wash behind the
426/// glyphs whose source falls inside it) and hands back the `id` when the
427/// reader activates it; what the range *means* is entirely the host's.
428///
429/// Ranges are source bytes, like the caret and the selection, so a host that
430/// anchors quotes against the source ([`Doc::selection_quote`] is the other
431/// half of that loop) can paint what it found without any coordinate
432/// conversion. A range that drifts off the text it meant is the host's to
433/// re-anchor; leaf draws what it is told.
434#[derive(Debug, Clone, PartialEq, Eq)]
435pub struct Highlight {
436 /// Byte offset in the source where the wash begins.
437 pub start: usize,
438 /// Byte offset where it ends (exclusive).
439 pub end: usize,
440 /// The host's name for it, handed back on activation. Opaque to leaf.
441 pub id: String,
442 /// A rendering hint the frontend maps — a `#RRGGBB` hex string, or
443 /// nothing for the theme's default wash.
444 pub color: Option<String>,
445 /// A margin glyph's name, or nothing for wash-only ink. A highlight with
446 /// a marker gets a small glyph in the margin beside its first line, and
447 /// the glyph — not the wash — is what activates it: the wash is ink, the
448 /// marker is the control, which is what lets a reader put a caret in (or
449 /// copy from) annotated text without a card leaping at them. The name is
450 /// opaque to leaf; an Apple frontend reads it as an SF Symbol, a web one
451 /// as a class.
452 pub marker: Option<String>,
453}
454
455impl Highlight {
456 /// The range covering source `offset` in a list [`Doc::set_highlights`]
457 /// sorted, first by start where several overlap — the one place that
458 /// question is answered, for the frontends that paint by asking it as well
459 /// as for [`Doc::highlight_at`].
460 ///
461 /// The list is sorted by `(start, end)`, so the scan can stop at the first
462 /// range starting past `offset` rather than running to the end. A painter
463 /// asking once per glyph wants [`HighlightCursor`] instead; this is the
464 /// one-shot form, for the host asking what the reader just activated.
465 pub fn covering(highlights: &[Highlight], offset: usize) -> Option<&Highlight> {
466 highlights
467 .iter()
468 .take_while(|h| h.start <= offset)
469 .find(|h| offset < h.end)
470 }
471}
472
473/// [`Highlight::covering`] for a caller walking the document in order — which
474/// is every painter, since a frontend draws rows top to bottom and glyphs left
475/// to right.
476///
477/// The one-shot form is a scan from the front of the list per glyph, and a
478/// document with two hundred search hits pays that two hundred times a row. A
479/// range that ends at or before an offset can never cover that offset *or any
480/// later one*, so the cursor retires those permanently and each glyph costs the
481/// ranges that actually reach it. The answer is identical to
482/// [`Highlight::covering`]'s, offset for offset — this is the same scan with
483/// the part that was being redone dropped, not a cheaper approximation.
484///
485/// Offsets are expected to arrive non-decreasing. One that goes backwards is
486/// still answered correctly: the cursor re-seats to the front, since a painter
487/// that revisits a row is asking a question the retired ranges may own again.
488pub struct HighlightCursor<'a> {
489 highlights: &'a [Highlight],
490 /// The first range not yet retired.
491 at: usize,
492 /// The last offset asked about, to notice a caller going backwards.
493 last: usize,
494}
495
496impl<'a> HighlightCursor<'a> {
497 pub fn new(highlights: &'a [Highlight]) -> Self {
498 HighlightCursor {
499 highlights,
500 at: 0,
501 last: 0,
502 }
503 }
504
505 /// The range covering `offset`, advancing the cursor past every range that
506 /// can no longer cover anything.
507 pub fn at(&mut self, offset: usize) -> Option<&'a Highlight> {
508 if offset < self.last {
509 self.at = 0;
510 }
511 self.last = offset;
512 while self
513 .highlights
514 .get(self.at)
515 .is_some_and(|h| h.end <= offset)
516 {
517 self.at += 1;
518 }
519 Highlight::covering(&self.highlights[self.at..], offset)
520 }
521}
522
523/// The identity of a built [`VisualMap`] — see [`Doc::visual_key`]. Opaque on
524/// purpose: the only useful question is whether two of them are the same map,
525/// and the tuple behind it (revision, wrap, reveal line) is core's business.
526#[derive(Clone, PartialEq, Eq, Debug)]
527pub struct VisualKey(Option<(u64, Option<usize>, Option<Range<usize>>)>);
528
529pub struct Doc {
530 editor: Editor,
531 pub format: Format,
532 pub path: PathBuf,
533 /// Current source, refreshed from the editor after every successful edit.
534 pub source: String,
535 /// The caret, as a byte offset into `source` (always on a char boundary).
536 pub caret: usize,
537 /// The selection's fixed end, if a selection is active; the moving end is
538 /// the caret. `None` means no selection.
539 pub anchor: Option<usize>,
540 pub dirty: bool,
541 pub status: Option<String>,
542 pub view: View,
543 /// Whether the document refuses to change — a *reading* surface over the
544 /// same rendering, selection, and navigation the editor has.
545 ///
546 /// Enforced here rather than by each frontend hiding its input paths,
547 /// because every mutation funnels through a few doors —
548 /// [`splice_exact`](Self::splice_exact), [`undo`](Self::undo),
549 /// [`redo`](Self::redo), and the handful of inserts that go to twig's own
550 /// verbs directly rather than through the splice (a typed literal, a link,
551 /// an image, a rule, a footnote, a cell's line break) — and guarded doors
552 /// are a guarantee where a frontend's suppressed keyboard is a hope. A
553 /// gated door reports exactly like a rolled-back splice, a path every
554 /// caller already handles. `a_read_only_document_refuses_every_door` is
555 /// the list; a new `self.editor.insert_*` call belongs on it.
556 read_only: bool,
557 /// The host-painted ranges, kept sorted by start — see [`Highlight`].
558 /// State like the selection rather than like the text: no edit history,
559 /// no dirty bit, redrawn from whatever the host last set.
560 highlights: Vec<Highlight>,
561 /// How much of the source markup the rich view exposes — a frontend preference (see
562 /// [`MarkupMode`]). Its two axes are read apart: the rendering one by
563 /// [`reveal_line`](Self::reveal_line), the editing one by
564 /// [`insert`](Self::insert).
565 markup_mode: MarkupMode,
566 /// Whether soft breaks fold into the reflowed paragraph or render where
567 /// they were written (see [`LineFlow`]) — an independent frontend
568 /// preference the WYSIWYG builder consults when it lays out a block.
569 line_flow: LineFlow,
570 /// The kind of the last edit, for coalescing: twig owns the undo *history*
571 /// (see `undo`/`redo`), but "what counts as one undo step" is a frontend-UX
572 /// call, so leaf decides when a run continues and tells twig to coalesce.
573 last_edit_kind: Option<EditKind>,
574 /// The inline marks the user has toggled *at a collapsed caret* with no
575 /// selection — "start typing bold here". Held as the XOR delta from the marks
576 /// already in force at [`pending_at`](Self::pending_at): a set bit means
577 /// "flip this kind for the next typed text", so it both turns a mark on where
578 /// none is (type into bold) and off where one already covers the caret (type
579 /// past the bold you're standing in). [`Doc::insert`] realises it onto the
580 /// freshly typed text and then clears it — a mark once realised is carried by
581 /// the caret sitting inside the run, not by this delta.
582 pending_marks: InlineMarks,
583 /// The caret offset [`pending_marks`](Self::pending_marks) applies to. The
584 /// delta is live only while the caret still stands here with no selection;
585 /// any motion or edit ([`move_to`](Self::move_to), a splice, a click) drops
586 /// it, so a toggled-but-never-typed format doesn't leak onto text elsewhere.
587 pending_at: Option<usize>,
588 /// The source as of the last open/save — `dirty` is `source != clean_source`,
589 /// so undoing back to the saved state correctly clears the modified flag.
590 clean_source: String,
591 /// A hash of the bytes leaf last read from `path` or wrote to it; `None`
592 /// while the document has no file behind it. [`Doc::disk_state`] compares
593 /// the file against this to catch an edit made *outside* leaf before a save
594 /// silently overwrites it — `clean_source` only knows what leaf itself did.
595 ///
596 /// A hash, not an mtime: mtime is the cheap answer and the wrong one — two
597 /// writes inside one filesystem timestamp tick are indistinguishable, a
598 /// clock that steps backwards (or a writer that restores an mtime) hides a
599 /// real change, and a `touch` invents one. The whole point of the watermark
600 /// is to not clobber someone's work, so it reads the bytes and compares what
601 /// is actually there. That costs a file read per question, which is why the
602 /// question is asked on a user event (focus, save) and not every frame.
603 disk_hash: Option<u64>,
604 /// The "sticky" display column vertical motion aims for, in the active
605 /// view's grid. Set on the first `move_up`/`move_down` of a run and
606 /// reused by every subsequent one in that run, so passing through a
607 /// shorter line doesn't permanently forget the original column. Any
608 /// horizontal motion or edit clears it.
609 ///
610 /// A column, not a character index: dropping down a line of `你好` onto one
611 /// of ASCII has to land under the glyph the caret was drawn beneath, which
612 /// is the only thing the user can see to aim by. Where the goal falls inside
613 /// a wide character on the target line, the mapping resolves it to that
614 /// character — the caret lands on it rather than between its cells.
615 goal_col: Option<usize>,
616 /// The rendered map for the WYSIWYG view; empty in the source view. Movement
617 /// and clicks read it to stay in visible space.
618 pub vmap: VisualMap,
619 /// The syntax map for the source view; empty in the WYSIWYG view, which
620 /// styles resolved glyphs instead. Built by [`Doc::build_source`] — a
621 /// frontend that never calls it paints raw source unstyled, which is what
622 /// every frontend did before this map existed.
623 pub smap: SourceMap,
624 /// The revision `smap` was built from, or `None` before the first build.
625 /// The map is a pure function of the text alone — no width, no caret, no
626 /// reveal line — so unlike [`vmap_key`](Self::vmap_key) the revision is the
627 /// whole key.
628 smap_key: Option<u64>,
629 /// Everything the map is built from, as one number: bumped whenever the
630 /// document's text changes, and never by a motion, a selection, or a save.
631 /// A frontend can hold work against it — see [`Doc::revision`].
632 revision: u64,
633 /// How many history steps stand behind the caret, and how many ahead of
634 /// it — the answer to a native Edit menu's "may Undo be enabled?", which
635 /// twig's history does not ask itself. Counted at [`refresh`](Self::refresh),
636 /// the funnel every edit comes through, and moved back and forth by
637 /// [`undo`](Self::undo)/[`redo`](Self::redo). An upper bound rather than
638 /// an exact depth: a coalesced run of typing is one of twig's steps but
639 /// several of these, and twig's own cap on history is not mirrored here.
640 /// Neither error can make `can_undo` false while a step remains, which is
641 /// the only property a menu needs; the one place the bound can be wrong the
642 /// other way — the cap has retired every step — is reconciled the moment
643 /// twig reports nothing to undo.
644 undo_steps: usize,
645 redo_steps: usize,
646 /// What `vmap` was built from, or `None` before the first build. The map is
647 /// a pure function of `(revision, wrap, reveal line)`, so when those haven't
648 /// moved, rebuilding it produces the identical map — see
649 /// [`Doc::build_visual`].
650 ///
651 /// The reveal line ([`Doc::reveal_line`]) is the caret's, and is `None` in
652 /// every mode but [`MarkupMode::Full`] — so outside that mode the key is
653 /// text and width alone, and a caret motion still rebuilds nothing.
654 vmap_key: Option<(u64, Option<usize>, Option<Range<usize>>)>,
655 /// Per-block row cache backing the incremental rebuild: when the text
656 /// changes, only the top-level blocks whose bytes moved are re-rendered and
657 /// the rest are reused shifted (see [`wysiwyg::BlockCache`]). Persists across
658 /// builds; a pure accelerator, so it's never read for correctness.
659 block_cache: wysiwyg::BlockCache,
660 /// How many visual rows each block image reserves, keyed by its destination —
661 /// set by the frontend through [`Doc::set_media_rows`] once it has decoded and
662 /// measured the pictures. Core does no image I/O, so this is the only way it
663 /// learns a picture's height; a destination not in the map reserves the bare
664 /// one-row placeholder. Threaded into the builder so [`wysiwyg::build_cached`]
665 /// sizes each placeholder, and folded into `vmap_key` so a height change
666 /// rebuilds the map.
667 media_rows: HashMap<String, usize>,
668
669 // View geometry the renderer stamps each frame, so mouse events can map a
670 // screen cell back to a byte offset.
671 pub scroll: usize,
672 pub body_origin: (u16, u16),
673 /// Width of the body rectangle last painted by the frontend. Zero means
674 /// unknown (used by tests or a frontend that has not drawn yet).
675 pub body_width: u16,
676 pub body_height: u16,
677 /// The caret as of the last frame drawn, or `None` before the first.
678 ///
679 /// Scrolling is the viewport's business, not the caret's: the view follows
680 /// the caret when the caret *moves*, but a wheel that doesn't touch the
681 /// caret has to be free to scroll away from it — otherwise the view is
682 /// pinned to the caret and stops dead at the edge of the document you can
683 /// see. Comparing against this is what tells the two apart, and it catches a
684 /// caret set by any route, including a frontend assigning the field itself.
685 pub drawn_caret: Option<usize>,
686}
687
688/// The Markdown extensions every leaf document is parsed with — four of them,
689/// each departing from twig's defaults for a reason leaf can state.
690///
691/// `html_elements` promotes embedded raw HTML (`<img>`, `<picture>`,
692/// `<source>`, …) into semantic AST nodes, so a picture becomes a real `image`
693/// node the frontends can frame and rasterize instead of opaque `raw_block`
694/// text. `directives` turns on generic `:::name{.class}` fenced-div containers
695/// (`directive` nodes), which a host app uses for its own semantics (diaryx's
696/// `:::vis{.audience}` visibility blocks) — core renders any directive as a
697/// plain tinted container, agnostic of `name`.
698///
699/// `highlight` and `highlight_colors` are the pair that makes Markdown read
700/// `==text==` as a `mark` node, and `==🔴 text==` as one carrying a
701/// `data-color`. leaf already had somewhere to put both: the
702/// [`Mark`](crate::Role::Mark) role and the ⌘⇧M highlight button predate them,
703/// and until twig 3.3 a Markdown document could only ever *receive* a highlight
704/// from a Djot one it was converted from — the button wrote `==…==` and the
705/// reparse read it straight back as text.
706/// They are on together because a colour is inert without the highlight itself,
707/// and a document that writes `==🔴 x==` means the colour by it.
708///
709/// Every flag is inert for non-Markdown formats, so it's safe to pass them
710/// unconditionally. Threading this through every constructor (not just `open`)
711/// keeps `from_source`, `blank`, and `reload` parsing the same document the same
712/// way — twig reparses with these same flags after each edit.
713pub(crate) fn parse_extensions() -> MarkdownExtensions {
714 MarkdownExtensions {
715 html_elements: true,
716 directives: true,
717 highlight: true,
718 highlight_colors: true,
719 ..Default::default()
720 }
721}
722
723/// Build an editor over `bytes` in `format` with leaf's [`parse_extensions`],
724/// mapping twig's error into the `anyhow` context every constructor shares.
725fn new_editor(bytes: &[u8], format: Format) -> Result<Editor> {
726 Editor::new_ext(bytes, format, parse_extensions()).map_err(|e| anyhow!("twig parse: {e}"))
727}
728
729/// Does `format` spell a table as a **pipe table** — the one grid twig's table
730/// editor knows how to emit?
731///
732/// This is the single capability leaf still has to answer for itself, and the
733/// only hand-maintained format list left in this file. Every other gesture is
734/// [`Format::supports`], which is twig's own answer read across the C ABI — but
735/// twig deliberately leaves the table ops out of that query, because they read
736/// no `Syntax` table at all. They rewrite a grid that is already in the source
737/// and refuse on *position*, never on format. Handed a caret inside an HTML
738/// `<table>`, `table_insert_row` therefore re-emits the whole element as
739/// `| a | b |` and reports success — a real splice, a clean reparse, an honest
740/// `dirty` flag, and nothing downstream able to tell it from a good edit.
741///
742/// So the list is narrow on purpose. `Format` is `#[non_exhaustive]`, and the
743/// wildcard answers "no" for a format leaf has never heard of: a new twig
744/// language that *does* spell pipe tables loses its grid controls until this
745/// line is updated, which shows up as a missing button. The other default hands
746/// it to [`Doc::table_op`], which rewrites documents it cannot spell.
747fn spells_pipe_tables(format: Format) -> bool {
748 matches!(format, Format::Markdown | Format::Djot)
749}
750
751/// Which of leaf's authoring controls this document's format can actually
752/// spell — one flag per toolbar button, resolved once so a frontend can build
753/// its chrome instead of discovering each refusal on a click.
754///
755/// Every field but [`table`](Self::table) is `Format::supports` on the gesture
756/// the matching [`Doc`] method calls, so this record cannot drift from what the
757/// ops do; `table` is [`spells_pipe_tables`], the one answer twig doesn't
758/// export.
759///
760/// **The formats are ragged, and that is the point.** A single per-document
761/// boolean was enough while the two authorable formats were Markdown and djot
762/// and everything else spelled nothing. HTML is neither: it writes seven of the
763/// eight inline marks as a tag pair, plus `<code>`, `<hr>` and an in-cell
764/// `<br>`, and spells no heading marker, no line prefix, no fence, no task box,
765/// no link — because its versions of those have a different *shape*, not a
766/// different alphabet. So ⌘B works in an HTML document and ⌘1 does not, and no
767/// one flag can say that. Markdown and djot differ from each other too:
768/// `==mark==` is djot-only, and an in-cell `<br>` is Markdown-only.
769#[derive(Clone, Copy, Debug, Eq, PartialEq)]
770pub struct Capabilities {
771 /// ⌘B — `InlineKind::Strong`.
772 pub bold: bool,
773 /// ⌘I — `InlineKind::Emph`.
774 pub italic: bool,
775 /// Inline code — `InlineKind::Verbatim`.
776 pub code: bool,
777 /// Highlight — `InlineKind::Mark`. Djot spells it; Markdown does not.
778 pub mark: bool,
779 /// ⌘U — `InlineKind::Insert`, which every format that marks at all spells.
780 pub underline: bool,
781 /// Strikethrough — `InlineKind::Delete`.
782 pub strike: bool,
783 pub superscript: bool,
784 pub subscript: bool,
785 /// Heading levels and "make this a paragraph" — [`Doc::set_block`].
786 pub heading: bool,
787 pub blockquote: bool,
788 pub bullet_list: bool,
789 pub ordered_list: bool,
790 /// The checkbox controls: giving an item a box, and ticking one.
791 pub task: bool,
792 pub link: bool,
793 /// Covers [`Doc::insert_media`] too — see the note there on why the three
794 /// media kinds stand or fall together.
795 pub image: bool,
796 /// The horizontal-rule button. HTML spells this one (`<hr>`).
797 pub thematic_break: bool,
798 /// The footnote button — [`Doc::insert_footnote`]. Markdown and djot spell
799 /// the pair; HTML has no footnote of its own, so the button goes away rather
800 /// than writing brackets that would render as brackets.
801 pub footnote: bool,
802 /// Setting a fenced block's language — a control only ever offered with the
803 /// caret already in a fence.
804 pub code_language: bool,
805 /// The grid controls: insert/delete/move a row or column, set a column's
806 /// alignment. Pair with [`Doc::caret_in_table`], which asks the other
807 /// question — an HTML `<table>` holds the caret and still can't be edited.
808 pub table: bool,
809 /// Shift+Return inside a cell. Markdown and HTML spell it; djot has no
810 /// idiomatic in-cell break.
811 pub cell_line_break: bool,
812}
813
814impl Capabilities {
815 /// Resolve every flag for `format`. Pure and cheap — twig computes each from
816 /// a static table — but a frontend that wants to hold them can.
817 pub fn of(format: Format) -> Self {
818 let inline = |k| format.supports(Gesture::ToggleInline(k));
819 let container = |k| format.supports(Gesture::ToggleBlockContainer(k));
820 Self {
821 bold: inline(InlineKind::Strong),
822 italic: inline(InlineKind::Emph),
823 code: inline(InlineKind::Verbatim),
824 mark: inline(InlineKind::Mark),
825 underline: inline(InlineKind::Insert),
826 strike: inline(InlineKind::Delete),
827 superscript: inline(InlineKind::Superscript),
828 subscript: inline(InlineKind::Subscript),
829 heading: format.supports(Gesture::SetBlock),
830 blockquote: container(BlockContainerKind::BlockQuote),
831 bullet_list: container(BlockContainerKind::BulletList),
832 ordered_list: container(BlockContainerKind::OrderedList),
833 // Both halves of the checkbox story, and leaf offers no control that
834 // needs only one: the item gesture mints the box, the checked one
835 // ticks it, and a format spelling a `task_marker` spells both.
836 task: format.supports(Gesture::ToggleTaskItem)
837 && format.supports(Gesture::ToggleTaskChecked),
838 link: format.supports(Gesture::InsertLink),
839 image: format.supports(Gesture::InsertImage),
840 thematic_break: format.supports(Gesture::InsertThematicBreak),
841 footnote: format.supports(Gesture::InsertFootnote),
842 code_language: format.supports(Gesture::SetCodeLanguage),
843 table: spells_pipe_tables(format),
844 cell_line_break: format.supports(Gesture::InsertLineBreak),
845 }
846 }
847}
848
849impl Doc {
850 #[cfg(feature = "fs")]
851 pub fn open(path: PathBuf) -> Result<Self> {
852 let bytes = std::fs::read(&path).with_context(|| format!("reading {}", path.display()))?;
853 Self::from_disk_bytes(path, bytes)
854 }
855
856 /// An empty document *named* `path`, for a file that isn't there yet — what
857 /// every other terminal editor gives you when you name a file that doesn't
858 /// exist. It is a real named document, not a [`Doc::blank`]: `is_untitled`
859 /// is false, so ⌘S writes straight to `path` with no Save As detour, and
860 /// the header shows the name the user asked for.
861 ///
862 /// The format comes from the extension, exactly as [`Doc::open`] reads it —
863 /// so `leaf notes.dj` starts a djot buffer rather than the Markdown
864 /// [`Doc::blank`] has to assume for want of a name. An extension leaf can't
865 /// parse is still an error: a mistyped flag or a stray argument should say
866 /// so, not open a buffer promising to save somewhere.
867 ///
868 /// The watermark is the hash of *no bytes*, not `None`, and that is the
869 /// whole trick: `None` means untitled, and would leave [`Doc::disk_state`]
870 /// answering [`DiskState::Untitled`] for a document that has a path and
871 /// intends to write to it. Hashing `""` instead makes the answers the true
872 /// ones — [`DiskState::Missing`] while the file still isn't there (a save
873 /// recreates it, which is exactly what this is for), and
874 /// [`DiskState::Changed`] if somebody creates it underneath us between
875 /// launch and save, so the frontend's overwrite prompt guards a new file as
876 /// it guards an opened one.
877 ///
878 /// Nothing is written here. A buffer that is never typed into never touches
879 /// the filesystem, and a `path` whose directory doesn't exist is allowed to
880 /// open — the write is where that fails, and it says so then.
881 #[cfg(feature = "fs")]
882 pub fn create(path: PathBuf) -> Result<Self> {
883 Self::from_disk_bytes(path, Vec::new())
884 }
885
886 /// [`Doc::open`] when the file is there, [`Doc::create`] when it isn't —
887 /// the call a CLI frontend wants for its path argument.
888 ///
889 /// The decision is made from the failed read itself rather than a `exists()`
890 /// check first, so there is no window between the two for the file to appear
891 /// or vanish in. Only `NotFound` opens a new buffer: a permissions error or
892 /// a directory in the way is still an error, because pretending those are
893 /// "no file yet" would offer to save over something leaf couldn't read.
894 #[cfg(feature = "fs")]
895 pub fn open_or_create(path: PathBuf) -> Result<Self> {
896 match std::fs::read(&path) {
897 Ok(bytes) => Self::from_disk_bytes(path, bytes),
898 Err(e) if e.kind() == std::io::ErrorKind::NotFound => Self::create(path),
899 Err(e) => Err(e).with_context(|| format!("reading {}", path.display())),
900 }
901 }
902
903 /// The shared body of [`Doc::open`] and [`Doc::create`]: bytes that are (or
904 /// stand in for) the file at `path`, parsed as the format its extension
905 /// names. Keeping the two on one path is what makes a new file's document
906 /// identical in every respect to an opened one but its contents.
907 #[cfg(feature = "fs")]
908 fn from_disk_bytes(path: PathBuf, bytes: Vec<u8>) -> Result<Self> {
909 let format = detect_format(&path)?;
910 let editor = new_editor(&bytes, format)?;
911 let source = String::from_utf8(bytes).map_err(|_| anyhow!("document is not UTF-8"))?;
912 let disk_hash = Some(hash_bytes(source.as_bytes()));
913 // Store the document's *absolute* path. A relative one (`leaf README.md`)
914 // has an empty parent, so a frontend can't resolve a relative image
915 // destination (``) against the document's directory and the
916 // picture silently falls back to its text placeholder. `absolute` is
917 // purely lexical — it prefixes the current directory and normalizes, but
918 // reads nothing and resolves no symlinks — so `file_name` and save are
919 // unchanged; it only gives `path.parent()` something to join against.
920 let path = std::path::absolute(&path).unwrap_or(path);
921 Ok(Doc::from_parts(editor, format, path, source, disk_hash))
922 }
923
924 /// Build a document from an in-memory string, the format named explicitly —
925 /// the portable, filesystem-free counterpart to [`Doc::open`] (which reads a
926 /// path and sniffs the format from its extension). A wasm or FFI host, which
927 /// has no path to read, uses this: it hands over bytes it fetched however it
928 /// could, and later persists [`Doc::source`] however it can (a browser
929 /// download, `localStorage`, a backend `PUT`) and calls [`Doc::mark_saved`].
930 ///
931 /// No file backs the result, so it starts untitled ([`Doc::is_untitled`] is
932 /// true) exactly like a [`Doc::blank`] that has been given content.
933 pub fn from_source(source: String, format: Format) -> Result<Self> {
934 let editor = new_editor(source.as_bytes(), format)?;
935 Ok(Doc::from_parts(
936 editor,
937 format,
938 PathBuf::new(),
939 source,
940 None,
941 ))
942 }
943
944 /// An untitled, empty document — the `+` button and a `leaf` launched with
945 /// no file argument. Nothing on disk backs it until a [`Doc::save_as`].
946 ///
947 /// It is Markdown, because a format has to be chosen before a name exists to
948 /// read one from: `detect_format` reads the extension and an untitled
949 /// document has neither. Markdown is what leaf's own files are, what its
950 /// block markers are already written for (`insert_block_prefix`), and the
951 /// extension a Save As will overwhelmingly pick — a wrong guess here would
952 /// mean typing djot into a buffer parsing it as Markdown. Note that Save As
953 /// *doesn't* revisit this: see [`Doc::save_as`].
954 pub fn blank() -> Result<Self> {
955 let format = Format::Markdown;
956 let editor = new_editor(b"", format)?;
957 // An empty `path` is the untitled marker (`path` is a public `PathBuf`
958 // field two frontends already read; making it an `Option` to say this
959 // would break both). `is_untitled` is the question to ask, not the
960 // representation to copy.
961 Ok(Doc::from_parts(
962 editor,
963 format,
964 PathBuf::new(),
965 String::new(),
966 None,
967 ))
968 }
969
970 /// The fields every constructor agrees on, so `open` and `blank` can't drift
971 /// apart in the ones neither of them has an opinion about.
972 fn from_parts(
973 editor: Editor,
974 format: Format,
975 path: PathBuf,
976 source: String,
977 disk_hash: Option<u64>,
978 ) -> Self {
979 Doc {
980 editor,
981 format,
982 path,
983 disk_hash,
984 clean_source: source.clone(),
985 source,
986 caret: 0,
987 anchor: None,
988 dirty: false,
989 status: None,
990 read_only: false,
991 highlights: Vec::new(),
992 // leaf opens in the rich-text (WYSIWYG) view by default — the
993 // markup-resolved surface is leaf's differentiator. Frontends can
994 // still start in source view explicitly (e.g. a CLI flag), and ⌘e/⌥w
995 // toggles at runtime.
996 view: View::Wysiwyg,
997 // `None` by default — the clean surface Diaryx ships, with typed
998 // syntax kept literal; a markup-fluent frontend can climb the
999 // ladder to `Shortcuts` or `Full`.
1000 markup_mode: MarkupMode::default(),
1001 // Fold by default — flowing prose that reflows to the viewport, the
1002 // behaviour every frontend had before this preference existed.
1003 line_flow: LineFlow::default(),
1004 last_edit_kind: None,
1005 pending_marks: InlineMarks::empty(),
1006 pending_at: None,
1007 goal_col: None,
1008 vmap: VisualMap::default(),
1009 smap: SourceMap::default(),
1010 // No map yet — the first `build_source` always builds.
1011 smap_key: None,
1012 revision: 0,
1013 undo_steps: 0,
1014 redo_steps: 0,
1015 // No map yet — the first `build_visual` always builds.
1016 vmap_key: None,
1017 block_cache: wysiwyg::BlockCache::default(),
1018 media_rows: HashMap::new(),
1019 scroll: 0,
1020 body_origin: (0, 0),
1021 body_width: 0,
1022 body_height: 0,
1023 drawn_caret: None,
1024 }
1025 }
1026
1027 /// Whether this document has no file behind it yet — a [`Doc::blank`] that
1028 /// has never been saved. The question a ⌘S handler asks to know it should
1029 /// open a Save As picker instead ([`Doc::save`] won't guess a name), and the
1030 /// header asks to know the name it shows is a placeholder.
1031 pub fn is_untitled(&self) -> bool {
1032 self.path.as_os_str().is_empty()
1033 }
1034
1035 pub fn toggle_view(&mut self) {
1036 self.view = match self.view {
1037 View::Source => View::Wysiwyg,
1038 View::Wysiwyg => View::Source,
1039 };
1040 self.scroll = 0;
1041 self.status = None;
1042 // Entering WYSIWYG, the caret may be sitting in now-hidden frontmatter;
1043 // lift it to the first rendered offset.
1044 self.clamp_caret();
1045 }
1046
1047 /// The current markup-exposure preference (see [`MarkupMode`]).
1048 pub fn markup_mode(&self) -> MarkupMode {
1049 self.markup_mode
1050 }
1051
1052 /// Set the markup-exposure preference. Both of its axes take effect at
1053 /// once: the editing one on the next [`insert`](Self::insert), and the
1054 /// rendering one on the next build — which is why this drops the cached
1055 /// visual map and the per-block render cache, exactly as
1056 /// [`set_line_flow`](Self::set_line_flow) does.
1057 pub fn set_markup_mode(&mut self, mode: MarkupMode) {
1058 if self.markup_mode == mode {
1059 return;
1060 }
1061 self.markup_mode = mode;
1062 // Neither cache is keyed on the mode, and moving between `Full` and the
1063 // hidden modes changes every row the caret's line renders to — so
1064 // invalidate both explicitly.
1065 self.vmap_key = None;
1066 self.block_cache = wysiwyg::BlockCache::default();
1067 }
1068
1069 /// The source byte range of the line the caret sits on, when that line
1070 /// should render its raw delimiters — `None` in every mode and view that
1071 /// hides them, which is what the builder reads as "reveal nothing".
1072 ///
1073 /// A *source* line (newline to newline), not a visual row: a wrapped
1074 /// paragraph and a `LineFlow::Preserve` soft break both split one source
1075 /// line across several rows, and revealing half a delimiter pair because the
1076 /// other half wrapped would be worse than revealing neither. The range
1077 /// excludes the terminating newline and is empty-but-present on a blank
1078 /// line, which reveals nothing but still keys the caches correctly.
1079 ///
1080 /// Only in [`View::Wysiwyg`]: source view already shows every byte, so
1081 /// there is nothing there to reveal.
1082 pub(crate) fn reveal_line(&self) -> Option<Range<usize>> {
1083 if !self.markup_mode.reveals_caret_line() || self.view != View::Wysiwyg {
1084 return None;
1085 }
1086 Some(source_line_range(&self.source, self.caret))
1087 }
1088
1089 /// The current soft-break flow preference (see [`LineFlow`]).
1090 pub fn line_flow(&self) -> LineFlow {
1091 self.line_flow
1092 }
1093
1094 /// Set the soft-break flow preference. The mode changes how every block lays
1095 /// out, so a change drops the cached visual map and the per-block render
1096 /// cache, forcing the next [`build_visual`] to rebuild under the new flow.
1097 ///
1098 /// [`build_visual`]: Self::build_visual
1099 pub fn set_line_flow(&mut self, mode: LineFlow) {
1100 if self.line_flow == mode {
1101 return;
1102 }
1103 self.line_flow = mode;
1104 // Both caches are keyed on `(revision, wrap)`, neither of which moved —
1105 // so invalidate them explicitly, or the next build would reuse rows laid
1106 // out under the old flow.
1107 self.vmap_key = None;
1108 self.block_cache = wysiwyg::BlockCache::default();
1109 }
1110
1111 pub fn view_name(&self) -> &'static str {
1112 match self.view {
1113 View::Source => "source",
1114 View::Wysiwyg => "wysiwyg",
1115 }
1116 }
1117
1118 /// Rebuild the WYSIWYG visual map for the current tree at `width` columns
1119 /// (called by the renderer each frame it's in the WYSIWYG view).
1120 /// Build the WYSIWYG map, wrapped at `width` display columns.
1121 ///
1122 /// Cheap to call every frame, which is what both frontends do: the map is a
1123 /// pure function of the document and the wrap width, so a call that would
1124 /// rebuild the same map returns the one already built. Only an edit (or a
1125 /// resize) pays.
1126 ///
1127 /// That isn't a micro-optimisation. A frontend repaints for reasons that have
1128 /// nothing to do with the text — a blinking caret, a scroll, a focus change —
1129 /// and rebuilding here is O(document): 23 ms on a 1 MB file, of which 5 ms is
1130 /// marshalling twig's AST across the C ABI. Paid twice a second by the GUI's
1131 /// blink timer, that was 14% of a core spent redrawing an unchanged document.
1132 /// (`cargo run --release -p leaf-core --example bench` for the numbers.)
1133 pub fn build_visual(&mut self, width: usize) {
1134 self.build_map(Some(width));
1135 }
1136
1137 /// Build the WYSIWYG map with each block as a single unwrapped row — for a
1138 /// frontend (the GUI) that wraps at its own proportional pixel width rather
1139 /// than a fixed character column.
1140 pub fn build_visual_unwrapped(&mut self) {
1141 self.build_map(None);
1142 }
1143
1144 /// Build the source view's syntax map ([`Doc::smap`]) — the styling for
1145 /// [`View::Source`], the way [`build_visual`](Self::build_visual) is the
1146 /// styling for [`View::Wysiwyg`].
1147 ///
1148 /// A frontend calls this before painting raw source. One that doesn't gets
1149 /// an empty map and paints unstyled text, so this is additive: nothing
1150 /// breaks by not calling it.
1151 ///
1152 /// Built at most once per revision, and the revision is the whole key — the
1153 /// map has no width and no caret in it, so it survives every resize, every
1154 /// motion, and every selection change.
1155 ///
1156 /// The builds it does do cost a whole-arena marshal, which is precisely what
1157 /// the WYSIWYG path works to avoid, so this has no incremental path where
1158 /// that one has two. From `cargo run --release -p leaf-core --example
1159 /// bench`, per keystroke, against the WYSIWYG build the source view is
1160 /// *not* doing:
1161 ///
1162 /// | size | nodes | marshal | `source::build` | (`wysiwyg::build`) |
1163 /// |------:|-------:|--------:|----------------:|-------------------:|
1164 /// | 10 KB| 613 | 0.16 ms| 0.07 ms | 0.28 ms |
1165 /// | 100 KB| 6 097 | 0.84 ms| 0.38 ms | 2.43 ms |
1166 /// | 1 MB| 60 601 | 5.67 ms| 3.12 ms | 23.39 ms |
1167 ///
1168 /// Linear, two thirds of it the marshal, and the build itself five to seven
1169 /// times cheaper than the one it stands in for at every size. Comfortable
1170 /// well past any document a person edits in a terminal — a megabyte is where
1171 /// it would want [`Editor::dirty_range`] and the same splice treatment
1172 /// `build_spliced` gives the other map. The door is open; nothing has needed
1173 /// it yet.
1174 pub fn build_source(&mut self) {
1175 if self.smap_key == Some(self.revision) {
1176 return;
1177 }
1178 let nodes = self.nodes();
1179 self.smap = source::build(&nodes, &self.source);
1180 self.smap_key = Some(self.revision);
1181 }
1182
1183 /// Tell the model how many visual rows each block image should reserve, keyed
1184 /// by the image's destination. A terminal frontend calls this once it has
1185 /// decoded and measured its pictures — core does no image I/O, so this is the
1186 /// only way it learns a height — and the next [`Doc::build_visual`] lays each
1187 /// placeholder out that tall (the label row plus blank filler rows the
1188 /// frontend paints the raster over). A destination left out of the map falls
1189 /// back to the bare one-row placeholder, which is also what a frontend that
1190 /// can't draw pictures (or lays them out in its own units, like the GUI) gets
1191 /// by never calling this.
1192 ///
1193 /// Cheap to call every frame with the same map: only a *change* invalidates
1194 /// the built map (and the block-row cache, since a height isn't part of a
1195 /// block's bytes and so wouldn't otherwise re-render it). Steady state is a
1196 /// no-op, so a frontend can just hand over its current measurements each frame.
1197 pub fn set_media_rows(&mut self, rows: HashMap<String, usize>) {
1198 if self.media_rows == rows {
1199 return;
1200 }
1201 self.media_rows = rows;
1202 // A height lives outside the block's source bytes, so the content-keyed
1203 // block cache would hand back the old-height rows on a hit. Drop it (and
1204 // the splice layout it carries) so the next build re-renders every block
1205 // at the new heights, and force that build by clearing the map key.
1206 self.block_cache = wysiwyg::BlockCache::default();
1207 self.vmap_key = None;
1208 }
1209
1210 /// The revision the document's text is at — bumped by every edit, undo,
1211 /// redo, and reload, and by nothing else. A frontend caches against this to
1212 /// tell a repaint that needs new work from one that doesn't.
1213 ///
1214 /// It counts *edits*, not distinct texts: typing `x` and deleting it again
1215 /// lands on the same text two revisions later. Work is only ever rebuilt
1216 /// needlessly, never wrongly reused.
1217 pub fn revision(&self) -> u64 {
1218 self.revision
1219 }
1220
1221 /// The identity of the map presently in [`vmap`](Self::vmap) — what the last
1222 /// [`build_visual`](Self::build_visual) built it from, or the identity of an
1223 /// unbuilt map before the first one.
1224 ///
1225 /// This is *not* [`revision`](Self::revision). The revision says where the
1226 /// text is; this says where the map is, and the two part company the moment
1227 /// an edit lands, until something rebuilds. A frontend that keeps its own
1228 /// copy of the map — leaf-ratatui stashes core's before splicing filler rows
1229 /// under an oversized heading — compares this against the value it held when
1230 /// it took the copy, and learns whether `vmap` is still the map it stashed
1231 /// or one somebody else has since rebuilt. Restoring a copy over a newer
1232 /// map would paint a stale document; restoring nothing hands core's
1233 /// incremental rebuild a map it never built.
1234 pub fn visual_key(&self) -> VisualKey {
1235 VisualKey(self.vmap_key.clone())
1236 }
1237
1238 /// The map, built at most once per `(revision, wrap)`. `clamp_caret` still
1239 /// runs on every call: the caret moves without the document changing, and
1240 /// keeping it on a legal stop is this function's job either way.
1241 fn build_map(&mut self, wrap: Option<usize>) {
1242 // Under `MarkupMode::Full` the map is a function of the caret's *line*
1243 // as well as the text, so the line joins the key: moving within a line
1244 // still reuses the map, and crossing into another one rebuilds it. In
1245 // every other mode `reveal_line` is `None` and the key is what it was,
1246 // so caret motion goes on costing nothing.
1247 let reveal = self.reveal_line();
1248 let key = (self.revision, wrap, reveal.clone());
1249 if self.vmap_key.as_ref() != Some(&key) {
1250 // Enumerate the top-level blocks cheaply — no whole-arena marshal.
1251 // A subtree is pulled only for the block(s) that actually changed, so
1252 // the FFI marshal shrinks from O(document) to O(edited block).
1253 let top = self.top_blocks();
1254
1255 // Fast path: when twig reports a dirty byte range, try to patch the
1256 // previous map in place — a single-block edit moves the prefix,
1257 // shifts the suffix, and re-renders only one block. `build_spliced`
1258 // returns `None` (and we fall back to the always-correct full rebuild)
1259 // whenever the edit reshaped the block structure, hit a table, or
1260 // there's no previous map to patch.
1261 // Preserve soft breaks as written when the flow preference asks for
1262 // it — the builder renders each as its own visual row instead of
1263 // folding it into the reflowed paragraph.
1264 let preserve_soft = self.line_flow == LineFlow::Preserve;
1265 let spliced = match self.editor.dirty_range() {
1266 Some(dirty) => {
1267 let prev = std::mem::take(&mut self.vmap);
1268 let source = &self.source;
1269 let cache = &mut self.block_cache;
1270 let media_rows = &self.media_rows;
1271 let editor = &mut self.editor;
1272 wysiwyg::build_spliced(
1273 prev,
1274 source,
1275 wrap,
1276 preserve_soft,
1277 &top,
1278 dirty,
1279 media_rows,
1280 reveal.clone(),
1281 cache,
1282 |id| editor.subtree(NodeId(id)).unwrap_or_default(),
1283 )
1284 }
1285 None => None,
1286 };
1287 self.vmap = spliced.unwrap_or_else(|| {
1288 let source = &self.source;
1289 let cache = &mut self.block_cache;
1290 let media_rows = &self.media_rows;
1291 let editor = &mut self.editor;
1292 wysiwyg::build_cached(
1293 &top,
1294 source,
1295 wrap,
1296 preserve_soft,
1297 media_rows,
1298 reveal,
1299 cache,
1300 |id| editor.subtree(NodeId(id)).unwrap_or_default(),
1301 )
1302 });
1303 // Acknowledge the dirty range so the next edit's range starts fresh.
1304 self.editor.clear_dirty();
1305 self.vmap_key = Some(key);
1306 }
1307 self.clamp_caret();
1308 }
1309
1310 fn nodes(&mut self) -> Vec<FlatNode> {
1311 self.editor.nodes().unwrap_or_default()
1312 }
1313
1314 /// The document's top-level blocks for the incremental render. See
1315 /// [`wysiwyg::top_blocks`] for why this isn't simply `child_spans(None)`.
1316 fn top_blocks(&mut self) -> Vec<QueryMatch> {
1317 wysiwyg::top_blocks(&mut self.editor)
1318 }
1319
1320 pub fn format_name(&self) -> &'static str {
1321 // `Format` is `#[non_exhaustive]` as of twig 3.0, so the wildcard is
1322 // required. It also covers `Asciidoc`, which twig parses but cannot
1323 // serialize — leaf never opens a document in it (see `Doc::open`).
1324 match self.format {
1325 Format::Djot => "djot",
1326 Format::Markdown => "markdown",
1327 Format::Xml => "xml",
1328 Format::Html => "html",
1329 _ => "unknown",
1330 }
1331 }
1332
1333 /// Whether this document's format offers *any* door in — `false` only for a
1334 /// wholly parse-only format (XML, AsciiDoc), where every gesture refuses and
1335 /// a frontend may as well open the file read-only.
1336 ///
1337 /// This is a much weaker claim than the name suggests, and driving per-button
1338 /// state from it is exactly the mistake to avoid: HTML answers `true` because
1339 /// it spells the inline marks with a tag pair (`<strong>`, `<em>`, `<code>`)
1340 /// while a heading, a quote, a list, a task box, a link and a code fence all
1341 /// remain unspellable there. Ask [`capabilities`](Self::capabilities) — or
1342 /// [`supports`](Self::supports) — per control.
1343 pub fn authorable(&self) -> bool {
1344 self.format.is_authorable()
1345 }
1346
1347 /// Whether this document's format can spell `gesture`, which is twig's own
1348 /// answer rather than a copy of it: `Format::supports` reads the same
1349 /// `Syntax` table the `Editor` method consults before refusing.
1350 ///
1351 /// It is a fact about the *format*, not about the caret. `true` does not
1352 /// promise the gesture succeeds where it is standing — a link over a table
1353 /// border still fails — only that it will not fail with
1354 /// `UnsupportedFormat`. Gray out on `false`; don't read `true` as "this
1355 /// will work here".
1356 pub fn supports(&self, gesture: Gesture) -> bool {
1357 self.format.supports(gesture)
1358 }
1359
1360 /// Every control's enabled state in one read — what a toolbar builds itself
1361 /// from when a document opens or its format changes. See [`Capabilities`].
1362 pub fn capabilities(&self) -> Capabilities {
1363 Capabilities::of(self.format)
1364 }
1365
1366 /// Refuse a gesture this document's format cannot spell, saying so in the
1367 /// status line. `true` means the caller must return without calling twig.
1368 ///
1369 /// Most of these refusals duplicate one twig would make anyway, and they are
1370 /// made here regardless because a message naming the *document's* format
1371 /// reads better than one naming twig's internals. Two of them are not
1372 /// duplicates and are the reason this is a guard rather than an error
1373 /// translation:
1374 ///
1375 /// - The table family (see [`table_op`](Self::table_op)) consults no
1376 /// `Syntax` table, so twig does not refuse it at all.
1377 /// - [`toggle`](Self::toggle) at a collapsed caret never reaches twig — it
1378 /// arms a sticky mark for text not yet typed, which is a promise `insert`
1379 /// could not keep.
1380 fn refuse_unsupported(&mut self, what: &str, gesture: Gesture) -> bool {
1381 self.refuse_unless(what, self.supports(gesture))
1382 }
1383
1384 /// [`refuse_unsupported`](Self::refuse_unsupported) against a capability leaf
1385 /// answers itself — today only [`spells_pipe_tables`].
1386 fn refuse_unless(&mut self, what: &str, supported: bool) -> bool {
1387 if supported {
1388 return false;
1389 }
1390 self.status = Some(format!("{what}: not supported in {}", self.format_name()));
1391 true
1392 }
1393
1394 /// The name to show for this document. An untitled one has no file to name
1395 /// it, and both frontends put this straight on screen — an empty path
1396 /// renders as an empty header, so it says so instead.
1397 pub fn file_name(&self) -> String {
1398 if self.is_untitled() {
1399 return "untitled".into();
1400 }
1401 self.path
1402 .file_name()
1403 .map(|s| s.to_string_lossy().into_owned())
1404 .unwrap_or_else(|| self.path.display().to_string())
1405 }
1406
1407 /// The selection as an ordered `[start, end)` byte range, or `None` when the
1408 /// caret and anchor coincide (an empty selection is no selection).
1409 pub fn selection(&self) -> Option<(usize, usize)> {
1410 self.anchor
1411 .map(|a| (a.min(self.caret), a.max(self.caret)))
1412 .filter(|(s, e)| s != e)
1413 }
1414
1415 /// The selected text, or `None` when there's no selection — the source
1416 /// slice a copy/cut hands to the system clipboard.
1417 pub fn selected_text(&self) -> Option<&str> {
1418 self.selection().map(|(s, e)| &self.source[s..e])
1419 }
1420
1421 /// The selection as a quote with a little of what surrounds it — the shape
1422 /// a host that cites, annotates, or searches for a passage wants, cut from
1423 /// the **source** rather than from anything rendered, so the quote is
1424 /// findable in the document again by plain string search.
1425 ///
1426 /// `context` is a count of characters (not bytes) on each side, clipped at
1427 /// the document's edges; the slices land on char boundaries by
1428 /// construction. `None` when nothing is selected.
1429 pub fn selection_quote(&self, context: usize) -> Option<Quote> {
1430 let (start, end) = self.selection()?;
1431 let mut before = start;
1432 for _ in 0..context {
1433 match self.source[..before].chars().next_back() {
1434 Some(c) => before -= c.len_utf8(),
1435 None => break,
1436 }
1437 }
1438 let mut after = end;
1439 for _ in 0..context {
1440 match self.source[after..].chars().next() {
1441 Some(c) => after += c.len_utf8(),
1442 None => break,
1443 }
1444 }
1445 Some(Quote {
1446 exact: self.source[start..end].to_string(),
1447 prefix: self.source[before..start].to_string(),
1448 suffix: self.source[end..after].to_string(),
1449 start,
1450 end,
1451 })
1452 }
1453
1454 /// Whether the document refuses to change — see the field.
1455 pub fn read_only(&self) -> bool {
1456 self.read_only
1457 }
1458
1459 /// Turn the read-only gate on or off. A frontend preference like
1460 /// [`set_markup_mode`](Self::set_markup_mode): nothing about the document
1461 /// itself changes, only what may be done to it from here on.
1462 pub fn set_read_only(&mut self, on: bool) {
1463 self.read_only = on;
1464 }
1465
1466 /// The host-painted ranges, sorted by start — see [`Highlight`].
1467 pub fn highlights(&self) -> &[Highlight] {
1468 &self.highlights
1469 }
1470
1471 /// Replace the host-painted ranges wholesale. The whole set each time,
1472 /// rather than add/remove verbs: the host owns the list (it derives it
1473 /// from its own state — annotations, search hits), and a replace can
1474 /// never leave the two disagreeing about what should be on screen.
1475 pub fn set_highlights(&mut self, mut highlights: Vec<Highlight>) {
1476 highlights.retain(|h| h.start < h.end);
1477 highlights.sort_by_key(|h| (h.start, h.end));
1478 self.highlights = highlights;
1479 }
1480
1481 /// The highlight covering source `offset`, if one does — first by start
1482 /// when several overlap, which makes overlapping washes resolvable rather
1483 /// than undefined. What a frontend asks when the reader activates a spot.
1484 ///
1485 /// [`Highlight::covering`] is the whole of it: the frontends paint by
1486 /// asking the same question per glyph, against a slice they were handed
1487 /// rather than against a `Doc`, and one answer for both is what keeps a
1488 /// wash and an activation agreeing about which range a spot is in.
1489 pub fn highlight_at(&self, offset: usize) -> Option<&Highlight> {
1490 Highlight::covering(&self.highlights, offset)
1491 }
1492
1493 /// The AST breadcrumb at the caret (root → deepest), e.g.
1494 /// `doc › para › strong`. Read live from twig via `ancestors_at`.
1495 pub fn breadcrumb(&mut self) -> String {
1496 match self.editor.ancestors_at(self.caret) {
1497 Ok(chain) => chain
1498 .iter()
1499 .map(|m| m.kind.as_str())
1500 .collect::<Vec<_>>()
1501 .join(" › "),
1502 Err(_) => String::new(),
1503 }
1504 }
1505
1506 // ── editing ──────────────────────────────────────────────────────────────
1507
1508 /// Replace the byte range `[start, end)` with `text`, re-anchoring the caret
1509 /// after it. The public form of the internal splice — a pixel frontend that
1510 /// hit-tests to a byte offset (or an IME that hands back an explicit range)
1511 /// edits through this, the same twig `edit_range` the caret ops use.
1512 pub fn edit(&mut self, start: usize, end: usize, text: &str) {
1513 self.splice(start, end, text, EditKind::Other);
1514 }
1515
1516 /// Insert typed `text` at the caret, replacing the selection if there is one.
1517 /// A single typed character coalesces with the run of typing before it; a
1518 /// newline or a multi-character insert is its own undo step.
1519 ///
1520 /// Typed input only — clipboard text goes through [`paste`](Self::paste).
1521 pub fn insert(&mut self, text: &str) {
1522 // The read-only gate, up front: the paths below reach twig by several
1523 // verbs, not all of them through the splice — see the field.
1524 if self.read_only {
1525 return;
1526 }
1527 // Typing against a block picture would dissolve it — see
1528 // `open_paragraph_at_block_media`. Give the text a paragraph first, so
1529 // what the caret was standing beside stays a picture.
1530 self.open_paragraph_at_block_media(text);
1531 // Armed sticky marks (⌘b with no selection) turn the next typed text
1532 // bold/italic/… and then retire — see `insert_with_marks`. Whitespace is
1533 // the exception: it takes no mark of its own and keeps the delta armed
1534 // for the character behind it — see `insert_space_with_marks`.
1535 let pending = self.pending_here();
1536 if !pending.is_empty() && self.selection().is_none() && !text.is_empty() {
1537 if text.trim().is_empty() {
1538 self.insert_space_with_marks(self.caret, text, pending);
1539 } else {
1540 self.insert_with_marks(self.caret, text, pending);
1541 }
1542 return;
1543 }
1544 // `MarkupMode::None`: typed syntax stays literal — twig escapes
1545 // anything that would open markup, so a Diaryx user never mints
1546 // formatting by keyboard (it comes from commands instead). The other two
1547 // rungs of the ladder author markup from what you type, which is the
1548 // whole difference between them and this one. Only in the rendered view
1549 // (source view is for typing raw markup) and only where the format has a
1550 // literal spelling at all: escaping is a backslash before a byte from the
1551 // format's own alphabet, and a format with no such alphabet (HTML escapes
1552 // with entities, XML spells nothing) would have `\&` written into it,
1553 // which is two literal characters and not an escape. Marks (⌘b) still
1554 // format — that path returned above; and leaf's own structural inserts go
1555 // through `insert_raw`, never here, so a list marker or quote gutter is
1556 // written as the markup it is.
1557 if !self.markup_mode.authors()
1558 && self.view == View::Wysiwyg
1559 && !text.is_empty()
1560 && self.supports(Gesture::InsertLiteral)
1561 {
1562 self.insert_literal_typed(text);
1563 return;
1564 }
1565 self.insert_raw(text);
1566 }
1567
1568 /// Insert `text` verbatim at the caret (replacing any selection) — the plain
1569 /// path with no Hidden-mode literal escaping. leaf's own structural inserts
1570 /// (a list marker, a quote gutter, an in-cell `<br>`) call this: they ARE
1571 /// markup by design and must not be escaped.
1572 fn insert_raw(&mut self, text: &str) {
1573 let (s, e) = self.selection().unwrap_or((self.caret, self.caret));
1574 self.splice(s, e, text, typed_edit_kind(text));
1575 }
1576
1577 /// Open a paragraph for text about to be inserted at one of a block media's
1578 /// two caret stops, and leave the caret standing in it.
1579 ///
1580 /// A block image is a paragraph whose entire content is the picture, and the
1581 /// caret's only homes on it are in front of it and just past it (see
1582 /// [`VisualMap::block_media_stop`]). Text inserted at either offset joins
1583 /// *that* paragraph — and a paragraph holding anything besides the image is
1584 /// no longer a block image but a line of text with an inline one in it. The
1585 /// frontend that was painting a photo there paints a text run instead; the
1586 /// picture is still in the file, and nothing said a word. Those two offsets
1587 /// are also exactly where a click on the picture lands, so the whole accident
1588 /// is one tap and one keystroke.
1589 ///
1590 /// So the break goes in first and the text lands in the new empty paragraph —
1591 /// what pressing Return before typing would have done, which is a habit no
1592 /// one should have to learn from losing a photo. A no-op everywhere else, and
1593 /// over a selection (which is replaced, not joined into).
1594 ///
1595 /// A picture inside a quote or a list leaves its container, because `\n\n`
1596 /// ends the block. The alternative is worse: the `\n> ` / next-item
1597 /// continuation [`newline`](Self::newline) writes stays in the same
1598 /// *paragraph*, which is the thing being prevented.
1599 ///
1600 /// Only in the rendered view. Source view is for typing raw markup, where
1601 /// putting a character against an image is exactly what it looks like.
1602 fn open_paragraph_at_block_media(&mut self, text: &str) {
1603 if self.view != View::Wysiwyg || text.is_empty() || text == "\n" {
1604 return;
1605 }
1606 if self.selection().is_some() {
1607 return;
1608 }
1609 // The map may be a revision behind (nothing has drawn since the last
1610 // edit), and this asks it about offsets — a stale answer would splice a
1611 // break into the wrong place. Free when it is already current, which it
1612 // is whenever a frontend drew a frame between keystrokes.
1613 self.rebuild_map();
1614 let at = self.caret;
1615 let Some((side, _)) = self.vmap.block_media_stop(at) else {
1616 return;
1617 };
1618 if !self.splice(at, at, "\n\n", EditKind::Other) {
1619 return;
1620 }
1621 // The break is part of the keystroke, not an edit of its own: leave the
1622 // run marked as typing so the character about to arrive folds into it and
1623 // one undo puts the document back the way it was found. (A paste, or a
1624 // multi-character insert, is `EditKind::Other` and stays its own step —
1625 // as it would have been anywhere else in the document.)
1626 self.last_edit_kind = Some(EditKind::Insert);
1627 if side == MediaStop::Before {
1628 // The break went in above the picture and the caret rode to the end
1629 // of it — which is still hard against the picture. Step back onto the
1630 // blank line it opened, so the text lands above rather than in front.
1631 self.caret = at;
1632 }
1633 }
1634
1635 /// A delete key pressed at one of a block picture's two caret stops, handled
1636 /// as the picture being an *atom* rather than a run of bytes. Returns whether
1637 /// the key was consumed.
1638 ///
1639 /// The caret rests in front of a block image and just past it, never inside
1640 /// its markup — which the rendered view doesn't show. So the byte a delete
1641 /// key nominally takes there is one the writer cannot see, and taking it
1642 /// leaves the picture as broken markup rather than as anything anyone asked
1643 /// for: Backspace at the stop past `` removes the closing paren, and
1644 /// a photo becomes the literal text `
1647 /// prevents from the typing side, and it cost this repository's own test vault
1648 /// a photo before it was found.
1649 ///
1650 /// So the key aimed *at* the picture deletes the picture, whole — Backspace
1651 /// when it is behind the caret, Delete when it is in front — which is what
1652 /// every editor does with an embed, and one undo away. The key aimed *away*
1653 /// from it would otherwise delete the paragraph break and merge a neighbour
1654 /// into the picture's own paragraph, which dissolves it just as surely; it
1655 /// steps the caret over the boundary instead and leaves the
1656 /// next press to delete in the block it has reached — the same "first press
1657 /// steps out of the atom, second press deletes" every delete key here gets,
1658 /// word-deletes included (⌥⌫ in front of a picture is aimed at the prose
1659 /// above, and reaches it on the second press rather than taking the break and
1660 /// the picture with it on the first).
1661 fn delete_around_block_media(&mut self, forward: bool) -> bool {
1662 // The map answers about offsets, so it has to be this revision's — see
1663 // the same call in `open_paragraph_at_block_media`.
1664 self.rebuild_map();
1665 let Some((side, span)) = self.vmap.block_media_stop(self.caret) else {
1666 return false;
1667 };
1668 let aimed_at_it = side
1669 == if forward {
1670 MediaStop::Before
1671 } else {
1672 MediaStop::After
1673 };
1674 if !aimed_at_it {
1675 let over = if forward {
1676 self.vmap.stop_after(self.caret)
1677 } else {
1678 self.vmap.stop_before(self.caret)
1679 };
1680 if let Some(off) = over.filter(|&o| o >= self.caret_floor()) {
1681 self.caret = off;
1682 self.anchor = None;
1683 self.goal_col = None;
1684 }
1685 return true;
1686 }
1687 // Take the break that held the picture apart from its neighbour with it,
1688 // so the delete doesn't leave a blank paragraph standing where the
1689 // picture was. The last arm is a picture that is the whole document.
1690 let (from, to) = if self.source[..span.start].ends_with("\n\n") {
1691 (span.start - 2, span.end)
1692 } else if self.source[span.end..].starts_with("\n\n") {
1693 (span.start, span.end + 2)
1694 } else {
1695 (span.start, span.end)
1696 };
1697 self.splice(from.max(self.caret_floor()), to, "", EditKind::Other);
1698 true
1699 }
1700
1701 /// The Hidden-mode typing path: replace any selection, then insert `text`
1702 /// escaped so it stays literal. When it replaces a selection the two edits
1703 /// fold into one undo step, so an overwrite undoes atomically (and restores
1704 /// the selection) exactly as a plain one does.
1705 fn insert_literal_typed(&mut self, text: &str) {
1706 let kind = typed_edit_kind(text);
1707 match self.selection() {
1708 Some((s, e)) => {
1709 if !self.splice(s, e, "", EditKind::Other) {
1710 return;
1711 }
1712 // Typing over a whole marked run takes its delimiters with it
1713 // (the empty content couldn't hold them — see
1714 // `repair_mark_edges`) and leaves its marks armed at the caret.
1715 // The text taking the run's place inherits them, exactly as it
1716 // would have by landing inside a run that survived.
1717 let pending = self.pending_here();
1718 if !pending.is_empty() && !text.trim().is_empty() {
1719 self.insert_with_marks(self.caret, text, pending);
1720 return;
1721 }
1722 self.insert_literal_at(self.caret, text, kind, true);
1723 }
1724 None => {
1725 self.insert_literal_at(self.caret, text, kind, false);
1726 }
1727 }
1728 }
1729
1730 /// The sticky-mark delta that is live right now: the marks armed by [`toggle`]
1731 /// at a collapsed caret, but only while the caret still stands where they
1732 /// were armed and nothing is selected. Empty otherwise, so a stale delta
1733 /// never styles text it wasn't meant for.
1734 fn pending_here(&self) -> InlineMarks {
1735 if self.anchor.is_none() && self.pending_at == Some(self.caret) {
1736 self.pending_marks
1737 } else {
1738 InlineMarks::empty()
1739 }
1740 }
1741
1742 /// Drop the armed sticky marks — any caret motion, selection, or edit does
1743 /// this, so "start bold here" only ever applies at the exact spot it was
1744 /// asked for.
1745 fn clear_pending(&mut self) {
1746 self.pending_marks = InlineMarks::empty();
1747 self.pending_at = None;
1748 }
1749
1750 /// Insert `text` at `at` carrying the armed sticky `marks`: a mark not yet in
1751 /// force is wrapped around the freshly typed text; a mark the caret already
1752 /// stands inside is *shed* — the text is inserted past the run's end so it
1753 /// lands unmarked ("type normally again"). The caret comes to rest inside any
1754 /// added runs, so continued typing inherits the marks with no re-wrapping,
1755 /// and the delta is cleared: the marks now live in the document, not here.
1756 fn insert_with_marks(&mut self, at: usize, text: &str, marks: InlineMarks) {
1757 let base = self.mark_spans_at(at);
1758 let base_set: InlineMarks = base.iter().map(|(k, _)| *k).collect();
1759 // Nothing to shed, and a run of exactly these marks standing just behind
1760 // the caret: carry on writing *that* run rather than opening a second
1761 // one beside it.
1762 if base_set.is_empty() && self.rejoin_run(at, text, marks) {
1763 return;
1764 }
1765 // Shed the marks we're turning off: step the insertion point past the
1766 // end of each run the caret sits in, so the new text falls outside it.
1767 let mut ins_at = at;
1768 for (kind, span) in &base {
1769 if marks.contains(*kind) {
1770 ins_at = ins_at.max(span.end);
1771 }
1772 }
1773 if !self.splice_exact(ins_at, ins_at, text, EditKind::Other) {
1774 return;
1775 }
1776 // The plain splice inserted exactly `text` at `ins_at`; that byte range
1777 // is the content every added mark wraps.
1778 let (mut cs, mut ce) = (ins_at, ins_at + text.len());
1779 for kind in marks.iter() {
1780 if !base_set.contains(kind) {
1781 let (ncs, nce) = self.wrap_span(cs, ce, kind);
1782 cs = ncs;
1783 ce = nce;
1784 }
1785 }
1786 self.caret = ce.min(self.source.len());
1787 self.anchor = None;
1788 self.last_edit_kind = None;
1789 // Realised: the marks are in the document now, and the caret sits inside
1790 // them, so there is no delta left to carry. Arm nothing, but remember the
1791 // spot so a *further* toggle before typing starts a clean delta here.
1792 self.pending_marks = InlineMarks::empty();
1793 self.pending_at = Some(self.caret);
1794 self.clamp_caret();
1795 self.record_caret();
1796 }
1797
1798 /// Carry on the marked run just behind `at` — moving its closing delimiters
1799 /// out past the new text — instead of opening a second run of the same marks
1800 /// beside it. Returns whether it did.
1801 ///
1802 /// This is the far half of the mark-edge rule (see [`splice`](Self::splice)).
1803 /// A space typed after a bold word steps the caret out of the run, because
1804 /// `**bold **` is not bold; the next character has to step back *in*, or the
1805 /// writer who typed one bold phrase is left with `**bold** **and**` — two
1806 /// runs that read the same to a reader but spell the file in a way nobody
1807 /// wrote. Only whitespace may stand in the gap (a run doesn't reach across
1808 /// words it isn't marking), and the marks behind it must be exactly the ones
1809 /// armed — a run of *some* other kind is a neighbour, not this phrase.
1810 fn rejoin_run(&mut self, at: usize, text: &str, marks: InlineMarks) -> bool {
1811 if text.is_empty() || text.trim() != text {
1812 return false;
1813 }
1814 let gap_at = self.source[..at].trim_end_matches([' ', '\t']).len();
1815 // Walk in through the delimiters stacked at that point, innermost last:
1816 // `***both*** ` closes two runs with one `***`, and rejoining means
1817 // getting behind all of them.
1818 let (mut cut, mut kinds) = (gap_at, InlineMarks::empty());
1819 while let Some((kind, content_end)) = self
1820 .editor
1821 .ancestors_at(prev_boundary(&self.source, cut))
1822 .unwrap_or_default()
1823 .into_iter()
1824 .filter(|m| m.span.end == cut)
1825 .find_map(|m| Some((inline_kind(&m.kind)?, m.content_span.clone()?.end)))
1826 {
1827 if content_end >= cut {
1828 break; // a mark with no closing delimiter to step behind
1829 }
1830 kinds.insert(kind);
1831 cut = content_end;
1832 }
1833 if cut == gap_at || kinds != marks {
1834 return false;
1835 }
1836 // Re-spell the tail: the gap, then the new text, then the delimiters that
1837 // used to close in front of them — read out of the document rather than
1838 // written from a table, so whatever twig spells them with is what moves.
1839 let tail = format!(
1840 "{}{text}{}",
1841 &self.source[gap_at..at],
1842 &self.source[cut..gap_at]
1843 );
1844 if !self.splice_exact(cut, at, &tail, EditKind::Other) {
1845 return false;
1846 }
1847 self.caret = (cut + (at - gap_at) + text.len()).min(self.source.len());
1848 self.anchor = None;
1849 self.last_edit_kind = None;
1850 self.pending_marks = InlineMarks::empty();
1851 self.pending_at = Some(self.caret);
1852 self.clamp_caret();
1853 self.record_caret();
1854 true
1855 }
1856
1857 /// Insert typed whitespace at a caret with sticky marks armed. Whitespace is
1858 /// never itself wrapped: a mark around a space draws nothing a reader can
1859 /// see, and in Markdown and Djot it draws its own delimiters instead
1860 /// (`** **`). So the space goes in unmarked — outside any run the armed
1861 /// marks are shedding — and the marks stay armed for the character after it,
1862 /// which rejoins the run (see [`rejoin_run`](Self::rejoin_run)).
1863 fn insert_space_with_marks(&mut self, at: usize, text: &str, marks: InlineMarks) {
1864 let base = self.mark_spans_at(at);
1865 // What the *next* character carries: the armed delta resolved against the
1866 // marks in force here, which the space must not quietly drop.
1867 let want = base
1868 .iter()
1869 .map(|(k, _)| *k)
1870 .collect::<InlineMarks>()
1871 .xor(marks);
1872 let mut ins_at = at;
1873 for (kind, span) in &base {
1874 if marks.contains(*kind) {
1875 ins_at = ins_at.max(span.end);
1876 }
1877 }
1878 if !self.splice(ins_at, ins_at, text, typed_edit_kind(text)) {
1879 return;
1880 }
1881 self.rearm(want);
1882 self.record_caret();
1883 }
1884
1885 /// Wrap `[s, e)` in `kind` via twig and return the byte span the *content*
1886 /// (not the delimiters) occupies afterwards. Markdown/Djot inline delimiters
1887 /// are symmetric (`**`…`**`, `_`…`_`, `` ` ``…`` ` ``), so the bytes twig
1888 /// added split evenly around the content — half the growth on each side.
1889 fn wrap_span(&mut self, s: usize, e: usize, kind: InlineKind) -> (usize, usize) {
1890 // The read-only gate — this door reaches twig without the splice.
1891 if self.read_only {
1892 return (s, e);
1893 }
1894 match self.editor.toggle_inline(s, e, kind) {
1895 Ok(change) => {
1896 self.last_edit_kind = None;
1897 self.refresh();
1898 self.dirty = self.source != self.clean_source;
1899 let added = (change.new.end - change.new.start).saturating_sub(e - s);
1900 let half = added / 2;
1901 (change.new.start + half, change.new.end - half)
1902 }
1903 // Unsupported here (e.g. mark on Markdown): leave the text unwrapped
1904 // rather than lose the keystroke.
1905 Err(e2) => {
1906 self.status = Some(format!("{kind:?}: {e2}"));
1907 (s, e)
1908 }
1909 }
1910 }
1911
1912 /// The safe offset to splice a block-level break at, given a caret that may
1913 /// sit exactly between an inline mark's content and its own closing
1914 /// delimiter (`content_span.end == off < span.end` for some enclosing mark
1915 /// — the WYSIWYG caret's natural resting place at the end of `**bold**`
1916 /// with nothing following it on the line: the closing `**` renders no
1917 /// glyph of its own, so the caret's "end of line" offset lands right
1918 /// before it). Splicing a paragraph/list/quote break at `off` itself would
1919 /// sever the delimiter from its content, stranding it alone on the new
1920 /// line. Walks out to the *outermost* such mark's `span.end` instead, so
1921 /// nested marks closing at the same point (`**_x_**`) all clear together.
1922 /// A no-op everywhere else — mid-run, or past real trailing content, no
1923 /// mark's `content_span` ends exactly at `off`.
1924 fn skip_trailing_close_delims(&mut self, off: usize) -> usize {
1925 let off = off.min(self.source.len());
1926 self.editor
1927 .ancestors_at(off)
1928 .unwrap_or_default()
1929 .into_iter()
1930 .filter(|m| inline_kind(&m.kind).is_some())
1931 .filter(|m| off < m.span.end && m.content_span.as_ref().is_some_and(|c| c.end == off))
1932 .map(|m| m.span.end)
1933 .max()
1934 .unwrap_or(off)
1935 }
1936
1937 /// The offset a *delete* aimed at the character before `off` should stop at,
1938 /// when `off` is the start of a run's text and the bytes behind it are that
1939 /// run's opening delimiter. The rich view draws no glyph for a `**`, so the
1940 /// byte behind the caret at the start of a bold word is not a character the
1941 /// writer can see, let alone one they aimed Backspace at: taking it leaves
1942 /// `a *bold** c` — the styling gone and a literal asterisk in its place. The
1943 /// delete steps over the whole delimiter to the visible character in front of
1944 /// it instead. Walks out to the *outermost* mark opening there, so
1945 /// `**_x_**` clears every delimiter at once, and is a no-op anywhere else.
1946 fn skip_leading_open_delims(&mut self, off: usize) -> usize {
1947 let off = off.min(self.source.len());
1948 self.editor
1949 .ancestors_at(off)
1950 .unwrap_or_default()
1951 .into_iter()
1952 .filter(|m| inline_kind(&m.kind).is_some())
1953 .filter(|m| {
1954 m.span.start < off && m.content_span.as_ref().is_some_and(|c| c.start == off)
1955 })
1956 .map(|m| m.span.start)
1957 .min()
1958 .unwrap_or(off)
1959 }
1960
1961 /// `off` moved *inside* the run whose closing delimiters end there — the
1962 /// other offset the rich view draws in the same place, since a `**` renders
1963 /// no glyph of its own. `**bold**` has a caret home on each side of its
1964 /// closing delimiter, one column apart on screen and eight bytes and a whole
1965 /// run apart in the file, and a plain ← lands on the outer one whenever a
1966 /// space follows the phrase. The inner one is what the writer is pointing at
1967 /// there: the end of their bold word. Walks in through every mark closing at
1968 /// that point, innermost last, so `***both***` lands inside both. A no-op
1969 /// anywhere else — mid-run, or in prose, no mark's span ends at `off`.
1970 fn step_inside_close_delims(&mut self, off: usize) -> usize {
1971 let mut off = off.min(self.source.len());
1972 loop {
1973 let inner = self
1974 .editor
1975 .ancestors_at(prev_boundary(&self.source, off))
1976 .unwrap_or_default()
1977 .into_iter()
1978 .filter(|m| inline_kind(&m.kind).is_some() && m.span.end == off)
1979 .filter_map(|m| m.content_span.clone().map(|c| c.end))
1980 .filter(|&end| end < off)
1981 .max();
1982 match inner {
1983 Some(end) => off = end,
1984 None => return off,
1985 }
1986 }
1987 }
1988
1989 /// The mirror at the opening edge: `off` moved inside the run whose
1990 /// delimiters *start* there, onto the first character of its text. See
1991 /// [`step_inside_close_delims`](Self::step_inside_close_delims).
1992 fn step_inside_open_delims(&mut self, off: usize) -> usize {
1993 let mut off = off.min(self.source.len());
1994 loop {
1995 let inner = self
1996 .editor
1997 .ancestors_at(off)
1998 .unwrap_or_default()
1999 .into_iter()
2000 .filter(|m| inline_kind(&m.kind).is_some() && m.span.start == off)
2001 .filter_map(|m| m.content_span.clone().map(|c| c.start))
2002 .filter(|&start| start > off)
2003 .min();
2004 match inner {
2005 Some(start) => off = start,
2006 None => return off,
2007 }
2008 }
2009 }
2010
2011 /// The inline mark kinds whose span covers `off`, each with that span — the
2012 /// span-carrying sibling of [`marks_at`](Self::marks_at), which reports node
2013 /// ids instead. Used to shed a mark by stepping past the end of its run.
2014 fn mark_spans_at(&mut self, off: usize) -> Vec<(InlineKind, std::ops::Range<usize>)> {
2015 let off = off.min(self.source.len());
2016 self.editor
2017 .ancestors_at(off)
2018 .unwrap_or_default()
2019 .into_iter()
2020 .filter(|m| off < m.span.end)
2021 .filter_map(|m| inline_kind(&m.kind).map(|k| (k, m.span.clone())))
2022 .collect()
2023 }
2024
2025 /// Insert clipboard `text` at the caret, replacing the selection if there is
2026 /// one — always its own undo step, whatever its length.
2027 ///
2028 /// Provenance is the whole point, and only the caller has it. `insert` reads
2029 /// a lone character as a keystroke and folds it into the run around it,
2030 /// which is right for typing and wrong for a one-character paste: that paste
2031 /// would vanish mid-run on an undo it was never part of, and the characters
2032 /// the user actually typed would go with it. Length can't tell the two
2033 /// apart — `⌘V` of `x` and typing `x` are the same string — so the door the
2034 /// caller comes through is what says which happened.
2035 pub fn paste(&mut self, text: &str) {
2036 // Pasting against a block picture dissolves it exactly as typing does,
2037 // and for the same reason — see `open_paragraph_at_block_media`.
2038 self.open_paragraph_at_block_media(text);
2039 let (s, e) = self.selection().unwrap_or((self.caret, self.caret));
2040 self.splice(s, e, text, EditKind::Other);
2041 }
2042
2043 /// Replace `[start, end)` with `text` as one step of an IME composition —
2044 /// the same splice as [`edit`](Self::edit), but marked so the run of steps
2045 /// folds into a single undo.
2046 ///
2047 /// A composition is *one* act of writing. Typing `かんじ` and picking 感じ is a
2048 /// dozen calls here, each replacing the last one's provisional bytes, and an
2049 /// undo step per call means undoing a word means pressing ⌘Z until the reading
2050 /// unspools backwards through kana — the intermediate states were never text
2051 /// the user wrote. Only the frontend knows a call is provisional (the bytes
2052 /// look like any other edit), so the door the caller comes through is what
2053 /// says so, exactly as it is for [`paste`](Self::paste) versus
2054 /// [`insert`](Self::insert).
2055 ///
2056 /// Pair with [`end_composition`](Self::end_composition), or the *next*
2057 /// composition folds into this one.
2058 pub fn edit_composing(&mut self, start: usize, end: usize, text: &str) {
2059 self.splice(start, end, text, EditKind::Compose);
2060 }
2061
2062 /// Close the open composition run, so the next one is its own undo step.
2063 /// Call when the IME commits or withdraws a composition.
2064 ///
2065 /// Only clears a *composition* run: a frontend that reports an end it never
2066 /// began (some IMEs unmark unprompted) would otherwise split the run of
2067 /// typing around it into two undo steps for no reason the user can see.
2068 pub fn end_composition(&mut self) {
2069 if self.last_edit_kind == Some(EditKind::Compose) {
2070 self.last_edit_kind = None;
2071 }
2072 }
2073
2074 // ── the clipboard's rich flavor ──────────────────────────────────────────
2075
2076 /// The selection rendered as HTML, for the clipboard's `text/html` flavor —
2077 /// what lets a paste into Docs/Mail/Slack keep its formatting. `None` when
2078 /// nothing is selected, or when the selection doesn't render (the caller
2079 /// still has [`selected_text`](Self::selected_text), which is what to publish
2080 /// as `text/plain` either way).
2081 ///
2082 /// **The fragment is a source substring, and that is the honest limit here.**
2083 /// It's parsed standalone, so a selection whose meaning depends on its
2084 /// surroundings converts as what it literally says rather than what it looks
2085 /// like on screen: half a list item is a paragraph, a row torn out of a table
2086 /// is the text of a row, the `**` of a bold run selected without its closing
2087 /// `**` is two asterisks. Every one of those still *renders* — there's no
2088 /// error to report — it just renders as the fragment and not as the document.
2089 /// Widening the range to whole blocks would publish text the user didn't
2090 /// select, which is a worse lie than a fragment being a fragment; the plain
2091 /// flavor has the same substring, so the two flavors at least agree.
2092 pub fn selection_html(&mut self) -> Option<String> {
2093 let (start, end) = self.selection()?;
2094 let inline = self.selection_is_inline(start, end);
2095 let html = html::render_fragment(&self.source[start..end], self.format)?;
2096 Some(match inline {
2097 true => html::strip_sole_paragraph(html),
2098 false => html,
2099 })
2100 }
2101
2102 /// Paste the clipboard's `text/html` flavor, converting it to this document's
2103 /// format first. Its own undo step, like any [`paste`](Self::paste).
2104 ///
2105 /// Returns whether it landed. `false` means the HTML didn't convert to
2106 /// anything worth pasting — the caller should fall back to the plain flavor
2107 /// rather than treat it as an error. The `html` module has the full list of
2108 /// what that covers: a table twig won't build, markup it doesn't recognise,
2109 /// an empty result.
2110 pub fn paste_html(&mut self, html: &str) -> bool {
2111 match html::parse_fragment(html, self.format) {
2112 Some(source) => {
2113 self.paste(&source);
2114 true
2115 }
2116 None => false,
2117 }
2118 }
2119
2120 /// Does the selection live *inside* a single top-level block?
2121 ///
2122 /// The question [`selection_html`](Self::selection_html) needs and the
2123 /// fragment can't answer: `**bold**` renders as `<p><strong>bold</strong></p>`
2124 /// whether the user selected one word of a sentence or a whole paragraph, and
2125 /// only the document knows which. Selecting a word and pasting into Docs
2126 /// should extend the line you paste into; selecting the paragraph should make
2127 /// a paragraph. So a selection strictly within one block is inline (its `<p>`
2128 /// is an artifact of standalone parsing), and one that covers a whole block —
2129 /// or spans two — keeps its structure.
2130 ///
2131 /// Reads the block from twig rather than guessing from the bytes:
2132 /// `ancestors_at` is `[doc, block, …inline]`, so index 1 is the top-level
2133 /// block containing an offset, and two ends inside the same one cannot have
2134 /// crossed a block boundary.
2135 fn selection_is_inline(&mut self, start: usize, end: usize) -> bool {
2136 // The last *character*, not `end - 1`: the selection's end is exclusive
2137 // and may sit mid-codepoint's-worth of bytes past the last char.
2138 let Some((off, _)) = self.source[start..end].char_indices().next_back() else {
2139 return false;
2140 };
2141 let (Some(head), Some(tail)) =
2142 (self.top_block_span(start), self.top_block_span(start + off))
2143 else {
2144 return false;
2145 };
2146 head == tail && !(start <= head.start && end >= head.end)
2147 }
2148
2149 /// The byte span of the top-level block containing `offset`, or `None` at an
2150 /// offset that belongs to no block (the blank line between two of them).
2151 fn top_block_span(&mut self, offset: usize) -> Option<std::ops::Range<usize>> {
2152 self.editor
2153 .ancestors_at(offset)
2154 .ok()?
2155 .get(1)
2156 .map(|m| m.span.clone())
2157 }
2158
2159 // ── indentation ──────────────────────────────────────────────────────────
2160
2161 /// One indent level.
2162 ///
2163 /// Two spaces, not the four both frontends type for Tab today, because in a
2164 /// markdown document four columns isn't a width — it's a *meaning*. Four
2165 /// spaces at the head of a line is markdown's indented-code-block marker, so
2166 /// one Tab on a paragraph would reparse it into code and style it as such;
2167 /// two cannot, and the line stays the prose it was. Two is also exactly
2168 /// where a `- ` bullet's content starts, so an indented line lands under its
2169 /// parent item's text instead of beside it — the column a list-aware indent
2170 /// has to hit anyway, which keeps this width from being relitigated later.
2171 const INDENT: &'static str = " ";
2172
2173 /// Indent the selected lines — or the caret's line, with no selection — by
2174 /// one level (Tab).
2175 pub fn indent(&mut self) {
2176 self.reindent(true);
2177 // Nesting changes an ordered list's numbering (the nested item restarts,
2178 // its old siblings resume) — keep the source markers in step.
2179 self.renumber_here();
2180 // Nesting an empty `-` item under a text line reparses that text as a
2181 // setext heading; swap the dash for a `*` before it can (a no-op unless
2182 // the collapse actually happened).
2183 self.avoid_setext_collapse();
2184 }
2185
2186 /// Take one indent level back off the selected lines, or the caret's line
2187 /// (Shift+Tab). A line with no indentation is left exactly as it is.
2188 ///
2189 /// A line with *less* than a full level gives back what it has rather than
2190 /// refusing: outdent's job is to walk a line left, and real documents — hand
2191 /// written, or reflowed by some other editor — are full of indentation that
2192 /// was never a clean multiple of anything. Refusing there would strand the
2193 /// line at a depth Shift+Tab couldn't undo.
2194 pub fn outdent(&mut self) {
2195 self.reindent(false);
2196 self.renumber_here();
2197 }
2198
2199 /// The body of [`indent`](Self::indent) / [`outdent`](Self::outdent).
2200 ///
2201 /// One splice across the whole line range, never one per line: a Tab is one
2202 /// thing the user did, so it has to be one undo step and one reparse. Per
2203 /// line, twig would reparse the document once per line and leave a stack of
2204 /// steps that Shift+⌘Z walks back one line at a time.
2205 fn reindent(&mut self, add: bool) {
2206 let (sel_start, sel_end) = self.selection().unwrap_or((self.caret, self.caret));
2207 let start = source_line_range(&self.source, sel_start).start;
2208 let end = source_line_range(&self.source, sel_end).end;
2209 let region = self.source[start..end].to_string();
2210 let lines: Vec<&str> = region.split('\n').collect();
2211 // A blank line has no text to move, and padding it would leave nothing
2212 // but trailing whitespace — but Tab on a blank line *is* a request for
2213 // indentation to type into, so the skip only applies where the op has
2214 // other lines to do real work on.
2215 let skip_blank = add && lines.len() > 1;
2216
2217 let mut out = String::with_capacity(region.len() + lines.len() * Self::INDENT.len());
2218 let mut deltas: Vec<isize> = Vec::with_capacity(lines.len());
2219 let mut line_off = start;
2220 for (i, full) in lines.iter().enumerate() {
2221 if i > 0 {
2222 out.push('\n');
2223 }
2224 // A list item moves by having its whole leading prefix *replaced*,
2225 // never by having spaces pushed in front of the line. twig spells
2226 // both prefixes, so the quote markers, the parent's indent and an
2227 // ordered marker's extra column all come out right without leaf
2228 // measuring any of them — and a line that only looks like an item
2229 // (a Djot continuation) reports no marker and is left to the plain
2230 // path, where a Tab is just a Tab.
2231 let marker = self.list_marker_on_line(line_off);
2232 let own = marker
2233 .as_ref()
2234 .map(|m| m.marker_start - m.line_start)
2235 .unwrap_or(0);
2236 let delta = if add {
2237 if skip_blank && full.trim().is_empty() {
2238 out.push_str(full);
2239 0
2240 } else if marker.is_some() && self.first_item_of_list(line_off) {
2241 // The first item of a list has no preceding sibling to nest
2242 // under, so a Tab here can't spell a sub-list — twig would
2243 // reparse the shoved-over marker as the same list, only
2244 // indented, which Shift+Tab then can't cleanly undo. Leave the
2245 // item where it is, the way every list editor refuses to
2246 // over-indent a list's first line.
2247 out.push_str(full);
2248 0
2249 } else if marker.is_some() {
2250 // Nesting means standing where a *continuation* of this line
2251 // would stand: past the parent's marker, inside its content
2252 // column. That is `continuation_prefix`, less a checkbox.
2253 let new = self.nesting_prefix_at(line_off);
2254 let delta = new.len() as isize - own as isize;
2255 out.push_str(&new);
2256 out.push_str(&full[own..]);
2257 delta
2258 } else {
2259 out.push_str(Self::INDENT);
2260 out.push_str(full);
2261 Self::INDENT.len() as isize
2262 }
2263 } else if marker.is_some() {
2264 // Unnesting is the mirror: stand where the parent item's own
2265 // line starts, which drops exactly the level it contributed.
2266 let new = self.outdent_prefix_at(line_off);
2267 let delta = new.len() as isize - own as isize;
2268 out.push_str(&new);
2269 out.push_str(&full[own..]);
2270 delta
2271 } else {
2272 // A plain line gives back the ordinary step.
2273 let strip = outdent_width(full, Self::INDENT.len());
2274 out.push_str(&full[strip..]);
2275 -(strip as isize)
2276 };
2277 deltas.push(delta);
2278 line_off += full.len() + 1;
2279 }
2280 // Nothing to give back. Returning before the splice keeps an outdent at
2281 // column zero from spending an undo step on a document it never changed.
2282 if deltas.iter().all(|d| *d == 0) {
2283 return;
2284 }
2285
2286 // Every line's text keeps its offset *within the line*, so the caret is
2287 // remapped by its column, not by its byte offset — which the prefixes on
2288 // the lines above it have already invalidated.
2289 let remap = |off: usize| -> usize {
2290 let (mut old_ls, mut new_ls) = (start, start);
2291 for (line, delta) in lines.iter().zip(&deltas) {
2292 let old_le = old_ls + line.len();
2293 let new_len = (line.len() as isize + delta) as usize;
2294 if off <= old_le {
2295 let col = (off - old_ls) as isize;
2296 return new_ls + ((col + delta).max(0) as usize).min(new_len);
2297 }
2298 old_ls = old_le + 1;
2299 new_ls += new_len + 1;
2300 }
2301 start + out.len()
2302 };
2303 let placed = match self.selection() {
2304 // Keep the rewritten region selected, the way a container toggle
2305 // keeps its own: it leaves a second Tab aimed at the same lines
2306 // rather than at whatever the shifted offsets now happen to cover.
2307 Some(_) => (start + out.len(), Some(start)),
2308 None => (remap(self.caret), None),
2309 };
2310
2311 // A rolled-back splice leaves the old source in place, where every offset
2312 // computed above addresses text that was never written.
2313 if !self.splice(start, end, &out, EditKind::Other) {
2314 return;
2315 }
2316 // `splice` re-anchors to the end of the `Change`, which for a whole-region
2317 // rewrite is the last line's end — nowhere the caret was. Place it, then
2318 // re-record the caret so this is the state redo restores, not the one
2319 // `splice` left behind from the `Change`.
2320 self.caret = placed.0.min(self.source.len());
2321 self.anchor = placed.1;
2322 self.clamp_caret();
2323 self.record_caret();
2324 }
2325
2326 /// The Enter key.
2327 ///
2328 /// In source view it's a literal newline. In WYSIWYG it's **AST-aware**: a
2329 /// bare `\n` is only a markdown soft break (same paragraph), so the block the
2330 /// caret is in decides what actually gets written.
2331 ///
2332 /// - paragraph → twig's [`Editor::split_block`], which parts the
2333 /// block at the caret and reopens its container
2334 /// - list item → likewise: the next item, its indent, quote
2335 /// prefix and `[ ]` box all reproduced by twig —
2336 /// except an *empty* item, which exits the list
2337 /// - block quote → likewise: a new paragraph inside the quote
2338 /// - heading → a new *paragraph*, not another heading
2339 /// - code block → a literal newline (stay in the block)
2340 /// - blank line → a literal newline (one Backspace undoes it)
2341 /// - [`LineFlow::Preserve`] → a single soft break, which renders as a
2342 /// visible line
2343 ///
2344 /// Where `split_block` is used it replaces markup leaf used to spell by hand,
2345 /// and it is better at it: it drops the whitespace the caret was sitting in
2346 /// front of instead of stranding it at the head of the second half, and it
2347 /// knows continuations leaf's marker scan never covered — a checklist item
2348 /// continues as an *unchecked* checklist item rather than a plain bullet.
2349 ///
2350 /// The exceptions above are exceptions because `split_block` is either wrong
2351 /// there or refuses: parting a fence yields two fences with the code split
2352 /// between them, parting a heading yields a second heading where every editor
2353 /// gives a paragraph, and a blank line, an empty item, a setext heading and a
2354 /// table all report an error rather than a split.
2355 pub fn newline(&mut self) {
2356 if self.view == View::Source {
2357 self.insert_raw("\n");
2358 return;
2359 }
2360 // Enter over a selection replaces it with a paragraph break.
2361 if let Some((s, e)) = self.selection() {
2362 self.splice(s, e, "\n\n", EditKind::Other);
2363 return;
2364 }
2365 // A caret resting exactly between an inline mark's content and its own
2366 // closing delimiter (`**bold**` with nothing after it on the line —
2367 // the WYSIWYG caret's natural end-of-line position) must not splice a
2368 // block break there: every path below eventually does via
2369 // `insert_raw`/`self.caret`, and splicing before the hidden closing
2370 // delimiter would strand it alone on the new line.
2371 self.caret = self.skip_trailing_close_delims(self.caret);
2372 // The block the caret is in. `block_offset_for_caret` nudges off a line
2373 // end (where the caret sits at the doc level); on a bare line (e.g. an
2374 // empty list item) fall back to the caret so the enclosing list/quote is
2375 // still visible in the ancestors.
2376 let off = self.block_offset_for_caret().unwrap_or(self.caret);
2377 let kinds: Vec<Kind> = self
2378 .editor
2379 .ancestors_at(off)
2380 .map(|c| c.into_iter().map(|m| m.kind).collect())
2381 .unwrap_or_default();
2382 let has = |k: Kind| kinds.contains(&k);
2383
2384 if has(Kind::CodeBlock) {
2385 self.insert_raw("\n");
2386 return;
2387 }
2388 // An *empty* list item exits the list — the standard double-Enter — which
2389 // `split_block` reports as an error rather than a split (there is no
2390 // content to part), so it stays leaf's. `list_marker_on_line` is itself
2391 // the AST gate — it answers from the tree, so a `- ` that reads as a
2392 // marker byte-for-byte but opens no item (a setext underline, a Djot
2393 // continuation line) never reaches here.
2394 if let Some(marker) = self.list_marker_on_line(self.caret)
2395 && self.item_is_empty(&marker)
2396 {
2397 self.exit_list(&marker);
2398 return;
2399 }
2400 // On an *empty* paragraph line, a lone Enter should add a single blank line,
2401 // not another full paragraph break — so it moves down one line and one
2402 // Backspace undoes it, not two. (`split_block` errors here too.)
2403 let line_start = self.source[..self.caret].rfind('\n').map_or(0, |i| i + 1);
2404 let line_end = self.source[self.caret..]
2405 .find('\n')
2406 .map_or(self.source.len(), |i| self.caret + i);
2407 if self.source[line_start..line_end].trim().is_empty() {
2408 self.insert_raw("\n");
2409 return;
2410 }
2411 // In `Preserve` flow a soft break is a *visible* line the author means to
2412 // make, so Enter writes a single `\n` and typing continues the same
2413 // paragraph on the next line — the behaviour of an ordinary text editor.
2414 // A second Enter then lands on the blank line above and takes the
2415 // empty-line branch, so double-Enter still promotes to a full paragraph
2416 // break; and Backspace, which deletes a lone `\n` over a soft break,
2417 // undoes a single Enter symmetrically. In `Fold` flow a lone `\n` would
2418 // render as an invisible space, so Enter keeps making the paragraph break
2419 // that actually shows.
2420 //
2421 // Only in running prose. A list or a quote has a continuation of its own
2422 // to write, and a `\n` there is not a soft line but a lost container.
2423 let in_container = has(Kind::ListItem) || has(Kind::TaskListItem) || has(Kind::BlockQuote);
2424 if self.line_flow == LineFlow::Preserve && !in_container {
2425 self.insert_raw("\n");
2426 return;
2427 }
2428 // A heading gets a *paragraph*, never a second heading: Enter at the end
2429 // of a title is how every editor is asked for the body under it, and
2430 // `split_block` would repeat the `#` instead. Whitespace at the split
2431 // point goes with the break rather than opening the new paragraph, which
2432 // is what `split_block` does everywhere else.
2433 if has(Kind::Heading) {
2434 let mut end = self.caret;
2435 while self.source.as_bytes().get(end) == Some(&b' ') {
2436 end += 1;
2437 }
2438 self.splice(self.caret, end, "\n\n", EditKind::Other);
2439 return;
2440 }
2441 self.split_block_here();
2442 }
2443
2444 /// Part the block at the caret with twig's [`Editor::split_block`], leaving
2445 /// the caret in the second half.
2446 ///
2447 /// twig reopens whatever the first half was inside of — the bullet with its
2448 /// indent, the quote's `>`, a checklist item's `[ ]` — which is the whole
2449 /// reason this replaced the markup leaf used to spell from the line's bytes.
2450 /// It renumbers nothing, though: a new item mid-list is written with its
2451 /// neighbour's number, so [`renumber_here`](Self::renumber_here) still runs
2452 /// behind it, folded into the same undo step.
2453 ///
2454 /// Falls back to a plain paragraph break if twig declines, so an unhandled
2455 /// shape still moves the caret down rather than swallowing the keystroke.
2456 fn split_block_here(&mut self) {
2457 // The read-only gate — this door reaches twig without the splice.
2458 if self.read_only {
2459 return;
2460 }
2461 match self.editor.split_block(self.caret) {
2462 Ok(change) => {
2463 self.last_edit_kind = None;
2464 self.refresh();
2465 self.anchor = None;
2466 self.caret = change.new.end;
2467 self.dirty = self.source != self.clean_source;
2468 self.status = None;
2469 self.clamp_caret();
2470 self.record_caret();
2471 // Aimed at the new block's *start*: the caret twig leaves is one
2472 // past the marker it wrote, where there is no list in reach.
2473 self.renumber_at(change.new.start);
2474 }
2475 Err(_) => self.insert_raw("\n\n"),
2476 }
2477 }
2478
2479 /// Whether the item on the marker's line carries no content — the shape
2480 /// double-Enter reads as "I'm done with this list."
2481 fn item_is_empty(&self, line: &ListMarker) -> bool {
2482 let content_start = line.content_start().min(self.source.len());
2483 let line_end = self.source[self.caret..]
2484 .find('\n')
2485 .map(|i| self.caret + i)
2486 .unwrap_or(self.source.len());
2487 self.source[content_start..line_end.max(content_start)]
2488 .trim()
2489 .is_empty()
2490 }
2491
2492 /// Leave the list: replace the empty item's marker with a blank line, so the
2493 /// caret lands in a fresh paragraph below it.
2494 ///
2495 /// Inside a quote the blank line has to stay quoted (a bare one would end the
2496 /// quote), and the caret's new line keeps the `> ` it was already behind —
2497 /// leaving the list without also leaving the quote.
2498 fn exit_list(&mut self, line: &ListMarker) {
2499 let prefix = self.quote_prefix_at(line.marker_start);
2500 let blank = prefix.trim_end();
2501 self.splice(
2502 line.line_start,
2503 self.caret,
2504 &format!("{blank}\n{prefix}"),
2505 EditKind::Other,
2506 );
2507 }
2508
2509 /// What a line continuing the containers at `off` has to open with — the
2510 /// quote markers reproduced, each enclosing item's marker as its width in
2511 /// spaces. Also the column a nested item's marker stands in, which is what
2512 /// makes it Tab's answer.
2513 fn continuation_prefix_at(&mut self, off: usize) -> String {
2514 self.editor
2515 .document()
2516 .and_then(|mut d| d.continuation_prefix(off))
2517 .map(|p| p.text)
2518 .unwrap_or_default()
2519 }
2520
2521 /// The column a *nested list* may open at inside the item at `off` — which
2522 /// is not always where the item's own text continues.
2523 ///
2524 /// twig counts a task item's `[ ] ` box as part of its marker, correctly:
2525 /// it is markup a rich view hides, and the item's own wrapped text does
2526 /// stand past it. But a nested list may only open at the *list* marker's
2527 /// column, and four columns further in is an indented continuation of the
2528 /// paragraph instead — `- [ ] a` + ` - [ ] b` is one item, not two.
2529 /// So the box's own width goes back.
2530 ///
2531 /// The one place leaf still reads a checkbox's spelling. It goes when twig
2532 /// reports the list marker's column apart from the box; `checked` is what
2533 /// says a box is there at all, so only its width is being measured here.
2534 fn nesting_prefix_at(&mut self, off: usize) -> String {
2535 let cont = self.continuation_prefix_at(off);
2536 let Some(item) = self.innermost_list_item(off) else {
2537 return cont;
2538 };
2539 if item.checked.is_none() {
2540 return cont;
2541 }
2542 let box_width = item
2543 .marker_span
2544 .and_then(|m| self.source.get(m))
2545 .and_then(|marker| marker.rfind('[').map(|i| marker.len() - i))
2546 .unwrap_or(0);
2547 // The trailing columns are the ones the item's own marker contributed,
2548 // so trimming from the end leaves any quote prefix standing.
2549 cont[..cont.len().saturating_sub(box_width)].to_string()
2550 }
2551
2552 /// Where the line of the item *containing* the item at `off` begins — the
2553 /// prefix Shift+Tab moves back to, which gives up exactly the level the
2554 /// parent contributed. The quote prefix alone for a top-level item, which
2555 /// has no level left to give.
2556 fn outdent_prefix_at(&mut self, off: usize) -> String {
2557 let items: Vec<usize> = self
2558 .editor
2559 .document()
2560 .and_then(|mut d| d.ancestors_at_caret(off))
2561 .map(|c| {
2562 c.into_iter()
2563 .filter(|m| m.kind == Kind::ListItem || m.kind == Kind::TaskListItem)
2564 .map(|m| m.span.start)
2565 .collect()
2566 })
2567 .unwrap_or_default();
2568 // The second-innermost item is the parent; its own line's indent is the
2569 // target. `list_marker_on_line` gives that line's prefix directly.
2570 let parent = items.len().checked_sub(2).map(|i| items[i]);
2571 match parent.and_then(|p| self.list_marker_on_line(p)) {
2572 Some(m) => self.source[m.line_start..m.marker_start].to_string(),
2573 None => self.quote_prefix_at(off),
2574 }
2575 }
2576
2577 /// The block-quote prefix in force at `off` — `""` outside a quote, `"> "`
2578 /// inside one, `"> > "` inside two.
2579 ///
2580 /// Assembled from each enclosing quote's own [`FlatNode::marker_span`], so
2581 /// the `>` and the space after it are twig's spelling rather than leaf's.
2582 /// The whole line prefix can't answer this: it also carries the indent of
2583 /// whatever the quote holds, which a blank separator line must *not* repeat.
2584 fn quote_prefix_at(&mut self, off: usize) -> String {
2585 let Ok(chain) = self
2586 .editor
2587 .document()
2588 .and_then(|mut d| d.ancestors_at_caret(off))
2589 else {
2590 return String::new();
2591 };
2592 let quotes: Vec<usize> = chain
2593 .iter()
2594 .filter(|m| m.kind == Kind::BlockQuote)
2595 .map(|m| m.node_id as usize)
2596 .collect();
2597 let Ok(nodes) = self.editor.nodes() else {
2598 return String::new();
2599 };
2600 quotes
2601 .iter()
2602 .filter_map(|id| nodes.get(*id)?.marker_span.clone())
2603 .filter_map(|s| self.source.get(s))
2604 .collect()
2605 }
2606
2607 /// Whether the item at `off` sits inside another one — the test Backspace
2608 /// uses to choose between outdenting and dropping the marker.
2609 ///
2610 /// Counted from the AST rather than from the line's leading whitespace,
2611 /// which is indentation in Markdown and, in Djot, may be nothing at all.
2612 fn item_is_nested(&mut self, off: usize) -> bool {
2613 self.editor
2614 .document()
2615 .and_then(|mut d| d.ancestors_at_caret(off))
2616 .map(|c| {
2617 c.into_iter()
2618 .filter(|m| m.kind == Kind::ListItem || m.kind == Kind::TaskListItem)
2619 .count()
2620 > 1
2621 })
2622 .unwrap_or(false)
2623 }
2624
2625 /// The innermost list item containing `probe`, under twig's **caret**
2626 /// containment rule — a block's end is inside it.
2627 ///
2628 /// Half-open containment can't answer this. An empty item's span is exactly
2629 /// its marker, so the caret sitting after `- ` is one past the end and the
2630 /// item it is plainly in tests as out of reach; that is the shape
2631 /// double-Enter has to recognise to leave the list.
2632 fn innermost_list_item(&mut self, probe: usize) -> Option<FlatNode> {
2633 let chain = self
2634 .editor
2635 .document()
2636 .and_then(|mut d| d.ancestors_at_caret(probe))
2637 .ok()?;
2638 let id = chain
2639 .iter()
2640 .rev()
2641 .find(|m| m.kind == Kind::ListItem || m.kind == Kind::TaskListItem)?
2642 .node_id as usize;
2643 self.editor.nodes().ok()?.get(id).cloned()
2644 }
2645
2646 /// The list marker opening `off`'s line, per twig — `None` when that line
2647 /// opens no list item.
2648 ///
2649 /// [`Document::line_prefix`] is the whole hidden run from the line start:
2650 /// `> 1. ` is a quote's marker, an indent, and an item's marker together,
2651 /// and it is `None` on a *continuation* line, which opens nothing. That last
2652 /// case is the one leaf could never get right by reading bytes. `- a\n - b`
2653 /// is two items in Markdown and one in Djot, where a marker cannot interrupt
2654 /// a paragraph and ` - b` is literal text — identical bytes, and only the
2655 /// parser knows which document it is looking at.
2656 ///
2657 /// The item's own marker is separated out via its
2658 /// [`FlatNode::marker_span`], so `marker_start` splits the prefix into what
2659 /// the containers around it contribute and what the item does.
2660 fn list_marker_on_line(&mut self, off: usize) -> Option<ListMarker> {
2661 let off = off.min(self.source.len());
2662 let prefix = self.editor.document().ok()?.line_prefix(off).ok()??;
2663 // The prefix belongs to a list only when an item's marker closes it —
2664 // a heading's `# ` or a bare quote's `> ` is a prefix too.
2665 let item = self.innermost_list_item(prefix.end.min(self.source.len()))?;
2666 let marker = item.marker_span.clone()?;
2667 if marker.end != prefix.end {
2668 return None;
2669 }
2670 Some(ListMarker {
2671 line_start: prefix.start,
2672 marker_start: marker.start,
2673 text: self.source.get(prefix)?.to_string(),
2674 })
2675 }
2676
2677 /// Whether the list item on `line_start`'s line is the **first item** of its
2678 /// list — the one Tab must not nest, because nesting needs a preceding
2679 /// sibling to become the new parent and a first item has none. `false` for a
2680 /// line that isn't a list item, and for an item with a sibling above it (the
2681 /// one Tab *can* nest). Gated on the AST, not the marker bytes: `- ` reads
2682 /// the same in a setext underline that opens no list at all.
2683 fn first_item_of_list(&mut self, line_start: usize) -> bool {
2684 let Some(marker) = self.list_marker_on_line(line_start) else {
2685 return false;
2686 };
2687 // Probe just inside the marker, where the item's own node is in reach —
2688 // the marker offset itself can resolve to the enclosing list, not the
2689 // `list_item`, whose span starts at the marker.
2690 let probe = marker.content_start().min(self.source.len());
2691 let Some(item) = self.innermost_list_item(probe) else {
2692 return false;
2693 };
2694 let Ok(nodes) = self.editor.nodes() else {
2695 return false;
2696 };
2697 match item.parent {
2698 // First when the parent list opens with this very item.
2699 Some(pid) => nodes
2700 .get(pid.0 as usize)
2701 .is_some_and(|p| p.first_child == Some(item.id)),
2702 // A parentless item is trivially the first (and only) one.
2703 None => true,
2704 }
2705 }
2706
2707 pub fn backspace(&mut self) {
2708 if let Some((s, e)) = self.selection() {
2709 self.splice(s, e, "", EditKind::Other);
2710 return;
2711 }
2712 // WYSIWYG: Backspace at the very start of a list item's content is a
2713 // structural key, not a character delete — it walks the "un-indent, then
2714 // un-list" ladder every list editor gives that keystroke (outdent a
2715 // nested item, strip a top-level one's marker to a paragraph). In source
2716 // view the `- ` is visible text the user is deleting a byte of, so it
2717 // keeps its literal meaning there, like Enter does.
2718 if self.view != View::Source && self.backspace_list_start() {
2719 return;
2720 }
2721 // WYSIWYG: and the same at the start of a heading's content — the `# `
2722 // there is markup the rich view hides, not text the user typed.
2723 if self.view != View::Source && self.backspace_heading_start() {
2724 return;
2725 }
2726 // WYSIWYG: at a block picture's stops, a byte-at-a-time delete would take
2727 // the markup apart under a caret that cannot see it — see
2728 // `delete_around_block_media`.
2729 if self.view != View::Source && self.delete_around_block_media(false) {
2730 return;
2731 }
2732 // WYSIWYG: Backspace on a *blank line* deletes back to the previous caret
2733 // stop, not a single newline. On a line with no text of its own, the byte
2734 // before the caret is a `\n` that spells part of a block boundary — the gap
2735 // between two blocks, drawn but never a caret home. Removing just it strands
2736 // the caret in that gap and leaves an odd blank line the eye reads as one
2737 // separator but the caret can't land on: the "extra newline" left behind
2738 // after leaving a list (Enter, Enter) or a paragraph and pressing Backspace.
2739 // Deleting to the previous stop instead collapses the whole break at once,
2740 // landing the caret at the end of the block above. Two blank lines in a row
2741 // are one stop apart, so this still removes exactly one — the lone-Enter /
2742 // lone-Backspace symmetry the empty-line case is built on is untouched.
2743 if self.view != View::Source
2744 && self.caret > self.caret_floor()
2745 && self.caret_on_blank_line()
2746 && let Some(stop) = self.vmap.stop_before(self.caret)
2747 {
2748 let stop = stop.max(self.caret_floor());
2749 if stop < self.caret {
2750 self.splice(stop, self.caret, "", EditKind::Delete);
2751 return;
2752 }
2753 }
2754 if self.caret > self.caret_floor() {
2755 // An in-cell `<br>` draws as one newline glyph, so Backspace over it
2756 // takes the whole tag — a single-byte step would leave a broken `<br`
2757 // showing in the cell. Rich view only (source view edits the literal).
2758 if self.view != View::Source
2759 && let Some((start, end)) = self.cell_break_at(BreakEdge::Backward)
2760 {
2761 let start = start.max(self.caret_floor());
2762 if start < end {
2763 self.splice(start, end, "", EditKind::Delete);
2764 return;
2765 }
2766 }
2767 // Aim the delete at the character the writer can *see* behind the
2768 // caret, never at a delimiter the rich view drew nothing for. Two
2769 // steps, and either can apply: from the far side of a run's closing
2770 // `**` step back into the run (the caret is drawn at the end of its
2771 // word), and at the start of a run's text step out past its opening
2772 // `**` to the character in front of it, leaving the run standing.
2773 // Without them a plain Backspace unspells the phrase it is editing
2774 // and leaves a literal asterisk on screen.
2775 let end = if self.view == View::Source {
2776 self.caret
2777 } else {
2778 let inside = self.step_inside_close_delims(self.caret);
2779 self.skip_leading_open_delims(inside)
2780 .max(self.caret_floor())
2781 };
2782 // Never delete back across the floor — that would eat hidden
2783 // frontmatter the WYSIWYG caret can't even see.
2784 let mut prev = prev_boundary(&self.source, end).max(self.caret_floor());
2785 // Take a hidden escape backslash with the char it escapes: the rich
2786 // view draws `\*` as a single `*`, so Backspace over it must delete
2787 // both bytes, never strand the `\` as a lone visible backslash (the
2788 // mirror of the Hidden-mode typing that wrote the escape). Source view
2789 // shows the `\`, so there it is an ordinary character.
2790 if self.view != View::Source
2791 && prev > self.caret_floor()
2792 && self.is_hidden_escape(prev - 1)
2793 {
2794 prev -= 1;
2795 }
2796 if prev < end {
2797 self.splice(prev, end, "", EditKind::Delete);
2798 }
2799 }
2800 }
2801
2802 /// Whether the caret's own source line holds nothing but whitespace — an
2803 /// empty paragraph, or the blank line a block boundary is spelled with. The
2804 /// test for [`backspace`](Self::backspace)'s stop-wise delete: such a line has
2805 /// no text of its own, so the newline before the caret belongs to the gap
2806 /// between blocks rather than to any word the caret is editing.
2807 fn caret_on_blank_line(&self) -> bool {
2808 let line_start = self.source[..self.caret].rfind('\n').map_or(0, |i| i + 1);
2809 let line_end = self.source[self.caret..]
2810 .find('\n')
2811 .map_or(self.source.len(), |i| self.caret + i);
2812 self.source[line_start..line_end].trim().is_empty()
2813 }
2814
2815 /// The source span of an in-cell hard break (`<br>`) touching the caret on the
2816 /// `edge` side — the byte range to delete whole. A table row is one source
2817 /// line, so its break is spelled `<br>` yet drawn as a single newline glyph
2818 /// (see `wysiwyg.rs`); a delete over it must take every byte, or a one-byte
2819 /// step strands a broken `<br` in the cell. `Backward` matches a break ending
2820 /// at the caret (Backspace), `Forward` one starting at it (Delete). `None`
2821 /// when no such break is adjacent. Only the in-cell break is spelled `<br>`
2822 /// (an ordinary hard break is ` \n`), so the leading `<` alone tells them
2823 /// apart — no ancestor walk needed. Rich view only; source view shows the
2824 /// literal tag and deletes it a byte at a time.
2825 fn cell_break_at(&mut self, edge: BreakEdge) -> Option<(usize, usize)> {
2826 let caret = self.caret;
2827 let nodes = self.nodes();
2828 let src = self.source.as_bytes();
2829 nodes
2830 .iter()
2831 .find(|n| {
2832 n.kind == Kind::HardBreak
2833 && n.span.start < n.span.end
2834 && src.get(n.span.start) == Some(&b'<')
2835 && match edge {
2836 BreakEdge::Backward => n.span.end == caret,
2837 BreakEdge::Forward => n.span.start == caret,
2838 }
2839 })
2840 .map(|n| (n.span.start, n.span.end))
2841 }
2842
2843 /// Whether the source byte at `off` is a backslash twig consumed as an escape
2844 /// (hidden in the rich view), as against a literal backslash (drawn). A
2845 /// backslash escapes exactly an ASCII-punctuation character (the CommonMark /
2846 /// Djot rule twig follows), so `\` + punctuation is the whole test — no AST
2847 /// round-trip needed.
2848 fn is_hidden_escape(&self, off: usize) -> bool {
2849 let b = self.source.as_bytes();
2850 b.get(off) == Some(&b'\\') && b.get(off + 1).is_some_and(u8::is_ascii_punctuation)
2851 }
2852
2853 /// Backspace's list behaviour: when the caret sits exactly at the start of a
2854 /// list item's content (right after its marker), outdent the item if it's
2855 /// nested, else strip the marker so it becomes a paragraph. Returns whether
2856 /// it acted — `false` leaves Backspace its ordinary character delete.
2857 fn backspace_list_start(&mut self) -> bool {
2858 let Some(marker) = self.list_marker_on_line(self.caret) else {
2859 return false;
2860 };
2861 // Only right after the marker. That the line opens a real item is
2862 // already settled: `list_marker_on_line` answers from the tree.
2863 if self.caret != marker.content_start() {
2864 return false;
2865 }
2866 if self.item_is_nested(marker.marker_start) {
2867 // Nested: give back one level, keeping the marker and carrying the
2868 // caret with it.
2869 self.outdent();
2870 } else {
2871 // Top level: drop the marker, leaving a paragraph, then renumber the
2872 // siblings the removed item was counted among. Only the marker goes —
2873 // a quote prefix in front of it still has a quote to hold up.
2874 self.splice(marker.marker_start, self.caret, "", EditKind::Other);
2875 self.renumber_here();
2876 }
2877 true
2878 }
2879
2880 /// Backspace's heading behaviour: with the caret exactly at the start of an
2881 /// ATX heading's content — right after the `#` marker the rich view hides —
2882 /// strip the marker so the line becomes a paragraph. The peer of
2883 /// [`backspace_list_start`](Self::backspace_list_start)'s ladder, and the same
2884 /// reasoning: hidden block markup is structure, so the keystroke over it is
2885 /// structural.
2886 ///
2887 /// Without this the ordinary delete takes the space out of `# Title` and
2888 /// leaves `#Title`, which is no longer a heading at all — the hash the view
2889 /// had been hiding surfaces as literal text the user has to delete a second
2890 /// time, having never typed it. A closing sequence (`# Title #`, hidden at the
2891 /// other end) goes with the marker for the same reason.
2892 ///
2893 /// Returns whether it acted; `false` leaves Backspace its character delete.
2894 fn backspace_heading_start(&mut self) -> bool {
2895 let caret = self.caret;
2896 // The heading whose content opens exactly at the caret. A bare `#` has no
2897 // content span at all — its content starts (and ends) where the line does.
2898 let Some((span, content_end, marker)) = self.nodes().iter().find_map(|n| {
2899 let (start, end) = match &n.content_span {
2900 Some(c) => (c.start, c.end),
2901 None => (n.span.end, n.span.end),
2902 };
2903 (n.kind == Kind::Heading && start == caret)
2904 .then(|| (n.span.clone(), end, n.marker_span.clone()))
2905 }) else {
2906 return false;
2907 };
2908 // twig reports the marker's own extent, so there is nothing to walk back
2909 // over and no `#` in this file. A setext heading has no marker — its
2910 // content opens the line — so it falls through to the ordinary delete,
2911 // as does anything else sitting at a content start.
2912 // `m.end == caret` is what excludes a setext heading, whose marker is the
2913 // underline *after* the content rather than a prefix before it.
2914 let Some(marker) = marker.filter(|m| m.end == caret) else {
2915 return false;
2916 };
2917 let start = marker.start;
2918 // A closing `#` sequence is hidden too, so it can't be left behind. Only
2919 // when the tail really is one: trailing spaces alone are nothing to strip.
2920 let tail = &self.source[content_end..span.end];
2921 if tail.contains('#') && tail.chars().all(|c| c == '#' || c.is_whitespace()) {
2922 let kept = self.source[caret..content_end].to_string();
2923 self.splice(start, span.end, &kept, EditKind::Other);
2924 // The splice leaves the caret past the text it re-wrote; the caret
2925 // belongs where the content now starts, which is where it already was.
2926 self.caret = start;
2927 self.record_caret();
2928 } else {
2929 self.splice(start, caret, "", EditKind::Other);
2930 }
2931 true
2932 }
2933
2934 pub fn delete_forward(&mut self) {
2935 if let Some((s, e)) = self.selection() {
2936 self.splice(s, e, "", EditKind::Other);
2937 } else if self.caret < self.source.len() {
2938 // The mirror of Backspace's: forward-delete in front of a picture
2939 // would eat the `!` off its markup and leave a link where a photo was.
2940 if self.view != View::Source && self.delete_around_block_media(true) {
2941 return;
2942 }
2943 // Delete forward over an in-cell `<br>` takes the whole tag, the mirror
2944 // of Backspace's swallow (see `cell_break_at`) — else a byte-step
2945 // strands a broken `<br` in the cell.
2946 if self.view != View::Source
2947 && let Some((start, end)) = self.cell_break_at(BreakEdge::Forward)
2948 {
2949 self.splice(start, end, "", EditKind::Delete);
2950 return;
2951 }
2952 // The mirror of Backspace's two steps: from in front of a run's
2953 // opening `**` step into it, onto the first letter of its text, and
2954 // at the end of a run's text step out past its closing `**` to the
2955 // character beyond. Either way Delete takes the character it looks
2956 // like it is pointing at, and never a delimiter drawn as nothing.
2957 // The caret then settles back inside the run it was standing in —
2958 // see `settle_inside_close_delims`.
2959 let from = if self.view == View::Source {
2960 self.caret
2961 } else {
2962 let inside = self.step_inside_open_delims(self.caret);
2963 self.skip_trailing_close_delims(inside)
2964 };
2965 let next = next_boundary(&self.source, from);
2966 if from < next {
2967 self.splice(from, next, "", EditKind::Delete);
2968 }
2969 }
2970 }
2971
2972 /// Delete from the caret back to the start of the previous word (⌥⌫ /
2973 /// Ctrl+⌫). Deletes the selection instead when one is active.
2974 pub fn delete_word_back(&mut self) {
2975 if let Some((s, e)) = self.selection() {
2976 self.splice(s, e, "", EditKind::Other);
2977 } else {
2978 // A word back from just past a picture is a word *of its markup*, and
2979 // a word back from in front of one runs through the paragraph break
2980 // into the prose above — dissolving the picture either way. See
2981 // `delete_around_block_media`.
2982 if self.view != View::Source && self.delete_around_block_media(false) {
2983 return;
2984 }
2985 let start = self.word_left_from(self.caret).max(self.caret_floor());
2986 if start < self.caret {
2987 let (s, e) = self.widen_over_emptied_inlines(start, self.caret);
2988 self.splice(s, e, "", EditKind::Delete);
2989 }
2990 }
2991 }
2992
2993 /// Delete from the caret forward to the end of the next word (⌥⌦ /
2994 /// Ctrl+Del). Deletes the selection instead when one is active.
2995 pub fn delete_word_forward(&mut self) {
2996 if let Some((s, e)) = self.selection() {
2997 self.splice(s, e, "", EditKind::Other);
2998 } else {
2999 // The mirror: a word forward from in front of a picture is its markup.
3000 if self.view != View::Source && self.delete_around_block_media(true) {
3001 return;
3002 }
3003 let end = self.word_right_from(self.caret);
3004 if end > self.caret {
3005 let (s, e) = self.widen_over_emptied_inlines(self.caret, end);
3006 self.splice(s, e, "", EditKind::Delete);
3007 }
3008 }
3009 }
3010
3011 /// Delete from the caret back to the start of its line (⌘⌫). Deletes the
3012 /// selection instead when one is active, as every other delete here does.
3013 ///
3014 /// The line is the view's own — the one Home and End work on, so in WYSIWYG
3015 /// a soft-wrapped row is a line. It is not Home's *target*, though: Home
3016 /// stops at the first character and this takes the indentation with it, the
3017 /// way Cocoa's `deleteToBeginningOfLine:` does. Stopping at the text would
3018 /// leave an indent behind that nothing can then ask to delete, where a caret
3019 /// left at column 0 is one press of Home away from either.
3020 pub fn delete_to_line_start(&mut self) {
3021 if let Some((s, e)) = self.selection() {
3022 self.splice(s, e, "", EditKind::Other);
3023 return;
3024 }
3025 // Never back across the floor: hidden frontmatter isn't on this line, or
3026 // on any line the WYSIWYG caret can see.
3027 let (start, _) = self.line_span();
3028 let start = start.max(self.caret_floor());
3029 if start < self.caret {
3030 let (s, e) = self.widen_over_emptied_inlines(start, self.caret);
3031 self.splice(s, e, "", EditKind::Delete);
3032 }
3033 }
3034
3035 /// Kill from the caret to the end of its line (^K). Deletes the selection
3036 /// instead when one is active.
3037 ///
3038 /// At the end of the line it does nothing, rather than pulling the line
3039 /// below up into this one. Joining has no meaning to give it in both views
3040 /// at once: a WYSIWYG line ends at a soft wrap as often as at a newline, and
3041 /// there is nothing there to delete, while the newline a *source* line ends
3042 /// with is only half of the blank line that separates two paragraphs —
3043 /// deleting one leaves a soft break, which is not the join it looks like.
3044 /// The views agreeing is worth more than emacs' second press, and Delete is
3045 /// already the key that joins.
3046 pub fn delete_to_line_end(&mut self) {
3047 if let Some((s, e)) = self.selection() {
3048 self.splice(s, e, "", EditKind::Other);
3049 return;
3050 }
3051 let (_, end) = self.line_span();
3052 if end > self.caret {
3053 let (s, e) = self.widen_over_emptied_inlines(self.caret, end);
3054 self.splice(s, e, "", EditKind::Delete);
3055 }
3056 }
3057
3058 /// Grow a WYSIWYG word-delete to swallow any inline node it empties.
3059 ///
3060 /// A glyph-space range covers what the user can see, which for `**bold**` is
3061 /// the word and never the delimiters around it — so deleting the word on its
3062 /// own leaves `a **** c`, markup wrapped around nothing. They asked for the
3063 /// word, and the styling was the word's; the two go together. Only the
3064 /// node's delimiters are taken, and those are hidden here anyway, so nothing
3065 /// visible outside the range is lost.
3066 ///
3067 /// Repeated to a fixed point: emptying `***bold***` empties the emph inside
3068 /// the strong, and only then is the strong empty too.
3069 fn widen_over_emptied_inlines(&mut self, start: usize, end: usize) -> (usize, usize) {
3070 if self.view == View::Source {
3071 return (start, end);
3072 }
3073 let nodes = self.nodes();
3074 let (mut s, mut e) = (start, end);
3075 loop {
3076 let mut grew = false;
3077 for n in nodes.iter().filter(|n| wysiwyg::is_inline(n)) {
3078 let Some(text) = inline_content_span(n, &self.source) else {
3079 continue;
3080 };
3081 // Some of its text survives, so the node still has a job.
3082 if text.start < s || text.end > e {
3083 continue;
3084 }
3085 if n.span.start < s || n.span.end > e {
3086 s = s.min(n.span.start);
3087 e = e.max(n.span.end);
3088 grew = true;
3089 }
3090 }
3091 if !grew {
3092 return (s, e);
3093 }
3094 }
3095 }
3096
3097 /// One splice of document text, keeping the **mark-edge rule**: an inline
3098 /// mark's content never begins or ends with whitespace. In Markdown and Djot
3099 /// a delimiter standing against a space is not a delimiter at all — `**bold **`
3100 /// is four literal asterisks around a word, and a rich view drawing the
3101 /// document faithfully has no choice but to show them. That is correct
3102 /// rendering of what the file says, and nobody typing a space after a bold
3103 /// word meant to say it.
3104 ///
3105 /// So the space goes *outside* the run instead — `**bold** ` — which is the
3106 /// same document to a reader and a live one to a parser. The caret follows it
3107 /// out and keeps the marks armed (see [`rearm`](Self::rearm)), so the next
3108 /// character rejoins the run (see [`rejoin_run`](Self::rejoin_run)) and the
3109 /// writer sees one unbroken bold phrase, never a flash of raw syntax.
3110 ///
3111 /// Every ordinary edit — typing, deleting, pasting, an IME step — comes
3112 /// through here, so the rule holds however the whitespace arrives at the
3113 /// edge. The repair is decided *after* the plain edit, by asking whether the
3114 /// mark actually died: a code span's backticks aren't whitespace-sensitive
3115 /// (`` `code ` `` is still code), and nothing is re-spelled when nothing broke.
3116 fn splice(&mut self, start: usize, end: usize, text: &str, kind: EditKind) -> bool {
3117 let fix = self.mark_edge_fix(start, end, text);
3118 if !self.splice_exact(start, end, text, kind) {
3119 return false;
3120 }
3121 if let Some(fix) = fix {
3122 self.repair_mark_edges(fix);
3123 }
3124 if text.is_empty() && end > start {
3125 self.settle_inside_close_delims();
3126 }
3127 true
3128 }
3129
3130 /// After a delete, take a caret left standing past a run's closing delimiters
3131 /// back inside the run.
3132 ///
3133 /// A delete leaves the caret where the deleted bytes began, and when those
3134 /// bytes were the last thing after a marked phrase — the space the mark-edge
3135 /// rule pushed out of `**bold** `, say — that spot is the far side of the
3136 /// closing `**`. The rich view has nothing to draw there: the delimiters are
3137 /// hidden, so the caret shows at the end of the word either way, and the two
3138 /// offsets are one place on screen with two different meanings. Typing at the
3139 /// outer one lands past the run, so the writer who backspaced a space out of
3140 /// their bold phrase watches the next character come out plain, and the
3141 /// toolbar button go dark, with the caret never appearing to move.
3142 ///
3143 /// The end of the run's text is the caret's home there — a delete that took
3144 /// away everything after a phrase leaves the caret at the end of that phrase,
3145 /// which is inside it — so it settles onto that
3146 /// ([`step_inside_close_delims`](Self::step_inside_close_delims) does the
3147 /// walk, through every mark closing at the point): the word stays bold, the
3148 /// button stays lit, and the next character carries on the phrase.
3149 ///
3150 /// Rich view only, and only where a mark really closes at the caret — mid-run
3151 /// or in plain prose no span ends there and the caret stays put. The opening
3152 /// edge is left alone on purpose: a caret in front of a run inherits from the
3153 /// text on its left, which is the plain text outside.
3154 fn settle_inside_close_delims(&mut self) {
3155 if self.view != View::Wysiwyg {
3156 return;
3157 }
3158 let at = self.step_inside_close_delims(self.caret);
3159 if at != self.caret {
3160 self.caret = at;
3161 self.clear_pending();
3162 self.record_caret();
3163 }
3164 }
3165
3166 /// The splice exactly as asked, with no mark-edge repair — for the callers
3167 /// that are *writing* the delimiters themselves ([`insert_with_marks`](Self::insert_with_marks)
3168 /// and [`rejoin_run`](Self::rejoin_run)) and place their own offsets around
3169 /// the bytes they inserted.
3170 ///
3171 /// One `edit_range` through twig, then re-anchor the caret from the returned
3172 /// `Change` and refresh the cached source. A reparse-breaking edit (rare for
3173 /// Markdown/Djot) leaves the document untouched and reports.
3174 ///
3175 /// Returns whether the edit landed — for a caller that has offsets of its
3176 /// own to place afterwards, which a rolled-back splice would leave pointing
3177 /// into text that never came to exist.
3178 fn splice_exact(&mut self, start: usize, end: usize, text: &str, kind: EditKind) -> bool {
3179 // The read-only gate, for every edit at once — see the field.
3180 if self.read_only {
3181 return false;
3182 }
3183 // twig records an undo step for every edit; when this one continues a
3184 // run of the same kind (typing, deleting), tell twig to fold it into the
3185 // step before it so the whole run undoes at once.
3186 let coalesce = kind != EditKind::Other && self.last_edit_kind == Some(kind);
3187 // Hand twig the pre-edit caret before the splice, so the undo step it
3188 // retires carries where the caret was standing.
3189 self.record_caret();
3190 match self.editor.edit_range(start, end, text) {
3191 Ok(change) => {
3192 if coalesce {
3193 let _ = self.editor.coalesce_last_undo();
3194 }
3195 self.last_edit_kind = Some(kind);
3196 self.refresh();
3197 self.caret = change.new.end;
3198 self.anchor = None;
3199 self.goal_col = None;
3200 self.clear_pending();
3201 self.dirty = self.source != self.clean_source;
3202 self.status = None;
3203 // And the post-edit caret, so a later redo restores it.
3204 self.record_caret();
3205 true
3206 }
3207 // The edit was rolled back, so twig's history did not move and
3208 // neither may ours: pushing here would leave a step with no edit
3209 // under it and shift every later undo onto the wrong caret.
3210 Err(e) => {
3211 self.status = Some(format!("edit: {e}"));
3212 false
3213 }
3214 }
3215 }
3216
3217 /// The re-spelling that would keep the mark-edge rule for the edit
3218 /// `[start, end)` → `text`, or `None` when the edit leaves no whitespace
3219 /// against a delimiter and the plain splice is already right. Computed
3220 /// *before* the edit, while the run's spans and delimiters can still be read
3221 /// off the document; applied afterwards, and only if the mark really died —
3222 /// see [`repair_mark_edges`](Self::repair_mark_edges).
3223 ///
3224 /// Rich view only. Source view is for typing raw markup, where a space put
3225 /// against a `**` is exactly the character it looks like.
3226 fn mark_edge_fix(&mut self, start: usize, end: usize, text: &str) -> Option<MarkEdgeFix> {
3227 if self.view != View::Wysiwyg || start > end || end > self.source.len() {
3228 return None;
3229 }
3230 // Every inline mark standing over the edit, outermost first, with the
3231 // content span that says where its delimiters are.
3232 let chain: Vec<(InlineKind, std::ops::Range<usize>, std::ops::Range<usize>)> = self
3233 .editor
3234 .ancestors_at(start)
3235 .unwrap_or_default()
3236 .into_iter()
3237 .filter_map(|m| {
3238 let kind = inline_kind(&m.kind)?;
3239 let content = m.content_span.clone()?;
3240 Some((kind, m.span.clone(), content))
3241 })
3242 .collect();
3243 // The innermost run whose *content* holds the whole edit: the one whose
3244 // text is being changed, rather than one the edit merely sits under.
3245 let (kind, span, content) = chain
3246 .iter()
3247 .rev()
3248 .find(|(_, _, c)| c.start <= start && end <= c.end)?
3249 .clone();
3250 // What that content becomes. Whitespace at either end of it is what
3251 // would put out the mark.
3252 let body = format!(
3253 "{}{text}{}",
3254 &self.source[content.start..start],
3255 &self.source[end..content.end]
3256 );
3257 let (lead, trail) = if body.trim().is_empty() {
3258 // Nothing but whitespace left: there is no content to mark at all,
3259 // and the delimiters go with it rather than closing on a space.
3260 (body.len(), 0)
3261 } else {
3262 (
3263 body.len() - body.trim_start().len(),
3264 body.len() - body.trim_end().len(),
3265 )
3266 };
3267 // Nothing against a delimiter, and something still between them: the
3268 // plain edit stands. An emptied run is broken just as surely (`**b**`
3269 // with the `b` deleted is the literal `****`) and is re-spelt as the
3270 // nothing it now says.
3271 if lead == 0 && trail == 0 && !body.is_empty() {
3272 return None;
3273 }
3274 // Marks that open or close exactly where this one does — `***both***` is
3275 // two runs sharing an edge — spell their delimiters as one run of bytes,
3276 // so the whitespace has to clear all of them together.
3277 let (mut open_at, mut close_at) = (span.start, span.end);
3278 for _ in 0..chain.len() {
3279 match chain.iter().find(|(_, _, c)| c.start == open_at) {
3280 Some((_, s, _)) => open_at = s.start,
3281 None => break,
3282 }
3283 }
3284 for _ in 0..chain.len() {
3285 match chain.iter().find(|(_, _, c)| c.end == close_at) {
3286 Some((_, s, _)) => close_at = s.end,
3287 None => break,
3288 }
3289 }
3290 let open = &self.source[open_at..content.start];
3291 let close = &self.source[content.end..close_at];
3292 let core = &body[lead..body.len() - trail];
3293 let respelt = if core.is_empty() {
3294 body.clone()
3295 } else {
3296 format!(
3297 "{}{open}{core}{close}{}",
3298 &body[..lead],
3299 &body[body.len() - trail..]
3300 )
3301 };
3302 // The caret sits just past the inserted text within the new content —
3303 // which, when that lands in the whitespace, is now outside the delimiters.
3304 let pos = (start - content.start) + text.len();
3305 let caret = if core.is_empty() || pos <= lead {
3306 open_at + pos
3307 } else if pos >= lead + core.len() {
3308 open_at + lead + open.len() + core.len() + close.len() + (pos - lead - core.len())
3309 } else {
3310 open_at + lead + open.len() + (pos - lead)
3311 };
3312 Some(MarkEdgeFix {
3313 kind,
3314 probe: content.start,
3315 start: open_at,
3316 end: close_at + text.len() - (end - start),
3317 text: respelt,
3318 caret,
3319 // The marks in force here, resolved against any armed sticky delta —
3320 // what the writer is typing in, and so what has to still be true on
3321 // the far side of the delimiter the caret just stepped over.
3322 want: chain
3323 .iter()
3324 .filter(|(_, s, _)| start < s.end)
3325 .map(|(k, _, _)| *k)
3326 .collect::<InlineMarks>()
3327 .xor(self.pending_here()),
3328 })
3329 }
3330
3331 /// Apply a [`MarkEdgeFix`] — but only if the edit it was computed for really
3332 /// did break the mark. Whether whitespace at a delimiter is fatal is the
3333 /// format's business, not leaf's: `**bold **` is no longer strong, while
3334 /// `` `code ` `` is still perfectly good verbatim, and Djot's braced spellings
3335 /// don't care either. Asking the parser afterwards settles it for every kind
3336 /// and format at once, and costs a re-spelling only where one is due.
3337 ///
3338 /// The repair rides along with the edit that caused it — one undo step puts
3339 /// back what the writer typed, not a delimiter shuffle they never saw.
3340 fn repair_mark_edges(&mut self, fix: MarkEdgeFix) {
3341 if fix.end > self.source.len() {
3342 return;
3343 }
3344 if self.marks_at(fix.probe).iter().any(|(k, _)| *k == fix.kind) {
3345 return; // still a mark: these delimiters don't mind the whitespace
3346 }
3347 let resumed = self.last_edit_kind;
3348 if !self.splice_exact(fix.start, fix.end, &fix.text, EditKind::Other) {
3349 return;
3350 }
3351 let _ = self.editor.coalesce_last_undo();
3352 // The keystroke owns the undo step, so the run of typing it belongs to
3353 // keeps coalescing over the repair rather than breaking in two here.
3354 self.last_edit_kind = resumed;
3355 self.caret = fix.caret.min(self.source.len());
3356 self.anchor = None;
3357 self.goal_col = None;
3358 self.rearm(fix.want);
3359 self.clamp_caret();
3360 self.record_caret();
3361 }
3362
3363 /// Arm whatever sticky delta reproduces `want` at the caret — the marks the
3364 /// writer is typing in, carried across an edit that moved the caret out of
3365 /// the run holding them. Arms nothing when the caret already stands in
3366 /// exactly those marks, but still remembers the spot, so a further ⌘b starts
3367 /// a clean delta here (see [`toggle`](Self::toggle)).
3368 fn rearm(&mut self, want: InlineMarks) {
3369 let here: InlineMarks = self
3370 .marks_at(self.caret)
3371 .into_iter()
3372 .map(|(k, _)| k)
3373 .collect();
3374 self.pending_marks = want.xor(here);
3375 self.pending_at = Some(self.caret);
3376 }
3377
3378 /// Insert `text` at `at` as a *literal* run via twig's `insert_literal`,
3379 /// which backslash-escapes any character that would otherwise open markup in
3380 /// this format and position (`*` → `\*`, a line-start `#` → `\#`). The mirror
3381 /// of [`splice`](Self::splice) for the Hidden reveal mode's typing path, with
3382 /// the same caret re-anchor, coalescing, and rollback contract. `at` must be
3383 /// a collapsed point — a selection is deleted by the caller first, since
3384 /// `insert_literal` inserts rather than replaces.
3385 fn insert_literal_at(
3386 &mut self,
3387 at: usize,
3388 text: &str,
3389 kind: EditKind,
3390 force_coalesce: bool,
3391 ) -> bool {
3392 // The read-only gate: this door goes to twig directly, not through
3393 // `splice_exact`, so it guards itself — see the field.
3394 if self.read_only {
3395 return false;
3396 }
3397 // `force_coalesce` folds this into the immediately preceding edit (the
3398 // selection-delete of an overwrite) so the pair is one undo step; else it
3399 // coalesces only when it continues a run of the same-kind typing.
3400 let coalesce =
3401 force_coalesce || (kind != EditKind::Other && self.last_edit_kind == Some(kind));
3402 // The mark-edge rule holds for typed text however it is spelled — see
3403 // `splice`. Only an insert twig passed through unchanged can use it,
3404 // since a fix is measured in the bytes that actually land, and an escape
3405 // adds bytes this couldn't have counted.
3406 let fix = self.mark_edge_fix(at, at, text);
3407 self.record_caret();
3408 match self.editor.insert_literal(at, text) {
3409 Ok(change) => {
3410 if coalesce {
3411 let _ = self.editor.coalesce_last_undo();
3412 }
3413 self.last_edit_kind = Some(kind);
3414 self.refresh();
3415 self.caret = change.new.end;
3416 self.anchor = None;
3417 self.goal_col = None;
3418 self.clear_pending();
3419 self.dirty = self.source != self.clean_source;
3420 self.status = None;
3421 self.record_caret();
3422 if let Some(fix) = fix.filter(|_| change.new.end - change.new.start == text.len()) {
3423 self.repair_mark_edges(fix);
3424 }
3425 true
3426 }
3427 Err(e) => {
3428 self.status = Some(format!("edit: {e}"));
3429 false
3430 }
3431 }
3432 }
3433
3434 /// After a structural list edit (a new item, a nest/unnest), renumber the
3435 /// ordered list the caret sits in so its source markers run `1, 2, 3, …`
3436 /// again — a raw splice leaves them stale (`1. 2. 2. 3.`). twig does the
3437 /// renumber as its own edit; fold it into the edit that triggered it so the
3438 /// two undo as one, and only when it actually changed the source (a no-op or
3439 /// a caret outside any ordered list must not coalesce the real edit into the
3440 /// step before it).
3441 fn renumber_here(&mut self) {
3442 self.renumber_at(self.caret);
3443 }
3444
3445 /// [`renumber_here`](Self::renumber_here) aimed somewhere other than the
3446 /// caret — for an edit that leaves the caret one past the item it just wrote,
3447 /// where twig resolves no list to renumber.
3448 fn renumber_at(&mut self, off: usize) {
3449 // The read-only gate — this door reaches twig without the splice.
3450 if self.read_only {
3451 return;
3452 }
3453 let before = self.source.clone();
3454 if self.editor.renumber_ordered_lists(off).is_err() {
3455 return; // not inside an ordered list — nothing to renumber
3456 }
3457 self.refresh();
3458 if self.source != before {
3459 let _ = self.editor.coalesce_last_undo();
3460 self.dirty = self.source != self.clean_source;
3461 self.clamp_caret();
3462 self.record_caret();
3463 }
3464 }
3465
3466 /// Repair the one trap a list edit can spring on itself. An *empty* `-`
3467 /// sub-item written directly beneath a text line reparses that text as a
3468 /// setext heading — `- hello\n - ` is `<h2>hello</h2>`, because a lone `-`
3469 /// is also a setext-H2 underline (twig is right; pandoc agrees). `*` and `+`
3470 /// bullets can't underline anything, so swap the dash for a `*`: the item
3471 /// stays an empty nested bullet, the parent stays prose, and the source
3472 /// round-trips instead of hiding a heading the user never asked for. Folded
3473 /// into the triggering edit's undo step, the way renumbering is.
3474 ///
3475 /// Gated on the collapse having actually happened (the swapped dash was
3476 /// swallowed into a `heading`), so a real setext heading the author wrote —
3477 /// or a `- x` with content, which can't underline anything — is never
3478 /// touched. This has to live in the *edit*, not the renderer: leaving the
3479 /// hazardous bytes on disk and only painting over them would ship a file
3480 /// every other CommonMark tool reads as a heading.
3481 ///
3482 /// This one keeps its own byte scan, and has to: the hazard is precisely
3483 /// that the dash stopped being a list marker, so [`list_marker_on_line`] —
3484 /// which asks twig which lines open an item — reports nothing here. There is
3485 /// no node to ask about. It is also the last Markdown spelling leaf writes on
3486 /// purpose rather than for want of an answer; once twig spells continuations
3487 /// itself, avoiding the trap becomes twig's, and this goes.
3488 ///
3489 /// [`list_marker_on_line`]: Self::list_marker_on_line
3490 fn avoid_setext_collapse(&mut self) {
3491 let caret = self.caret.min(self.source.len());
3492 let line_start = self.source[..caret].rfind('\n').map_or(0, |i| i + 1);
3493 let bytes = self.source.as_bytes();
3494 let mut dash = line_start;
3495 while matches!(bytes.get(dash), Some(b' ' | b'\t')) {
3496 dash += 1;
3497 }
3498 // A dash bullet is the only marker that doubles as a setext underline.
3499 if bytes.get(dash) != Some(&b'-') {
3500 return;
3501 }
3502 // Only an *empty* item is a bare underline; `- x` carries content and
3503 // can't fold the line above into a heading.
3504 let line_end = self.source[dash..]
3505 .find('\n')
3506 .map_or(self.source.len(), |i| dash + i);
3507 if !self.source[dash + 1..line_end].trim().is_empty() {
3508 return;
3509 }
3510 // The tell: that dash was swallowed into a `heading`. A properly nested
3511 // empty item sits under a `list_item`, with no heading in reach. Probe
3512 // the dash byte itself (well inside the heading), not the caret, whose
3513 // end-of-line offset can fall on the half-open span boundary.
3514 let collapsed = self
3515 .editor
3516 .ancestors_at(dash)
3517 .map(|c| c.into_iter().any(|m| m.kind == Kind::Heading))
3518 .unwrap_or(false);
3519 if !collapsed {
3520 return;
3521 }
3522 let caret = self.caret;
3523 if self.splice(dash, dash + 1, "*", EditKind::Other) {
3524 // Same width, so the caret keeps its column; fold into the edit that
3525 // triggered this so Tab stays one undo step.
3526 let _ = self.editor.coalesce_last_undo();
3527 self.caret = caret.min(self.source.len());
3528 self.clamp_caret();
3529 self.record_caret();
3530 }
3531 }
3532
3533 fn snapshot(&self) -> CaretState {
3534 CaretState {
3535 caret: self.caret,
3536 anchor: self.anchor,
3537 }
3538 }
3539
3540 /// Hand twig the current caret and selection as the blob for the live
3541 /// document state. Called before an edit — so the step twig retires records
3542 /// where the caret was, and undo can restore it — and again once the op has
3543 /// placed the caret, so redo restores where the edit left it.
3544 ///
3545 /// This is the whole of leaf's undo-caret bookkeeping now. twig carries the
3546 /// caret through its own history, so coalescing falls out for free (folding
3547 /// two twig steps into one drops the intermediate blob, keeping the run's
3548 /// first) and the parallel stacks that had to march in lockstep — and could
3549 /// silently drift out of it — are gone.
3550 fn record_caret(&mut self) {
3551 let _ = self.editor.set_caret_blob(&self.snapshot().to_blob());
3552 }
3553
3554 /// Toggle an inline mark over the selection (Bold / Italic / Code / …). Keeps
3555 /// the toggled region selected so a second press cleanly reverses it.
3556 pub fn toggle(&mut self, kind: InlineKind) {
3557 // The read-only gate — this door reaches twig without the splice.
3558 if self.read_only {
3559 return;
3560 }
3561 // Ahead of the no-selection branch below: arming a mark for text not yet
3562 // typed is a promise `insert` cannot keep in a format with no delimiters
3563 // to spell it with. Per *kind*, not per format — Markdown spells three
3564 // of the eight marks, djot all eight, HTML seven.
3565 if self.refuse_unsupported(&format!("{kind:?}"), Gesture::ToggleInline(kind)) {
3566 return;
3567 }
3568 let Some((s, e)) = self.selection() else {
3569 // No selection: arm the mark for the next text typed here, the way a
3570 // word processor does. `⌘b`, type, `⌘b` again toggles bold on and off
3571 // in the flow of typing without ever selecting anything — the delta
3572 // is realised onto the freshly typed text by `insert`. A fresh caret
3573 // position starts the delta over from the marks actually in force.
3574 if self.pending_at != Some(self.caret) {
3575 self.pending_marks = InlineMarks::empty();
3576 self.pending_at = Some(self.caret);
3577 }
3578 self.pending_marks.flip(kind);
3579 self.status = None;
3580 return;
3581 };
3582 // Whitespace at the edge of a selection is not part of what was chosen —
3583 // a double-click takes the space after the word with it — and a mark
3584 // cannot close against one anyway: `**word **` is four literal asterisks
3585 // (the mark-edge rule, see `splice`). Mark the words, leave the spaces.
3586 let picked = &self.source[s..e];
3587 let (s, e) = (
3588 s + (picked.len() - picked.trim_start().len()),
3589 e - (picked.len() - picked.trim_end().len()),
3590 );
3591 if s >= e {
3592 self.status = Some(format!("{kind:?}: nothing selected to mark"));
3593 return;
3594 }
3595 // Styling a selection is a one-shot act, not a sticky mode.
3596 self.clear_pending();
3597 self.record_caret();
3598 match self.editor.toggle_inline(s, e, kind) {
3599 Ok(change) => {
3600 self.last_edit_kind = None; // structural edit is its own undo step
3601 self.refresh();
3602 self.anchor = Some(change.new.start);
3603 self.caret = change.new.end;
3604 self.dirty = self.source != self.clean_source;
3605 self.status = None;
3606 self.record_caret();
3607 }
3608 Err(e) => self.status = Some(format!("{kind:?}: {e}")),
3609 }
3610 }
3611
3612 /// Convert the block at the caret to a heading level or paragraph.
3613 pub fn set_block(&mut self, kind: BlockKind) {
3614 // The read-only gate — this door reaches twig without the splice.
3615 if self.read_only {
3616 return;
3617 }
3618 if self.refuse_unsupported(&format!("{kind:?}"), Gesture::SetBlock) {
3619 return;
3620 }
3621 self.record_caret();
3622 // A blank line has no node to convert, and twig opens a block there
3623 // rather than declining — so the caret's own offset is the right thing
3624 // to hand it when `block_offset_for_caret` finds nothing.
3625 let offset = self.block_offset_for_caret().unwrap_or(self.caret);
3626 match self.editor.set_block(offset, kind) {
3627 Ok(change) => {
3628 self.last_edit_kind = None;
3629 self.refresh();
3630 // Opening a block on a blank line writes a marker the caret
3631 // belongs *after*; converting an existing one moves nothing.
3632 self.caret = self.caret.max(change.new.end);
3633 self.clamp_caret();
3634 self.anchor = None;
3635 self.dirty = self.source != self.clean_source;
3636 self.status = None;
3637 self.record_caret();
3638 }
3639 Err(e) => self.status = Some(format!("{kind:?}: {e}")),
3640 }
3641 }
3642
3643 /// Whether `off` is inside a text block (paragraph, heading, code block…).
3644 fn has_block_at(&mut self, off: usize) -> bool {
3645 self.editor.ancestors_at(off).ok().is_some_and(|chain| {
3646 chain
3647 .iter()
3648 .any(|m| !wysiwyg::is_inline_kind(&m.kind) && !is_block_container(&m.kind))
3649 })
3650 }
3651
3652 /// The offset to hand twig's `set_block`: the caret when it is already inside
3653 /// a block, otherwise nudged onto the previous character (a caret at a line
3654 /// end sits at the doc level, outside the block). `None` when the caret is on
3655 /// a blank line — a new paragraph with no block node to convert.
3656 fn block_offset_for_caret(&mut self) -> Option<usize> {
3657 let caret = self.caret.min(self.source.len());
3658 if self.has_block_at(caret) {
3659 return Some(caret);
3660 }
3661 // Nudge to the previous character — but never across a newline: that would
3662 // target the previous block, and a blank line genuinely has no block.
3663 if let Some((i, ch)) = self.source[..caret].char_indices().next_back()
3664 && ch != '\n'
3665 && self.has_block_at(i)
3666 {
3667 return Some(i);
3668 }
3669 None
3670 }
3671
3672 /// The heading level of the text block at the caret, or `None` when that
3673 /// block is not a heading.
3674 pub fn current_heading_level(&mut self) -> Option<u32> {
3675 let caret = self.caret;
3676 self.nodes()
3677 .into_iter()
3678 .filter(|n| n.kind == Kind::Heading)
3679 .find(|n| n.span.start <= caret && caret <= n.span.end)
3680 .and_then(|n| n.level)
3681 }
3682
3683 /// The inline marks in force at the caret (or over the selection) — what a
3684 /// toolbar draws lit, and the block-level [`Doc::current_heading_level`]'s
3685 /// inline counterpart. Cheap enough to call every frame: one twig
3686 /// `ancestors_at` query per caret (two with a selection), each walking root
3687 /// → deepest node at one offset. It never snapshots the tree the way
3688 /// `current_heading_level` does, and the returned set is a `Copy` bitset, so
3689 /// the only allocation is twig's own small ancestor `Vec`.
3690 ///
3691 /// **A selection reports a mark only when the mark covers *all* of it.**
3692 /// That's what every real toolbar means by an active button — Bold lit over
3693 /// a half-bold selection would claim a press turns bold *off*, when
3694 /// [`Doc::toggle`] hands the range to twig and gets the whole thing bolded.
3695 /// Whole-coverage is asked as "is the same mark node standing over both the
3696 /// first and the last character?": inline nodes are contiguous, so one node
3697 /// covering both ends covers every byte between them. Two touching runs
3698 /// (`**a****b**`) are two nodes, and correctly light nothing.
3699 ///
3700 /// At a bare caret a mark is active when the caret stands inside the mark's
3701 /// span — `span.start <= caret < span.end`, delimiters included, which is
3702 /// what makes the boundaries behave. In `a **bold** b` the offsets from the
3703 /// opening `*` (2) through the last byte of the closing `**` (9) are all
3704 /// bold, so the WYSIWYG caret both before `b` and after `d` (the delimiters
3705 /// are hidden, and those offsets are 4 and 8) reports bold — matching where
3706 /// typing would actually land inside the marked run. The offset one past the
3707 /// mark (10) is the text after it and reports nothing, at the end of the
3708 /// buffer exactly as in the middle.
3709 pub fn active_inline_marks(&mut self) -> InlineMarks {
3710 let Some((start, end)) = self.selection() else {
3711 // The marks actually in force at the caret, flipped by any armed
3712 // sticky delta — so `⌘b` at a bare caret lights the Bold button
3713 // immediately, before a single character is typed.
3714 let base: InlineMarks = self
3715 .marks_at(self.caret)
3716 .into_iter()
3717 .map(|(k, _)| k)
3718 .collect();
3719 return base.xor(self.pending_here());
3720 };
3721 // The selection's *last character*, not its exclusive end: `end` is the
3722 // offset one past the selection, which for a selection ending exactly at
3723 // a mark's close is already outside it (`[4,10)` of `a **bold** b` is
3724 // entirely bold, but offset 10 is the space after).
3725 let last = prev_boundary(&self.source, end);
3726 let head = self.marks_at(start);
3727 let tail = self.marks_at(last);
3728 head.into_iter()
3729 .filter(|m| tail.contains(m))
3730 .map(|(k, _)| k)
3731 .collect()
3732 }
3733
3734 /// The inline marks whose span covers `off`, each with the id of the node
3735 /// carrying it — the id is what lets a selection tell one mark node from
3736 /// another of the same kind.
3737 fn marks_at(&mut self, off: usize) -> Vec<(InlineKind, u32)> {
3738 let off = off.min(self.source.len());
3739 self.editor
3740 .ancestors_at(off)
3741 .unwrap_or_default()
3742 .into_iter()
3743 // `span.end` is the offset one *past* the mark, so it isn't in it.
3744 // twig already resolves a boundary to whatever starts there — in
3745 // `**bold** x` offset 8 is the following text, not the strong — but
3746 // when nothing follows, the tie has nobody to break for and the
3747 // chain still ends at the mark. That would make the answer at the
3748 // last offset of the document depend on whether the file happens to
3749 // end in a newline; the rule is `span.start <= off < span.end`, and
3750 // it's the same rule at the end of a buffer as in the middle.
3751 .filter(|m| off < m.span.end)
3752 .filter_map(|m| inline_kind(&m.kind).map(|k| (k, m.node_id)))
3753 .collect()
3754 }
3755
3756 /// Toggle a heading at the caret: if the block is already this heading level,
3757 /// revert it to a paragraph; otherwise convert it to this heading level.
3758 /// This gives the heading commands the same toggle feel as bold/italic/code —
3759 /// re-applying a heading a line already has turns it back into body text.
3760 pub fn toggle_heading(&mut self, level: u32) {
3761 if self.current_heading_level() == Some(level) {
3762 self.set_block(BlockKind::Paragraph);
3763 } else {
3764 self.set_block(BlockKind::Heading(level));
3765 }
3766 }
3767
3768 /// Toggle a block quote around the selection, or around the block at the
3769 /// caret — the toolbar's Quote button.
3770 pub fn toggle_blockquote(&mut self) {
3771 self.toggle_container(BlockContainerKind::BlockQuote);
3772 }
3773
3774 /// Toggle a numbered (`ordered`) or bulleted list over the selection, or
3775 /// over the block at the caret — one op with the kind as a flag, the way
3776 /// `toggle_heading` takes its level, so a frontend needs no twig type to
3777 /// name the two buttons.
3778 ///
3779 /// Pressing the *other* list's button while in a list converts in place
3780 /// rather than nesting, so the pair reads as one three-state control
3781 /// (bulleted / numbered / neither) rather than two independent wrappers.
3782 pub fn toggle_list(&mut self, ordered: bool) {
3783 self.toggle_container(if ordered {
3784 BlockContainerKind::OrderedList
3785 } else {
3786 BlockContainerKind::BulletList
3787 });
3788 }
3789
3790 // ── Task list items ──────────────────────────────────────────────────────
3791 // The checkbox in `- [x] done`. twig owns all three gestures: the box is
3792 // inline content of the item's first paragraph rather than part of its
3793 // marker, so adding or removing one must leave the item's continuation
3794 // indentation alone, and an item inside a quote is found past the quote
3795 // markers. leaf names the gesture and the offset; the spelling is twig's.
3796
3797 /// Whether the list item at the caret carries a checkbox, and which way it
3798 /// faces — `Some(true)` ticked, `Some(false)` empty, `None` for a plain list
3799 /// item or no item at all. What a toolbar reads to light its checkbox button.
3800 pub fn task_checked_at_caret(&mut self) -> Option<bool> {
3801 self.task_checked_at(self.caret)
3802 }
3803
3804 /// [`task_checked_at_caret`](Self::task_checked_at_caret) for an arbitrary
3805 /// offset — what a frontend asks before deciding a click landed on a box.
3806 pub fn task_checked_at(&mut self, offset: usize) -> Option<bool> {
3807 self.innermost_list_item(offset.min(self.source.len()))?
3808 .checked
3809 }
3810
3811 /// Tick or untick the task item at the caret (the checkbox's keyboard half).
3812 /// A no-op with a reported reason when the caret is in no task item — minting
3813 /// a box here is [`toggle_task_item`](Self::toggle_task_item)'s job.
3814 pub fn toggle_task_checked(&mut self) {
3815 self.toggle_task_at(self.caret);
3816 }
3817
3818 /// Tick or untick the task item covering `offset` — what a *click* on a
3819 /// rendered checkbox is. Separate from the caret form because a click carries
3820 /// its own offset and must not first move the caret there: ticking a box
3821 /// three paragraphs away should not take the cursor with it.
3822 pub fn toggle_task_at(&mut self, offset: usize) {
3823 // The read-only gate — this door reaches twig without the splice.
3824 if self.read_only {
3825 return;
3826 }
3827 if self.refuse_unsupported("task", Gesture::ToggleTaskChecked) {
3828 return;
3829 }
3830 let offset = offset.min(self.source.len());
3831 self.record_caret();
3832 match self.editor.toggle_task_checked(offset) {
3833 Ok(_) => self.after_task_edit(),
3834 Err(e) => self.status = Some(format!("task: {e}")),
3835 }
3836 }
3837
3838 /// Give the list item at the caret a checkbox, or take its checkbox away —
3839 /// the gesture that converts between a plain bullet and a task. A new box
3840 /// arrives unticked.
3841 pub fn toggle_task_item(&mut self) {
3842 // The read-only gate — this door reaches twig without the splice.
3843 if self.read_only {
3844 return;
3845 }
3846 if self.refuse_unsupported("task", Gesture::ToggleTaskItem) {
3847 return;
3848 }
3849 let caret = self.caret.min(self.source.len());
3850 self.record_caret();
3851 match self.editor.toggle_task_item(caret) {
3852 Ok(_) => self.after_task_edit(),
3853 Err(e) => self.status = Some(format!("task: {e}")),
3854 }
3855 }
3856
3857 /// Settle after a task gesture. The caret rides its old byte offset and is
3858 /// clamped back in: a box is three or four bytes on the item's first line, so
3859 /// text after it shifts by that much at most, and `clamp_caret` lands it on a
3860 /// real stop either way.
3861 fn after_task_edit(&mut self) {
3862 self.last_edit_kind = None;
3863 self.refresh();
3864 self.anchor = None;
3865 self.dirty = self.source != self.clean_source;
3866 self.status = None;
3867 self.clamp_caret();
3868 self.record_caret();
3869 }
3870
3871 // ── Tables ───────────────────────────────────────────────────────────────
3872 // A table is a grid, and twig edits it as one — add/remove/move a row or
3873 // column, set a column's alignment — re-spelling the whole table in a single
3874 // splice. Every gesture is anchored at the caret's cell. leaf just names the
3875 // gesture and re-reads the result; the whole table's numbering, borders, and
3876 // delimiter are twig's to keep straight.
3877
3878 /// Whether the caret is inside a table — what a frontend asks to enable or
3879 /// disable its table controls.
3880 ///
3881 /// An HTML `<table>` still answers `true`: the caret really is in a table,
3882 /// and the reason the grid controls stay dark there is
3883 /// [`Capabilities::table`], which is a fact about the document's format
3884 /// rather than about the caret. A frontend needs both.
3885 pub fn caret_in_table(&mut self) -> bool {
3886 let caret = self.caret.min(self.source.len());
3887 self.editor
3888 .ancestors_at(caret)
3889 .map(|c| c.into_iter().any(|m| m.kind == Kind::Table))
3890 .unwrap_or(false)
3891 }
3892
3893 /// One grid op, guarded and settled — the shared body of the seven below.
3894 ///
3895 /// The guard is why this exists rather than seven copies of the same three
3896 /// lines, and it is the one guard leaf cannot delegate to twig. The table
3897 /// editor is the gesture family that consults no `Syntax` table (it spells a
3898 /// grid, not a delimiter) and therefore the one twig's `Format::supports`
3899 /// deliberately has no variant for: handed an HTML `<table>` it rebuilds the
3900 /// grid as a *pipe table* and reports success, swapping the element out for
3901 /// `| a | b |` and taking the rest of the document's markup with it. Nothing
3902 /// downstream could tell that from a successful edit — the splice is real,
3903 /// the reparse succeeds, `dirty` is honest — which is what makes it worth
3904 /// stopping at the door rather than detecting after the fact. See
3905 /// [`spells_pipe_tables`].
3906 fn table_op(
3907 &mut self,
3908 what: &str,
3909 op: impl FnOnce(&mut Editor, usize) -> Result<(), twig::Error>,
3910 ) {
3911 if self.refuse_unless(what, spells_pipe_tables(self.format)) {
3912 return;
3913 }
3914 self.record_caret();
3915 let at = self.caret;
3916 let r = op(&mut self.editor, at);
3917 self.apply_table(r, what);
3918 }
3919
3920 /// Insert an empty row below (`below`) or above the caret's row.
3921 pub fn table_insert_row(&mut self, below: bool) {
3922 self.table_op("table row", |e, at| e.table_insert_row(at, below));
3923 }
3924
3925 /// Delete the caret's row (not the header, not the last body row).
3926 pub fn table_delete_row(&mut self) {
3927 self.table_op("table row", |e, at| e.table_delete_row(at));
3928 }
3929
3930 /// Insert an empty column right (`right`) or left of the caret's column.
3931 pub fn table_insert_column(&mut self, right: bool) {
3932 self.table_op("table column", |e, at| e.table_insert_column(at, right));
3933 }
3934
3935 /// Delete the caret's column (unless it is the only one).
3936 pub fn table_delete_column(&mut self) {
3937 self.table_op("table column", |e, at| e.table_delete_column(at));
3938 }
3939
3940 /// Set the caret's column to `alignment`.
3941 pub fn table_set_alignment(&mut self, alignment: Alignment) {
3942 self.table_op("table alignment", |e, at| {
3943 e.table_set_alignment(at, alignment)
3944 });
3945 }
3946
3947 /// Move the caret's row one place down (`down`) or up, within the body rows.
3948 pub fn table_move_row(&mut self, down: bool) {
3949 self.table_op("table row", |e, at| e.table_move_row(at, down));
3950 }
3951
3952 /// Move the caret's column one place right (`right`) or left.
3953 pub fn table_move_column(&mut self, right: bool) {
3954 self.table_op("table column", |e, at| e.table_move_column(at, right));
3955 }
3956
3957 /// Settle the caret and document flags after a table op (or report its
3958 /// error). twig re-spells the whole table, so the caret rides its old byte
3959 /// offset and is clamped back into the rebuilt bytes — near enough to where
3960 /// it was, since the op preserves the cells' content and order around it.
3961 fn apply_table(&mut self, result: Result<(), twig::Error>, what: &str) {
3962 match result {
3963 Ok(()) => {
3964 self.last_edit_kind = None;
3965 self.refresh();
3966 self.anchor = None;
3967 self.clamp_caret();
3968 self.dirty = self.source != self.clean_source;
3969 self.status = None;
3970 self.record_caret();
3971 }
3972 Err(e) => self.status = Some(format!("{what}: {e}")),
3973 }
3974 }
3975
3976 /// One `toggle_block_container` over the block-level target.
3977 ///
3978 /// leaf says *where*; twig decides everything else — which blocks the range
3979 /// covers, whether that means wrapping, unwrapping, nesting or converting,
3980 /// and how this document's format spells the prefix. The rule that a
3981 /// container only comes off when the range covers every block it holds is
3982 /// what the re-anchoring below is built around.
3983 fn toggle_container(&mut self, kind: BlockContainerKind) {
3984 // The read-only gate — this door reaches twig without the splice.
3985 if self.read_only {
3986 return;
3987 }
3988 if self.refuse_unsupported(&format!("{kind:?}"), Gesture::ToggleBlockContainer(kind)) {
3989 return;
3990 }
3991 let selected = self.selection();
3992 // A blank line holds no block, and twig opens an *empty* container on one
3993 // — since 3.2.0; it used to decline the range with `NotFound`, which is
3994 // why this used to lend it a scratch paragraph to wrap. Worth knowing
3995 // here because the line-for-line caret mapping below cannot describe it:
3996 // opening one under a paragraph writes the blank line the format needs
3997 // above the marker too, so the rewritten region has a line the old one
3998 // didn't, and "the same line, the same distance from its end" lands on
3999 // that new blank instead of in the container.
4000 let opened_empty = selected.is_none() && self.block_offset_for_caret().is_none();
4001 // Without a selection the target is the caret's own block, resolved the
4002 // way `set_block` resolves it — a caret at a line end sits at the doc
4003 // level and has to be nudged back onto the block it looks like it's in.
4004 // An empty range is enough: twig widens to the whole lines it touches.
4005 let (start, end) = match selected {
4006 Some(range) => range,
4007 None => {
4008 let off = self.block_offset_for_caret().unwrap_or(self.caret);
4009 (off, off)
4010 }
4011 };
4012 self.record_caret();
4013 match self.editor.toggle_block_container(start, end, kind) {
4014 Ok(change) => {
4015 // Read the caret's place out of the *pre-edit* source, before
4016 // `refresh` swaps that source out from under it.
4017 let place = (selected.is_none() && !opened_empty)
4018 .then(|| self.caret_line_tail(&change.old));
4019 self.last_edit_kind = None; // structural edit is its own undo step
4020 self.refresh();
4021 match place {
4022 // Both land the caret at the far end of what twig wrote, and
4023 // differ only in what they leave selected.
4024 //
4025 // From a selection: select what the container now holds, the
4026 // way `toggle` keeps its marked region selected — and for a
4027 // stronger reason than symmetry: a container comes *off* only
4028 // a range covering every block it holds, so a selection left
4029 // on its old bytes (now short by a prefix per line) would nest
4030 // on the second press instead of reversing the first.
4031 //
4032 // From a blank line: nothing to select, and the end of the
4033 // region is exactly past the bare `> ` / `- ` twig wrote —
4034 // the caret standing inside the container that was asked for.
4035 None => {
4036 self.anchor = (!opened_empty).then_some(change.new.start);
4037 self.caret = change.new.end;
4038 }
4039 Some(place) => {
4040 self.anchor = None;
4041 self.caret = self.line_tail_offset(&change.new, place);
4042 }
4043 }
4044 self.dirty = self.source != self.clean_source;
4045 self.status = None;
4046 self.clamp_caret();
4047 self.record_caret();
4048 }
4049 Err(e) => self.status = Some(format!("{kind:?}: {e}")),
4050 }
4051 }
4052
4053 /// The caret's place inside the region a container toggle is rewriting, in
4054 /// the only terms the rewrite preserves: which of the region's lines it sits
4055 /// on, and how many bytes of that line lie ahead of it.
4056 ///
4057 /// A container's markup goes in at column 0 and never touches what follows
4058 /// on the line, so that pair survives the edit exactly where a byte offset
4059 /// does not — a caret left on its old offset slides back by one prefix per
4060 /// line above it, which on a hard-wrapped paragraph parks it *inside* the
4061 /// `> ` it just asked for.
4062 fn caret_line_tail(&self, old: &std::ops::Range<usize>) -> (usize, usize) {
4063 let caret = self.caret.clamp(old.start, old.end);
4064 let line = self.source[old.start..caret].matches('\n').count();
4065 let end = self.source[caret..old.end]
4066 .find('\n')
4067 .map_or(old.end, |i| caret + i);
4068 (line, end - caret)
4069 }
4070
4071 /// [`caret_line_tail`](Self::caret_line_tail) undone against the rewritten
4072 /// region: the offset `tail` bytes back from the end of the region's `line`.
4073 ///
4074 /// Both walks are clamped rather than trusted, because the one op that does
4075 /// *not* keep a region's lines one-to-one is stripping a list — twig blows
4076 /// the items back apart with blank lines between them — and a caret landing
4077 /// on the nearest line of the right item beats one landing out of the region
4078 /// entirely.
4079 fn line_tail_offset(
4080 &self,
4081 new: &std::ops::Range<usize>,
4082 (line, tail): (usize, usize),
4083 ) -> usize {
4084 let region = &self.source[new.start.min(self.source.len())..new.end.min(self.source.len())];
4085 let mut start = 0;
4086 for _ in 0..line {
4087 match region[start..].find('\n') {
4088 Some(i) => start += i + 1,
4089 None => break,
4090 }
4091 }
4092 let end = region[start..]
4093 .find('\n')
4094 .map_or(region.len(), |i| start + i);
4095 new.start + end.saturating_sub(tail).max(start)
4096 }
4097
4098 /// Link the selection to `destination` — the toolbar's Link button. With no
4099 /// selection it acts at the caret, which re-points a link the caret is
4100 /// already standing in (twig replaces an existing link's destination and
4101 /// keeps its text) and otherwise spells a link that has no text of its own:
4102 /// an autolink (`<https://x.dev>`) where the destination is one, and
4103 /// `[destination](destination)` where it isn't.
4104 ///
4105 /// `destination` reaches twig raw. Escaping it is format knowledge and the
4106 /// two formats genuinely disagree — Markdown ends a destination at the first
4107 /// space and moves it into `<…>`, djot reads that `<…>` as part of the URL
4108 /// itself — so the side holding the document is the side that gets to spell
4109 /// it. A destination twig can't carry at all (one with a newline) comes back
4110 /// as an error rather than a quietly rewritten URL.
4111 pub fn insert_link(&mut self, destination: &str) {
4112 if self.read_only || self.refuse_unsupported("link", Gesture::InsertLink) {
4113 return;
4114 }
4115 let (start, end) = self.selection().unwrap_or((self.caret, self.caret));
4116 self.record_caret();
4117 match self.editor.insert_link(start, end, destination) {
4118 Ok(change) => {
4119 self.last_edit_kind = None;
4120 self.refresh();
4121 match self.link_text_span(change.new.start) {
4122 // A link with text of its own: select it, so typing replaces
4123 // a `[dest](dest)`'s stand-in label and a second press
4124 // re-points what the first one linked.
4125 Some(text) => {
4126 self.anchor = (text.start != text.end).then_some(text.start);
4127 self.caret = text.end;
4128 }
4129 // An autolink is finished the moment it's written — its text
4130 // *is* the URL. Leaving it selected would aim the next press
4131 // at the one shape twig still wraps instead of re-points.
4132 None => {
4133 self.anchor = None;
4134 self.caret = change.new.end;
4135 }
4136 }
4137 self.dirty = self.source != self.clean_source;
4138 self.status = None;
4139 self.clamp_caret();
4140 self.record_caret();
4141 }
4142 Err(e) => self.status = Some(format!("link: {e}")),
4143 }
4144 }
4145
4146 /// Insert a block-level image at the caret: ``. Any
4147 /// selection becomes the alt text (so "select a caption, insert image" labels
4148 /// it); with no selection, `alt` is used — empty for none. The caret lands
4149 /// just past the inserted image.
4150 ///
4151 /// Both halves go through twig (`insert_literal` for the alt text,
4152 /// `insert_image` for the image), so neither is spelled here. That used to be a
4153 /// `format!`, and it was wrong the first time an app inserted a real filename:
4154 /// Markdown ends a destination at the first space, so `` is
4155 /// not an image at all — and the fix is per-format, since moving into the
4156 /// `<…>` form is exactly wrong for Djot, where `<…>` becomes the URL itself.
4157 pub fn insert_image(&mut self, destination: &str, alt: &str) {
4158 if self.read_only || self.refuse_unsupported("image", Gesture::InsertImage) {
4159 return;
4160 }
4161 let (start, end) = self.selection().unwrap_or((self.caret, self.caret));
4162 self.record_caret();
4163 // With no selection and an explicit `alt`, the alt text has to exist in the
4164 // document before it can be the image's — and it is raw caller input, so
4165 // it goes in through `insert_literal`, which escapes it for the format
4166 // rather than letting a `]` in someone's caption close the image early.
4167 let (start, end) = if start == end && !alt.is_empty() {
4168 match self.editor.insert_literal(start, alt) {
4169 Ok(change) => (change.new.start, change.new.end),
4170 Err(e) => {
4171 self.status = Some(format!("image: {e}"));
4172 return;
4173 }
4174 }
4175 } else {
4176 (start, end)
4177 };
4178 match self.editor.insert_image(start, end, destination) {
4179 Ok(change) => {
4180 self.last_edit_kind = None;
4181 self.refresh();
4182 // Just past the image, nothing selected — where a caret belongs
4183 // after inserting one.
4184 self.anchor = None;
4185 self.caret = change.new.end;
4186 self.dirty = self.source != self.clean_source;
4187 self.status = None;
4188 self.clamp_caret();
4189 self.record_caret();
4190 }
4191 Err(e) => self.status = Some(format!("image: {e}")),
4192 }
4193 }
4194
4195 /// Insert a block-level image, video, or audio at the caret. The image case
4196 /// is [`insert_image`](Self::insert_image); video and audio are spelled as
4197 /// HTML elements, which is the only spelling Markdown and Djot have for them:
4198 ///
4199 /// ```text
4200 /// <video src="clip.mp4" controls>alt</video>
4201 /// <audio src="take.mp3" controls>alt</audio>
4202 /// ```
4203 ///
4204 /// HTML rather than a `::video{…}` directive deliberately. A directive means
4205 /// something only to an app that knows the vocabulary, so the document would
4206 /// read as literal punctuation everywhere else; `<video>` is what every other
4207 /// renderer already understands, and what leaf's own reader picks back up
4208 /// through `html_elements` promotion (see [`parse_extensions`]).
4209 ///
4210 /// The one-line spelling needs twig ≥ 2.5.1, which widened CommonMark's
4211 /// HTML-block tag list to cover `<video>`/`<audio>`/`<picture>` under
4212 /// `html_elements`. Before that only the multi-line form parsed as a block at
4213 /// all, and this wrote three lines to work around it.
4214 ///
4215 /// `controls` is always written: a player with no transport is a still frame
4216 /// the reader can't do anything with. Any selection becomes the element's
4217 /// fallback text, exactly as it becomes an image's alt.
4218 ///
4219 /// The same verbatim-insertion caveat as [`insert_image`](Self::insert_image)
4220 /// applies, and bites harder here: a `"` in `destination` closes the
4221 /// attribute. A frontend taking these from a file picker is fine; one taking
4222 /// them from free text should keep them tame.
4223 ///
4224 /// [`MediaInfo`]: crate::MediaInfo
4225 pub fn insert_media(&mut self, kind: MediaKind, destination: &str, alt: &str) {
4226 if kind == MediaKind::Image {
4227 return self.insert_image(destination, alt);
4228 }
4229 // Gated on the *image* gesture, not on one of its own — there isn't one,
4230 // since the bytes below are spelled here rather than by twig, and an HTML
4231 // document would in fact parse them. The button is one control with three
4232 // kinds behind it, and two of them working in a format where the third
4233 // cannot is a worse surface than three that agree — especially as
4234 // `insert_image` is the kind anyone reaches for first.
4235 if self.refuse_unsupported("media", Gesture::InsertImage) {
4236 return;
4237 }
4238 let (start, end) = self.selection().unwrap_or((self.caret, self.caret));
4239 let alt_text = self
4240 .selected_text()
4241 .map(str::to_string)
4242 .unwrap_or_else(|| alt.to_string());
4243 let tag = match kind {
4244 MediaKind::Audio => "audio",
4245 _ => "video",
4246 };
4247 let markup = format!("<{tag} src=\"{destination}\" controls>{alt_text}</{tag}>");
4248 self.edit(start, end, &markup);
4249 }
4250
4251 /// Insert a thematic break at the caret — the toolbar's Horizontal Rule
4252 /// button. Spelling and placement are both twig's; leaf used to write `---`
4253 /// itself, which was the Markdown spelling in a djot document too.
4254 ///
4255 /// A rule is a block, so `insert_thematic_break` alone has nowhere to put one
4256 /// mid-paragraph and lands it after the caret's whole block. To get a rule
4257 /// *at* the caret — the paragraph parted in two around it, which is what a
4258 /// rule button is understood to do — the paragraph is first divided with
4259 /// `split_block` and the rule then aimed at the **first** half. Aiming it at
4260 /// the offset `split_block` returns puts the rule after the *second* half
4261 /// instead, which is a rule in the right document and the wrong place.
4262 ///
4263 /// Only a plain paragraph is split. Everywhere else the rule simply lands
4264 /// after the block, which is both twig's own answer and the better one:
4265 /// splitting a fenced code block would leave two fences with a rule between
4266 /// them, and splitting a list item would mint an item nobody asked for on the
4267 /// way to a rule that lands after the list regardless. A table and a setext
4268 /// heading refuse the split outright, so they take the same path by
4269 /// themselves.
4270 pub fn insert_thematic_break(&mut self) {
4271 if self.read_only || self.refuse_unsupported("thematic break", Gesture::InsertThematicBreak)
4272 {
4273 return;
4274 }
4275 self.caret = self.skip_trailing_close_delims(self.caret);
4276 // A selection is replaced by the rule, so collapse it first and let the
4277 // split-and-rule below run from the caret it leaves behind.
4278 if let Some((s, e)) = self.selection() {
4279 self.splice(s, e, "", EditKind::Other);
4280 }
4281 self.anchor = None;
4282 self.record_caret();
4283 let at = self.caret;
4284 if self.caret_in_bare_paragraph() {
4285 // A failure here is not fatal: the rule still lands after the block,
4286 // which is exactly what this call was trying to improve on.
4287 let _ = self.editor.split_block(at);
4288 }
4289 match self.editor.insert_thematic_break(at) {
4290 Ok(change) => {
4291 self.last_edit_kind = None;
4292 self.refresh();
4293 self.anchor = None;
4294 self.caret = change.new.end;
4295 self.dirty = self.source != self.clean_source;
4296 self.status = None;
4297 self.clamp_caret();
4298 self.record_caret();
4299 }
4300 Err(e) => self.status = Some(format!("thematic break: {e}")),
4301 }
4302 }
4303
4304 /// Whether the caret sits in a paragraph and nothing else — no list item, no
4305 /// quote, no fence, no table. The one shape where parting the block around
4306 /// the caret is unambiguously what a rule button means; see
4307 /// [`insert_thematic_break`](Self::insert_thematic_break) for why every other
4308 /// container is left to take the rule after itself.
4309 fn caret_in_bare_paragraph(&mut self) -> bool {
4310 let caret = self.caret.min(self.source.len());
4311 let Ok(chain) = self.editor.ancestors_at(caret) else {
4312 return false;
4313 };
4314 let mut in_para = false;
4315 for m in chain {
4316 match m.kind {
4317 Kind::Para => in_para = true,
4318 Kind::ListItem
4319 | Kind::TaskListItem
4320 | Kind::BlockQuote
4321 | Kind::CodeBlock
4322 | Kind::Table => return false,
4323 _ => {}
4324 }
4325 }
4326 in_para
4327 }
4328
4329 /// The destination of the link under the caret — what a Link prompt shows so
4330 /// ⌘K on an existing link edits its URL instead of asking for it again.
4331 /// `None` when the caret stands in no link.
4332 ///
4333 /// An autolink carries no separate destination: its text *is* the URL, so
4334 /// that's what comes back for one.
4335 pub fn link_destination_at_caret(&mut self) -> Option<String> {
4336 self.link_destination_at(self.caret)
4337 }
4338
4339 /// The destination of the link at `off`.
4340 /// [`link_destination_at_caret`](Self::link_destination_at_caret) for a place
4341 /// the caret isn't.
4342 ///
4343 /// The offset form exists for the same reason
4344 /// [`footnote_at`](Self::footnote_at)'s does: a frontend drawing a *piece* of
4345 /// the document somewhere else — a footnote's text in a popover, say — has
4346 /// rows and runs but no caret in them, and still needs to know which of those
4347 /// runs a reader can follow.
4348 pub fn link_destination_at(&mut self, off: usize) -> Option<String> {
4349 self.nodes()
4350 .into_iter()
4351 .filter(|n| matches!(n.kind.as_str(), "link" | "url" | "email"))
4352 .filter(|n| n.span.start <= off && off < n.span.end)
4353 .max_by_key(|n| n.span.start)
4354 .and_then(|n| n.destination.or(n.text))
4355 }
4356
4357 /// Where the locator `id` lands in this document — the `#v2` half of a
4358 /// `chapter.dj#v2`, resolved to the block it names. `None` when nothing here
4359 /// answers to it.
4360 ///
4361 /// The other end of a link, and the reason this exists: without it a
4362 /// destination has only file granularity, so following a citation into a
4363 /// chapter drops the reader at the top of it to hunt for the verse. Which is
4364 /// also why it is a *document* query rather than a caret one — the document
4365 /// being asked is usually not the one the reader is in.
4366 ///
4367 /// Three readings, tried in order, because the same `#some-heading` is
4368 /// written three ways across the formats leaf opens:
4369 ///
4370 /// 1. **A declared id**, exactly as written: djot's `{#v1}` on a block, and
4371 /// the auto-ids djot mints for its headings. The only exact answer, so it
4372 /// goes first — a document that says `{#v1}` has settled the question.
4373 /// 2. **A declared id, slugged.** djot spells a heading's auto-id
4374 /// `Some-Heading-Here`; nearly every tool that *writes* a link to one
4375 /// spells it `#some-heading-here`. Comparing slugs is what lets a link
4376 /// authored anywhere land on a djot heading.
4377 /// 3. **A heading's text, slugged.** Markdown has no ids at all — twig mints
4378 /// none and `{#custom}` is literal text in a Markdown heading — so for
4379 /// the format most vaults are written in, the heading's own words are the
4380 /// only thing a fragment can name. This is the rule every Markdown
4381 /// renderer already follows, which is what makes `#a-heading` mean in
4382 /// diaryx what it means on the web.
4383 ///
4384 /// Ties go to the earliest match, then to the widest: a duplicated id is the
4385 /// document's mistake and the first one is the answer every anchor
4386 /// implementation gives, while preferring the wider span picks the section
4387 /// over the heading that opens it — more for a peek to show, same place to
4388 /// land.
4389 pub fn locate(&mut self, id: &str) -> Option<Landing> {
4390 let id = id.trim();
4391 if id.is_empty() {
4392 return None;
4393 }
4394 let nodes = self.nodes();
4395
4396 // Earliest wins, then widest. `Reverse` on the end because `min_by_key`
4397 // is picking, among nodes that start together, the one that ends last.
4398 let pick = |matches: &mut dyn Iterator<Item = &FlatNode>| {
4399 matches
4400 .min_by_key(|n| (n.span.start, std::cmp::Reverse(n.span.end)))
4401 .map(|n| Landing {
4402 start: n.span.start,
4403 end: n.span.end,
4404 })
4405 };
4406
4407 if let Some(landing) = pick(&mut nodes.iter().filter(|n| declared_id(n) == Some(id))) {
4408 return Some(landing);
4409 }
4410 let want = slug(id);
4411 if want.is_empty() {
4412 return None;
4413 }
4414 if let Some(landing) = pick(
4415 &mut nodes
4416 .iter()
4417 .filter(|n| declared_id(n).map(slug).as_deref() == Some(&*want)),
4418 ) {
4419 return Some(landing);
4420 }
4421
4422 // A heading by its words. Its span is one line, so the end comes from
4423 // where the *section* it opens gives out — the next heading that is not
4424 // under it, or the end of the document. A Markdown heading has no
4425 // section node to ask (twig only builds those for djot), and a peek that
4426 // showed the heading alone would answer "what does that say" with the
4427 // title of the thing it says.
4428 let heading = nodes
4429 .iter()
4430 .filter(|n| n.kind == Kind::Heading)
4431 .filter(|n| {
4432 n.content_span
4433 .clone()
4434 .and_then(|s| self.source.get(s))
4435 .is_some_and(|text| slug(text) == want)
4436 })
4437 .min_by_key(|n| n.span.start)?;
4438 let level = heading.level.unwrap_or(u32::MAX);
4439 let end = nodes
4440 .iter()
4441 .filter(|n| n.kind == Kind::Heading)
4442 .filter(|n| n.span.start > heading.span.start)
4443 .filter(|n| n.level.unwrap_or(u32::MAX) <= level)
4444 .map(|n| n.span.start)
4445 .min()
4446 .unwrap_or(self.source.len());
4447 Some(Landing {
4448 start: heading.span.start,
4449 end,
4450 })
4451 }
4452
4453 /// Write a footnote at the caret — the toolbar's Footnote button, and the
4454 /// one gesture in the footnote story that *authors* rather than follows.
4455 ///
4456 /// Both halves go in as one twig edit: the `[^1]` where the caret is, and
4457 /// the `[^1]:` definition at the end of the document. Half a footnote is not
4458 /// a footnote — a bare reference with nothing defining it renders as literal
4459 /// brackets — so a single button that wrote only the reference would leave
4460 /// the author to hand-spell the other half in a document that had just
4461 /// stopped showing them what the first half meant. One edit also means one
4462 /// undo takes both back.
4463 ///
4464 /// The definition's body is left empty and **the caret lands in it**, which
4465 /// is the whole point of pressing the button: nobody wants a reference to a
4466 /// note they have not written yet. Getting back to where they were writing
4467 /// is [`footnote_definition_at_caret`](Self::footnote_definition_at_caret) —
4468 /// the same return leg a reader following a reference already uses, so the
4469 /// author is left standing on the near end of a round trip that works.
4470 ///
4471 /// A selection collapses to its *end* rather than being replaced: a
4472 /// reference annotates the words before it, so "select the claim, add a
4473 /// footnote" should mark that claim, not consume it.
4474 pub fn insert_footnote(&mut self) {
4475 if self.read_only || self.refuse_unsupported("footnote", Gesture::InsertFootnote) {
4476 return;
4477 }
4478 let at = self.selection().map_or(self.caret, |(_, end)| end);
4479 self.anchor = None;
4480 self.caret = at;
4481 self.record_caret();
4482 let label = self.next_footnote_label();
4483 match self.editor.insert_footnote(at, &label) {
4484 Ok(change) => {
4485 self.last_edit_kind = None;
4486 self.refresh();
4487 self.anchor = None;
4488 // `change.new` runs from the reference to the end of the
4489 // document, so its start is the `[^1]` just written and
4490 // `footnote_at` resolves it to the note the same way a reader's
4491 // tap does — and to the note's *body*, which is already a caret
4492 // stop even when it is empty (the `[^1]:` marker draws as `[1] `
4493 // and has none), so this needs no snap on top. The fallback is
4494 // the reference's own offset: a format that spelled the pair some
4495 // way leaf can't read back should still leave the caret on the
4496 // edit rather than at the far end of a document it just grew.
4497 self.caret = self
4498 .footnote_at(change.new.start)
4499 .and_then(|note| note.offset)
4500 .unwrap_or(change.new.start);
4501 self.dirty = self.source != self.clean_source;
4502 self.status = None;
4503 self.clamp_caret();
4504 self.record_caret();
4505 }
4506 Err(e) => self.status = Some(format!("footnote: {e}")),
4507 }
4508 }
4509
4510 /// The label to give a footnote the author has not named: the lowest counting
4511 /// number no footnote in the document is already wearing.
4512 ///
4513 /// twig takes the label rather than minting one, because it holds no opinion
4514 /// about what a document's footnotes should be called — and it is right not
4515 /// to. Numbering them is what every author of a numbered note expects, and
4516 /// re-using a taken number would silently point the new reference at somebody
4517 /// else's note (twig reuses an existing definition rather than appending a
4518 /// second one, which is the right rule for citing a note twice on purpose and
4519 /// exactly the wrong accident to have by default).
4520 ///
4521 /// *References* are counted alongside definitions, not just definitions: a
4522 /// document carrying a dangling `[^2]` has a 2 that means something to
4523 /// whoever wrote it, and minting a definition for it here would answer a
4524 /// question nobody asked. Non-numeric labels (`[^why]`) are left out of the
4525 /// count entirely — they take no number, so they block none.
4526 fn next_footnote_label(&mut self) -> String {
4527 let mut taken: Vec<u32> = wysiwyg::footnote_definitions(&mut self.editor)
4528 .into_iter()
4529 .filter_map(|note| wysiwyg::footnote_label(&self.source, note.span.start))
4530 .filter_map(|label| label.parse().ok())
4531 .collect();
4532 taken.extend(
4533 self.nodes()
4534 .into_iter()
4535 .filter(|n| n.kind == Kind::FootnoteReference)
4536 .filter_map(|n| wysiwyg::footnote_reference_label(&self.source, n.span))
4537 .filter_map(|label| label.parse::<u32>().ok()),
4538 );
4539 (1..).find(|n| !taken.contains(n)).unwrap_or(1).to_string()
4540 }
4541
4542 /// The footnote reference under the caret, resolved to the note it names.
4543 /// [`footnote_at`](Self::footnote_at) at the caret's offset.
4544 pub fn footnote_at_caret(&mut self) -> Option<FootnoteRef> {
4545 self.footnote_at(self.caret)
4546 }
4547
4548 /// The footnote reference at `off`, resolved to the note it names — what a
4549 /// frontend shows when a reader activates a `[^1]`.
4550 ///
4551 /// A reference is not a link node, so
4552 /// [`link_destination_at_caret`](Self::link_destination_at_caret) does not
4553 /// (and should not) answer for one: a link names a destination to leave for,
4554 /// a reference names a note that is already in this document. Following one
4555 /// is a move within the page, which is why this hands back an `offset`
4556 /// rather than something to open.
4557 ///
4558 /// Offset-based rather than caret-only because the gesture that wants this
4559 /// most is the one that must not move the caret: a pointer hovering a `[1]`
4560 /// asks what note it names without disturbing where the reader was typing.
4561 /// The caret is just the offset a click already placed —
4562 /// [`footnote_at_caret`](Self::footnote_at_caret) passes it.
4563 ///
4564 /// `None` when `off` stands in no reference. A reference whose note the
4565 /// document never defines is *not* `None` — it answers with the label it
4566 /// looked for and no text, which is what lets a frontend say so instead of
4567 /// silently doing nothing.
4568 pub fn footnote_at(&mut self, off: usize) -> Option<FootnoteRef> {
4569 // Innermost-wins by latest start, the rule its link sibling uses.
4570 let span = self
4571 .nodes()
4572 .into_iter()
4573 .filter(|n| n.kind == Kind::FootnoteReference)
4574 .filter(|n| n.span.start <= off && off < n.span.end)
4575 .max_by_key(|n| n.span.start)?
4576 .span;
4577 let label = wysiwyg::footnote_reference_label(&self.source, span)?.to_string();
4578
4579 // The note itself. Definitions are roots beside `doc` rather than
4580 // children of it, so they're asked for directly — see
4581 // `wysiwyg::footnote_definitions`.
4582 let note = wysiwyg::footnote_definitions(&mut self.editor)
4583 .into_iter()
4584 .find(|m| wysiwyg::footnote_label(&self.source, m.span.start) == Some(&label));
4585 let Some(note) = note else {
4586 return Some(FootnoteRef {
4587 label,
4588 text: None,
4589 offset: None,
4590 end: None,
4591 });
4592 };
4593 let body = wysiwyg::footnote_body_span(&self.source, note.span.clone());
4594 Some(FootnoteRef {
4595 label,
4596 text: body
4597 .clone()
4598 .and_then(|b| self.source.get(b))
4599 .map(str::to_string),
4600 // The body's start, not the definition's — see `FootnoteRef::offset`.
4601 offset: body.clone().map(|b| b.start),
4602 end: body.map(|b| b.end),
4603 })
4604 }
4605
4606 /// The footnote *definition* the caret stands in, and where the reference
4607 /// that names it is. [`footnote_definition_at`](Self::footnote_definition_at)
4608 /// at the caret's offset.
4609 pub fn footnote_definition_at_caret(&mut self) -> Option<FootnoteDef> {
4610 self.footnote_definition_at(self.caret)
4611 }
4612
4613 /// The footnote definition spanning `off`, and where the reference that
4614 /// names it is — the return leg of [`footnote_at`](Self::footnote_at).
4615 ///
4616 /// The mirror image, deliberately: the same gesture that takes a reader from
4617 /// `[1]` down to the note takes them from the note back up to `[1]`, so
4618 /// following a footnote is a round trip rather than a fall. It needs no
4619 /// memory of how the reader arrived — the document says where the reference
4620 /// is — which is what makes it work for a reader who scrolled to the notes
4621 /// themselves, and what keeps it right after an edit moves either end.
4622 ///
4623 /// `None` when `off` stands in no definition. A definition nothing cites is
4624 /// *not* `None`, for [`FootnoteRef`]'s reason in reverse: it answers with
4625 /// its label and no offset, so a frontend can say "nothing refers to this"
4626 /// rather than offer a jump that goes nowhere.
4627 pub fn footnote_definition_at(&mut self, off: usize) -> Option<FootnoteDef> {
4628 // Definitions are roots beside `doc`, so `nodes()` — which walks the
4629 // document body — never reports one. They're asked for directly, the way
4630 // `footnote_at` asks for the note it resolves to.
4631 //
4632 // Closed at the end, unlike the half-open test its neighbours use. A
4633 // definition's span stops at its last content byte — the newline ending
4634 // the line is outside it — so `span.end` is the caret stop at the end of
4635 // the note's own row, not the first byte of anything after. Excluding it
4636 // meant the one caret an author is guaranteed to have, the one left
4637 // sitting at the end of the note they just typed, was in no definition at
4638 // all: writing a note and then asking to go back to its reference
4639 // answered nothing. Two definitions in a row still can't both match —
4640 // there is a blank line between them — and `max_by_key` decides anyway.
4641 let note = wysiwyg::footnote_definitions(&mut self.editor)
4642 .into_iter()
4643 .filter(|m| m.span.start <= off && off <= m.span.end)
4644 .max_by_key(|m| m.span.start)?;
4645 let label = wysiwyg::footnote_label(&self.source, note.span.start)?.to_string();
4646
4647 // The earliest reference carrying this label. `min` rather than a `find`,
4648 // because `nodes()` reports a flattened walk whose order is twig's
4649 // business, not document order. Bound first: the walk needs `&mut self`
4650 // and reading the labels back out needs `&self.source`.
4651 let nodes = self.nodes();
4652 let offset = nodes
4653 .into_iter()
4654 .filter(|n| n.kind == Kind::FootnoteReference)
4655 .filter(|n| {
4656 wysiwyg::footnote_reference_label(&self.source, n.span.clone()) == Some(&*label)
4657 })
4658 // Past the `[^`, onto the label — see `FootnoteDef::offset`.
4659 .map(|n| n.span.start + 2)
4660 .min();
4661 Some(FootnoteDef { label, offset })
4662 }
4663
4664 /// The destination of the image under the caret — what an image prompt shows
4665 /// so editing an existing image starts from its current URL instead of blank,
4666 /// the image analogue of [`link_destination_at_caret`](Self::link_destination_at_caret).
4667 /// `None` when the caret stands in no image. A caret resting just after a
4668 /// block image (its trailing stop) is still "in" it — the half-open span test
4669 /// excludes that offset, which is the intended precision: past the image is
4670 /// past it.
4671 pub fn image_destination_at_caret(&mut self) -> Option<String> {
4672 let off = self.caret;
4673 self.nodes()
4674 .into_iter()
4675 .filter(|n| n.kind == Kind::Image)
4676 .filter(|n| n.span.start <= off && off < n.span.end)
4677 .max_by_key(|n| n.span.start)
4678 .and_then(|n| n.destination)
4679 }
4680
4681 /// The language of the fenced code block the caret stands in — what a
4682 /// language prompt shows so editing it starts from the current value rather
4683 /// than blank. `None` when the caret is in no code block, or in one whose
4684 /// fence carries no language (or an indented block, which has no fence).
4685 pub fn code_language_at_caret(&mut self) -> Option<String> {
4686 let start = self.code_block_start_at_caret()?;
4687 wysiwyg::code_language(&self.source, start)
4688 }
4689
4690 /// Whether the caret stands in a fenced code block — the one a language
4691 /// prompt could edit. A frontend gates its "set language" affordance on this
4692 /// (an indented block, which can't carry a language, reports `false`).
4693 pub fn caret_in_fenced_code(&mut self) -> bool {
4694 self.code_block_start_at_caret()
4695 .is_some_and(|start| wysiwyg::code_info_span(&self.source, start).is_some())
4696 }
4697
4698 /// Set (or clear, with `""`) the language of the fenced code block the caret
4699 /// is in — the prompt's confirm. A no-op when the caret is in no fenced
4700 /// block, and a reported error for a language the format's fence cannot
4701 /// carry.
4702 ///
4703 /// twig rewrites the info string, so the fence's own width — measured
4704 /// against a body neither side touches — is kept, and a language holding a
4705 /// space, a line end or the fence character is refused rather than written
4706 /// out to reparse as something else. Leaf used to splice over the info span
4707 /// itself and `trim()` the input, which handled the one bad case it had
4708 /// thought of.
4709 pub fn set_code_language(&mut self, lang: &str) {
4710 // The read-only gate — this door reaches twig without the splice.
4711 if self.read_only {
4712 return;
4713 }
4714 if self.refuse_unsupported("code language", Gesture::SetCodeLanguage) {
4715 return;
4716 }
4717 if self.code_block_start_at_caret().is_none() {
4718 return;
4719 }
4720 let lang = lang.trim();
4721 // `None` clears the info string; `Some("")` asks for an empty one. Both
4722 // write a bare fence, and the prompt's empty value means "clear".
4723 let want = (!lang.is_empty()).then_some(lang);
4724 self.record_caret();
4725 match self.editor.set_code_language(self.caret, want) {
4726 Ok(_) => {
4727 self.last_edit_kind = None;
4728 self.refresh();
4729 self.anchor = None;
4730 self.dirty = self.source != self.clean_source;
4731 self.status = None;
4732 self.clamp_caret();
4733 self.record_caret();
4734 }
4735 Err(e) => self.status = Some(format!("code language: {e}")),
4736 }
4737 }
4738
4739 /// The `span.start` of the code block covering the caret — the anchor
4740 /// [`wysiwyg::code_info_span`] reads the fence from. `None` when the caret is
4741 /// in none.
4742 fn code_block_start_at_caret(&mut self) -> Option<usize> {
4743 let off = self.caret;
4744 self.nodes()
4745 .into_iter()
4746 .filter(|n| n.kind == Kind::CodeBlock && n.span.start <= off && off <= n.span.end)
4747 .max_by_key(|n| n.span.start)
4748 .map(|n| n.span.start)
4749 }
4750
4751 /// The source range of the text inside the link covering `off` — what sits
4752 /// between its `[` and `]`. `None` when twig reports no link there.
4753 fn link_text_span(&mut self, off: usize) -> Option<std::ops::Range<usize>> {
4754 self.nodes()
4755 .into_iter()
4756 // Two links can touch (`[a](x)[b](y)`), and then one's `span.end` is
4757 // the other's `span.start`; the link that starts latest at or before
4758 // `off` is the one `off` is actually in.
4759 .filter(|n| n.kind == Kind::Link && n.span.start <= off && off < n.span.end)
4760 .max_by_key(|n| n.span.start)
4761 .and_then(|n| n.content_span)
4762 }
4763
4764 // ── undo / redo ───────────────────────────────────────────────────────────
4765 // twig owns the history of *bytes* (it owns the buffer) and now carries the
4766 // caret through it too: `record_caret` stashes each state's caret in twig's
4767 // opaque per-step blob, and undo/redo hand it back with the source they
4768 // restore. So leaf keeps no history of its own — no parallel stacks to march
4769 // in lockstep and silently drift out of it.
4770
4771 /// Undo the last edit step (⌘Z / ^Z), putting the caret and selection back
4772 /// where they were when that step began.
4773 pub fn undo(&mut self) {
4774 if self.read_only {
4775 return;
4776 }
4777 let (undone, redoable) = (self.undo_steps, self.redo_steps);
4778 match self.editor.undo() {
4779 Ok(Some(change)) => {
4780 self.after_history(change);
4781 // `refresh` counted the restore as an edit; it was a step back.
4782 self.undo_steps = undone.saturating_sub(1);
4783 self.redo_steps = redoable + 1;
4784 }
4785 Ok(None) => {
4786 self.undo_steps = 0;
4787 self.status = Some("nothing to undo".into());
4788 }
4789 Err(e) => self.status = Some(format!("undo: {e}")),
4790 }
4791 }
4792
4793 /// Redo the last undone edit step (⇧⌘Z / ^Y), putting the caret and
4794 /// selection back where that step originally left them.
4795 pub fn redo(&mut self) {
4796 if self.read_only {
4797 return;
4798 }
4799 let (undone, redoable) = (self.undo_steps, self.redo_steps);
4800 match self.editor.redo() {
4801 Ok(Some(change)) => {
4802 self.after_history(change);
4803 // `refresh` counted the restore as an edit; it was a step forward.
4804 self.undo_steps = undone + 1;
4805 self.redo_steps = redoable.saturating_sub(1);
4806 }
4807 Ok(None) => {
4808 self.redo_steps = 0;
4809 self.status = Some("nothing to redo".into());
4810 }
4811 Err(e) => self.status = Some(format!("redo: {e}")),
4812 }
4813 }
4814
4815 /// Refresh the cached source and put the caret back where the step being
4816 /// undone/redone had it, clearing any active run.
4817 ///
4818 /// The caret comes from twig's blob for the restored state (what
4819 /// `record_caret` stored). `change` is only the fallback for a state with no
4820 /// blob — a caret at the end of the restored text, which is where this always
4821 /// landed before the blobs were kept. It is the edit site, not where the user
4822 /// was standing, so it's a floor and not the behaviour: undoing should hand
4823 /// back the document *and* the place you were working, which for an edit made
4824 /// anywhere but under the caret are two different places.
4825 fn after_history(&mut self, change: Change) {
4826 self.refresh();
4827 match self
4828 .editor
4829 .caret_blob()
4830 .ok()
4831 .and_then(|b| CaretState::from_blob(&b))
4832 {
4833 Some(state) => {
4834 self.caret = state.caret.min(self.source.len());
4835 self.anchor = state.anchor.map(|a| a.min(self.source.len()));
4836 }
4837 None => {
4838 self.caret = change.new.end.min(self.source.len());
4839 self.anchor = None;
4840 }
4841 }
4842 self.goal_col = None;
4843 self.last_edit_kind = None;
4844 self.dirty = self.source != self.clean_source;
4845 self.status = None;
4846 self.clamp_caret();
4847 }
4848
4849 // ── the file ──────────────────────────────────────────────────────────────
4850
4851 #[cfg(feature = "fs")]
4852 pub fn save(&mut self) {
4853 if self.is_untitled() {
4854 // No path to write and no name to invent: ⌘S on an untitled document
4855 // is a Save As, and only a frontend has a picker to ask with. Say so
4856 // rather than failing at the filesystem with an empty path.
4857 self.status = Some("untitled — save as…".into());
4858 return;
4859 }
4860 let path = self.path.clone();
4861 if self.write(&path) {
4862 self.mark_saved();
4863 }
4864 }
4865
4866 /// Save As: write the document to `path` and *move* it there — `self.path`
4867 /// becomes `path`, and every later [`Doc::save`] writes the new file. That's
4868 /// what Save As means; a copy would leave the user editing a document whose
4869 /// name is no longer where their keystrokes go.
4870 ///
4871 /// The move only happens if the bytes actually landed. A failed write leaves
4872 /// the path, `dirty`, and the disk watermark exactly as they were, with the
4873 /// same `save failed: …` status a failed [`Doc::save`] sets — the document
4874 /// must never come away believing it was saved.
4875 ///
4876 /// An existing `path` is overwritten, and the caller is the one that knows
4877 /// whether to ask first: a Save As picker has already run that prompt, and a
4878 /// second confirmation from down here would be the same question twice.
4879 ///
4880 /// `format` does **not** follow the new extension. The buffer is parsed as
4881 /// the format it was opened with, and re-reading it as another one is a
4882 /// conversion — a different, lossy operation that would throw away the undo
4883 /// history — not a rename. So `notes.md` saved as `notes.dj` holds Markdown
4884 /// in a `.dj` file, and `format_name()` keeps honestly saying `markdown`
4885 /// until it's reopened.
4886 #[cfg(feature = "fs")]
4887 pub fn save_as(&mut self, path: PathBuf) {
4888 if !self.write(&path) {
4889 return;
4890 }
4891 self.path = path;
4892 self.mark_saved();
4893 }
4894
4895 /// Put `source` on disk at `path`, reporting whether it got there. The one
4896 /// place leaf writes a document, so a save and a Save As can't disagree
4897 /// about what a failure looks like.
4898 #[cfg(feature = "fs")]
4899 fn write(&mut self, path: &Path) -> bool {
4900 match std::fs::write(path, self.source.as_bytes()) {
4901 Ok(()) => true,
4902 Err(e) => {
4903 self.status = Some(format!("save failed: {e}"));
4904 false
4905 }
4906 }
4907 }
4908
4909 /// Re-base the document's saved watermark to the current bytes: clears
4910 /// `dirty`, records `source` as the new clean state (so undoing back to here
4911 /// clears the flag again), and re-stamps the on-disk hash.
4912 ///
4913 /// [`Doc::save`]/[`Doc::save_as`] call this after a write lands. It is also
4914 /// the hook a **filesystem-free host** calls itself once it has persisted
4915 /// [`Doc::source`] its own way (a browser download, `localStorage`, a backend
4916 /// `PUT`) — which is why it is public and touches no filesystem: the bytes
4917 /// are already where that host wants them, and this just tells the model they
4918 /// are safe.
4919 pub fn mark_saved(&mut self) {
4920 self.clean_source = self.source.clone();
4921 self.dirty = false;
4922 // The bytes on disk are now ours, so this is the new watermark: without
4923 // re-stamping it, every save would report its own work as an external
4924 // change forever after.
4925 self.disk_hash = Some(hash_bytes(self.source.as_bytes()));
4926 self.status = Some(format!("saved {}", self.file_name()));
4927 }
4928
4929 /// What the file looks like now against the bytes leaf last read or wrote.
4930 ///
4931 /// Reads the file and hashes it (see `disk_hash` for why it isn't an mtime),
4932 /// so this is a filesystem round-trip, not a per-frame question — ask it
4933 /// when a window regains focus, on a timer, or before a save.
4934 ///
4935 /// This *only* reports the file. Whether the document also has unsaved edits
4936 /// is `dirty`, and the interesting case is the conjunction: `dirty` plus
4937 /// [`DiskState::Changed`] means a save overwrites someone's work and a
4938 /// [`Doc::reload`] discards the user's. leaf-core deliberately won't choose —
4939 /// it has no way to ask — so it hands a frontend both halves and lets it put
4940 /// the question to the person who can answer it.
4941 #[cfg(feature = "fs")]
4942 pub fn disk_state(&self) -> DiskState {
4943 let Some(want) = self.disk_hash else {
4944 return DiskState::Untitled;
4945 };
4946 match std::fs::read(&self.path) {
4947 Ok(bytes) if hash_bytes(&bytes) == want => DiskState::Unchanged,
4948 Ok(_) => DiskState::Changed,
4949 Err(e) if e.kind() == std::io::ErrorKind::NotFound => DiskState::Missing,
4950 Err(_) => DiskState::Unreadable,
4951 }
4952 }
4953
4954 /// Re-read the file and replace the document with what's there — the other
4955 /// answer to a [`DiskState::Changed`].
4956 ///
4957 /// **Discards unsaved changes, unconditionally.** It doesn't check `dirty`
4958 /// first: a frontend that wants to protect unsaved work asks (`dirty` +
4959 /// [`Doc::disk_state`]) *before* calling this, and one reloading a clean
4960 /// document shouldn't have to argue with a guard.
4961 ///
4962 /// **The undo history survives, and the reload is one step in it.** The
4963 /// whole buffer is spliced with the file's bytes through the same door every
4964 /// other edit goes through, as an [`EditKind::Other`] that coalesces with
4965 /// nothing on either side — so ^Z after a formatter or a `git checkout` has
4966 /// swapped the document out from under a reader gives them back what they
4967 /// were looking at, marked dirty, and ^Z again carries on into whatever they
4968 /// had done before it. This used to build a fresh parse and drop the stack,
4969 /// on the reasoning that twig's history belongs to the buffer and these are
4970 /// different bytes; that is true of *rebasing* a step onto them and not of
4971 /// recording the swap itself as one, which is all this is. A splice twig
4972 /// won't take falls back to the fresh parse, and only that path still costs
4973 /// the history.
4974 ///
4975 /// The caret keeps its byte offset, clamped to the new length; the selection
4976 /// is dropped. Anything cleverer would be a lie: leaf doesn't know how the
4977 /// file changed, so it can't know where the caret "still" is. Clamping keeps
4978 /// it where the user left it in the common case (a change further down the
4979 /// file, or none in the text they're sitting in), and never puts it
4980 /// somewhere invalid. A selection has two such offsets and no such excuse —
4981 /// silently reinterpreting one over changed bytes would arm the *next*
4982 /// keystroke to delete something the user never selected.
4983 ///
4984 /// Nothing is touched unless the whole reload succeeds; a failure leaves the
4985 /// document alone with a status.
4986 #[cfg(feature = "fs")]
4987 pub fn reload(&mut self) {
4988 if self.is_untitled() {
4989 self.status = Some("no file to reload".into());
4990 return;
4991 }
4992 let bytes = match std::fs::read(&self.path) {
4993 Ok(b) => b,
4994 Err(e) => {
4995 self.status = Some(format!("reload failed: {e}"));
4996 return;
4997 }
4998 };
4999 let Ok(source) = String::from_utf8(bytes) else {
5000 self.status = Some("reload failed: file is not UTF-8".into());
5001 return;
5002 };
5003 // Already these bytes — someone saved a file back unchanged, or leaf's
5004 // own write is being read back. Re-baseline against it and stop: a
5005 // splice of the text onto itself would put an undo step on the stack for
5006 // something nobody did.
5007 if source == self.source {
5008 self.disk_hash = Some(hash_bytes(source.as_bytes()));
5009 self.clean_source = source;
5010 self.dirty = false;
5011 self.status = Some(format!("reloaded {}", self.file_name()));
5012 return;
5013 }
5014 let caret = self.caret;
5015 // The pre-reload caret, so undoing the swap puts it back where the
5016 // reader was standing — the same bracketing `splice_exact` does.
5017 self.record_caret();
5018 if self
5019 .editor
5020 .edit_range(0, self.source.len(), &source)
5021 .is_ok()
5022 {
5023 self.refresh();
5024 } else {
5025 // twig wouldn't take the splice. Start over from the bytes, which is
5026 // what this always did, and is the one path that still costs the
5027 // history — `format` is the format this document *is*, not what the
5028 // (unchanged) name now says, see `save_as`.
5029 match new_editor(source.as_bytes(), self.format) {
5030 Ok(editor) => {
5031 self.editor = editor;
5032 self.source = source.clone();
5033 // Not going through `refresh`, so the revision has to move
5034 // here or every frontend keeps painting the old file from
5035 // cache.
5036 self.revision += 1;
5037 }
5038 Err(e) => {
5039 self.status = Some(format!("reload failed: {e}"));
5040 return;
5041 }
5042 }
5043 }
5044 self.disk_hash = Some(hash_bytes(source.as_bytes()));
5045 self.clean_source = self.source.clone();
5046 self.caret = caret.min(self.source.len());
5047 self.anchor = None;
5048 self.goal_col = None;
5049 self.last_edit_kind = None;
5050 self.dirty = false;
5051 self.status = Some(format!("reloaded {}", self.file_name()));
5052 self.clamp_caret();
5053 // And the post-reload caret, so a redo restores it.
5054 self.record_caret();
5055 }
5056
5057 /// Re-read the source from twig after it has changed the document. The one
5058 /// funnel every edit, undo, and redo comes through — so it's where the
5059 /// revision moves, and anything cached against the text dies here.
5060 fn refresh(&mut self) {
5061 if let Ok(s) = self.editor.source_str() {
5062 self.source = s;
5063 }
5064 self.revision += 1;
5065 // An edit is a step onto the history and the end of anything undone;
5066 // `undo`/`redo` come through here too and correct this after.
5067 self.undo_steps += 1;
5068 self.redo_steps = 0;
5069 self.clamp_caret();
5070 }
5071
5072 /// Whether [`undo`](Self::undo) has a step to take back — for a native
5073 /// Edit menu to enable its item by. See the note on `undo_steps` for what
5074 /// "has" means here.
5075 pub fn can_undo(&self) -> bool {
5076 !self.read_only && self.undo_steps > 0
5077 }
5078
5079 /// Whether [`redo`](Self::redo) has an undone step to restore.
5080 pub fn can_redo(&self) -> bool {
5081 !self.read_only && self.redo_steps > 0
5082 }
5083
5084 // ── caret movement ─────────────────────────────────────────────────────────
5085 // `extend` grows the selection (Shift+motion): it pins the anchor on the
5086 // first extended step and moves only the caret; an un-extended motion drops
5087 // the selection.
5088
5089 /// Place the caret at byte `offset` (clamped to a char boundary), extending
5090 /// the selection when `extend` is set. The public form of `move_to`, for a
5091 /// frontend that hit-tests pixels straight to a source offset.
5092 pub fn place_caret(&mut self, offset: usize, extend: bool) {
5093 self.goal_col = None;
5094 let before = self.caret;
5095 // A pixel hit-test can land between the visible caret stops — in the
5096 // blank gap a paragraph break is drawn with, or inside a hidden delimiter.
5097 // Snap to the nearest real stop so the caret can't come to rest where it
5098 // would draw in one place and type in another. The `(row, col)` click
5099 // path (`click`) already snaps this way through `offset_of_pos`; the
5100 // source view reaches every byte, so it snaps to nothing.
5101 let target = match self.view {
5102 View::Wysiwyg => self.vmap.snap_to_stop(offset.min(self.source.len())),
5103 // The source view reaches every byte, so there is no stop to snap
5104 // to — but "every byte" still means every *character* boundary. A
5105 // caret resting inside a multi-byte character draws nowhere real
5106 // and panics the next time anything slices there.
5107 View::Source => self.char_boundary_at_or_before(offset),
5108 };
5109 self.move_to(target, extend);
5110 self.clamp_caret();
5111 self.debug_assert_on_a_stop(before);
5112 }
5113
5114 /// Select the whole document (⌘A / Ctrl+A) — everything reachable in the
5115 /// active view, so in WYSIWYG it starts below hidden frontmatter (copy won't
5116 /// grab the metadata) while the source view still selects the literal whole.
5117 pub fn select_all(&mut self) {
5118 self.anchor = Some(self.caret_floor());
5119 self.caret = self.source.len();
5120 self.goal_col = None;
5121 self.last_edit_kind = None;
5122 self.status = None;
5123 }
5124
5125 /// Select the word (or whitespace / punctuation run) at `offset` — the
5126 /// double-click gesture. Anchors on the run's start with the caret at its
5127 /// end so a following Shift-motion extends from the far edge.
5128 pub fn select_word_at(&mut self, offset: usize) {
5129 let (s, e) = word_range_at(&self.source, offset.min(self.source.len()));
5130 self.anchor = Some(s);
5131 self.caret = e;
5132 self.goal_col = None;
5133 self.last_edit_kind = None;
5134 self.status = None;
5135 self.clamp_caret();
5136 }
5137
5138 /// Select the whole enclosing text block (paragraph, heading, list item's
5139 /// text…) at `offset` — the triple-click gesture. Reads the range straight
5140 /// from the AST (twig's `content_span`), so it selects the entire *logical*
5141 /// paragraph even when that paragraph soft-wraps across several visual rows —
5142 /// where a visual-row-based select breaks down, because one source offset at
5143 /// a wrap boundary belongs to two rows at once.
5144 pub fn select_block_at(&mut self, offset: usize) {
5145 let off = offset.min(self.source.len());
5146 let range = self
5147 .editor
5148 .ancestors_at(off)
5149 .ok()
5150 .and_then(|chain| {
5151 // Ancestors run root → deepest; the deepest node that is neither
5152 // an inline span nor a multi-block container is the text block
5153 // the caret sits in (a paragraph, a heading, a code block…).
5154 chain
5155 .into_iter()
5156 .rev()
5157 .find(|m| !wysiwyg::is_inline_kind(&m.kind) && !is_block_container(&m.kind))
5158 .map(|m| m.content_span.unwrap_or(m.span))
5159 })
5160 .unwrap_or_else(|| source_line_range(&self.source, off));
5161 self.anchor = Some(range.start.min(self.source.len()));
5162 self.caret = range.end.min(self.source.len());
5163 self.goal_col = None;
5164 self.last_edit_kind = None;
5165 self.status = None;
5166 self.clamp_caret();
5167 }
5168
5169 /// Select the exact source range `[start, end)` — anchor at `start`, caret
5170 /// at `end` — without snapping either end to a visible caret stop.
5171 ///
5172 /// The one caret verb that takes a range it was *handed* rather than one it
5173 /// worked out, for a host that already knows the bytes it means: a search
5174 /// hit, an annotation's footprint, a quote re-anchored through
5175 /// [`Doc::selection_quote`]. [`place_caret`](Self::place_caret) is the
5176 /// wrong tool for that, and not by a little — it snaps to the nearest
5177 /// *visible* stop, and where a range butts up against a hidden delimiter
5178 /// the nearest stop is the one before it, so selecting the "needle" of
5179 /// `**needle**` comes back with "needl" and an edit against it strands the
5180 /// "e".
5181 ///
5182 /// What `place_caret` does that is bookkeeping rather than snapping still
5183 /// happens here, because a host handing in a range is not asking to opt out
5184 /// of the invariants:
5185 ///
5186 /// - both ends are clamped into the document and up to
5187 /// [`caret_floor`](Self::caret_floor) — in WYSIWYG the leading
5188 /// frontmatter is hidden, and a caret parked in it draws nowhere and
5189 /// types into the metadata;
5190 /// - both land on character boundaries, so nothing slices a `é` in half;
5191 /// - the sticky vertical goal column is dropped, and any armed inline mark
5192 /// disarmed, since a range from outside inherits neither.
5193 ///
5194 /// An empty range is a caret rather than a selection —
5195 /// [`selection`](Self::selection) reports `None` for it, as it does for any
5196 /// anchor that has met the caret.
5197 pub fn select_range(&mut self, start: usize, end: usize) {
5198 let floor = self.caret_floor();
5199 let anchor = self.char_boundary_at_or_before(start.clamp(floor, self.source.len()));
5200 let caret = self.char_boundary_at_or_before(end.clamp(floor, self.source.len()));
5201 self.anchor = Some(anchor);
5202 self.caret = caret;
5203 self.goal_col = None;
5204 self.status = None;
5205 self.last_edit_kind = None;
5206 self.clear_pending();
5207 }
5208
5209 /// `offset` itself if it is a character boundary, else the boundary before
5210 /// it. An offset that isn't one draws nowhere real and panics the next time
5211 /// anything slices there.
5212 fn char_boundary_at_or_before(&self, offset: usize) -> usize {
5213 let mut o = offset.min(self.source.len());
5214 while o > 0 && !self.source.is_char_boundary(o) {
5215 o -= 1;
5216 }
5217 o
5218 }
5219
5220 /// The lowest source offset the caret may occupy in the active view. In
5221 /// WYSIWYG, leading frontmatter is hidden and unreachable, so the floor is
5222 /// the first rendered offset; the source view reaches everything, so it's 0.
5223 fn caret_floor(&self) -> usize {
5224 match self.view {
5225 View::Wysiwyg => self.vmap.content_start.min(self.source.len()),
5226 View::Source => 0,
5227 }
5228 }
5229
5230 /// Land in a table cell with its whole content selected — the anchor at the
5231 /// cell's start, the caret at its end — so a Tab/Return hop into a cell reads
5232 /// like tabbing into a form field: the text comes up selected, so typing
5233 /// replaces it and an arrow collapses to an edge. An empty cell (`start ==
5234 /// end`) collapses to a plain caret home (an empty selection is no selection).
5235 fn select_cell(&mut self, start: usize, end: usize) {
5236 self.select_range(start, end);
5237 }
5238
5239 fn move_to(&mut self, offset: usize, extend: bool) {
5240 if extend {
5241 if self.anchor.is_none() {
5242 self.anchor = Some(self.caret);
5243 }
5244 } else {
5245 self.anchor = None;
5246 }
5247 self.caret = offset.min(self.source.len()).max(self.caret_floor());
5248 self.status = None;
5249 // A caret move ends the current typing/deletion run, so the next edit
5250 // starts a fresh undo group rather than coalescing across the gap.
5251 self.last_edit_kind = None;
5252 // Moving away disarms any sticky mark — "start bold" applies only where
5253 // it was asked for, not wherever the caret next lands.
5254 self.clear_pending();
5255 }
5256
5257 // In the source view, motion walks source bytes / source lines. In the
5258 // WYSIWYG view it walks the rendered glyph grid (the visual map), which is
5259 // what steps the caret cleanly over hidden delimiters.
5260
5261 pub fn move_left(&mut self, extend: bool) {
5262 self.goal_col = None;
5263 if !extend && let Some((s, _e)) = self.selection() {
5264 self.move_to(s, false);
5265 return;
5266 }
5267 let target = match self.view {
5268 View::Source => {
5269 if self.caret > 0 {
5270 prev_boundary(&self.source, self.caret)
5271 } else {
5272 0
5273 }
5274 }
5275 // Walks caret *stops*, not columns: decoration (a table border, a
5276 // cell's padding) is stepped over in one press, and a hidden
5277 // delimiter never holds the caret up.
5278 View::Wysiwyg => self.vmap.stop_before(self.caret).unwrap_or(self.caret),
5279 };
5280 let before = self.caret;
5281 self.move_to(target, extend);
5282 self.debug_assert_on_a_stop(before);
5283 }
5284
5285 pub fn move_right(&mut self, extend: bool) {
5286 self.goal_col = None;
5287 if !extend && let Some((_s, e)) = self.selection() {
5288 self.move_to(e, false);
5289 return;
5290 }
5291 let target = match self.view {
5292 View::Source => {
5293 if self.caret < self.source.len() {
5294 next_boundary(&self.source, self.caret)
5295 } else {
5296 self.caret
5297 }
5298 }
5299 View::Wysiwyg => self.vmap.stop_after(self.caret).unwrap_or(self.caret),
5300 };
5301 let before = self.caret;
5302 self.move_to(target, extend);
5303 self.debug_assert_on_a_stop(before);
5304 }
5305
5306 /// Move to the start of the previous word (⌥← / Ctrl+←).
5307 pub fn move_word_left(&mut self, extend: bool) {
5308 self.goal_col = None;
5309 let before = self.caret;
5310 let target = self.word_left_from(self.caret);
5311 self.move_to(target, extend);
5312 self.debug_assert_on_a_stop(before);
5313 }
5314
5315 /// Move to the end of the next word (⌥→ / Ctrl+→).
5316 pub fn move_word_right(&mut self, extend: bool) {
5317 self.goal_col = None;
5318 let before = self.caret;
5319 let target = self.word_right_from(self.caret);
5320 self.move_to(target, extend);
5321 self.debug_assert_on_a_stop(before);
5322 }
5323
5324 // Word boundaries are found in the space the *view* is in. The source view
5325 // walks the source, because there the source is what's rendered. WYSIWYG
5326 // walks the rendered text instead: `**` is invisible to the user, so it has
5327 // to be invisible to word motion too — a caret parked inside one draws in
5328 // the column after `bold` and types two bytes earlier, and a word-delete
5329 // that stops there shreds the markup into `a ** c`.
5330
5331 /// The word boundary to the left of `off` in the active view's space.
5332 fn word_left_from(&self, off: usize) -> usize {
5333 match self.view {
5334 View::Source => prev_word(&self.source, off),
5335 View::Wysiwyg => self.glyph_word_left(off),
5336 }
5337 }
5338
5339 /// The word boundary to the right of `off` in the active view's space.
5340 fn word_right_from(&self, off: usize) -> usize {
5341 match self.view {
5342 View::Source => next_word(&self.source, off),
5343 View::Wysiwyg => self.glyph_word_right(off),
5344 }
5345 }
5346
5347 /// The character class of the glyph drawn at stop `off`.
5348 ///
5349 /// Read from the source, because a stop points at the source byte its glyph
5350 /// came from — the source *is* where the rendered character is written. What
5351 /// makes the walk glyph space rather than source space is that it only ever
5352 /// visits stops, and the hidden bytes between them have none.
5353 fn class_at(&self, off: usize) -> Class {
5354 self.source
5355 .get(off..)
5356 .and_then(|s| s.chars().next())
5357 .map_or(Class::Space, classify)
5358 }
5359
5360 /// [`next_word`] in glyph space: skip any leading separators, then consume
5361 /// the following word run, with the stop table standing in for the source's
5362 /// characters.
5363 fn glyph_word_right(&self, from: usize) -> usize {
5364 let Some(mut off) = self.vmap.stop_at_or_after(from) else {
5365 return from;
5366 };
5367 let mut in_word = false;
5368 loop {
5369 match self.class_at(off) {
5370 Class::Word => in_word = true,
5371 _ if in_word => return off,
5372 _ => {}
5373 }
5374 match self.vmap.stop_after(off) {
5375 Some(next) => off = next,
5376 None => return off,
5377 }
5378 }
5379 }
5380
5381 /// [`prev_word`] in glyph space: skip separators walking left, then consume
5382 /// the preceding word run.
5383 fn glyph_word_left(&self, from: usize) -> usize {
5384 let Some(mut off) = self.vmap.stop_at_or_before(from) else {
5385 return from;
5386 };
5387 let mut in_word = false;
5388 while let Some(prev) = self.vmap.stop_before(off) {
5389 match self.class_at(prev) {
5390 Class::Word => in_word = true,
5391 _ if in_word => return off,
5392 _ => {}
5393 }
5394 off = prev;
5395 }
5396 off
5397 }
5398
5399 /// After a motion that walks the visual map, the caret must be *on* the map.
5400 /// A stop is the only offset where the caret draws and edits in the same
5401 /// place, and it's the invariant both a caret parked inside an emoji and one
5402 /// parked inside a `**` were quietly breaking.
5403 ///
5404 /// Only when the caret actually moved: a walk with nowhere to go leaves it
5405 /// where it was, which is wherever the floor or a frontend put it rather
5406 /// than somewhere this motion chose.
5407 fn debug_assert_on_a_stop(&self, before: usize) {
5408 debug_assert!(
5409 self.view != View::Wysiwyg
5410 || self.vmap.num_rows() == 0
5411 || self.caret == before
5412 || self.vmap.is_stop(self.caret),
5413 "motion left the caret at {}, which is not a caret stop: it would draw in \
5414 one place and type in another",
5415 self.caret
5416 );
5417 }
5418
5419 // Up and Down run off the ends of the document rather than stopping dead at
5420 // them: Up from the first row lands at the document's start, Down from the
5421 // last at its end. That's Cocoa's rule (`moveUp:`/`moveDown:` past the edge
5422 // are `moveToBeginningOfDocument:`/`moveToEndOfDocument:`), and holding ↓
5423 // reaching the end of the text is what a reader means by it.
5424 //
5425 // The views used to disagree here by accident rather than by decision: the
5426 // source view fell into the edge behaviour through `row_col_to_offset`
5427 // clamping an out-of-range row to the end of the string, while WYSIWYG had
5428 // no row below to walk to and did nothing at all. They share the rule now,
5429 // each in its own space — the source view reaches every byte, WYSIWYG only
5430 // the offsets it draws.
5431
5432 pub fn move_up(&mut self, extend: bool) {
5433 let (row, col) = self.caret_pos();
5434 let goal = self.goal_col.unwrap_or(col);
5435 let target = match self.view {
5436 View::Source => match row.checked_sub(1) {
5437 Some(r) => row_col_to_offset(&self.source, r, goal),
5438 None => self.reachable_start(),
5439 },
5440 // A table's border rules are drawn but hold no caret, so Up steps
5441 // over them to the row that does.
5442 View::Wysiwyg => match self.vmap.navigable_above(row) {
5443 Some(r) => self.row_target(r, goal),
5444 None => self.reachable_start(),
5445 },
5446 };
5447 self.step_vertical(target, goal, extend);
5448 }
5449
5450 pub fn move_down(&mut self, extend: bool) {
5451 let (row, col) = self.caret_pos();
5452 let goal = self.goal_col.unwrap_or(col);
5453 let target = match self.view {
5454 View::Source => match self.source_row_below(row) {
5455 Some(r) => row_col_to_offset(&self.source, r, goal),
5456 None => self.reachable_end(),
5457 },
5458 View::Wysiwyg => match self.vmap.navigable_below(row) {
5459 Some(r) => self.row_target(r, goal),
5460 None => self.reachable_end(),
5461 },
5462 };
5463 self.step_vertical(target, goal, extend);
5464 }
5465
5466 /// Land a vertical motion at `target`, latching the `goal` column it aimed
5467 /// with so the rest of the run keeps aiming there.
5468 ///
5469 /// A motion with nowhere to go changes *nothing*, the goal column included:
5470 /// the latch used to run before the early return at the top of the document,
5471 /// so an Up that did nothing still armed a column, and the next Down aimed
5472 /// at one the caret had never been in.
5473 fn step_vertical(&mut self, target: usize, goal: usize, extend: bool) {
5474 let before = self.caret;
5475 if target == before {
5476 return;
5477 }
5478 self.goal_col = Some(goal);
5479 self.move_to(target, extend);
5480 self.debug_assert_on_a_stop(before);
5481 }
5482
5483 /// The source line below `row`, or `None` when `row` is the last one. Lines
5484 /// are counted by newline, so a trailing one leaves a real, empty last line
5485 /// for the caret to sit on — the document ends below it, not on it.
5486 fn source_row_below(&self, row: usize) -> Option<usize> {
5487 let last = self.source.bytes().filter(|&b| b == b'\n').count();
5488 (row < last).then_some(row + 1)
5489 }
5490
5491 /// Where a vertical motion aiming at the `goal` column lands on visual row
5492 /// `r`: the column clamped to the row, mapped to its offset, then held
5493 /// inside the row's own [bounds](Self::row_bounds) — a wrapped row's last
5494 /// column belongs to the row below, and a gutter's column 0 points at the
5495 /// block rather than at this row.
5496 fn row_target(&self, r: usize, goal: usize) -> usize {
5497 let (start, end) = self.row_bounds(r);
5498 self.vmap
5499 .offset_of_pos(r, goal.min(self.vmap.row_width(r)))
5500 .clamp(start, end)
5501 }
5502
5503 /// The first and last offsets the caret can reach in the active view.
5504 ///
5505 /// Not the same span in both: the source view shows every byte, so it can
5506 /// reach every byte. WYSIWYG reaches only what it draws — hidden frontmatter
5507 /// sits below the first stop, and a document's trailing newline is drawn
5508 /// nowhere and so sits past the last.
5509 fn reachable_start(&self) -> usize {
5510 match self.view {
5511 View::Source => 0,
5512 View::Wysiwyg => self.vmap.stop_at_or_after(0).unwrap_or(self.caret),
5513 }
5514 }
5515
5516 fn reachable_end(&self) -> usize {
5517 match self.view {
5518 View::Source => self.source.len(),
5519 View::Wysiwyg => self
5520 .vmap
5521 .stop_at_or_before(self.source.len())
5522 .unwrap_or(self.caret),
5523 }
5524 }
5525
5526 /// The `[start, end]` offsets visual row `r` *draws* — everything on it,
5527 /// including the space a soft wrap ate off its end, which is drawn on this
5528 /// row however much the offset past it belongs to the next one.
5529 fn row_span(&self, r: usize) -> (usize, usize) {
5530 let start = self
5531 .vmap
5532 .row_start(r)
5533 .unwrap_or_else(|| self.vmap.offset_of_pos(r, 0));
5534 let end = self.vmap.offset_of_pos(r, self.vmap.row_width(r));
5535 (start.min(end), end)
5536 }
5537
5538 /// [`row_span`](Self::row_span) narrowed to where the caret can stand: a
5539 /// soft wrap's shared offset opens the row below (see `pos_of_offset`), so
5540 /// this row's last position is the one before it — the offset before the
5541 /// space the wrap ate, where the caret draws just past the row's last word
5542 /// and types there too.
5543 ///
5544 /// Aiming at the shared offset instead is what stalled End: it is the row's
5545 /// last *column*, so End pressed on the row reached it and then read back as
5546 /// the row below's start, where a second press ran on to that row's end and
5547 /// the next to the one after — End walking down the paragraph a row a press.
5548 fn row_bounds(&self, r: usize) -> (usize, usize) {
5549 let (start, end) = self.row_span(r);
5550 let wraps = self
5551 .vmap
5552 .navigable_below(r)
5553 .and_then(|b| self.vmap.row_start(b))
5554 .is_some_and(|off| off == end);
5555 match wraps {
5556 true => (start, self.vmap.stop_before(end).unwrap_or(end).max(start)),
5557 false => (start, end),
5558 }
5559 }
5560
5561 /// The `[start, end]` of the line Home and End aim at: the visual row in
5562 /// WYSIWYG, the logical line in the source view. Both ends are caret stops.
5563 ///
5564 /// A soft-wrapped row is a line here, because it is one to the eye and the
5565 /// eye is what these keys are aimed by — a reader pressing End means the end
5566 /// of the line they can see. (`select_block_at` wants the opposite and reads
5567 /// the AST for it: a triple-click grabs the whole paragraph, however many
5568 /// rows it folds into.)
5569 fn line_bounds(&self) -> (usize, usize) {
5570 let (row, _) = self.caret_pos();
5571 match self.view {
5572 View::Source => {
5573 let start = line_start(&self.source, row);
5574 (start, line_end_from(&self.source, start))
5575 }
5576 View::Wysiwyg => self.row_bounds(row),
5577 }
5578 }
5579
5580 /// The same line as [`line_bounds`](Self::line_bounds), as far as it is
5581 /// *drawn* — what a kill takes.
5582 ///
5583 /// The two part only at a soft wrap, over the space the wrap ate: the caret
5584 /// can't stand after it (that offset opens the row below, and End stopping
5585 /// there would walk), but it is on this row, and a kill that spared it would
5586 /// leave a double space behind where the row's text had been. Deleting it
5587 /// joins nothing — a wrap is drawn, not written.
5588 fn line_span(&self) -> (usize, usize) {
5589 let (row, _) = self.caret_pos();
5590 match self.view {
5591 View::Source => self.line_bounds(),
5592 View::Wysiwyg => self.row_span(row),
5593 }
5594 }
5595
5596 /// The first offset in `[start, end]` holding something other than
5597 /// whitespace, or `end` when the line holds nothing else — where Home aims.
5598 ///
5599 /// Walks the space the view is in, as word motion does: WYSIWYG steps stops,
5600 /// so a hidden delimiter is never taken for the line's first character (nor
5601 /// landed on), and the source view steps the source it is showing.
5602 fn first_non_space(&self, start: usize, end: usize) -> usize {
5603 let mut off = start;
5604 while off < end {
5605 if self.class_at(off) != Class::Space {
5606 return off;
5607 }
5608 off = match self.view {
5609 View::Source => next_boundary(&self.source, off),
5610 View::Wysiwyg => match self.vmap.stop_after(off) {
5611 Some(next) => next,
5612 None => return end,
5613 },
5614 };
5615 }
5616 end
5617 }
5618
5619 /// Home: to the first character on the line, or to column 0 when the caret
5620 /// is already on it — the two-press toggle every editor spells this way.
5621 /// The indentation is somewhere the caret has to be able to reach and almost
5622 /// never where a reader is headed, so it costs the second press.
5623 pub fn move_home(&mut self, extend: bool) {
5624 self.goal_col = None;
5625 let (start, end) = self.line_bounds();
5626 let text = self.first_non_space(start, end);
5627 let target = if self.caret == text { start } else { text };
5628 let before = self.caret;
5629 self.move_to(target, extend);
5630 self.debug_assert_on_a_stop(before);
5631 }
5632
5633 /// End: to the end of the line.
5634 pub fn move_end(&mut self, extend: bool) {
5635 self.goal_col = None;
5636 let (_, end) = self.line_bounds();
5637 let before = self.caret;
5638 self.move_to(end, extend);
5639 self.debug_assert_on_a_stop(before);
5640 }
5641
5642 /// Hop to the next (Tab) or previous (Shift+Tab) table cell, landing with the
5643 /// cell's whole content selected (see [`Self::select_cell`]). Returns `false`
5644 /// when the caret isn't in a table, or is already in the last/first cell — the
5645 /// frontend then does whatever Tab normally does (indent), so Tab keeps its
5646 /// meaning everywhere else.
5647 pub fn cell_hop(&mut self, forward: bool) -> bool {
5648 let Some((grid, r, c)) = self.table_grid_at(self.caret) else {
5649 return false;
5650 };
5651 // Flatten to document (row-major) order and step one cell either way.
5652 let i: usize = grid[..r].iter().map(Vec::len).sum::<usize>() + c;
5653 let flat: Vec<(usize, usize)> = grid.into_iter().flatten().collect();
5654 let next = if forward {
5655 i.checked_add(1)
5656 } else {
5657 i.checked_sub(1)
5658 };
5659 let Some(&(start, end)) = next.and_then(|j| flat.get(j)) else {
5660 return false; // at the table's edge; leave Tab to the frontend
5661 };
5662 self.select_cell(start, end);
5663 true
5664 }
5665
5666 /// Move the caret to the cell directly above (`down == false`) or below in
5667 /// the same column, landing with the cell's whole content selected (see
5668 /// [`Self::select_cell`]). Returns `false` at the grid's top/bottom edge (or
5669 /// when the caret isn't in a table), so the frontend can fall through — the
5670 /// vertical counterpart of [`Self::cell_hop`].
5671 ///
5672 /// A ragged row that is short a column clamps to its last cell, so Down never
5673 /// falls out of the table over a gap the row above happened to have.
5674 pub fn cell_move_vertical(&mut self, down: bool) -> bool {
5675 let Some((grid, r, c)) = self.table_grid_at(self.caret) else {
5676 return false;
5677 };
5678 let target = match down {
5679 true => r + 1,
5680 false if r == 0 => return false,
5681 false => r - 1,
5682 };
5683 let Some(row) = grid.get(target) else {
5684 return false;
5685 };
5686 let Some(&(start, end)) = row.get(c).or_else(|| row.last()) else {
5687 return false;
5688 };
5689 self.select_cell(start, end);
5690 true
5691 }
5692
5693 /// The table containing `off` as a row-major grid of `(start, end)` cell
5694 /// caret homes, plus the `(row, col)` the caret sits in — `None` when `off`
5695 /// isn't in a table. Read straight off the visual map's laid-out grid, so
5696 /// every cell (an empty one included, whose derived home twig gives no
5697 /// `content_span` for) is present and in the order Tab walks them.
5698 // Grid, row, column — three returns that only ever travel together, and a
5699 // named type for the pair of them would be read at one call site.
5700 #[allow(clippy::type_complexity)]
5701 fn table_grid_at(&self, off: usize) -> Option<(Vec<Vec<(usize, usize)>>, usize, usize)> {
5702 for t in &self.vmap.tables {
5703 let mut pos = None;
5704 let grid: Vec<Vec<(usize, usize)>> = t
5705 .grid
5706 .iter()
5707 .enumerate()
5708 .map(|(r, row)| {
5709 row.cells
5710 .iter()
5711 .enumerate()
5712 .map(|(c, cell)| {
5713 if pos.is_none() && off >= cell.start && off <= cell.end {
5714 pos = Some((r, c));
5715 }
5716 (cell.start, cell.end)
5717 })
5718 .collect()
5719 })
5720 .collect();
5721 if let Some((r, c)) = pos {
5722 return Some((grid, r, c));
5723 }
5724 }
5725 None
5726 }
5727
5728 // ── table key policy ──────────────────────────────────────────────────────
5729 // The three keys a table gives its own meaning — Tab, Return, Shift+Return —
5730 // as one policy every frontend shares, rather than each re-deriving it. Each
5731 // reports whether it acted *as a table key*; a `false` hands the key back to
5732 // the frontend's ordinary handling (indent, newline) so it keeps its meaning
5733 // everywhere else.
5734
5735 /// Tab / Shift+Tab inside a table. Tab steps to the next cell, appending a
5736 /// fresh row and entering it when it runs off the last one; Shift+Tab steps
5737 /// back and simply stays put at the very first cell. `false` when the caret
5738 /// isn't in a table.
5739 pub fn cell_tab(&mut self, forward: bool) -> bool {
5740 if !self.caret_in_table() {
5741 return false;
5742 }
5743 if self.cell_hop(forward) {
5744 return true;
5745 }
5746 // Off the last cell: grow the table by a row and step into its first
5747 // cell. (Shift+Tab at the first cell has nowhere to go and just holds.)
5748 if forward {
5749 self.append_row_and_enter(0);
5750 }
5751 true
5752 }
5753
5754 /// Return inside a table: drop to the cell below in the same column,
5755 /// appending a new row when the caret is already in the last one. `false`
5756 /// when the caret isn't in a table, so the frontend inserts a newline.
5757 pub fn cell_return(&mut self) -> bool {
5758 if !self.caret_in_table() {
5759 return false;
5760 }
5761 if self.cell_move_vertical(true) {
5762 return true;
5763 }
5764 // Already on the last row: grow one below and drop into the same column.
5765 let col = self.table_grid_at(self.caret).map_or(0, |(_, _, c)| c);
5766 self.append_row_and_enter(col);
5767 true
5768 }
5769
5770 /// Append a row below the caret's (last) row and land in `col` of it. The
5771 /// caret is in the last row, so twig's "insert below" makes the fresh row the
5772 /// table's new last — but twig re-spells the whole table, moving every byte,
5773 /// so the destination is read back from the rebuilt grid by the table's
5774 /// position (stable across a row insert), not from the pre-edit caret.
5775 fn append_row_and_enter(&mut self, col: usize) {
5776 let table = self.caret_table_index();
5777 self.table_insert_row(true);
5778 self.rebuild_map();
5779 let Some((start, end)) = table
5780 .and_then(|ti| self.vmap.tables.get(ti))
5781 .and_then(|t| t.grid.last())
5782 .and_then(|row| row.cells.get(col.min(row.cells.len().saturating_sub(1))))
5783 .map(|cell| (cell.start, cell.end))
5784 else {
5785 return;
5786 };
5787 self.select_cell(start, end);
5788 }
5789
5790 /// The index, among the document's tables, of the one the caret sits in —
5791 /// `None` when it's in none. Used to re-find a table after an edit re-spells
5792 /// it (a row insert leaves the table order unchanged).
5793 fn caret_table_index(&self) -> Option<usize> {
5794 let off = self.caret;
5795 self.vmap.tables.iter().position(|t| {
5796 t.grid
5797 .iter()
5798 .any(|row| row.cells.iter().any(|c| off >= c.start && off <= c.end))
5799 })
5800 }
5801
5802 /// Shift+Return inside a table: insert a hard line break *within* the current
5803 /// cell, via twig's `insert_line_break`. `false` when the caret isn't in a
5804 /// table, so the frontend inserts an ordinary line break.
5805 ///
5806 /// A table row is a single source line, so the newline-spelled hard break
5807 /// can't live in a cell. twig spells the in-cell break the format's way
5808 /// (`<br>` for Markdown) and reparses it as a *semantic* `hard_break`, so the
5809 /// break round-trips as structure the renderer reads back as a line — not the
5810 /// opaque raw HTML the old raw-splice left behind.
5811 ///
5812 /// Djot has no idiomatic in-cell break, so twig refuses it
5813 /// (`UnsupportedFormat`) rather than emit a `<br>` that any other djot reader
5814 /// would render as the literal text `<br>`. The gesture is still *consumed*
5815 /// there — returning `false` would let the frontend insert a real newline,
5816 /// which splits the one-line row — it just leaves the cell unchanged and says
5817 /// so on the status line. A rollback (`EditConflict`) is swallowed the same.
5818 ///
5819 /// Which formats refuse is [`Capabilities::cell_line_break`], and the two
5820 /// have to be read together: djot is not the only `false`, and naming it in
5821 /// the message was already a guess that HTML — which spells the break as its
5822 /// own `<br>` — would have made wrong.
5823 pub fn cell_line_break(&mut self) -> bool {
5824 if self.read_only || !self.caret_in_table() {
5825 return false;
5826 }
5827 self.record_caret();
5828 match self.editor.insert_line_break(self.caret) {
5829 Ok(change) => {
5830 self.last_edit_kind = None;
5831 self.refresh();
5832 self.caret = change.new.end;
5833 self.anchor = None;
5834 self.goal_col = None;
5835 self.clamp_caret();
5836 self.dirty = self.source != self.clean_source;
5837 self.status = None;
5838 self.record_caret();
5839 }
5840 Err(twig::Error::UnsupportedFormat) => {
5841 self.status = Some(format!(
5842 "in-cell line breaks aren't supported in {}",
5843 self.format_name()
5844 ));
5845 }
5846 Err(_) => {}
5847 }
5848 true
5849 }
5850
5851 /// Rebuild the visual map at the width the last build used. A structural edit
5852 /// bumps the revision and swaps the source in, but leaves the *map* stale;
5853 /// when a single gesture edits and then moves over the result (Tab appending
5854 /// a row, then stepping into it), the move needs the map to already show the
5855 /// edit rather than waiting for the frontend's next frame.
5856 fn rebuild_map(&mut self) {
5857 let wrap = self.vmap_key.as_ref().and_then(|(_, w, _)| *w);
5858 self.build_map(wrap);
5859 }
5860
5861 /// Move the caret to the very start of the document (⌘↑ on macOS,
5862 /// Ctrl+Home on Windows/Linux).
5863 pub fn move_doc_start(&mut self, extend: bool) {
5864 self.goal_col = None;
5865 self.move_to(0, extend);
5866 }
5867
5868 /// Move the caret to the very end of the document (⌘↓ on macOS,
5869 /// Ctrl+End on Windows/Linux).
5870 pub fn move_doc_end(&mut self, extend: bool) {
5871 self.goal_col = None;
5872 let end = self.source.len();
5873 self.move_to(end, extend);
5874 }
5875
5876 /// Point the caret at the body cell `(row, col)` the mouse landed on —
5877 /// `col` being a cell of the terminal grid, which is what a display column
5878 /// is. A click on the far cell of a wide character lands at that
5879 /// character's start; the mapping's own doc-comments carry the rule.
5880 pub fn click(&mut self, row: usize, col: usize, extend: bool) {
5881 self.goal_col = None;
5882 let target = match self.view {
5883 View::Source => row_col_to_offset(&self.source, row, col),
5884 View::Wysiwyg => self.vmap.offset_of_pos(row, col),
5885 };
5886 let before = self.caret;
5887 self.move_to(target, extend);
5888 self.debug_assert_on_a_stop(before);
5889 }
5890
5891 /// Settle `scroll` for a frame about to be drawn: follow the caret onto the
5892 /// screen if it has moved since the last frame, and never scroll past the
5893 /// last of `rows`.
5894 ///
5895 /// Only if it has *moved* — that's the whole point. Revealing the caret on
5896 /// every frame ties the viewport to it, and a scroll wheel that fights the
5897 /// caret for the viewport loses: the view snaps back the instant it tries to
5898 /// pass the caret's row, so the document can't be scrolled beyond what's
5899 /// already on screen. A caret move is the frontend's cue to follow; a scroll
5900 /// with the caret sitting still is the reader's cue to leave it alone.
5901 pub fn follow_caret(&mut self, caret_row: usize, height: usize, rows: usize) {
5902 if self.drawn_caret != Some(self.caret) {
5903 if caret_row < self.scroll {
5904 self.scroll = caret_row;
5905 } else if height > 0 && caret_row >= self.scroll + height {
5906 self.scroll = caret_row + 1 - height;
5907 }
5908 self.drawn_caret = Some(self.caret);
5909 }
5910 self.scroll = self.scroll.min(rows.saturating_sub(1));
5911 }
5912
5913 /// The caret's screen position `(row, col)` in the active view's grid, with
5914 /// `col` a display column: the cell to draw the caret in, which on a line of
5915 /// `你好` or emoji is not the count of characters before it.
5916 pub fn caret_pos(&self) -> (usize, usize) {
5917 match self.view {
5918 View::Source => offset_to_row_col(&self.source, self.caret),
5919 View::Wysiwyg => self.vmap.pos_of_offset(self.caret),
5920 }
5921 }
5922
5923 fn clamp_caret(&mut self) {
5924 if self.caret > self.source.len() {
5925 self.caret = self.source.len();
5926 }
5927 // In WYSIWYG the caret can't sit inside hidden frontmatter; lift it (and
5928 // any selection anchor) to the first rendered offset.
5929 let floor = self.caret_floor();
5930 if self.caret < floor {
5931 self.caret = floor;
5932 }
5933 if let Some(a) = self.anchor
5934 && a < floor
5935 {
5936 self.anchor = Some(floor);
5937 }
5938 while self.caret > 0 && !self.source.is_char_boundary(self.caret) {
5939 self.caret -= 1;
5940 }
5941 }
5942}
5943
5944// ── byte-offset ⇄ (row, col) helpers ─────────────────────────────────────────
5945
5946// Left/right motion and backspace/delete step by *grapheme cluster*, not
5947// codepoint, so an emoji (a ZWJ sequence) or a base letter plus its combining
5948// marks moves and deletes as the single character a user sees. Grapheme
5949// boundaries are a superset of char boundaries, so the caret stays valid for twig.
5950
5951/// How an insert of `text` groups for undo: a single typed character folds into
5952/// the run of typing around it, while a newline or a multi-character insert is a
5953/// step of its own.
5954fn typed_edit_kind(text: &str) -> EditKind {
5955 if text.chars().take(2).count() == 1 && text != "\n" {
5956 EditKind::Insert
5957 } else {
5958 EditKind::Other
5959 }
5960}
5961
5962fn prev_boundary(s: &str, i: usize) -> usize {
5963 let mut cursor = GraphemeCursor::new(i, s.len(), true);
5964 cursor.prev_boundary(s, 0).ok().flatten().unwrap_or(0)
5965}
5966
5967fn next_boundary(s: &str, i: usize) -> usize {
5968 let mut cursor = GraphemeCursor::new(i, s.len(), true);
5969 cursor.next_boundary(s, 0).ok().flatten().unwrap_or(s.len())
5970}
5971
5972// ── word boundaries ──────────────────────────────────────────────────────────
5973// The shared primitive behind word-wise motion, word deletion, and
5974// double-click-to-select-a-word. A "word" is a maximal run of one character
5975// class; whitespace and punctuation are their own classes, so motion skips
5976// cleanly between them the way native text fields do.
5977
5978#[derive(PartialEq, Eq, Clone, Copy)]
5979enum Class {
5980 Word,
5981 Space,
5982 Other,
5983}
5984
5985/// The source range of an inline node's own visible text — the part of it a
5986/// WYSIWYG caret can reach, as against the delimiters that only spell it.
5987/// `None` for a node with no interior to empty (a `str`, a break).
5988///
5989/// twig reports no `content_span` for `verbatim`/`inline_math`, whose text sits
5990/// one delimiter in from the span — the same place the renderer maps it to. A
5991/// longer fence (`` ``a`` ``) breaks that assumption, so the guess is checked
5992/// against the source rather than trusted: a range guessed wrong here is text
5993/// deleted wrong.
5994fn inline_content_span(n: &FlatNode, source: &str) -> Option<std::ops::Range<usize>> {
5995 if let Some(span) = n.content_span.clone() {
5996 return Some(span);
5997 }
5998 match n.kind.as_str() {
5999 "verbatim" | "inline_math" => {
6000 let text = n.text.as_ref()?;
6001 let start = n.span.start + 1;
6002 let range = start..start + text.len();
6003 (source.get(range.clone()) == Some(text.as_str())).then_some(range)
6004 }
6005 _ => None,
6006 }
6007}
6008
6009/// The `id` a node declares, or `None` for one that declares none — the
6010/// attribute djot writes for a `{#v1}` and mints for a heading.
6011///
6012/// A bare attribute (`{#v1 hidden}`'s `hidden`) has no value, and a bare `id`
6013/// names nothing, so it reads as absent rather than as the empty string.
6014fn declared_id(n: &FlatNode) -> Option<&str> {
6015 n.attrs.iter().find(|(k, _)| k == "id")?.1.as_deref()
6016}
6017
6018/// A heading's words reduced to the form a link fragment spells them in:
6019/// lowercase, runs of anything else collapsed to a single `-`, with none left
6020/// dangling at either end. `## Some Heading Here` → `some-heading-here`.
6021///
6022/// The rule every Markdown renderer follows, and applied to djot's own auto-ids
6023/// too so that `#some-heading-here` and `#Some-Heading-Here` are one question.
6024/// Unicode-aware (`is_alphanumeric`, not an ASCII test), because a heading in
6025/// any other language is still a heading someone will link to. Underscores
6026/// survive for the same reason they do on the web: they are word characters
6027/// wherever identifiers are written.
6028fn slug(text: &str) -> String {
6029 let mut out = String::new();
6030 let mut pending = false;
6031 for c in text.chars() {
6032 if c.is_alphanumeric() || c == '_' {
6033 if pending && !out.is_empty() {
6034 out.push('-');
6035 }
6036 pending = false;
6037 out.extend(c.to_lowercase());
6038 } else {
6039 pending = true;
6040 }
6041 }
6042 out
6043}
6044
6045fn is_block_container(kind: &Kind) -> bool {
6046 matches!(
6047 kind,
6048 Kind::Doc
6049 | Kind::Section
6050 | Kind::BlockQuote
6051 | Kind::BulletList
6052 | Kind::OrderedList
6053 | Kind::TaskList
6054 | Kind::ListItem
6055 | Kind::TaskListItem
6056 // Every `container` — a directive in any of its three forms, or a
6057 // promoted HTML element. A *text* directive is really inline, so
6058 // claiming it here is a small overreach, and the deliberate one this
6059 // function's kind-only peer `is_inline_kind` documents: the pair is
6060 // consulted together, and answering "block container" for something
6061 // inline is what keeps an ancestor walk from stopping short of the
6062 // paragraph that actually holds it.
6063 | Kind::Container
6064 )
6065}
6066
6067/// The `[start, end)` byte range of the source line containing `off` (newline
6068/// excluded) — the fallback when `off` sits outside any AST block (e.g. a blank
6069/// line between paragraphs).
6070fn source_line_range(s: &str, off: usize) -> std::ops::Range<usize> {
6071 let off = off.min(s.len());
6072 let start = s[..off].rfind('\n').map(|p| p + 1).unwrap_or(0);
6073 let end = s[off..].find('\n').map(|p| off + p).unwrap_or(s.len());
6074 start..end
6075}
6076
6077/// How many leading bytes an outdent takes off `line`: a whole indent level
6078/// where the line has one, and whatever it has where it has less.
6079///
6080/// A leading tab counts as a level on its own. It's indentation some other
6081/// editor wrote, and one tab is one level everywhere it came from — measuring it
6082/// in spaces it doesn't contain would leave it untouchable.
6083fn outdent_width(line: &str, unit: usize) -> usize {
6084 if line.starts_with('\t') {
6085 return 1;
6086 }
6087 line.bytes().take(unit).take_while(|b| *b == b' ').count()
6088}
6089
6090/// A list marker found at the head of a line, together with everything before it
6091/// that a sibling line has to repeat.
6092///
6093/// The three offsets differ only inside a block quote, where `> - b` opens with
6094/// a `> ` quote marker the line's own text doesn't own. Outside one they collapse:
6095/// `line_start == marker_start`, and `text` is the plain `" - "`.
6096#[derive(Clone, Debug)]
6097struct ListMarker {
6098 /// The line's first byte.
6099 line_start: usize,
6100 /// Where the marker proper begins, past any quote prefix. The offset to hand
6101 /// the AST: a quoted item's span opens at its bullet, not at the `>`.
6102 marker_start: usize,
6103 /// `line_start` through the marker's trailing space — quote prefix, indent
6104 /// and bullet together, which is what the next item's line opens with.
6105 text: String,
6106}
6107
6108impl ListMarker {
6109 /// Where the item's content starts — one past the marker's trailing space.
6110 fn content_start(&self) -> usize {
6111 self.line_start + self.text.len()
6112 }
6113}
6114
6115fn classify(c: char) -> Class {
6116 if c == '_' || c.is_alphanumeric() {
6117 Class::Word
6118 } else if c.is_whitespace() {
6119 Class::Space
6120 } else {
6121 Class::Other
6122 }
6123}
6124
6125/// The offset at the end of the next word to the right of `i` (⌥→ / Ctrl+→):
6126/// skip any leading separators, then consume the following word run.
6127fn next_word(s: &str, i: usize) -> usize {
6128 let mut off = i;
6129 let mut in_word = false;
6130 for c in s[i..].chars() {
6131 if classify(c) == Class::Word {
6132 in_word = true;
6133 } else if in_word {
6134 break;
6135 }
6136 off += c.len_utf8();
6137 }
6138 off
6139}
6140
6141/// The offset at the start of the word to the left of `i` (⌥← / Ctrl+←):
6142/// skip separators walking left, then consume the preceding word run.
6143fn prev_word(s: &str, i: usize) -> usize {
6144 let mut off = i;
6145 let mut in_word = false;
6146 for c in s[..i].chars().rev() {
6147 if classify(c) == Class::Word {
6148 in_word = true;
6149 } else if in_word {
6150 break;
6151 }
6152 off -= c.len_utf8();
6153 }
6154 off
6155}
6156
6157/// The `[start, end)` run of same-class characters surrounding `off` — the
6158/// word (or whitespace/punctuation run) a double-click selects. At end-of-text
6159/// the run ending there is used.
6160fn word_range_at(s: &str, off: usize) -> (usize, usize) {
6161 if s.is_empty() {
6162 return (0, 0);
6163 }
6164 let off = off.min(s.len());
6165 let reference = if off < s.len() {
6166 s[off..].chars().next()
6167 } else {
6168 s[..off].chars().next_back()
6169 };
6170 let Some(rc) = reference else {
6171 return (off, off);
6172 };
6173 let class = classify(rc);
6174
6175 let mut start = off;
6176 for c in s[..start].chars().rev() {
6177 if classify(c) == class {
6178 start -= c.len_utf8();
6179 } else {
6180 break;
6181 }
6182 }
6183 let mut end = off;
6184 for c in s[end..].chars() {
6185 if classify(c) == class {
6186 end += c.len_utf8();
6187 } else {
6188 break;
6189 }
6190 }
6191 (start, end)
6192}
6193
6194/// `(row, col)` of byte offset `off`, `col` counted in *display columns* from
6195/// the line's start — terminal cells, not characters, so the column names the
6196/// cell the caret is drawn in even on a line of `你好` or emoji.
6197fn offset_to_row_col(s: &str, off: usize) -> (usize, usize) {
6198 let off = off.min(s.len());
6199 let mut row = 0;
6200 let mut line_start = 0;
6201 for (i, &b) in s.as_bytes().iter().enumerate() {
6202 if i >= off {
6203 break;
6204 }
6205 if b == b'\n' {
6206 row += 1;
6207 line_start = i + 1;
6208 }
6209 }
6210 (row, wysiwyg::text_width(&s[line_start..off]))
6211}
6212
6213/// The byte offset at display column `col` of `row` (clamped to that line's
6214/// end) — the inverse of [`offset_to_row_col`], which it has to agree with.
6215///
6216/// A column landing *inside* a character — the second cell of `你`, or any cell
6217/// but the first of an emoji — resolves to that character's start, which is the
6218/// column the caret would have been drawn at to begin with. So both cells of a
6219/// wide character mean the character, and every offset survives the round trip
6220/// out to a column and back. The walk steps by grapheme cluster for the same
6221/// reason the caret does: a cluster is the character, and the cells belong to it
6222/// rather than to the codepoints spelling it.
6223fn row_col_to_offset(s: &str, row: usize, col: usize) -> usize {
6224 let start = line_start(s, row);
6225 let end = line_end_from(s, start);
6226 let mut off = start;
6227 let mut at = 0; // the display column `off` sits at
6228 while off < end {
6229 let next = next_boundary(s, off).min(end);
6230 let cells = wysiwyg::text_width(&s[off..next]);
6231 if at + cells > col {
6232 break; // `col` is one of this cluster's own cells
6233 }
6234 at += cells;
6235 off = next;
6236 }
6237 off
6238}
6239
6240fn line_start(s: &str, row: usize) -> usize {
6241 if row == 0 {
6242 return 0;
6243 }
6244 let mut r = 0;
6245 for (i, &b) in s.as_bytes().iter().enumerate() {
6246 if b == b'\n' {
6247 r += 1;
6248 if r == row {
6249 return i + 1;
6250 }
6251 }
6252 }
6253 s.len()
6254}
6255
6256fn line_end_from(s: &str, start: usize) -> usize {
6257 s[start..].find('\n').map(|p| start + p).unwrap_or(s.len())
6258}
6259
6260/// twig's node-kind name for an inline mark, back to the [`InlineKind`] a
6261/// frontend names when it calls [`Doc::toggle`] — the inverse of the mapping
6262/// twig applies writing the mark out, so the toolbar can light the same button
6263/// that made the node.
6264///
6265/// `None` for every other kind, including the inline nodes that aren't marks at
6266/// all (`str`, `link`, `image`, the math and break kinds): they're things a
6267/// caret stands in, not formatting a button toggles.
6268fn inline_kind(kind: &Kind) -> Option<InlineKind> {
6269 Some(match kind {
6270 Kind::Strong => InlineKind::Strong,
6271 Kind::Emph => InlineKind::Emph,
6272 Kind::Verbatim => InlineKind::Verbatim,
6273 Kind::Mark => InlineKind::Mark,
6274 Kind::Superscript => InlineKind::Superscript,
6275 Kind::Subscript => InlineKind::Subscript,
6276 Kind::Insert => InlineKind::Insert,
6277 Kind::Delete => InlineKind::Delete,
6278 _ => return None,
6279 })
6280}
6281
6282/// A watermark for a file's contents (see `Doc::disk_hash`).
6283///
6284/// `DefaultHasher` is not stable across Rust releases, which doesn't matter: a
6285/// watermark is compared only against one taken by the same process moments
6286/// earlier, and never outlives it. 64 bits leaves a collision — an external edit
6287/// that hashes to exactly what leaf wrote — at odds no filesystem race gets near.
6288fn hash_bytes(bytes: &[u8]) -> u64 {
6289 use std::hash::{Hash, Hasher};
6290 let mut h = std::collections::hash_map::DefaultHasher::new();
6291 bytes.hash(&mut h);
6292 h.finish()
6293}
6294
6295#[cfg(feature = "fs")]
6296fn detect_format(path: &Path) -> Result<Format> {
6297 let ext = path
6298 .extension()
6299 .and_then(|e| e.to_str())
6300 .unwrap_or("")
6301 .to_ascii_lowercase();
6302 Ok(match ext.as_str() {
6303 "dj" | "djot" => Format::Djot,
6304 "md" | "markdown" => Format::Markdown,
6305 "xml" => Format::Xml,
6306 "html" | "htm" => Format::Html,
6307 other => return Err(anyhow!("unknown document extension: .{other}")),
6308 })
6309}
6310
6311#[cfg(test)]
6312mod tests {
6313 use super::*;
6314
6315 /// A document open in `view`. WYSIWYG motion reads the visual map, which the
6316 /// renderer stamps each frame, so the map is built here too — a WYSIWYG doc
6317 /// without one is a view no user is ever in.
6318 fn doc_in(view: View, name: &str, body: &str) -> Doc {
6319 // The fixture name doubles as the temp file's, so two tests picking the
6320 // same one raced under the parallel runner and read each other's body —
6321 // a green suite proving the wrong thing. The counter makes that
6322 // unreachable rather than asking every future caller to notice.
6323 static SEQ: std::sync::atomic::AtomicUsize = std::sync::atomic::AtomicUsize::new(0);
6324 let seq = SEQ.fetch_add(1, std::sync::atomic::Ordering::Relaxed);
6325 let mut p = std::env::temp_dir();
6326 p.push(format!("leaf_test_{name}_{seq}.md"));
6327 std::fs::write(&p, body).unwrap();
6328 let mut d = Doc::open(p).unwrap();
6329 d.view = view;
6330 if view == View::Wysiwyg {
6331 d.build_visual(80);
6332 }
6333 d
6334 }
6335
6336 // Source-view document for the source-behaviour tests. `Doc::open` now
6337 // defaults to WYSIWYG (leaf's default view), so pin the source view here;
6338 // `wysiwyg_doc` builds the rich-text variant on top of this.
6339 fn doc_with(name: &str, body: &str) -> Doc {
6340 doc_in(View::Source, name, body)
6341 }
6342
6343 /// Every visual row's drawn text — what the reader actually sees, which is
6344 /// the only thing the reveal preference is supposed to change.
6345 fn drawn_rows(d: &Doc) -> Vec<String> {
6346 d.vmap
6347 .rows
6348 .iter()
6349 .map(|r| r.glyphs.iter().map(|g| g.ch).collect())
6350 .collect()
6351 }
6352
6353 /// Put the caret at the first byte of `needle` and rebuild, so the row under
6354 /// it becomes the revealed line.
6355 fn caret_at(d: &mut Doc, needle: &str) {
6356 d.caret = d.source.find(needle).expect("needle in source");
6357 d.build_visual(80);
6358 }
6359
6360 #[test]
6361 fn blockquote_after_a_list_is_not_bulleted() {
6362 // twig nests a following top-level block quote under the `bullet_list`
6363 // (a direct child, not a `list_item`). The map must render it de-nested —
6364 // `│ quote`, never `• │ quote` — with a blank separator, like any block
6365 // that follows a list. Regression for the "combined list + blockquote" bug.
6366 let mut d = doc_in(View::Wysiwyg, "bq_after_list", "- item\n\n> quote\n");
6367 d.build_visual(80);
6368 let rows: Vec<String> = d
6369 .vmap
6370 .rows
6371 .iter()
6372 .map(|r| r.glyphs.iter().map(|g| g.ch).collect())
6373 .collect();
6374 assert!(
6375 rows.iter().any(|r| r == "│ quote"),
6376 "block quote should render on its own gutter, got rows: {rows:?}"
6377 );
6378 assert!(
6379 !rows.iter().any(|r| r.contains('•') && r.contains('│')),
6380 "no row should carry both a bullet and a quote gutter, got rows: {rows:?}"
6381 );
6382 }
6383
6384 // ── the map is built at most once per (revision, wrap) ───────────────────
6385 //
6386 // A frontend repaints for reasons that have nothing to do with the text — a
6387 // blinking caret, a scroll — and rebuilding the map is O(document). These
6388 // pin *that the cache fires*, which a passing suite can't tell you: a cache
6389 // that never hits is invisible to every other test in this file.
6390 //
6391 // The probe is to wreck the built map and ask for it again. A rebuild
6392 // repairs it; a cache hit hands the wreckage straight back. Nothing else
6393 // can distinguish the two from outside.
6394
6395 #[test]
6396 fn a_rebuild_with_nothing_changed_reuses_the_map() {
6397 let mut d = doc_in(View::Wysiwyg, "cache_hit", "# Title\n\nbody\n");
6398 d.build_visual(80);
6399 assert!(!d.vmap.rows.is_empty());
6400 d.vmap.rows.clear(); // wreck it
6401 d.build_visual(80);
6402 assert!(
6403 d.vmap.rows.is_empty(),
6404 "the map was rebuilt though nothing changed — the cache never fired"
6405 );
6406 }
6407
6408 #[test]
6409 fn an_edit_rebuilds_the_map() {
6410 let mut d = doc_in(View::Wysiwyg, "cache_edit", "# Title\n\nbody\n");
6411 d.build_visual(80);
6412 let before = d.revision();
6413 d.vmap.rows.clear();
6414 d.insert("x");
6415 d.build_visual(80);
6416 assert!(d.revision() > before, "an edit must move the revision");
6417 assert!(
6418 !d.vmap.rows.is_empty(),
6419 "an edited document must not paint from a stale map"
6420 );
6421 }
6422
6423 #[test]
6424 fn a_width_change_rebuilds_the_map() {
6425 // The map is a function of the wrap width too, so a resize is a miss
6426 // even though the text is untouched.
6427 let mut d = doc_in(
6428 View::Wysiwyg,
6429 "cache_width",
6430 "one two three four five six\n",
6431 );
6432 d.build_visual(80);
6433 d.vmap.rows.clear();
6434 d.build_visual(12);
6435 assert!(!d.vmap.rows.is_empty(), "a resize must rebuild the map");
6436 // And the unwrapped map is its own key, not the same as any width.
6437 d.vmap.rows.clear();
6438 d.build_visual_unwrapped();
6439 assert!(!d.vmap.rows.is_empty(), "unwrapped is a different map");
6440 }
6441
6442 #[test]
6443 fn a_motion_does_not_rebuild_the_map() {
6444 // The whole point: moving the caret changes nothing the map is built
6445 // from. If a motion bumped the revision, every arrow key would cost a
6446 // full rebuild and the cache would be worthless.
6447 let mut d = doc_in(View::Wysiwyg, "cache_motion", "# Title\n\nbody text\n");
6448 d.build_visual(80);
6449 let rev = d.revision();
6450 d.move_right(false);
6451 d.move_right(true);
6452 d.move_down(false);
6453 assert_eq!(d.revision(), rev, "a motion must not move the revision");
6454 d.vmap.rows.clear();
6455 d.build_visual(80);
6456 assert!(
6457 d.vmap.rows.is_empty(),
6458 "a motion should not rebuild the map"
6459 );
6460 }
6461
6462 #[test]
6463 fn saving_does_not_rebuild_the_map() {
6464 // Saving changes `dirty`, not the text.
6465 let mut d = doc_in(View::Wysiwyg, "cache_save", "# Title\n\nbody\n");
6466 d.insert("x");
6467 d.build_visual(80);
6468 let rev = d.revision();
6469 d.save();
6470 assert_eq!(d.revision(), rev, "a save must not move the revision");
6471 assert!(!d.dirty, "the save should have cleaned the document");
6472 }
6473
6474 #[test]
6475 fn a_reload_rebuilds_the_map() {
6476 // Reload replaces the text without going through `refresh`, so it has to
6477 // move the revision itself — else the editor paints the old file.
6478 let mut d = doc_in(View::Wysiwyg, "cache_reload", "# Title\n\nbody\n");
6479 d.build_visual(80);
6480 let rev = d.revision();
6481 std::fs::write(&d.path, "# Other\n\nwholly new\n").unwrap();
6482 d.reload();
6483 assert!(d.revision() > rev, "a reload must move the revision");
6484 d.build_visual(80);
6485 let text: String = d
6486 .vmap
6487 .rows
6488 .iter()
6489 .flat_map(|r| r.glyphs.iter().map(|g| g.ch))
6490 .collect();
6491 assert!(
6492 text.contains("wholly new"),
6493 "the reloaded text should be on screen, got {text:?}"
6494 );
6495 }
6496
6497 // ── golden-case harness ──────────────────────────────────────────────────
6498 // The pattern the whole parity suite can reuse: write a fixture with the
6499 // caret marked by `|`, run one action, and compare the rendered result —
6500 // also caret-marked — against the expected string. One readable line per
6501 // behavior, and it exercises the exact `Doc` ops both frontends call.
6502
6503 /// Split a `|`-marked fixture into `(source, caret_offset)`.
6504 fn parse_caret(marked: &str) -> (String, usize) {
6505 let caret = marked.find('|').expect("fixture needs a `|` caret marker");
6506 (marked.replacen('|', "", 1), caret)
6507 }
6508
6509 /// Render a doc's source with `|` at the caret (and `[`…`]` around any
6510 /// selection) so a result reads like the fixtures.
6511 fn render_caret(d: &Doc) -> String {
6512 // (offset, rank, char); rank keeps coincident markers ordered `[ | ]`
6513 // so the caret always renders inside its own selection.
6514 let mut marks: Vec<(usize, u8, char)> = vec![(d.caret, 1, '|')];
6515 if let Some((s, e)) = d.selection() {
6516 marks.push((s, 0, '['));
6517 marks.push((e, 2, ']'));
6518 }
6519 // Insert right-to-left: descending offset, then descending rank.
6520 marks.sort_by(|a, b| b.0.cmp(&a.0).then(b.1.cmp(&a.1)));
6521 let mut out = d.source.clone();
6522 for (at, _, ch) in marks {
6523 out.insert(at, ch);
6524 }
6525 out
6526 }
6527
6528 /// Load a `|`-marked fixture, run `action`, return the caret-marked result.
6529 fn golden(name: &str, marked: &str, action: impl FnOnce(&mut Doc)) -> String {
6530 golden_in(View::Source, name, marked, action)
6531 }
6532
6533 /// [`golden`] in a chosen view — the editing ops are the view's to share, so
6534 /// the same fixture has to read the same way in both.
6535 fn golden_in(view: View, name: &str, marked: &str, action: impl FnOnce(&mut Doc)) -> String {
6536 let (src, caret) = parse_caret(marked);
6537 let mut d = doc_in(view, name, &src);
6538 d.caret = caret;
6539 action(&mut d);
6540 render_caret(&d)
6541 }
6542
6543 #[test]
6544 fn word_motion_walks_word_by_word() {
6545 let g = |m, f: fn(&mut Doc)| golden("word_motion", m, f);
6546 assert_eq!(
6547 g("hello wor|ld", |d| d.move_word_left(false)),
6548 "hello |world"
6549 );
6550 assert_eq!(
6551 g("hello| world", |d| d.move_word_left(false)),
6552 "|hello world"
6553 );
6554 assert_eq!(
6555 g("hel|lo world", |d| d.move_word_right(false)),
6556 "hello| world"
6557 );
6558 assert_eq!(
6559 g("hello| world", |d| d.move_word_right(false)),
6560 "hello world|"
6561 );
6562 // Punctuation is its own class, so motion stops at the boundary.
6563 assert_eq!(g("|foo.bar", |d| d.move_word_right(false)), "foo|.bar");
6564 }
6565
6566 #[test]
6567 fn word_motion_extends_the_selection_when_asked() {
6568 assert_eq!(
6569 golden("word_sel", "hello |world", |d| d.move_word_right(true)),
6570 "hello [world|]"
6571 );
6572 }
6573
6574 #[test]
6575 fn delete_word_removes_a_whole_word() {
6576 let g = |m, f: fn(&mut Doc)| golden("del_word", m, f);
6577 assert_eq!(g("hello world|", |d| d.delete_word_back()), "hello |");
6578 assert_eq!(g("hello |world", |d| d.delete_word_forward()), "hello |");
6579 assert_eq!(g("foo |bar baz", |d| d.delete_word_back()), "|bar baz");
6580 }
6581
6582 // ── Home / End ───────────────────────────────────────────────────────────
6583
6584 #[test]
6585 fn home_toggles_between_the_line_s_text_and_its_margin() {
6586 // Source: the indentation is what the toggle is for. WYSIWYG resolves an
6587 // indent to the markup it spells everywhere it means one, so the fixture
6588 // with whitespace left to walk is a code block, which is verbatim.
6589 let g = |m, f: fn(&mut Doc)| golden("smart_home", m, f);
6590 assert_eq!(g(" inden|ted", |d| d.move_home(false)), " |indented");
6591 assert_eq!(g(" |indented", |d| d.move_home(false)), "| indented");
6592 assert_eq!(g("| indented", |d| d.move_home(false)), " |indented");
6593 // A line with no indentation has one place to go, so the toggle is a
6594 // no-op rather than a trip to nowhere.
6595 assert_eq!(g("hel|lo", |d| d.move_home(false)), "|hello");
6596 assert_eq!(g("|hello", |d| d.move_home(false)), "|hello");
6597
6598 let mut d = wysiwyg_doc("smart_home_wys", "```\n indented\n```\n");
6599 let indent = d.source.find(" indented").unwrap();
6600 d.caret = indent + 6; // inside "indented"
6601 d.move_home(false);
6602 assert_eq!(
6603 d.caret,
6604 indent + 4,
6605 "wysiwyg: Home aims at the code line's text"
6606 );
6607 d.move_home(false);
6608 assert_eq!(
6609 d.caret, indent,
6610 "wysiwyg: the second press takes the indent"
6611 );
6612 d.move_home(false);
6613 assert_eq!(d.caret, indent + 4, "wysiwyg: the toggle swaps back");
6614 }
6615
6616 #[test]
6617 fn end_takes_the_line_the_view_is_showing() {
6618 // The line differs by view for the same document, and that is the point:
6619 // a bare newline inside a paragraph is a soft break, which WYSIWYG draws
6620 // as a space on one row and the source view as two lines.
6621 let mut d = doc_with("end_src", "one two\nthree\n");
6622 d.caret = 1;
6623 d.move_end(false);
6624 assert_eq!(d.caret, 7, "source: the end of the source line");
6625
6626 let mut d = wysiwyg_doc("end_wys", "one two\nthree\n");
6627 d.caret = 1;
6628 d.move_end(false);
6629 assert_eq!(
6630 d.caret, 13,
6631 "wysiwyg: the end of the row, soft break and all"
6632 );
6633 }
6634
6635 #[test]
6636 fn home_and_end_extend_the_selection_when_asked() {
6637 for (view, tag) in VIEWS {
6638 let mut d = doc_in(view, &format!("home_end_ext_{tag}"), "hello world");
6639 d.caret = 6;
6640 d.move_end(true);
6641 assert_eq!(d.selection(), Some((6, 11)), "{tag}: End extends");
6642 let mut d = doc_in(view, &format!("home_ext_{tag}"), "hello world");
6643 d.caret = 6;
6644 d.move_home(true);
6645 assert_eq!(d.selection(), Some((0, 6)), "{tag}: Home extends");
6646 }
6647 }
6648
6649 // ── kill to the line's start / end ───────────────────────────────────────
6650
6651 #[test]
6652 fn kill_to_the_line_start_and_end_in_both_views() {
6653 for (view, tag) in VIEWS {
6654 // The gap that reads as a paragraph break in each view: the source
6655 // view's lines are the renderer's rows only where the source says so.
6656 let gap = if view == View::Source { "\n" } else { "\n\n" };
6657 let mut d = doc_in(
6658 view,
6659 &format!("kill_end_{tag}"),
6660 &format!("one two{gap}three\n"),
6661 );
6662 d.caret = 3;
6663 d.delete_to_line_end();
6664 assert_eq!(
6665 d.source,
6666 format!("one{gap}three\n"),
6667 "{tag}: ^K to the line's end"
6668 );
6669 assert_eq!(d.caret, 3, "{tag}: the caret stays where it kills from");
6670
6671 let mut d = doc_in(
6672 view,
6673 &format!("kill_start_{tag}"),
6674 &format!("one two{gap}three\n"),
6675 );
6676 d.caret = 7; // the end of the first line
6677 d.delete_to_line_start();
6678 assert_eq!(
6679 d.source,
6680 format!("{gap}three\n"),
6681 "{tag}: ⌘⌫ to the line's start"
6682 );
6683 assert_eq!(d.caret, 0, "{tag}");
6684 }
6685 }
6686
6687 #[test]
6688 fn a_kill_at_the_line_s_edge_leaves_the_lines_joined() {
6689 // The decision: at the boundary both kills do nothing, rather than
6690 // eating the line break. "Line" is the view's own — in WYSIWYG it ends
6691 // at a soft wrap as often as at a newline, where there is nothing
6692 // written to delete — and a source newline is only half of the blank
6693 // line between two paragraphs, so taking it leaves a soft break rather
6694 // than the join it looks like. Backspace and Delete are the keys for it.
6695 for (view, tag) in VIEWS {
6696 let gap = if view == View::Source { "\n" } else { "\n\n" };
6697 let src = format!("one{gap}three\n");
6698 let mut d = doc_in(view, &format!("kill_edge_end_{tag}"), &src);
6699 d.caret = 3; // the end of "one"
6700 d.delete_to_line_end();
6701 assert_eq!(
6702 d.source, src,
6703 "{tag}: ^K at the line's end joined it to the next"
6704 );
6705
6706 let mut d = doc_in(view, &format!("kill_edge_start_{tag}"), &src);
6707 d.caret = 3 + gap.len(); // the start of "three"
6708 d.delete_to_line_start();
6709 assert_eq!(
6710 d.source, src,
6711 "{tag}: ⌘⌫ at the line's start joined it to the last"
6712 );
6713 }
6714 }
6715
6716 #[test]
6717 fn a_kill_takes_the_selection_when_there_is_one() {
6718 // What every other delete here does with one, so these two as well.
6719 for (view, tag) in VIEWS {
6720 for (name, kill) in [
6721 (
6722 "end",
6723 (|d: &mut Doc| d.delete_to_line_end()) as fn(&mut Doc),
6724 ),
6725 ("start", |d: &mut Doc| d.delete_to_line_start()),
6726 ] {
6727 let mut d = doc_in(view, &format!("kill_sel_{name}_{tag}"), "one two three\n");
6728 d.anchor = Some(4);
6729 d.caret = 7; // "two"
6730 kill(&mut d);
6731 assert_eq!(
6732 d.source, "one three\n",
6733 "{tag}: {name} ignored the selection"
6734 );
6735 assert_eq!(d.selection(), None, "{tag}: {name}");
6736 }
6737 }
6738 }
6739
6740 #[test]
6741 fn a_kill_takes_the_markup_it_empties_with_it() {
6742 // The same hazard a word-delete has: a WYSIWYG range covers what the
6743 // user can see, which for `**bold**` is the word and never the
6744 // delimiters, so a kill that stopped at the text would leave `a ****` —
6745 // markup wrapped around nothing.
6746 let mut d = wysiwyg_doc("kill_widen", "a **bold**\n");
6747 d.caret = d.source.find("bold").unwrap();
6748 d.delete_to_line_end();
6749 assert_eq!(d.source, "a \n");
6750 }
6751
6752 #[test]
6753 fn a_kill_is_undone_in_one_step() {
6754 for (view, tag) in VIEWS {
6755 let mut d = doc_in(view, &format!("kill_undo_{tag}"), "one two three\n");
6756 d.caret = 3;
6757 d.delete_to_line_end();
6758 assert_eq!(d.source, "one\n", "{tag}");
6759 d.undo();
6760 assert_eq!(d.source, "one two three\n", "{tag}: a kill takes one undo");
6761 }
6762 }
6763
6764 #[test]
6765 fn select_block_grabs_the_whole_paragraph_from_any_wrapped_row() {
6766 // Regression: triple-click used move_home/move_end over visual rows, so
6767 // it only worked on a paragraph's first row (a wrap-boundary offset maps
6768 // to the earlier row). select_block_at reads the AST, so every offset in
6769 // the paragraph selects the whole thing.
6770 let body = "one two three four five six seven eight\n";
6771 let mut d = doc_with("sel_block", body);
6772 d.view = View::Wysiwyg;
6773 d.build_visual(12); // force the paragraph to wrap into several rows
6774 assert!(d.vmap.num_rows() > 1, "test needs a wrapped paragraph");
6775 let para = (0, "one two three four five six seven eight".len());
6776 for off in [0usize, 8, 19, 28, 38] {
6777 d.caret = 0;
6778 d.anchor = None;
6779 d.select_block_at(off);
6780 assert_eq!(
6781 d.selection(),
6782 Some(para),
6783 "offset {off} should select the paragraph"
6784 );
6785 }
6786 }
6787
6788 #[test]
6789 fn select_block_uses_content_span_for_a_heading() {
6790 let mut d = doc_with("sel_head", "# Title\n\nbody\n");
6791 d.select_block_at(4); // inside "Title"
6792 // content_span excludes the "# " marker.
6793 assert_eq!(d.selected_text(), Some("Title"));
6794 d.select_block_at(10); // inside "body"
6795 assert_eq!(d.selected_text(), Some("body"));
6796 }
6797
6798 #[test]
6799 fn select_all_spans_the_document() {
6800 let mut d = doc_with("sel_all", "abc\n\ndef\n");
6801 d.select_all();
6802 assert_eq!(d.selection(), Some((0, d.source.len())));
6803 }
6804
6805 #[test]
6806 fn select_word_at_picks_the_surrounding_word() {
6807 let mut d = doc_with("sel_word", "hello world\n");
6808 d.select_word_at(8); // inside "world"
6809 assert_eq!(d.selection(), Some((6, 11)));
6810 // Double-clicking at end-of-word still grabs the word to its left.
6811 d.select_word_at(5); // the space between the words
6812 assert_eq!(d.selection(), Some((5, 6)));
6813 }
6814
6815 #[test]
6816 fn word_helpers_respect_utf8_boundaries() {
6817 // "café" is 5 bytes ('é' is two); motion must land on char boundaries.
6818 assert_eq!(
6819 golden("utf8", "|café ok", |d| d.move_word_right(false)),
6820 "café| ok"
6821 );
6822 assert_eq!(golden("utf8b", "café |ok", |d| d.delete_word_back()), "|ok");
6823 }
6824
6825 #[test]
6826 fn typing_inserts_at_the_caret_and_advances_it() {
6827 let mut d = doc_with("type", "hello\n");
6828 d.insert("Hi ");
6829 assert_eq!(d.source, "Hi hello\n");
6830 assert_eq!(d.caret, 3);
6831 assert!(d.dirty);
6832 }
6833
6834 #[test]
6835 fn backspace_deletes_the_char_before_the_caret() {
6836 let mut d = doc_with("bs", "hello\n");
6837 d.caret = 3; // after "hel"
6838 d.backspace();
6839 assert_eq!(d.source, "helo\n");
6840 assert_eq!(d.caret, 2);
6841 }
6842
6843 #[test]
6844 fn typing_replaces_the_selection() {
6845 let mut d = doc_with("replace", "a word b\n");
6846 d.anchor = Some(2);
6847 d.caret = 6; // "word" selected
6848 d.insert("X");
6849 assert_eq!(d.source, "a X b\n");
6850 assert_eq!(d.caret, 3);
6851 assert_eq!(d.anchor, None);
6852 }
6853
6854 #[test]
6855 fn toggle_bold_wraps_then_unwraps_the_selection() {
6856 let mut d = doc_with("bold", "a word b\n");
6857 d.anchor = Some(2);
6858 d.caret = 6;
6859 d.toggle(InlineKind::Strong);
6860 assert_eq!(d.source, "a **word** b\n");
6861 // The toggled region stays selected, so a second toggle reverses it.
6862 d.toggle(InlineKind::Strong);
6863 assert_eq!(d.source, "a word b\n");
6864 d.toggle(InlineKind::Strong);
6865 assert_eq!(d.source, "a **word** b\n");
6866 }
6867
6868 #[test]
6869 fn toggle_code_wraps_then_unwraps_the_selection() {
6870 let mut d = doc_with("code_rt", "a word b\n");
6871 d.anchor = Some(2);
6872 d.caret = 6;
6873 d.toggle(InlineKind::Verbatim);
6874 assert_eq!(d.source, "a `word` b\n");
6875 d.toggle(InlineKind::Verbatim);
6876 assert_eq!(d.source, "a word b\n");
6877 }
6878
6879 #[test]
6880 fn sticky_bold_with_no_selection_wraps_the_next_typed_text() {
6881 // ⌘b at a bare caret, then type: the text comes out bold with no
6882 // selection ever made — the word-processor "start bold here" gesture.
6883 let mut d = doc_with("sticky_wrap", "xy\n");
6884 d.caret = 1; // between x and y
6885 d.toggle(InlineKind::Strong);
6886 assert_eq!(d.source, "xy\n", "arming a mark must not edit the document");
6887 d.insert("A");
6888 assert_eq!(d.source, "x**A**y\n");
6889 }
6890
6891 #[test]
6892 fn sticky_bold_lights_the_toolbar_before_any_typing() {
6893 // The button must light the instant ⌘b is pressed, or the mode is
6894 // invisible until the first character lands.
6895 let mut d = doc_with("sticky_light", "xy\n");
6896 d.caret = 1;
6897 assert!(!d.active_inline_marks().contains(InlineKind::Strong));
6898 d.toggle(InlineKind::Strong);
6899 assert!(d.active_inline_marks().contains(InlineKind::Strong));
6900 }
6901
6902 #[test]
6903 fn sticky_bold_toggled_off_types_normally_again() {
6904 // ⌘b, type, ⌘b, type: the first run is bold, the second is not — all
6905 // in the flow of typing, the exact sequence the user described.
6906 let mut d = doc_with("sticky_off", "\n");
6907 d.caret = 0;
6908 d.toggle(InlineKind::Strong);
6909 d.insert("a");
6910 d.insert("b"); // continues inside the run, no re-arming
6911 assert_eq!(d.source, "**ab**\n");
6912 d.toggle(InlineKind::Strong); // ⌘b again — shed bold
6913 d.insert("c");
6914 assert_eq!(d.source, "**ab**c\n");
6915 }
6916
6917 #[test]
6918 fn continued_typing_after_a_sticky_run_stays_in_the_run() {
6919 // Once a mark is realised the caret sits inside the run, so plain typing
6920 // extends it rather than starting a second, adjacent bold span.
6921 let mut d = doc_with("sticky_cont", "\n");
6922 d.caret = 0;
6923 d.toggle(InlineKind::Emph);
6924 d.insert("h");
6925 d.insert("i");
6926 assert_eq!(d.source, "*hi*\n");
6927 }
6928
6929 #[test]
6930 fn moving_the_caret_disarms_a_sticky_mark() {
6931 // Arming a mark and then moving away must not style text elsewhere.
6932 let mut d = doc_with("sticky_disarm", "xy\n");
6933 d.caret = 0;
6934 d.toggle(InlineKind::Strong);
6935 d.move_right(false); // caret 0 → 1, disarms
6936 assert!(!d.active_inline_marks().contains(InlineKind::Strong));
6937 d.insert("A");
6938 assert_eq!(d.source, "xAy\n", "the mark must not follow the caret");
6939 }
6940
6941 #[test]
6942 fn stacked_sticky_marks_apply_together() {
6943 // ⌘b then ⌘i before typing: the text comes out both bold and italic.
6944 let mut d = doc_with("sticky_stack", "\n");
6945 d.caret = 0;
6946 d.toggle(InlineKind::Strong);
6947 d.toggle(InlineKind::Emph);
6948 d.insert("x");
6949 // Land the caret on the styled character and confirm both marks are live.
6950 d.anchor = Some(d.source.find('x').unwrap());
6951 d.caret = d.anchor.unwrap() + 1;
6952 let marks = d.active_inline_marks();
6953 assert!(marks.contains(InlineKind::Strong), "bold: {}", d.source);
6954 assert!(marks.contains(InlineKind::Emph), "italic: {}", d.source);
6955 }
6956
6957 // ── the mark-edge rule (see `Doc::splice`) ───────────────────────────────
6958
6959 #[test]
6960 fn a_space_typed_in_a_bold_run_never_leaves_the_delimiters_showing() {
6961 // The reported bug, keystroke for keystroke: ⌘b, "bold", space, "hey".
6962 // The space inside the run made `**bold **`, which is *not* bold — four
6963 // literal asterisks — so the rich view drew them, correctly and
6964 // uselessly, until the next character happened to close the run again.
6965 let mut d = wysiwyg_doc("edge_typing", "a \n");
6966 d.caret = 2;
6967 d.toggle(InlineKind::Strong);
6968 for c in "bold".chars() {
6969 d.insert(&c.to_string());
6970 }
6971 assert_eq!(d.source, "a **bold**\n");
6972 d.insert(" ");
6973 assert_eq!(
6974 d.source, "a **bold** \n",
6975 "the space belongs outside the run"
6976 );
6977 assert!(
6978 d.active_inline_marks().contains(InlineKind::Strong),
6979 "bold is still what's being typed, so the button stays lit"
6980 );
6981 // What the writer is looking at while all this happens: their words.
6982 d.build_visual(80);
6983 let drawn: String = d.vmap.rows[0].glyphs.iter().map(|g| g.ch).collect();
6984 assert_eq!(drawn, "a bold ", "no delimiter ever surfaces: {}", d.source);
6985 for c in "hey".chars() {
6986 d.insert(&c.to_string());
6987 }
6988 assert_eq!(
6989 d.source, "a **bold hey**\n",
6990 "one bold phrase, not two runs"
6991 );
6992 }
6993
6994 #[test]
6995 fn typing_past_a_space_can_still_leave_the_bold_behind() {
6996 // The other half: the marks stay armed across the space, so ⌘b turns
6997 // them off again there and the next word is plain — the run isn't
6998 // rejoined by a caret that was told not to.
6999 let mut d = wysiwyg_doc("edge_shed", "\n");
7000 d.caret = 0;
7001 d.toggle(InlineKind::Strong);
7002 for c in "bold ".chars() {
7003 d.insert(&c.to_string());
7004 }
7005 assert_eq!(d.source, "**bold** \n");
7006 d.toggle(InlineKind::Strong);
7007 assert!(!d.active_inline_marks().contains(InlineKind::Strong));
7008 d.insert("x");
7009 assert_eq!(d.source, "**bold** x\n");
7010 }
7011
7012 #[test]
7013 fn a_space_typed_first_of_all_still_leaves_the_mark_armed() {
7014 // ⌘b and then a space before any word: the space is not marked (nothing
7015 // is), and the word after it is.
7016 let mut d = wysiwyg_doc("edge_space_first", "a\n");
7017 d.caret = 1;
7018 d.toggle(InlineKind::Strong);
7019 d.insert(" ");
7020 assert_eq!(d.source, "a \n");
7021 assert!(d.active_inline_marks().contains(InlineKind::Strong));
7022 d.insert("b");
7023 assert_eq!(d.source, "a **b**\n");
7024 }
7025
7026 #[test]
7027 fn a_space_typed_at_either_edge_of_an_existing_mark_steps_outside_it() {
7028 let mut d = wysiwyg_doc("edge_tail", "x **bold**\n");
7029 d.caret = 8; // the caret's home at the end of the run's text
7030 d.insert(" ");
7031 assert_eq!(
7032 d.source, "x **bold** \n",
7033 "the space lands past the delimiters"
7034 );
7035 assert_eq!(d.caret, 11, "and the caret stands past it, outside the run");
7036
7037 let mut d = wysiwyg_doc("edge_head", "x **bold** y\n");
7038 d.caret = 4; // in front of the "b"
7039 d.insert(" ");
7040 assert_eq!(d.source, "x **bold** y\n");
7041 assert_eq!(d.caret, 3, "in front of the run, where the space was typed");
7042 }
7043
7044 #[test]
7045 fn a_delete_that_backs_a_space_onto_a_delimiter_moves_the_delimiter() {
7046 // Backspace over the last letter of a bold phrase.
7047 let mut d = wysiwyg_doc("edge_bksp", "a **bold h**\n");
7048 d.caret = 10; // past the "h"
7049 d.backspace();
7050 assert_eq!(d.source, "a **bold** \n");
7051 assert_eq!(d.caret, 11, "the caret keeps the place on screen it had");
7052 assert!(d.active_inline_marks().contains(InlineKind::Strong));
7053 d.insert("x");
7054 assert_eq!(d.source, "a **bold x**\n", "and typing rejoins the run");
7055 }
7056
7057 #[test]
7058 fn deleting_the_last_of_a_run_takes_its_delimiters_with_it() {
7059 // `**b**` with the `b` gone is `****`: two delimiters with nothing to
7060 // mark, which is only text. The marks live on in the caret instead.
7061 let mut d = wysiwyg_doc("edge_empty", "a **b** c\n");
7062 d.caret = 5;
7063 d.backspace();
7064 assert_eq!(d.source, "a c\n");
7065 assert!(d.active_inline_marks().contains(InlineKind::Strong));
7066 d.insert("x");
7067 assert_eq!(d.source, "a **x** c\n");
7068 }
7069
7070 #[test]
7071 fn typing_over_a_whole_bold_word_keeps_it_bold() {
7072 let mut d = wysiwyg_doc("edge_replace", "a **bold** c\n");
7073 d.anchor = Some(4);
7074 d.caret = 8; // the word, not its delimiters
7075 d.insert("x");
7076 assert_eq!(d.source, "a **x** c\n");
7077 }
7078
7079 #[test]
7080 fn a_code_span_keeps_the_space_it_is_given() {
7081 // Backticks are not whitespace-sensitive the way `**` is: `` `code ` ``
7082 // is still verbatim, so nothing is re-spelt. The repair asks the parser
7083 // rather than a table of kinds, and this is the answer it gets.
7084 let mut d = wysiwyg_doc("edge_code", "a `code` c\n");
7085 d.caret = 7;
7086 d.insert(" ");
7087 assert_eq!(d.source, "a `code ` c\n");
7088 }
7089
7090 #[test]
7091 fn a_delete_from_a_runs_outer_edge_reaches_into_the_run() {
7092 // A run's closing delimiter has a caret home on each side of it, one
7093 // column apart on screen — and a plain ← off the space after a bold word
7094 // lands on the outer one. The character drawn behind the caret there is
7095 // still the last letter of the phrase, so that is what Backspace takes;
7096 // the byte behind it is a `*` nobody can see.
7097 let mut d = wysiwyg_doc("edge_outer_close", "**bold** x\n");
7098 d.caret = 9;
7099 d.move_left(false);
7100 assert_eq!(d.caret, 8, "← rests past the delimiters, not inside them");
7101 d.backspace();
7102 assert_eq!(
7103 d.source, "**bol** x\n",
7104 "a letter of the phrase, not its `*`"
7105 );
7106 assert_eq!(d.caret, 5);
7107
7108 // And the mirror in front of the opening delimiter, where Delete's
7109 // character is the first letter of the run.
7110 let mut d = wysiwyg_doc("edge_outer_open", "x**bold**\n");
7111 d.caret = 1;
7112 d.delete_forward();
7113 assert_eq!(d.source, "x**old**\n");
7114 assert_eq!(d.caret, 3, "inside the run, in front of what is left of it");
7115 }
7116
7117 #[test]
7118 fn a_delete_at_a_run_edge_never_eats_a_delimiter() {
7119 // The byte beside the caret at either edge of a bold word is a `*` the
7120 // rich view draws nothing for. Taking it is not the character delete the
7121 // key was pressed for — it unspells the run and puts a literal asterisk
7122 // on screen (`a *bold** c`). The visible character is the one that goes.
7123 let mut d = wysiwyg_doc("edge_open_bksp", "a **bold** c\n");
7124 d.caret = 4; // in front of the "b"
7125 d.backspace();
7126 assert_eq!(d.source, "a**bold** c\n", "the space goes, the run stands");
7127
7128 let mut d = wysiwyg_doc("edge_close_del", "a **bold** c\n");
7129 d.caret = 8; // past the "d"
7130 d.delete_forward();
7131 assert_eq!(d.source, "a **bold**c\n");
7132 assert_eq!(d.caret, 8, "and the caret stays inside the run");
7133 d.insert("x");
7134 assert_eq!(d.source, "a **boldx**c\n");
7135
7136 // A code span's backticks are hidden the same way, so they are covered
7137 // by the same rule and not by a list of kinds.
7138 let mut d = wysiwyg_doc("edge_open_code", "a `code` c\n");
7139 d.caret = 3;
7140 d.backspace();
7141 assert_eq!(d.source, "a`code` c\n");
7142 }
7143
7144 #[test]
7145 fn the_source_view_deletes_the_delimiter_byte_it_is_shown() {
7146 // The asterisks are on the screen there and the caret can stand between
7147 // them, so a delete takes exactly the byte it is aimed at.
7148 let mut d = doc_with("edge_open_src", "a **bold** c\n");
7149 d.caret = 4;
7150 d.backspace();
7151 assert_eq!(d.source, "a *bold** c\n");
7152
7153 let mut d = doc_with("edge_close_src", "a **bold** c\n");
7154 d.caret = 8;
7155 d.delete_forward();
7156 assert_eq!(d.source, "a **bold* c\n");
7157 }
7158
7159 #[test]
7160 fn backspacing_the_space_out_of_a_bold_phrase_leaves_the_caret_in_it() {
7161 // The reported bug, keystroke for keystroke: ⌘b, "bold", space, Backspace.
7162 // The space had stepped outside the run (the mark-edge rule), taking the
7163 // caret with it, so the delete put it back down on the far side of the
7164 // closing `**` — one place on screen, and the wrong side of it. Typing
7165 // came out plain and the toolbar went dark, with nothing to see.
7166 let mut d = wysiwyg_doc("edge_bksp_space", "\n");
7167 d.caret = 0;
7168 d.toggle(InlineKind::Strong);
7169 for c in "bold".chars() {
7170 d.insert(&c.to_string());
7171 }
7172 d.insert(" ");
7173 assert_eq!(d.source, "**bold** \n");
7174 d.backspace();
7175 assert_eq!(
7176 d.source, "**bold**\n",
7177 "the space goes, the delimiters stay"
7178 );
7179 assert_eq!(d.caret, 6, "and the caret comes back inside the run");
7180 assert!(
7181 d.active_inline_marks().contains(InlineKind::Strong),
7182 "so the button is still lit"
7183 );
7184 d.insert("x");
7185 assert_eq!(
7186 d.source, "**boldx**\n",
7187 "and the next character is still bold"
7188 );
7189 }
7190
7191 #[test]
7192 fn a_second_backspace_there_deletes_a_letter_of_the_phrase() {
7193 // What the stranded caret did next: the byte behind it was the closing
7194 // `*`, so a second press took that instead of a letter — `**bold*`, the
7195 // styling gone and an asterisk on the screen where the word had been.
7196 let mut d = wysiwyg_doc("edge_bksp_twice", "\n");
7197 d.caret = 0;
7198 d.toggle(InlineKind::Strong);
7199 for c in "bold ".chars() {
7200 d.insert(&c.to_string());
7201 }
7202 assert_eq!(d.source, "**bold** \n");
7203 d.backspace();
7204 d.backspace();
7205 assert_eq!(d.source, "**bol**\n", "the delete lands inside the run");
7206 assert_eq!(d.caret, 5);
7207 }
7208
7209 #[test]
7210 fn a_delete_that_ends_at_a_nested_run_settles_inside_every_delimiter() {
7211 // `***both***` closes two runs with one stack of asterisks: the caret has
7212 // to walk in through all of them, or it lands between the emph and the
7213 // strong and types half-marked.
7214 let mut d = wysiwyg_doc("edge_bksp_nested", "***both*** \n");
7215 d.caret = 11;
7216 d.backspace();
7217 assert_eq!(d.source, "***both***\n");
7218 assert_eq!(d.caret, 7, "past the last letter, inside both runs");
7219 d.insert("x");
7220 assert_eq!(d.source, "***bothx***\n");
7221 }
7222
7223 #[test]
7224 fn a_delete_that_ends_mid_run_leaves_the_caret_where_it_fell() {
7225 // The settle only moves a caret a run actually closed over. Ordinary
7226 // deletes — inside a run, or in plain prose — are untouched.
7227 let mut d = wysiwyg_doc("edge_bksp_mid", "a **bold** c\n");
7228 d.caret = 8;
7229 d.backspace();
7230 assert_eq!(d.source, "a **bol** c\n");
7231 assert_eq!(d.caret, 7);
7232
7233 let mut d = wysiwyg_doc("edge_bksp_plain", "plain\n");
7234 d.caret = 5;
7235 d.backspace();
7236 assert_eq!(d.source, "plai\n");
7237 assert_eq!(d.caret, 4);
7238 }
7239
7240 #[test]
7241 fn the_source_view_leaves_a_delete_where_it_landed() {
7242 // The delimiters are on the screen there, so the offset past them is a
7243 // place the caret can be seen to be — nothing to settle.
7244 let mut d = doc_with("edge_bksp_src", "**bold** \n");
7245 d.caret = 9;
7246 d.backspace();
7247 assert_eq!(d.source, "**bold**\n");
7248 assert_eq!(d.caret, 8);
7249 }
7250
7251 #[test]
7252 fn the_mark_edge_rule_clears_every_delimiter_of_a_nested_run() {
7253 // `***both***` closes two runs with one stack of asterisks; a space that
7254 // clears only the inner one lands against the outer's and breaks that
7255 // instead.
7256 let mut d = wysiwyg_doc("edge_nested", "a ***both***\n");
7257 d.caret = 9;
7258 d.insert(" ");
7259 assert_eq!(d.source, "a ***both*** \n");
7260 assert_eq!(d.caret, 13);
7261 d.insert("x");
7262 assert_eq!(d.source, "a ***both x***\n");
7263 }
7264
7265 #[test]
7266 fn the_mark_edge_repair_undoes_with_the_keystroke_that_caused_it() {
7267 // The delimiter shuffle is not an edit the writer made, so it is not a
7268 // step they have to undo past.
7269 let mut d = wysiwyg_doc("edge_undo", "a **bold**\n");
7270 d.caret = 8;
7271 d.insert(" ");
7272 assert_eq!(d.source, "a **bold** \n");
7273 d.undo();
7274 assert_eq!(d.source, "a **bold**\n");
7275 }
7276
7277 #[test]
7278 fn the_source_view_types_the_space_where_it_was_asked_to() {
7279 // The rule is a rich-view courtesy. In the source view the delimiters are
7280 // on the screen and the user is editing the bytes they can see.
7281 let mut d = doc_with("edge_src", "a **bold** c\n");
7282 d.caret = 8;
7283 d.insert(" ");
7284 assert_eq!(d.source, "a **bold ** c\n");
7285 }
7286
7287 #[test]
7288 fn toggling_a_mark_over_a_selection_leaves_its_edge_whitespace_out() {
7289 // Double-clicking a word takes the space after it; bolding that must not
7290 // spell `**word **`, which is not bold at all.
7291 let mut d = wysiwyg_doc("edge_sel", "a word b\n");
7292 d.anchor = Some(2);
7293 d.caret = 7; // "word "
7294 d.toggle(InlineKind::Strong);
7295 assert_eq!(d.source, "a **word** b\n");
7296 d.toggle(InlineKind::Strong);
7297 assert_eq!(d.source, "a word b\n");
7298 d.toggle(InlineKind::Strong);
7299 assert_eq!(
7300 d.source, "a **word** b\n",
7301 "reapplying the mark must not wrap stale delimiter offsets"
7302 );
7303 // And a selection of nothing but whitespace has no word to mark.
7304 let mut d = wysiwyg_doc("edge_sel_ws", "a word b\n");
7305 d.anchor = Some(6);
7306 d.caret = 7;
7307 d.toggle(InlineKind::Strong);
7308 assert_eq!(d.source, "a word b\n");
7309 assert!(d.status.is_some());
7310 }
7311
7312 #[test]
7313 fn set_block_turns_a_paragraph_into_a_heading_at_the_caret() {
7314 let mut d = doc_with("head_set", "hello\n");
7315 d.caret = 2; // caret inside the paragraph, no selection
7316 d.set_block(BlockKind::Heading(1));
7317 assert_eq!(d.source, "# hello\n");
7318 }
7319
7320 #[test]
7321 fn set_block_heading_works_in_wysiwyg_view() {
7322 // The app defaults to WYSIWYG; the caret is a source offset either way.
7323 let mut d = wysiwyg_doc("head_wys", "hello\n");
7324 d.caret = 2;
7325 d.set_block(BlockKind::Heading(1));
7326 assert_eq!(d.source, "# hello\n");
7327 }
7328
7329 #[test]
7330 fn toggle_heading_applies_switches_and_reverts() {
7331 let mut d = doc_with("head_toggle", "hello\n");
7332 d.caret = 2;
7333 d.toggle_heading(1);
7334 assert_eq!(d.source, "# hello\n"); // paragraph → H1
7335 d.toggle_heading(2);
7336 assert_eq!(d.source, "## hello\n"); // H1 → H2 (different level switches)
7337 d.toggle_heading(2);
7338 assert_eq!(d.source, "hello\n"); // same level reverts to paragraph
7339 }
7340
7341 #[test]
7342 fn preserve_enter_at_a_line_end_lands_the_caret_on_the_new_blank_line() {
7343 // Regression: Enter at the end of a soft-break line (mid-paragraph) opened
7344 // the blank line but the caret rendered on the *next* line, because the
7345 // separator was a non-navigable decoration row. In Preserve flow that
7346 // blank line is a real caret home — the caret must resolve onto it, and
7347 // typing there makes the soft break that continues the paragraph.
7348 let src = "line one:\nsecond line\n";
7349 let mut d = wysiwyg_doc("pre_enter_lineend", src);
7350 d.set_line_flow(LineFlow::Preserve);
7351 d.build_visual_unwrapped(); // the GUI path (pixel-wrapped)
7352 d.caret = 9; // the visual end of row 0, at the soft-break '\n'
7353 d.newline();
7354 d.build_visual_unwrapped();
7355 assert_eq!(d.source, "line one:\n\nsecond line\n");
7356 assert_eq!(
7357 d.caret, 10,
7358 "caret sits on the new blank line, not the next line"
7359 );
7360 // The blank line is row 1, and the caret resolves onto it — not row 2.
7361 assert_eq!(
7362 d.vmap.pos_of_offset(10),
7363 (1, 0),
7364 "caret renders on the blank row"
7365 );
7366 assert!(
7367 !d.vmap.rows[1].decoration,
7368 "the blank line is navigable in Preserve"
7369 );
7370 // Typing there makes a soft break: one paragraph, three lines.
7371 d.insert("new clause,");
7372 assert_eq!(d.source, "line one:\nnew clause,\nsecond line\n");
7373 }
7374
7375 #[test]
7376 fn preserve_enter_makes_a_soft_break_not_a_paragraph() {
7377 // Mid-paragraph: Enter splits the line with a single `\n`, a soft break
7378 // that keeps it one paragraph — where Fold would open a second paragraph.
7379 let mut d = wysiwyg_doc("pre_enter_mid", "abcdef\n");
7380 d.set_line_flow(LineFlow::Preserve);
7381 d.caret = 3;
7382 d.newline();
7383 assert_eq!(d.source, "abc\ndef\n", "mid-line Enter is a soft break");
7384
7385 // End-of-paragraph: Enter then typing continues the same paragraph on a
7386 // new line (a soft break), not a fresh paragraph.
7387 let mut d = wysiwyg_doc("pre_enter_end", "abc\n");
7388 d.set_line_flow(LineFlow::Preserve);
7389 d.caret = 3;
7390 d.newline();
7391 d.insert("def");
7392 assert_eq!(
7393 d.source, "abc\ndef\n",
7394 "end-of-line Enter + typing is a soft break"
7395 );
7396 }
7397
7398 #[test]
7399 fn preserve_double_enter_still_makes_a_paragraph() {
7400 // Two Enters in a row promote to a real paragraph break: the second lands
7401 // on the blank line the first opened and takes the empty-line branch.
7402 let mut d = wysiwyg_doc("pre_enter_dbl", "abc\n");
7403 d.set_line_flow(LineFlow::Preserve);
7404 d.caret = 3;
7405 d.newline();
7406 d.newline();
7407 d.insert("def");
7408 assert_eq!(
7409 d.source, "abc\n\ndef\n",
7410 "double Enter is a paragraph break"
7411 );
7412 }
7413
7414 #[test]
7415 fn preserve_backspace_joins_across_a_soft_break() {
7416 // Backspace is the symmetric undo of a Preserve Enter: over the `\n` of a
7417 // soft break it deletes the single newline and joins the two lines.
7418 let mut d = wysiwyg_doc("pre_bs", "abc\ndef\n");
7419 d.set_line_flow(LineFlow::Preserve);
7420 d.build_visual(80);
7421 d.caret = 4; // start of "def", just past the soft break
7422 d.backspace();
7423 assert_eq!(
7424 d.source, "abcdef\n",
7425 "Backspace joins across the soft break"
7426 );
7427 assert_eq!(d.caret, 3, "caret lands where the lines meet");
7428 }
7429
7430 #[test]
7431 fn fold_enter_still_starts_a_new_paragraph() {
7432 // The default flow is unchanged: a lone `\n` would render as an invisible
7433 // space, so Enter keeps opening the paragraph break that actually shows.
7434 let mut d = wysiwyg_doc("fold_enter", "abcdef\n");
7435 d.caret = 3;
7436 d.newline();
7437 assert_eq!(
7438 d.source, "abc\n\ndef\n",
7439 "Fold mid-line Enter is a paragraph break"
7440 );
7441 }
7442
7443 #[test]
7444 fn wysiwyg_one_enter_starts_a_new_paragraph() {
7445 // Regression: one Enter left the caret between the two newlines, so typing
7446 // made a soft break (one paragraph) and you needed a second Enter.
7447 let mut d = wysiwyg_doc("wys_enter", "abc\n");
7448 d.caret = 3;
7449 d.newline();
7450 d.insert("def");
7451 assert_eq!(d.source, "abc\n\ndef\n"); // two paragraphs, not "abc\ndef\n"
7452 }
7453
7454 #[test]
7455 fn enter_at_the_end_of_a_bold_run_keeps_its_closing_delimiter_attached() {
7456 // Regression: Enter at the caret's natural End-of-line resting place
7457 // after a bold run with nothing following it (on screen: right after
7458 // "bold", before the hidden closing "**") spliced the paragraph break
7459 // at that very byte offset — which sits *before* the closing "**" in
7460 // the source, since the delimiter is hidden and emits no glyph of its
7461 // own for `push_row`'s "end of row" fallback to count. That severed the
7462 // mark: "**bold**\n" became "**bold\n\n**\n", stranding the closing
7463 // "**" alone on the new line instead of leaving "**bold**" intact with
7464 // a fresh empty paragraph after it.
7465 let mut d = wysiwyg_doc("bold_eol_enter", "**bold**\n");
7466 d.move_end(false); // the WYSIWYG End key, from caret 0
7467 assert_eq!(
7468 d.caret, 6,
7469 "caret rests right after \"bold\", before the hidden \"**\""
7470 );
7471 d.newline();
7472 assert!(
7473 d.source.starts_with("**bold**"),
7474 "the closing ** must stay attached to \"bold\": got {:?}",
7475 d.source
7476 );
7477 assert_eq!(
7478 d.source, "**bold**\n\n\n",
7479 "a fresh empty paragraph follows the still-intact bold run"
7480 );
7481 }
7482
7483 #[test]
7484 fn source_view_enter_is_a_single_newline() {
7485 let mut d = doc_with("src_enter", "abc\n");
7486 d.caret = 3;
7487 d.newline();
7488 assert_eq!(d.source, "abc\n\n");
7489 }
7490
7491 #[test]
7492 fn heading_applies_at_the_end_of_a_paragraph() {
7493 // The caret at a line end sits at the doc level; set_block must still find
7494 // the block on that line.
7495 let mut d = doc_with("head_end", "abc\n");
7496 d.caret = 3; // end of "abc"
7497 d.toggle_heading(1);
7498 assert_eq!(d.source, "# abc\n");
7499 }
7500
7501 #[test]
7502 fn heading_on_an_empty_new_paragraph_creates_one() {
7503 let mut d = wysiwyg_doc("head_empty", "abc\n");
7504 d.caret = 3;
7505 d.newline(); // caret now on a fresh, empty paragraph
7506 d.toggle_heading(1);
7507 d.insert("Title");
7508 assert!(d.source.contains("# Title"), "got {:?}", d.source);
7509 }
7510
7511 #[test]
7512 fn a_heading_typed_on_a_blank_line_keeps_the_caret_on_its_own_row() {
7513 // The reported bug, end to end: click a blank line with another one under
7514 // it, press H1, type. The text landed in the heading and the caret's
7515 // offset was right (the source view drew it there), but the rich view
7516 // drew it two rows lower, on the trailing blank line — the empty `# `
7517 // heading had left every row below it short by the marker's two bytes,
7518 // and the blank line ended up claiming the heading's own end offset.
7519 let mut d = wysiwyg_doc("head_blank", "one\n\ntwo\n\n\n\n");
7520 d.build_visual_unwrapped();
7521 d.caret = d.vmap.offset_of_pos(4, 0); // the first of the two blank lines
7522 d.toggle_heading(1);
7523 for c in "title".chars() {
7524 d.insert(&c.to_string());
7525 d.build_visual_unwrapped(); // as a frontend does, one frame per key
7526 }
7527 assert_eq!(d.source, "one\n\ntwo\n\n# title\n\n");
7528 assert_eq!(
7529 d.caret_pos(),
7530 (4, 5),
7531 "the caret draws at the end of the heading"
7532 );
7533 }
7534
7535 #[test]
7536 fn clicking_an_empty_heading_types_after_its_marker() {
7537 // The same anchor from the other side: the empty heading's row is its own
7538 // caret home, so a click on it must land past the hidden `# `. Landing in
7539 // front of the hashes made the first keystroke un-heading the line.
7540 let mut d = wysiwyg_doc("head_click", "# \n");
7541 d.build_visual_unwrapped();
7542 d.caret = d.vmap.offset_of_pos(0, 0);
7543 d.insert("x");
7544 assert_eq!(d.source, "# x\n");
7545 }
7546
7547 #[test]
7548 fn wysiwyg_enter_after_a_heading_makes_a_paragraph() {
7549 let mut d = wysiwyg_doc("head_enter", "# Title\n");
7550 d.caret = 7; // end of the heading
7551 d.newline();
7552 d.insert("body");
7553 assert_eq!(d.source, "# Title\n\nbody\n");
7554 }
7555
7556 #[test]
7557 fn wysiwyg_enter_continues_a_bullet_list() {
7558 let mut d = wysiwyg_doc("wys_bullet", "- item\n");
7559 d.caret = 6; // end of "item"
7560 d.newline();
7561 d.insert("two");
7562 assert_eq!(d.source, "- item\n- two\n");
7563 }
7564
7565 #[test]
7566 fn wysiwyg_enter_increments_an_ordered_list() {
7567 let mut d = wysiwyg_doc("wys_ol", "1. one\n");
7568 d.caret = 6; // end of "one"
7569 d.newline();
7570 d.insert("two");
7571 assert_eq!(d.source, "1. one\n2. two\n");
7572 }
7573
7574 #[test]
7575 fn wysiwyg_backspace_after_leaving_a_list_collapses_the_gap_cleanly() {
7576 // Regression for the "extra newline" left between a list and the paragraph
7577 // below it. Enter, Enter leaves the list on a fresh empty paragraph
7578 // (`- item\n\n\n\nnext`, a navigable blank between the two blocks); one
7579 // Backspace should then take the caret cleanly back to the end of the list
7580 // item, `- item\n\nnext`, not delete a single newline and strand it on the
7581 // odd `- item\n\n\nnext` — a blank line the eye reads as one separator but
7582 // no caret can land on. The map is rebuilt between keystrokes exactly as a
7583 // frontend does, since Backspace reads the stop table to place the delete.
7584 let mut d = wysiwyg_doc("wys_exit_bksp", "- item\n\nnext\n");
7585 d.caret = 6; // end of "item"
7586 d.newline();
7587 d.build_visual(80);
7588 d.newline(); // leave the list onto a fresh empty paragraph
7589 d.build_visual(80);
7590 assert_eq!(
7591 d.source, "- item\n\n\n\nnext\n",
7592 "double-Enter opens the empty paragraph"
7593 );
7594 d.backspace();
7595 assert_eq!(
7596 d.source, "- item\n\nnext\n",
7597 "one Backspace collapses the whole gap"
7598 );
7599 assert_eq!(
7600 d.caret, 6,
7601 "and lands the caret back at the end of the list item"
7602 );
7603 }
7604
7605 #[test]
7606 fn wysiwyg_backspace_on_stacked_blank_lines_still_removes_just_one() {
7607 // The stop-wise delete must not over-reach when there is no block boundary
7608 // to cross: two blank lines in a row are one caret stop apart, so pressing
7609 // Enter on an empty line and then Backspace removes exactly the one newline
7610 // it added — the lone-Enter / lone-Backspace symmetry, preserved.
7611 let mut d = wysiwyg_doc("wys_stack", "abc\n\n\n");
7612 d.caret = 5; // the empty paragraph the first Enter already opened
7613 d.build_visual(80);
7614 d.newline();
7615 d.build_visual(80);
7616 assert_eq!(
7617 d.source, "abc\n\n\n\n",
7618 "Enter on the blank line adds one newline"
7619 );
7620 d.backspace();
7621 assert_eq!(
7622 d.source, "abc\n\n\n",
7623 "Backspace takes back exactly that one newline"
7624 );
7625 }
7626
7627 #[test]
7628 fn wysiwyg_enter_on_an_empty_list_item_exits_the_list() {
7629 let mut d = wysiwyg_doc("wys_exit", "- a\n- \n");
7630 d.caret = 6; // end of the empty "- " item
7631 d.newline();
7632 d.insert("p");
7633 assert_eq!(d.source, "- a\n\np\n");
7634 }
7635
7636 #[test]
7637 fn wysiwyg_enter_does_not_mistake_a_setext_underline_for_a_list() {
7638 // `text\n- \n` is a setext heading — the `- ` is its underline, not a
7639 // list item, though it reads as a `- ` marker byte-for-byte. Enter must
7640 // not take the list-exit path (which would splice the `- ` away as if
7641 // leaving an empty item); the AST guard sends it to a normal break and
7642 // leaves the underline intact.
7643 let mut d = wysiwyg_doc("wys_setext", "text\n- \n");
7644 assert!(
7645 d.nodes().iter().any(|n| n.kind == Kind::Heading),
7646 "precondition: twig parses this as a heading, not a list",
7647 );
7648 d.caret = 7; // on the `- ` underline line
7649 d.newline();
7650 assert!(
7651 d.source.contains("- "),
7652 "the setext underline survives, not spliced away as a list item: {:?}",
7653 d.source,
7654 );
7655 }
7656
7657 #[test]
7658 fn wysiwyg_enter_in_a_code_block_is_a_literal_newline() {
7659 let mut d = wysiwyg_doc("wys_code", "```\nabc\n```\n");
7660 d.caret = 7; // end of "abc" inside the fence
7661 d.newline();
7662 d.insert("def");
7663 assert_eq!(d.source, "```\nabc\ndef\n```\n");
7664 }
7665
7666 #[test]
7667 fn wysiwyg_enter_continues_a_block_quote() {
7668 // Enter opens a new *paragraph* inside the quote, not a second line of
7669 // the same one. `> quote\n> more` is a soft break, which under
7670 // `LineFlow::Fold` renders as a space — the keystroke would look like it
7671 // did nothing. The quoted blank line is what makes the break visible, and
7672 // it's the same thing Enter does in running prose.
7673 let mut d = wysiwyg_doc("wys_quote", "> quote\n");
7674 d.caret = 7; // end of "quote"
7675 d.newline();
7676 d.insert("more");
7677 assert_eq!(d.source, "> quote\n>\n> more\n");
7678 // Still one quote, now holding two paragraphs — not a quote and a stray
7679 // line that fell out of it.
7680 let quotes = d
7681 .nodes()
7682 .iter()
7683 .filter(|n| n.kind == Kind::BlockQuote)
7684 .count();
7685 assert_eq!(quotes, 1);
7686 }
7687
7688 #[test]
7689 fn set_block_makes_a_heading_at_the_caret() {
7690 let mut d = doc_with("head", "Title\n\nbody\n");
7691 d.caret = 0;
7692 d.set_block(BlockKind::Heading(2));
7693 assert_eq!(d.source, "## Title\n\nbody\n");
7694 d.set_block(BlockKind::Paragraph);
7695 assert_eq!(d.source, "Title\n\nbody\n");
7696 }
7697
7698 // ── block containers (quote / list) ──────────────────────────────────────
7699
7700 #[test]
7701 fn toggle_blockquote_wraps_the_block_at_the_caret_and_reverses() {
7702 let g = |m, f: fn(&mut Doc)| golden("quote", m, f);
7703 assert_eq!(g("hel|lo\n", |d| d.toggle_blockquote()), "> hel|lo\n");
7704 assert_eq!(g("> hel|lo\n", |d| d.toggle_blockquote()), "hel|lo\n");
7705 // A caret at a line end sits at the doc level; the block is still found.
7706 assert_eq!(g("hello|\n", |d| d.toggle_blockquote()), "> hello|\n");
7707 }
7708
7709 #[test]
7710 fn toggle_blockquote_keeps_the_caret_in_a_hard_wrapped_paragraph() {
7711 // Every source line of the paragraph gets its own `> `, so a caret left
7712 // on its old byte offset falls one prefix per line above it too far
7713 // back — inside the markup it just asked for rather than in its word.
7714 assert_eq!(
7715 golden("quote_wrap", "aaa\nb|bb\nccc\n", |d| d.toggle_blockquote()),
7716 "> aaa\n> b|bb\n> ccc\n"
7717 );
7718 }
7719
7720 #[test]
7721 fn toggle_blockquote_works_in_wysiwyg_view() {
7722 let g = |n, m, f: fn(&mut Doc)| golden_in(View::Wysiwyg, n, m, f);
7723 assert_eq!(
7724 g("q_wys", "hel|lo\n", |d| d.toggle_blockquote()),
7725 "> hel|lo\n"
7726 );
7727 assert_eq!(
7728 g("q_wys2", "> hel|lo\n", |d| d.toggle_blockquote()),
7729 "hel|lo\n"
7730 );
7731 }
7732
7733 #[test]
7734 fn toggle_list_makes_a_list_and_converts_between_the_kinds() {
7735 let g = |m, f: fn(&mut Doc)| golden("list", m, f);
7736 assert_eq!(g("hel|lo\n", |d| d.toggle_list(false)), "- hel|lo\n");
7737 assert_eq!(g("hel|lo\n", |d| d.toggle_list(true)), "1. hel|lo\n");
7738 // The *other* kind converts in place instead of nesting, which is what
7739 // makes the two buttons one three-state control.
7740 assert_eq!(g("- hel|lo\n", |d| d.toggle_list(true)), "1. hel|lo\n");
7741 assert_eq!(g("1. hel|lo\n", |d| d.toggle_list(false)), "- hel|lo\n");
7742 // Its own kind, over the only item the list holds, takes it off.
7743 assert_eq!(g("- hel|lo\n", |d| d.toggle_list(false)), "hel|lo\n");
7744 }
7745
7746 #[test]
7747 fn toggle_list_works_in_wysiwyg_view() {
7748 let g = |n, m, f: fn(&mut Doc)| golden_in(View::Wysiwyg, n, m, f);
7749 assert_eq!(
7750 g("l_wys", "hel|lo\n", |d| d.toggle_list(true)),
7751 "1. hel|lo\n"
7752 );
7753 assert_eq!(
7754 g("l_wys2", "1. hel|lo\n", |d| d.toggle_list(false)),
7755 "- hel|lo\n"
7756 );
7757 assert_eq!(
7758 g("l_wys3", "- hel|lo\n", |d| d.toggle_list(false)),
7759 "hel|lo\n"
7760 );
7761 }
7762
7763 #[test]
7764 fn a_list_over_a_selection_numbers_each_block_and_stays_selected() {
7765 // The selection has to grow with the markup: twig takes a container off
7766 // only a range covering every block it holds, so the second press can
7767 // reverse the first only if the result is what's selected.
7768 let mut d = doc_with("list_sel", "abc\n\ndef\n");
7769 d.select_all();
7770 d.toggle_list(true);
7771 assert_eq!(d.source, "1. abc\n\n2. def\n");
7772 assert_eq!(d.selection(), Some((0, d.source.len())));
7773 d.toggle_list(true);
7774 assert_eq!(d.source, "abc\n\ndef\n");
7775 }
7776
7777 #[test]
7778 fn toggle_blockquote_nests_a_partly_covered_quote() {
7779 // twig's rule: covering only some of a container's blocks nests, because
7780 // taking the quote off would drag its uncovered siblings out with it.
7781 let mut d = doc_with("quote_nest", "> a\n>\n> b\n");
7782 d.caret = 2; // in the first quoted paragraph only
7783 d.toggle_blockquote();
7784 assert_eq!(d.source, "> > a\n>\n> b\n");
7785 }
7786
7787 #[test]
7788 fn a_container_toggle_opens_an_empty_one_on_a_blank_line() {
7789 // A blank line used to be no block for twig to wrap —
7790 // `toggle_block_container` answered `NotFound` — so Quote and the list
7791 // buttons did nothing on the very line the H1 button works on, and leaf
7792 // lent twig a scratch paragraph to wrap and took it back out again.
7793 // twig 3.2.0 opens an empty container there itself, so what is left here
7794 // is where the caret lands: inside the marker that was just written.
7795 let mut d = doc_with("quote_blank", "\nabc\n");
7796 d.caret = 0;
7797 d.toggle_blockquote();
7798 assert_eq!(d.source, "> \nabc\n");
7799 assert_eq!(
7800 d.caret, 2,
7801 "the caret belongs inside the quote it just opened"
7802 );
7803 assert!(d.status.is_none(), "{:?}", d.status);
7804 assert!(d.dirty);
7805
7806 // And the paragraph below is still its own block: an empty container one
7807 // soft break from `abc` would take that paragraph into the quote with it.
7808 let mut d = wysiwyg_doc("quote_blank_rows", "\nabc\n");
7809 d.caret = 0;
7810 d.toggle_blockquote();
7811 d.build_visual(80);
7812 assert_eq!(drawn_rows(&d), ["│ ", "", "abc"]);
7813
7814 // The same from the other side: a blank line directly under a paragraph
7815 // earns the blank line an empty block needs, rather than being read as a
7816 // soft break inside that paragraph.
7817 let mut d = doc_with("list_blank_below", "abc\n");
7818 d.caret = 4;
7819 d.toggle_list(false);
7820 assert_eq!(d.source, "abc\n\n- ");
7821 assert_eq!(d.caret, 7);
7822 }
7823
7824 #[test]
7825 fn enter_at_the_end_of_a_quote_stays_in_the_quote() {
7826 // The gesture the rendering fix is for. `newline` inside a quote already
7827 // wrote the right source — `> a\n` becomes `> a\n>\n> \n`, twig's own
7828 // spelling — but the two marker lines it adds belonged to no node until
7829 // twig 3.2.0, so the gutter stopped at `a` and the line the writer had
7830 // just made drew as plain prose under the quote.
7831 let mut d = wysiwyg_doc("quote_enter", "> a\n");
7832 d.caret = 3; // past `a`, at the end of the quoted line
7833 d.newline();
7834 assert_eq!(d.source, "> a\n>\n> \n");
7835 d.build_visual(80);
7836 assert_eq!(drawn_rows(&d), ["│ a", "│ ", "│ "]);
7837 // And the caret is on the new line, not stranded on the old one.
7838 assert_eq!(d.caret, 8);
7839 }
7840
7841 #[test]
7842 fn opening_a_container_on_a_blank_line_is_one_undo_step() {
7843 // It was three edits — scratch, wrap, unscratch — coalesced into one, and
7844 // now it is twig's single edit. Either way one ⌘z has to put the blank
7845 // line back rather than undoing into a half-built document.
7846 for open in [
7847 &(|d: &mut Doc| d.toggle_blockquote()) as &dyn Fn(&mut Doc),
7848 &|d: &mut Doc| d.toggle_list(false),
7849 &|d: &mut Doc| d.toggle_list(true),
7850 ] {
7851 let mut d = doc_with("container_blank_undo", "a\n\n\n\nb\n");
7852 d.caret = 3;
7853 open(&mut d);
7854 assert_ne!(d.source, "a\n\n\n\nb\n");
7855 d.undo();
7856 assert_eq!(d.source, "a\n\n\n\nb\n");
7857 }
7858 }
7859
7860 #[test]
7861 fn a_container_toggle_is_one_undo_step() {
7862 let mut d = doc_with("quote_undo", "hello\n");
7863 d.caret = 3;
7864 d.insert("X"); // a typing run the structural edit must not fold into
7865 d.toggle_blockquote();
7866 assert_eq!(d.source, "> helXlo\n");
7867 d.undo();
7868 assert_eq!(d.source, "helXlo\n");
7869 }
7870
7871 // ── links ────────────────────────────────────────────────────────────────
7872
7873 #[test]
7874 fn insert_link_wraps_the_selection_and_leaves_its_text_selected() {
7875 let mut d = doc_with("link_sel", "word here\n");
7876 d.anchor = Some(0);
7877 d.caret = 4;
7878 d.insert_link("http://x.dev");
7879 assert_eq!(d.source, "[word](http://x.dev) here\n");
7880 // The text, not the destination — so a second press re-points the link
7881 // the first one made rather than nesting one inside it.
7882 assert_eq!(d.selected_text(), Some("word"));
7883 d.insert_link("http://y.dev");
7884 assert_eq!(d.source, "[word](http://y.dev) here\n");
7885 assert_eq!(d.selected_text(), Some("word"));
7886 }
7887
7888 #[test]
7889 fn insert_image_at_the_caret_spells_the_markup_and_lands_past_it() {
7890 let mut d = doc_with("img_caret", "before after\n");
7891 d.caret = 7; // between "before " and "after"
7892 d.insert_image("cat.png", "a cat");
7893 assert_eq!(d.source, "before after\n");
7894 // The caret sits just past the inserted image, nothing selected.
7895 assert_eq!(d.selection(), None);
7896 assert_eq!(d.caret, 7 + "".len());
7897 }
7898
7899 /// The bug a real vault hit: a filename with spaces in it. Markdown ends a
7900 /// destination at the first space, so the `format!` this used to be wrote
7901 /// something that was not an image at all — and the reader saw the markup as
7902 /// text. twig owns the spelling now, and moves it into the angle form.
7903 #[test]
7904 fn insert_image_spells_a_destination_with_spaces_so_it_stays_an_image() {
7905 let mut d = doc_with("img_space", "x\n");
7906 d.caret = 0;
7907 d.insert_image("Jesus Commands the Apostles to Rest.jpg", "");
7908 assert_eq!(
7909 d.source,
7910 "x\n"
7911 );
7912 // And it reads back as an image pointing at the unescaped path — the angle
7913 // brackets are spelling, not part of the destination.
7914 d.caret = 2;
7915 assert_eq!(
7916 d.image_destination_at_caret(),
7917 Some("Jesus Commands the Apostles to Rest.jpg".to_string())
7918 );
7919 }
7920
7921 /// A `)` in a caption or a filename must not close the image early.
7922 #[test]
7923 fn insert_image_escapes_a_paren_in_either_half() {
7924 let mut d = doc_with("img_paren", "x\n");
7925 d.caret = 0;
7926 d.insert_image("a)b.png", "");
7927 assert_eq!(d.source, "b.png)x\n");
7928 d.caret = 2;
7929 assert_eq!(d.image_destination_at_caret(), Some("a)b.png".to_string()));
7930 }
7931
7932 #[test]
7933 fn insert_image_uses_the_selection_as_alt_text() {
7934 let mut d = doc_with("img_sel", "caption here\n");
7935 d.anchor = Some(0);
7936 d.caret = 7; // "caption"
7937 d.insert_image("p.png", "ignored fallback");
7938 assert_eq!(d.source, " here\n");
7939 }
7940
7941 #[test]
7942 fn insert_image_with_no_alt_leaves_empty_brackets() {
7943 let mut d = doc_with("img_noalt", "\n");
7944 d.caret = 0;
7945 d.insert_image("logo.svg", "");
7946 assert_eq!(d.source, "\n");
7947 }
7948
7949 #[test]
7950 fn insert_media_spells_a_video_as_html_and_reads_it_back_as_a_block() {
7951 // The round trip is the point: it's no use writing markup the reader
7952 // can't pick up again. This is the pair that only holds from twig 2.5.1
7953 // on — before it, the one-line form went in fine and came back as a
7954 // paragraph of raw tags, publishing no media at all.
7955 let mut d = doc_with("vid_rt", "\n");
7956 d.caret = 0;
7957 d.insert_media(MediaKind::Video, "clip.mp4", "a clip");
7958 assert_eq!(
7959 d.source,
7960 "<video src=\"clip.mp4\" controls>a clip</video>\n"
7961 );
7962
7963 d.build_visual(80);
7964 assert_eq!(d.vmap.media.len(), 1, "reads back as one block media");
7965 assert_eq!(d.vmap.media[0].kind, MediaKind::Video);
7966 assert_eq!(d.vmap.media[0].destination, "clip.mp4");
7967 assert_eq!(d.vmap.media[0].alt, "a clip");
7968 }
7969
7970 #[test]
7971 fn insert_media_spells_audio_with_its_own_tag() {
7972 let mut d = doc_with("aud_rt", "\n");
7973 d.caret = 0;
7974 d.insert_media(MediaKind::Audio, "take.mp3", "");
7975 assert_eq!(d.source, "<audio src=\"take.mp3\" controls></audio>\n");
7976 d.build_visual(80);
7977 assert_eq!(d.vmap.media[0].kind, MediaKind::Audio);
7978 }
7979
7980 #[test]
7981 fn insert_media_uses_the_selection_as_fallback_text() {
7982 // The same courtesy `insert_image` does with alt: select a caption,
7983 // insert, and the caption labels the thing rather than being replaced.
7984 let mut d = doc_with("vid_sel", "the talk here\n");
7985 d.anchor = Some(0);
7986 d.caret = 8; // "the talk"
7987 d.insert_media(MediaKind::Video, "talk.mp4", "ignored fallback");
7988 assert_eq!(
7989 d.source,
7990 "<video src=\"talk.mp4\" controls>the talk</video> here\n"
7991 );
7992 }
7993
7994 #[test]
7995 fn insert_media_with_an_image_kind_is_just_insert_image() {
7996 let mut d = doc_with("img_via_media", "\n");
7997 d.caret = 0;
7998 d.insert_media(MediaKind::Image, "logo.svg", "x");
7999 assert_eq!(d.source, "\n");
8000 }
8001
8002 // ── thematic breaks ─────────────────────────────────────────────────────
8003
8004 /// The node the source parses as at `caret` — what confirms an inserted
8005 /// `---` actually reads back as a rule, not stray text or a setext heading.
8006 ///
8007 /// The *narrowest* node covering the offset. Every ancestor covers it too,
8008 /// and since twig 2.8 that includes the `doc` root, which now carries a real
8009 /// span (it reported none before, so taking the first match used to land on
8010 /// the block by luck and now always answers `"doc"`).
8011 fn kind_at(d: &mut Doc, caret: usize) -> Option<Kind> {
8012 d.nodes()
8013 .into_iter()
8014 .filter(|n| n.span.start <= caret && caret < n.span.end)
8015 .min_by_key(|n| n.span.end - n.span.start)
8016 .map(|n| n.kind)
8017 }
8018
8019 #[test]
8020 fn a_task_box_toggles_at_the_caret_and_reads_back() {
8021 let mut d = doc_with("task_toggle", "- [ ] todo\n- [x] done\n");
8022 d.caret = 8; // inside "todo"
8023 assert_eq!(d.task_checked_at_caret(), Some(false));
8024 d.toggle_task_checked();
8025 assert_eq!(d.source, "- [x] todo\n- [x] done\n");
8026 assert_eq!(d.task_checked_at_caret(), Some(true));
8027 d.toggle_task_checked();
8028 assert_eq!(d.source, "- [ ] todo\n- [x] done\n");
8029 }
8030
8031 #[test]
8032 fn a_click_toggles_a_box_without_taking_the_caret_with_it() {
8033 // The whole reason `toggle_task_at` exists apart from the caret form:
8034 // ticking a box elsewhere must not move the cursor out of what's being
8035 // typed.
8036 let mut d = doc_with("task_click", "- [ ] first\n- [ ] second\n");
8037 d.caret = 8; // inside "first"
8038 let second = d.source.find("second").unwrap();
8039 d.toggle_task_at(second);
8040 assert_eq!(d.source, "- [ ] first\n- [x] second\n");
8041 assert_eq!(d.caret, 8, "the caret stayed in the first item");
8042 }
8043
8044 #[test]
8045 fn a_plain_item_gains_and_loses_a_box() {
8046 let mut d = doc_with("task_mint", "- plain\n");
8047 d.caret = 4;
8048 assert_eq!(d.task_checked_at_caret(), None);
8049 d.toggle_task_item();
8050 assert_eq!(d.source, "- [ ] plain\n");
8051 assert_eq!(
8052 d.task_checked_at_caret(),
8053 Some(false),
8054 "a new box arrives unticked"
8055 );
8056 d.toggle_task_item();
8057 assert_eq!(d.source, "- plain\n");
8058 }
8059
8060 #[test]
8061 fn ticking_a_box_that_isnt_there_reports_rather_than_minting_one() {
8062 // `set checked` must not silently convert a bullet into a task — that is
8063 // `toggle_task_item`'s job, and twig refuses it here.
8064 let mut d = doc_with("task_none", "- plain\n");
8065 d.caret = 4;
8066 d.toggle_task_checked();
8067 assert_eq!(d.source, "- plain\n", "nothing written");
8068 assert!(
8069 d.status.is_some(),
8070 "the refusal should reach the status line"
8071 );
8072 }
8073
8074 #[test]
8075 fn a_task_item_in_a_quote_is_found_past_the_quote_marker() {
8076 let mut d = doc_with("task_quote", "> - [ ] nested\n");
8077 d.caret = d.source.find("nested").unwrap();
8078 assert_eq!(d.task_checked_at_caret(), Some(false));
8079 d.toggle_task_checked();
8080 assert_eq!(d.source, "> - [x] nested\n");
8081 }
8082
8083 #[test]
8084 fn insert_thematic_break_parts_the_paragraph_around_the_caret() {
8085 // A rule is a block, so twig's `insert_thematic_break` alone lands it
8086 // after the whole paragraph. `split_block` parts the paragraph first and
8087 // the rule is aimed at the *first* half, which is what a rule button is
8088 // understood to do — and what leaf spelled by hand until twig grew both
8089 // halves of the gesture.
8090 let mut d = doc_with("hr_mid", "before after\n");
8091 d.caret = 7; // between "before " and "after"
8092 d.insert_thematic_break();
8093 assert_eq!(d.source, "before \n\n---\n\nafter\n");
8094 assert_eq!(d.selection(), None);
8095 assert_eq!(
8096 kind_at(&mut d, "before \n\n".len()),
8097 Some(Kind::ThematicBreak)
8098 );
8099 }
8100
8101 #[test]
8102 fn insert_thematic_break_spells_the_rule_the_format_s_own_way() {
8103 // The whole point of delegating: `---` is Markdown's, `* * *` is djot's,
8104 // and leaf wrote the first into both until twig started spelling it.
8105 let mut md = doc_with("hr_md", "para\n");
8106 md.caret = 2;
8107 md.insert_thematic_break();
8108 assert_eq!(md.source, "pa\n\n---\n\nra\n");
8109
8110 let mut dj = Doc::from_source("para\n".into(), Format::Djot).unwrap();
8111 dj.caret = 2;
8112 dj.insert_thematic_break();
8113 assert_eq!(dj.source, "pa\n\n* * *\n\nra\n");
8114 }
8115
8116 #[test]
8117 fn clicking_below_a_final_thematic_break_can_type_after_it() {
8118 let mut d = wysiwyg_doc("hr_final_click", "---\n");
8119 d.build_visual(80);
8120 d.click(d.vmap.num_rows() + 2, 0, false);
8121 assert_eq!(d.caret, d.source.len(), "the caret belongs after the rule");
8122 d.insert("after");
8123 assert_eq!(d.source, "---\nafter");
8124 }
8125
8126 #[test]
8127 fn enter_in_a_nested_list_item_keeps_the_new_item_nested() {
8128 // The same bytes are two documents. In Markdown ` - b` is a nested item
8129 // and the next one belongs beside it, at its indent. In Djot a list
8130 // marker can't interrupt a paragraph, so those bytes are literal text in
8131 // item `a` and there is only one item — writing ` - ` under it would add
8132 // no item at all, just more text, and the new sibling has to go to
8133 // column zero. Both spellings come out of the *enclosing item's* line.
8134 let mut md = wysiwyg_doc("enter_nested_md", "- a\n - b\n");
8135 md.caret = "- a\n - b".len();
8136 md.newline();
8137 assert_eq!(md.source, "- a\n - b\n - \n");
8138 assert_eq!(list_items(&mut md), 3);
8139
8140 let mut dj = Doc::from_source("- a\n - b\n".into(), Format::Djot).unwrap();
8141 dj.view = View::Wysiwyg;
8142 dj.build_visual(80);
8143 dj.caret = "- a\n - b".len();
8144 dj.newline();
8145 assert_eq!(dj.source, "- a\n - b\n- \n");
8146 assert_eq!(list_items(&mut dj), 2);
8147
8148 // Where Djot's nesting is real — opened by a blank line — the indent is
8149 // reproduced there too, and the two formats agree again.
8150 let mut dj = Doc::from_source("- a\n\n - b\n".into(), Format::Djot).unwrap();
8151 dj.view = View::Wysiwyg;
8152 dj.build_visual(80);
8153 dj.caret = "- a\n\n - b".len();
8154 dj.newline();
8155 assert_eq!(dj.source, "- a\n\n - b\n - \n");
8156 assert_eq!(list_items(&mut dj), 3);
8157 }
8158
8159 #[test]
8160 fn tab_nests_an_item_at_the_column_its_own_marker_asks_for() {
8161 // Tab replaces the line's whole prefix with the one twig spells, so the
8162 // quote markers, the parent's indent and an ordered marker's extra
8163 // column are all its answer rather than leaf's arithmetic.
8164 for (name, body, caret, want) in [
8165 ("bullet", "- a\n- b\n", 6, "- a\n - b\n"),
8166 ("ordered", "1. a\n2. b\n", 8, "1. a\n 1. b\n"),
8167 ("quoted", "> - a\n> - b\n", 10, "> - a\n> - b\n"),
8168 // A checkbox is markup the item's own text wraps past, but a nested
8169 // list may only open at the *list* marker's column — four in from
8170 // there is a paragraph continuation, and `- [ ] a\n - [ ] b`
8171 // parses as one item, not two.
8172 ("task", "- [ ] a\n- [ ] b\n", 14, "- [ ] a\n - [ ] b\n"),
8173 (
8174 "quoted task",
8175 "> - [ ] a\n> - [ ] b\n",
8176 18,
8177 "> - [ ] a\n> - [ ] b\n",
8178 ),
8179 ] {
8180 let mut doc = wysiwyg_doc(name, body);
8181 doc.caret = caret;
8182 doc.indent();
8183 assert_eq!(doc.source, want, "{name}");
8184 // The nesting is real, not just indented text.
8185 assert_eq!(list_items(&mut doc), 2, "{name}");
8186 }
8187 }
8188
8189 #[test]
8190 fn backspace_only_outdents_where_the_format_says_there_is_an_item() {
8191 // The same bytes, the two formats disagreeing, and a gesture that used
8192 // to read the bytes. ` - b` is a nested item in Markdown, so Backspace
8193 // at its marker outdents. In Djot a marker can't interrupt a paragraph,
8194 // so those bytes are literal text inside item `a` — there is nothing to
8195 // outdent, and treating them as a marker turned one item into two, a
8196 // structural edit from a keystroke that should delete one character.
8197 //
8198 // twig's `line_prefix` is what tells them apart: it reports the marker
8199 // on the Markdown line and nothing on the Djot one, which is a
8200 // continuation. No byte scan can reach that answer.
8201 let src = "- a\n - b\n";
8202 let at = "- a\n - ".len();
8203
8204 let mut md = Doc::from_source(src.into(), Format::Markdown).unwrap();
8205 md.view = View::Wysiwyg;
8206 md.build_visual(80);
8207 md.caret = at;
8208 md.backspace();
8209 assert_eq!(md.source, "- a\n- b\n");
8210 assert_eq!(list_items(&mut md), 2);
8211
8212 let mut dj = Doc::from_source(src.into(), Format::Djot).unwrap();
8213 dj.view = View::Wysiwyg;
8214 dj.build_visual(80);
8215 dj.caret = at;
8216 dj.backspace();
8217 assert_eq!(dj.source, "- a\n -b\n"); // an ordinary character delete
8218 assert_eq!(list_items(&mut dj), 1); // and the structure is untouched
8219 }
8220
8221 #[test]
8222 fn enter_in_a_checklist_item_starts_another_unchecked_one() {
8223 // Leaf used to spell the next item from the marker bytes it scanned, and
8224 // its scanner stopped at the bullet — so Enter in a checklist wrote `- `
8225 // and dropped out of the checklist. twig reproduces the whole
8226 // continuation, and a fresh item is always unticked however the one above
8227 // it stands.
8228 for (name, body, want) in [
8229 ("unchecked", "- [ ] a\n", "- [ ] a\n- [ ] \n"),
8230 ("checked", "- [x] a\n", "- [x] a\n- [ ] \n"),
8231 ] {
8232 let mut doc = wysiwyg_doc(name, body);
8233 doc.caret = body.trim_end_matches('\n').len();
8234 doc.newline();
8235 assert_eq!(doc.source, want, "{name}");
8236 // Both items are checklist items — the new one is a box, not the
8237 // plain bullet the old marker scan left behind — and it is unticked
8238 // whichever way the one above it faces.
8239 let boxes: Vec<Option<bool>> = doc
8240 .nodes()
8241 .iter()
8242 .filter(|n| n.kind == Kind::TaskListItem)
8243 .map(|n| n.checked)
8244 .collect();
8245 assert_eq!(boxes.len(), 2, "{name}");
8246 assert_eq!(boxes[1], Some(false), "{name}");
8247 }
8248 }
8249
8250 #[test]
8251 fn a_split_takes_the_space_the_caret_was_in_front_of() {
8252 // Splicing a break at the caret strands the space the words were parted
8253 // at on the head of the second block, where it reads as an indent nobody
8254 // typed. twig's split consumes it.
8255 for (name, body, caret, want) in [
8256 ("para", "one two\n", 3, "one\n\ntwo\n"),
8257 ("item", "- one two\n", 5, "- one\n- two\n"),
8258 ("quote", "> one two\n", 5, "> one\n>\n> two\n"),
8259 // A heading takes leaf's own path, which has to match.
8260 ("heading", "# one two\n", 5, "# one\n\ntwo\n"),
8261 ] {
8262 let mut doc = wysiwyg_doc(name, body);
8263 doc.caret = caret;
8264 doc.newline();
8265 assert_eq!(doc.source, want, "{name}");
8266 }
8267 }
8268
8269 #[test]
8270 fn enter_at_the_end_of_a_heading_opens_a_paragraph() {
8271 // The one place leaf keeps its own break: `split_block` repeats the `#`,
8272 // and Enter after a title is how the body under it is asked for.
8273 let mut doc = wysiwyg_doc("head_enter", "# Title\n");
8274 doc.caret = "# Title".len();
8275 doc.newline();
8276 doc.insert("body");
8277 assert_eq!(doc.source, "# Title\n\nbody\n");
8278 assert_eq!(
8279 doc.nodes()
8280 .iter()
8281 .filter(|n| n.kind == Kind::Heading)
8282 .count(),
8283 1
8284 );
8285 }
8286
8287 #[test]
8288 fn enter_in_a_quoted_list_item_starts_the_next_quoted_item() {
8289 // A quoted item's marker doesn't open its line, so a scan that starts at
8290 // column zero finds a `>` where it wanted a bullet, calls the line "not a
8291 // list" and hands Enter to the plain-quote branch — which writes `> ` and
8292 // drops the list. The next item has to carry the whole prefix.
8293 for (name, body, want) in [
8294 ("flat", "> - a\n", "> - a\n> - \n"),
8295 ("sibling", "> - a\n> - b\n", "> - a\n> - b\n> - \n"),
8296 ("nested", "> - a\n> - b\n", "> - a\n> - b\n> - \n"),
8297 ("ordered", "> 1. a\n> 2. b\n", "> 1. a\n> 2. b\n> 3. \n"),
8298 ("twice quoted", "> > - a\n", "> > - a\n> > - \n"),
8299 ] {
8300 let mut doc = wysiwyg_doc(name, body);
8301 doc.caret = body.trim_end_matches('\n').len();
8302 doc.newline();
8303 assert_eq!(doc.source, want, "{name}");
8304 // The marker isn't just spelled right, it parses as an item.
8305 assert_eq!(list_items(&mut doc), body.lines().count() + 1, "{name}");
8306 }
8307 }
8308
8309 #[test]
8310 fn an_empty_quoted_item_leaves_the_list_and_stays_in_the_quote() {
8311 // Double-Enter exits the list. Unquoted that means a blank line, but a
8312 // *bare* blank line would end the quote too and drop the caret out of it,
8313 // so the separator keeps its `>` and the caret's line keeps its `> `.
8314 let mut doc = wysiwyg_doc("quoted_exit", "> - a\n> - \n");
8315 doc.caret = "> - a\n> - ".len();
8316 doc.newline();
8317 assert_eq!(doc.source, "> - a\n>\n> \n");
8318 assert_eq!(list_items(&mut doc), 1);
8319 // What "still in the quote" means for the next keystroke: the caret sits
8320 // behind the prefix, and what's typed there lands inside the quote as a
8321 // paragraph of its own — not as more of item `a`.
8322 doc.insert("x");
8323 assert_eq!(doc.source, "> - a\n>\n> x\n");
8324 assert!(
8325 doc.editor
8326 .ancestors_at(doc.caret - 1)
8327 .is_ok_and(|c| c.into_iter().any(|m| m.kind == Kind::BlockQuote))
8328 );
8329 }
8330
8331 #[test]
8332 fn backspace_at_a_quoted_marker_takes_the_marker_and_leaves_the_quote() {
8333 // The marker is hidden block markup, so Backspace over it is structural —
8334 // but only the marker is the list's. Splicing from the line start would
8335 // take the `>` with it and silently unquote the line.
8336 let mut doc = wysiwyg_doc("quoted_bksp", "> - a\n");
8337 doc.caret = "> - ".len();
8338 doc.backspace();
8339 assert_eq!(doc.source, "> a\n");
8340 assert_eq!(list_items(&mut doc), 0);
8341
8342 // A nested one outdents instead, moving the bullet within the quote
8343 // rather than moving the quote.
8344 let mut doc = wysiwyg_doc("quoted_outdent", "> - a\n> - b\n");
8345 doc.caret = "> - a\n> - ".len();
8346 doc.backspace();
8347 assert_eq!(doc.source, "> - a\n> - b\n");
8348 assert_eq!(list_items(&mut doc), 2);
8349 }
8350
8351 #[test]
8352 fn only_a_bare_paragraph_is_parted_around_the_caret() {
8353 // The split is deliberately narrow. Parting a fenced block would leave
8354 // two fences with a rule between them, and parting a list item would
8355 // mint an item nobody asked for on the way to a rule that lands after
8356 // the list either way — so both keep the whole block intact and take the
8357 // rule after it. A caret in a quote is likewise left alone.
8358 for (name, body, caret, want) in [
8359 (
8360 "code",
8361 "```\nfn x() {}\n```\n",
8362 8,
8363 "```\nfn x() {}\n```\n\n---\n",
8364 ),
8365 ("list", "- one two\n", 6, "- one two\n\n---\n"),
8366 ("quote", "> one two\n", 6, "> one two\n>\n> ---\n"),
8367 ] {
8368 let mut d = doc_with(&format!("hr_narrow_{name}"), body);
8369 d.caret = caret;
8370 d.insert_thematic_break();
8371 assert_eq!(d.source, want, "{name}: the block should stay whole");
8372 }
8373 }
8374
8375 #[test]
8376 fn insert_thematic_break_replaces_the_selection() {
8377 // Now that the rule lands *at* the caret again, replacing the selection
8378 // is coherent once more: the text goes, and the rule takes its place.
8379 // The space the deletion left leading the second half is consumed by the
8380 // split rather than opening the new paragraph with it.
8381 let mut d = doc_with("hr_sel", "one two three\n");
8382 d.anchor = Some(4);
8383 d.caret = 7; // "two"
8384 d.insert_thematic_break();
8385 assert_eq!(d.source, "one \n\n---\n\nthree\n");
8386 assert_eq!(d.selection(), None);
8387 }
8388
8389 #[test]
8390 fn insert_thematic_break_clears_a_code_block_and_a_table_rather_than_refusing() {
8391 // Both are blocks the rule lands *after*. Leaf used to refuse a fence,
8392 // because writing `---` into one is code, not a rule — twig now walks out
8393 // to the block that owns the caret's line, so there is nothing to refuse.
8394 let mut code = doc_with("hr_code", "```\nfn x() {}\n```\n");
8395 code.caret = 5; // inside the fenced code
8396 code.insert_thematic_break();
8397 assert_eq!(code.source, "```\nfn x() {}\n```\n\n---\n");
8398 assert_eq!(code.status, None, "no refusal to report any more");
8399
8400 let mut table = doc_with("hr_table", "| a | b |\n|---|---|\n| 1 | 2 |\n");
8401 table.caret = 3; // in the header row
8402 table.insert_thematic_break();
8403 assert_eq!(table.source, "| a | b |\n|---|---|\n| 1 | 2 |\n\n---\n");
8404 }
8405
8406 #[test]
8407 fn insert_thematic_break_in_a_list_item_ends_the_list() {
8408 // The un-indented rule cannot continue the list, so it closes the list
8409 // and lands at the top level rather than nested inside it.
8410 let mut d = doc_with("hr_list", "- one\n- two\n");
8411 d.caret = "- one\n- tw".len(); // mid "two"
8412 d.insert_thematic_break();
8413 d.build_visual(80);
8414 let rule_at = d.source.find("---").unwrap();
8415 assert_eq!(kind_at(&mut d, rule_at), Some(Kind::ThematicBreak));
8416 assert!(
8417 !d.nodes().iter().any(|n| n.kind == Kind::BulletList
8418 && n.span.start <= rule_at
8419 && rule_at < n.span.end),
8420 "the rule must not be nested inside the list"
8421 );
8422 }
8423
8424 #[test]
8425 fn insert_thematic_break_in_a_blockquote_stays_in_the_quote() {
8426 // Leaf used to end the quote. twig gives the rule the quote's own prefix,
8427 // which is the document the gesture was actually asked for.
8428 let mut d = doc_with("hr_quote", "> hello\n");
8429 d.caret = 4; // inside the quoted text
8430 d.insert_thematic_break();
8431 assert_eq!(d.source, "> hello\n>\n> ---\n");
8432 d.build_visual(80);
8433 let rule_at = d.source.find("---").unwrap();
8434 assert_eq!(kind_at(&mut d, rule_at), Some(Kind::ThematicBreak));
8435 assert!(
8436 d.nodes().iter().any(|n| n.kind == Kind::BlockQuote
8437 && n.span.start <= rule_at
8438 && rule_at < n.span.end),
8439 "the rule belongs to the quote it was asked for"
8440 );
8441 }
8442
8443 // ── typing against a block picture ────────────────────────────────────────
8444
8445 /// A rendered-view document with the caret parked on one of the picture's two
8446 /// stops, and the map already built — the state a frontend is in between
8447 /// drawing a frame and the next keystroke.
8448 fn doc_at_picture(name: &str, src: &str, side: MediaStop) -> Doc {
8449 let mut d = doc_in(View::Wysiwyg, name, src);
8450 d.build_visual_unwrapped();
8451 let start = src.find("".len(),
8455 };
8456 d
8457 }
8458
8459 /// The block media the map publishes, after rebuilding it — "is this still a
8460 /// picture, or has it become a line of text with an image in it?"
8461 fn media_count(d: &mut Doc) -> usize {
8462 d.build_visual_unwrapped();
8463 d.vmap.media.len()
8464 }
8465
8466 #[test]
8467 fn typing_past_a_block_picture_opens_a_paragraph_under_it() {
8468 // The accident this prevents: tap the blank page under a photo (which
8469 // lands on the picture's trailing stop), type, and `xy` is a
8470 // paragraph with an *inline* image — the photo stops being drawn.
8471 let mut d = doc_at_picture("pic_after", "hi\n\n\n", MediaStop::After);
8472 d.insert("xy");
8473 assert_eq!(d.source, "hi\n\n\n\nxy\n");
8474 assert_eq!(media_count(&mut d), 1, "still a picture");
8475 }
8476
8477 #[test]
8478 fn typing_in_front_of_a_block_picture_opens_a_paragraph_above_it() {
8479 let mut d = doc_at_picture("pic_before", "hi\n\n\n", MediaStop::Before);
8480 d.insert("xy");
8481 assert_eq!(d.source, "hi\n\nxy\n\n\n");
8482 assert_eq!(media_count(&mut d), 1);
8483 }
8484
8485 #[test]
8486 fn a_picture_that_opens_the_document_still_takes_a_paragraph_above_it() {
8487 let mut d = doc_at_picture("pic_first", "\n", MediaStop::Before);
8488 d.insert("x");
8489 assert_eq!(d.source, "x\n\n\n");
8490 assert_eq!(media_count(&mut d), 1);
8491 }
8492
8493 #[test]
8494 fn one_undo_puts_the_picture_back_the_way_it_was_found() {
8495 // The opened paragraph is part of the keystroke, not an edit the writer
8496 // made — so it undoes with the character, not a step later.
8497 let mut d = doc_at_picture("pic_undo", "hi\n\n\n", MediaStop::After);
8498 d.insert("x");
8499 assert_eq!(d.source, "hi\n\n\n\nx\n");
8500 d.undo();
8501 assert_eq!(d.source, "hi\n\n\n");
8502 }
8503
8504 #[test]
8505 fn pasting_against_a_block_picture_opens_a_paragraph_too() {
8506 // ⌘V dissolves the picture exactly as a keystroke does.
8507 let mut d = doc_at_picture("pic_paste", "hi\n\n\n", MediaStop::After);
8508 d.paste("pasted");
8509 assert_eq!(d.source, "hi\n\n\n\npasted\n");
8510 assert_eq!(media_count(&mut d), 1);
8511 }
8512
8513 #[test]
8514 fn typing_beside_an_inline_image_is_ordinary_editing() {
8515 // An inline image has no placeholder row and no stops of its own. Opening
8516 // a paragraph mid-sentence would be the bug, not the fix.
8517 let mut d = doc_in(View::Wysiwyg, "pic_inline", "see  here\n");
8518 d.build_visual_unwrapped();
8519 d.caret = "see ".len();
8520 d.insert("!");
8521 assert_eq!(d.source, "see ! here\n");
8522 }
8523
8524 #[test]
8525 fn source_view_types_raw_markup_against_an_image_untouched() {
8526 // Source view is for writing the markup itself; a break inserted behind
8527 // the writer's back there would be the editor arguing with them.
8528 let mut d = doc_in(View::Source, "pic_src", "\n");
8529 d.caret = "".len();
8530 d.insert("x");
8531 assert_eq!(d.source, "x\n");
8532 }
8533
8534 #[test]
8535 fn typing_over_a_selection_that_starts_at_a_picture_stop_replaces_it() {
8536 // A selection is replaced, not joined into, so there is nothing to
8537 // protect: the range takes the picture with it.
8538 let mut d = doc_at_picture("pic_sel", "hi\n\n\n", MediaStop::Before);
8539 d.anchor = Some(d.caret);
8540 d.caret = d.source.find("".len();
8541 d.insert("x");
8542 assert_eq!(d.source, "hi\n\nx\n");
8543 }
8544
8545 #[test]
8546 fn backspace_past_a_block_picture_deletes_the_picture_not_its_last_byte() {
8547 // What this actually cost: a real vault's photo, to one stray Backspace.
8548 // The caret past `` was deleting the closing paren — invisible
8549 // in the rendered view — and the photo became the text `\n", MediaStop::After);
8551 d.backspace();
8552 assert_eq!(d.source, "hi\n");
8553 assert_eq!(media_count(&mut d), 0, "the picture went, in one piece");
8554 d.undo();
8555 assert_eq!(
8556 d.source, "hi\n\n\n",
8557 "and comes back in one piece"
8558 );
8559 }
8560
8561 #[test]
8562 fn backspace_in_front_of_a_block_picture_steps_out_instead_of_merging_it() {
8563 // Deleting the break here would join the picture to the paragraph above,
8564 // where it is an *inline* image and stops being drawn. Step over the
8565 // boundary; the next press deletes in the paragraph the caret reached.
8566 let mut d = doc_at_picture("pic_bs_before", "hi\n\n\n", MediaStop::Before);
8567 d.backspace();
8568 assert_eq!(d.source, "hi\n\n\n", "nothing deleted");
8569 assert_eq!(d.caret, 2, "the caret stepped up to the end of `hi`");
8570 d.backspace();
8571 assert_eq!(d.source, "h\n\n\n", "and now it deletes there");
8572 assert_eq!(media_count(&mut d), 1, "the picture was never at risk");
8573 }
8574
8575 #[test]
8576 fn forward_delete_in_front_of_a_block_picture_deletes_the_picture() {
8577 // The mirror. A byte-step here eats the `!` and leaves a link.
8578 let mut d = doc_at_picture("pic_del", "hi\n\n\n\nbye\n", MediaStop::Before);
8579 d.delete_forward();
8580 assert_eq!(d.source, "hi\n\nbye\n");
8581 assert_eq!(media_count(&mut d), 0);
8582 }
8583
8584 #[test]
8585 fn forward_delete_past_a_block_picture_steps_over_the_boundary() {
8586 let mut d = doc_at_picture(
8587 "pic_del_after",
8588 "hi\n\n\n\nbye\n",
8589 MediaStop::After,
8590 );
8591 d.delete_forward();
8592 assert_eq!(d.source, "hi\n\n\n\nbye\n", "nothing deleted");
8593 assert_eq!(
8594 d.caret,
8595 d.source.find("bye").unwrap(),
8596 "the caret stepped down to `bye`"
8597 );
8598 }
8599
8600 #[test]
8601 fn a_picture_that_is_the_whole_document_still_deletes_cleanly() {
8602 let mut d = doc_at_picture("pic_only", "\n", MediaStop::After);
8603 d.backspace();
8604 assert_eq!(d.source, "\n");
8605 assert_eq!(media_count(&mut d), 0);
8606 }
8607
8608 #[test]
8609 fn a_word_delete_takes_the_picture_whole_or_steps_out_of_it() {
8610 // ⌥⌫ past a picture would otherwise eat a "word" of its markup.
8611 let mut d = doc_at_picture("pic_wordbs", "hi there\n\n\n", MediaStop::After);
8612 d.delete_word_back();
8613 assert_eq!(d.source, "hi there\n");
8614
8615 // And in front of one it runs *through* the paragraph break into the
8616 // prose above, which merges the picture inline — so it steps out first,
8617 // and the second press deletes the word it was aimed at.
8618 let mut d = doc_at_picture("pic_wordbs2", "hi there\n\n\n", MediaStop::Before);
8619 d.delete_word_back();
8620 assert_eq!(d.source, "hi there\n\n\n");
8621 d.delete_word_back();
8622 assert_eq!(
8623 d.source, "hi \n\n\n",
8624 "the word above went, the picture stayed"
8625 );
8626 assert_eq!(media_count(&mut d), 1);
8627 }
8628
8629 #[test]
8630 fn source_view_deletes_raw_markup_against_an_image_untouched() {
8631 let mut d = doc_in(View::Source, "pic_src_del", "\n");
8632 d.caret = "".len();
8633 d.backspace();
8634 assert_eq!(d.source, ";
8635 }
8636
8637 #[test]
8638 fn image_destination_at_caret_reads_the_image_under_the_caret() {
8639 let mut d = doc_with("img_read", "\n");
8640 d.caret = 3; // inside the image markup
8641 assert_eq!(d.image_destination_at_caret(), Some("cat.png".to_string()));
8642 // Past the image, the caret is in no image.
8643 d.caret = "".len();
8644 assert_eq!(d.image_destination_at_caret(), None);
8645 }
8646
8647 #[test]
8648 fn set_media_rows_reserves_blank_filler_rows_the_frontend_paints_over() {
8649 // The image is one placeholder row by default, and `set_media_rows` grows
8650 // it to the height the frontend measured: the label row plus blank
8651 // `decoration` fillers that hold the vertical space a raster is drawn into.
8652 let mut d = wysiwyg_doc("img_rows", "intro\n\n\n\nend\n");
8653 assert_eq!(d.vmap.media.len(), 1);
8654 let img_row = d.vmap.media[0].rows_span.start;
8655 assert_eq!(
8656 d.vmap.media[0].rows_span,
8657 img_row..img_row + 1,
8658 "default is one row"
8659 );
8660
8661 d.set_media_rows(HashMap::from([("cat.png".to_string(), 4)]));
8662 d.build_visual(80);
8663 assert_eq!(d.vmap.media.len(), 1, "still one image, now taller");
8664 let span = d.vmap.media[0].rows_span.clone();
8665 assert_eq!(span.end - span.start, 4, "reserves the four rows asked for");
8666 // The label row carries the mark and its glyphs; the three below are blank
8667 // decoration — drawn, but no caret and no text.
8668 assert!(
8669 d.vmap.rows[span.start].media.is_some(),
8670 "mark rides the first row"
8671 );
8672 for r in (span.start + 1)..span.end {
8673 assert!(d.vmap.rows[r].decoration, "filler row {r} is decoration");
8674 assert!(d.vmap.rows[r].glyphs.is_empty(), "filler row {r} is blank");
8675 assert!(
8676 d.vmap.rows[r].media.is_none(),
8677 "only the first row is marked"
8678 );
8679 }
8680 }
8681
8682 #[test]
8683 fn a_taller_image_adds_no_caret_stops_and_motion_steps_over_its_fillers() {
8684 // The extra rows are pure spacers: the caret's only homes stay the stop in
8685 // front of the image and the one just past it, so walking the document top
8686 // to bottom visits the same offsets whether the image is 1 row or 5.
8687 let body = "ab\n\n\n\ncd\n";
8688 let stops_at = |rows: usize| -> Vec<usize> {
8689 let mut d = wysiwyg_doc("img_stops", body);
8690 if rows > 1 {
8691 d.set_media_rows(HashMap::from([("p.png".to_string(), rows)]));
8692 d.build_visual(80);
8693 }
8694 d.caret = 0;
8695 let mut seen = vec![d.caret];
8696 loop {
8697 d.move_right(false);
8698 if *seen.last().unwrap() == d.caret {
8699 break;
8700 }
8701 seen.push(d.caret);
8702 }
8703 seen
8704 };
8705 assert_eq!(
8706 stops_at(1),
8707 stops_at(5),
8708 "reserving rows must not add stops"
8709 );
8710 }
8711
8712 #[test]
8713 fn insert_link_repoints_the_link_at_a_bare_caret() {
8714 let mut d = doc_with("link_repoint", "[word](http://x.dev)\n");
8715 d.caret = 3; // in the link's text, nothing selected
8716 d.insert_link("http://y.dev");
8717 assert_eq!(d.source, "[word](http://y.dev)\n");
8718 assert_eq!(d.selected_text(), Some("word"));
8719 }
8720
8721 #[test]
8722 fn insert_link_on_an_empty_range_autolinks_a_url() {
8723 // A link with no text of its own is an autolink, and twig spells it —
8724 // `<…>` is the canonical form and needs no text typed into it, so the
8725 // caret lands after it rather than selecting a finished link.
8726 let mut d = doc_with("link_empty", "\n");
8727 d.caret = 0;
8728 d.insert_link("http://x.dev");
8729 assert_eq!(d.source, "<http://x.dev>\n");
8730 assert_eq!(d.selection(), None);
8731 assert_eq!(d.caret, 14);
8732 }
8733
8734 #[test]
8735 fn insert_link_on_an_empty_range_falls_back_for_a_non_url() {
8736 // `<./notes.md>` is literal text in both formats and `<foo>` is raw HTML
8737 // in Markdown, so a destination that can't autolink doubles as the text
8738 // instead — which is then selected, ready to be typed over.
8739 let mut d = doc_with("link_rel", "\n");
8740 d.caret = 0;
8741 d.insert_link("./notes.md");
8742 assert_eq!(d.source, "[./notes.md](./notes.md)\n");
8743 assert_eq!(d.selection(), Some((1, 11)));
8744 d.insert("Notes");
8745 assert_eq!(d.source, "[Notes](./notes.md)\n");
8746 }
8747
8748 #[test]
8749 fn insert_link_repoints_the_autolink_the_caret_stands_in() {
8750 // The autolink's text is its URL, so re-pointing replaces the whole
8751 // node — the caret must not splice a second link inside the first.
8752 let mut d = doc_with("link_repoint_auto", "see <https://x.dev> ok\n");
8753 d.caret = 10;
8754 d.insert_link("https://y.dev");
8755 assert_eq!(d.source, "see <https://y.dev> ok\n");
8756 }
8757
8758 #[test]
8759 fn code_language_reads_and_edits_through_the_fence() {
8760 let mut d = doc_with("code_lang", "```rust\nlet x = 1;\n```\n");
8761 d.caret = 10; // inside the code body
8762 assert_eq!(d.code_language_at_caret().as_deref(), Some("rust"));
8763 assert!(d.caret_in_fenced_code());
8764
8765 d.set_code_language("python");
8766 assert!(
8767 d.source.starts_with("```python\n"),
8768 "source: {:?}",
8769 d.source
8770 );
8771 assert_eq!(d.code_language_at_caret().as_deref(), Some("python"));
8772
8773 // Clearing it leaves a bare fence and no label.
8774 d.set_code_language("");
8775 assert!(d.source.starts_with("```\n"), "source: {:?}", d.source);
8776 assert_eq!(d.code_language_at_caret(), None);
8777
8778 // A caret outside any code block edits nothing.
8779 let mut p = doc_with("code_lang_none", "just prose\n");
8780 assert!(!p.caret_in_fenced_code());
8781 p.set_code_language("rust");
8782 assert_eq!(p.source, "just prose\n");
8783 }
8784
8785 #[test]
8786 fn a_language_the_fence_cannot_carry_is_refused_not_written() {
8787 // Markdown's info string ends at whitespace, so `two words` would write
8788 // a fence that reads back with a different language than the one asked
8789 // for. twig refuses it; leaf reports that and leaves the source alone.
8790 // The old splice trimmed the ends and wrote whatever was left.
8791 let mut d = doc_with("code_lang_bad", "```rust\nx\n```\n");
8792 d.caret = 10;
8793 d.set_code_language("two words");
8794 assert_eq!(d.source, "```rust\nx\n```\n", "source should be untouched");
8795 assert!(d.status.is_some(), "the refusal should be reported");
8796 assert_eq!(d.code_language_at_caret().as_deref(), Some("rust"));
8797 }
8798
8799 #[test]
8800 fn link_destination_at_caret_reads_both_spellings() {
8801 let mut d = doc_with("link_dest", "see [t](https://x.dev) ok\n");
8802 d.caret = 5;
8803 assert_eq!(
8804 d.link_destination_at_caret().as_deref(),
8805 Some("https://x.dev")
8806 );
8807 d.caret = 0;
8808 assert_eq!(d.link_destination_at_caret(), None);
8809
8810 // An autolink has no `destination`; its text is the URL.
8811 let mut a = doc_with("link_dest_auto", "see <https://x.dev> ok\n");
8812 a.caret = 10;
8813 assert_eq!(
8814 a.link_destination_at_caret().as_deref(),
8815 Some("https://x.dev")
8816 );
8817 a.caret = 21;
8818 assert_eq!(a.link_destination_at_caret(), None);
8819 }
8820
8821 #[test]
8822 fn locate_finds_the_block_a_declared_id_names() {
8823 // The Book of Mormon shape: one document per chapter, one `{#v…}` per
8824 // verse. The locator has to land on the *verse*, which is the whole
8825 // reason a link carries one.
8826 let src = "{#v1}\nI, Nephi, having been born of goodly parents.\n\n\
8827 {#v2}\nYea, I make a record in the language of my father.\n";
8828 let mut d = Doc::from_source(src.to_string(), Format::Djot).unwrap();
8829 let v2 = d.locate("v2").expect("the document declares `{#v2}`");
8830 assert_eq!(
8831 d.source[v2.start..v2.end].trim_end(),
8832 "Yea, I make a record in the language of my father."
8833 );
8834 // The attribute line is not part of it: `start` is a place to put a
8835 // caret, and `{#v2}` is markup the caret has no business landing in.
8836 assert!(d.source[..v2.start].ends_with("{#v2}\n"));
8837 assert_eq!(d.locate("v99"), None);
8838 }
8839
8840 #[test]
8841 fn locate_reads_a_heading_by_its_words_when_the_format_mints_no_ids() {
8842 // Markdown has no ids at all — twig mints none, and `{#custom}` in a
8843 // Markdown heading is literal text. So `#the-second-part` can only be
8844 // the heading's own words, which is the rule every Markdown renderer
8845 // already follows and therefore the one a link was authored against.
8846 let src = "# Title\n\nintro\n\n## The Second Part\n\nbody\n\n## Third\n\nmore\n";
8847 let mut d = doc_with("locate_md", src);
8848 let hit = d.locate("the-second-part").expect("the heading's slug");
8849 assert!(d.source[hit.start..].starts_with("## The Second Part"));
8850 // Bounded by the next heading that isn't under it, so a peek shows the
8851 // section rather than only its title.
8852 assert_eq!(
8853 &d.source[hit.start..hit.end],
8854 "## The Second Part\n\nbody\n\n"
8855 );
8856
8857 // A subsection does not end its parent: `# Title` runs to `## Third`'s
8858 // sibling only because there is no other `#`, so it covers the lot.
8859 let title = d.locate("title").expect("the top heading");
8860 assert_eq!(title.end, d.source.len());
8861 }
8862
8863 #[test]
8864 fn locate_reads_a_djot_auto_id_however_the_link_spelled_it() {
8865 // djot mints `Some-Heading-Here`; a link to it is written
8866 // `#some-heading-here` by nearly everything that writes links. Both
8867 // spellings are one question.
8868 let src = "## Some Heading Here\n\nbody\n";
8869 let mut d = Doc::from_source(src.to_string(), Format::Djot).unwrap();
8870 let exact = d.locate("Some-Heading-Here").expect("djot's own spelling");
8871 let slugged = d.locate("some-heading-here").expect("the link's spelling");
8872 assert_eq!(exact, slugged);
8873 // The section, not the heading line — there is more to show than a title.
8874 assert_eq!(&d.source[exact.start..exact.end], src);
8875 }
8876
8877 #[test]
8878 fn locate_ignores_an_empty_locator_and_one_that_slugs_to_nothing() {
8879 let mut d = doc_with("locate_empty", "# Title\n\nbody\n");
8880 assert_eq!(d.locate(""), None);
8881 assert_eq!(d.locate(" "), None);
8882 // All punctuation: it names nothing, and must not be read as "match the
8883 // first heading whose slug is also empty".
8884 assert_eq!(d.locate("!!!"), None);
8885 }
8886
8887 #[test]
8888 fn locate_gives_a_duplicated_id_to_the_first_block_that_claims_it() {
8889 // The document's mistake, and the answer every other anchor
8890 // implementation gives — the alternative is for a link to mean whichever
8891 // of the two a walk happened to reach first.
8892 let src = "{#dup}\nfirst.\n\n{#dup}\nsecond.\n";
8893 let mut d = Doc::from_source(src.to_string(), Format::Djot).unwrap();
8894 let hit = d.locate("dup").expect("the first `{#dup}`");
8895 assert_eq!(d.source[hit.start..hit.end].trim_end(), "first.");
8896 }
8897
8898 #[test]
8899 fn insert_footnote_writes_both_halves_and_lands_the_caret_in_the_note() {
8900 // The button's whole job: a reference where the caret was, a definition
8901 // to give it meaning, and the caret waiting in the empty note so the
8902 // next keystroke is the note's first word.
8903 let mut d = doc_with("fn_insert", "A claim and more.\n");
8904 d.caret = 7; // just past "A claim"
8905 d.insert_footnote();
8906 assert!(
8907 d.source.starts_with("A claim[^1] and more."),
8908 "{:?}",
8909 d.source
8910 );
8911 assert!(
8912 d.source.contains("[^1]:"),
8913 "the definition too: {:?}",
8914 d.source
8915 );
8916 assert_eq!(d.status, None);
8917
8918 let reference = d.source.find("[^1]").unwrap();
8919 let note = d
8920 .footnote_at(reference + 2)
8921 .expect("the reference just written");
8922 assert_eq!(note.label, "1");
8923 assert_eq!(note.text.as_deref(), Some(""), "the note starts empty");
8924 assert_eq!(Some(d.caret), note.offset, "the caret waits in the note");
8925 // …and typing there is typing into the note, not near it.
8926 d.insert("the note");
8927 assert_eq!(
8928 d.footnote_at(reference + 2).and_then(|f| f.text),
8929 Some("the note".to_string())
8930 );
8931 }
8932
8933 #[test]
8934 fn insert_footnote_numbers_past_the_notes_already_written() {
8935 // A second press must not hand back a label somebody else is using: twig
8936 // reuses a defined label rather than appending a rival definition, so a
8937 // repeat of `1` would quietly point the new reference at the old note.
8938 let mut d = doc_with("fn_insert_number", "One[^1] two.\n\n[^1]: first\n");
8939 d.caret = 7; // past `[^1]`, before " two."
8940 d.insert_footnote();
8941 assert!(d.source.starts_with("One[^1][^2] two."), "{:?}", d.source);
8942 assert_eq!(d.source.matches("[^2]:").count(), 1);
8943 }
8944
8945 #[test]
8946 fn insert_footnote_counts_a_dangling_reference_and_ignores_a_named_one() {
8947 // `[^2]` with no definition is still a 2 that means something to whoever
8948 // wrote it — stepping over it would mint a note for their reference. A
8949 // word label takes no number, so it blocks none.
8950 let mut d = doc_with("fn_insert_dangling", "a[^2] b[^why] c\n\n[^why]: named\n");
8951 d.caret = d.source.find(" c").unwrap();
8952 d.insert_footnote();
8953 assert!(d.source.contains("[^1]:"), "1 is free: {:?}", d.source);
8954 assert!(
8955 d.source.starts_with("a[^2] b[^why][^1] c"),
8956 "{:?}",
8957 d.source
8958 );
8959 }
8960
8961 #[test]
8962 fn insert_footnote_marks_the_selection_rather_than_replacing_it() {
8963 // A reference annotates the words before it. Consuming the selection —
8964 // which is what an insert normally does — would delete the very claim
8965 // the author selected in order to footnote.
8966 let mut d = doc_with("fn_insert_sel", "A claim and more.\n");
8967 d.anchor = Some(2);
8968 d.caret = 7; // "claim" selected
8969 d.insert_footnote();
8970 assert!(
8971 d.source.starts_with("A claim[^1] and more."),
8972 "{:?}",
8973 d.source
8974 );
8975 }
8976
8977 #[test]
8978 fn a_note_just_written_still_knows_where_its_reference_is() {
8979 // The authoring loop in one test: press the button, type the note, ask to
8980 // go back. The caret ends at the note's last byte — which is the *end* of
8981 // the definition's span, the one offset the query used to exclude — so
8982 // this is where the round trip either works or doesn't.
8983 let mut d = doc_with("fn_insert_return", "A claim and more.\n");
8984 d.caret = 7;
8985 d.insert_footnote();
8986 d.insert("the note");
8987 assert_eq!(d.source, "A claim[^1] and more.\n\n[^1]: the note\n");
8988 let back = d
8989 .footnote_definition_at_caret()
8990 .expect("still in the note we just typed");
8991 assert_eq!(back.label, "1");
8992 // …and following it lands on the reference's label, where a reader's
8993 // return leg lands.
8994 assert_eq!(back.offset, Some(9));
8995 assert_eq!(&d.source[9..10], "1");
8996 }
8997
8998 #[test]
8999 fn insert_footnote_takes_one_undo_for_both_halves() {
9000 // twig writes the pair as a single edit; the point of that is here.
9001 let before = "A claim and more.\n";
9002 let mut d = doc_with("fn_insert_undo", before);
9003 d.caret = 7;
9004 d.insert_footnote();
9005 assert_ne!(d.source, before);
9006 d.undo();
9007 assert_eq!(d.source, before, "one undo takes back both halves");
9008 }
9009
9010 #[test]
9011 fn insert_footnote_refuses_a_format_that_cannot_spell_one() {
9012 // HTML is authorable — it spells the inline marks — and has no footnote.
9013 // The refusal says so rather than writing brackets that would render as
9014 // brackets.
9015 let src = "<p>A claim.</p>\n";
9016 let mut d = Doc::from_source(src.to_string(), Format::Html).unwrap();
9017 assert!(!Capabilities::of(Format::Html).footnote);
9018 d.caret = 5;
9019 d.insert_footnote();
9020 assert_eq!(d.source, src, "nothing written");
9021 assert!(d.status.is_some_and(|s| s.starts_with("footnote:")));
9022 }
9023
9024 #[test]
9025 fn insert_footnote_leaves_the_caret_on_a_real_stop_in_the_rich_view() {
9026 // The empty body is the one place this could go wrong: the definition
9027 // renders as a `[1] ` marker the caret cannot occupy, so a caret aimed a
9028 // byte early would draw up in the paragraph above the note it belongs to.
9029 let mut d = doc_in(View::Wysiwyg, "fn_insert_stop", "A claim and more.\n");
9030 d.place_caret(7, false);
9031 d.insert_footnote();
9032 d.build_visual(80); // the frame a frontend draws after the edit
9033 assert_eq!(
9034 d.vmap.snap_to_stop(d.caret),
9035 d.caret,
9036 "the caret sits on a stop"
9037 );
9038 let (row, _) = d.caret_pos();
9039 assert!(
9040 drawn_rows(&d)[row].contains("[1]"),
9041 "the caret is on the note's row, not above it: {:?}",
9042 drawn_rows(&d)
9043 );
9044 }
9045
9046 #[test]
9047 fn footnote_at_caret_resolves_a_reference_to_its_note() {
9048 // `[^1]` spans 7..11; its label byte is at 9. The definition follows a
9049 // blank line, as one has to.
9050 let mut d = doc_with("fn_at_caret", "A claim[^1] and more.\n\n[^1]: the note\n");
9051 d.caret = 9;
9052 let f = d
9053 .footnote_at_caret()
9054 .expect("the caret stands in a reference");
9055 assert_eq!(f.label, "1");
9056 assert_eq!(f.text.as_deref(), Some("the note"));
9057 // The offset points at the note's first word, not at the definition's
9058 // `[` — the marker is decoration with no caret stop on it.
9059 assert_eq!(f.offset, Some(29));
9060 assert_eq!(&d.source[29..37], "the note");
9061 // …and `end` closes the range, so a frontend can ask which rendered rows
9062 // the note occupies rather than re-deriving them from the text.
9063 assert_eq!(f.end, Some(37));
9064 assert_eq!(&d.source[f.offset.unwrap()..f.end.unwrap()], "the note");
9065 }
9066
9067 /// Two definitions in a row: each is its own note, and neither reaches into
9068 /// the other.
9069 ///
9070 /// A djot definition's span used to run past the blank line into the first
9071 /// byte of whatever followed, so this answered `"first note.\n\n["` — and the
9072 /// offsets named the *next* note's rows too, showing a reader two footnotes
9073 /// when they had asked about one. twig 3.1 ends the span after the block's
9074 /// own last line; the test outlives the workaround leaf carried for it.
9075 #[test]
9076 fn footnote_at_stops_a_note_at_the_definition_after_it() {
9077 let src = "Claim[^2a] and [^2b].\n\n[^2a]: first note.\n\n[^2b]: second note.\n";
9078 for format in [Format::Markdown, Format::Djot] {
9079 let mut d = Doc::from_source(src.to_string(), format).unwrap();
9080 d.caret = 7;
9081 let f = d.footnote_at_caret().expect("a reference");
9082 assert_eq!(f.text.as_deref(), Some("first note."), "in {format:?}");
9083 assert_eq!(
9084 &src[f.offset.unwrap()..f.end.unwrap()],
9085 "first note.",
9086 "in {format:?}"
9087 );
9088 }
9089 }
9090
9091 /// The other side of that boundary: a blank line *inside* a definition is
9092 /// interior to it, and the note keeps its second paragraph.
9093 ///
9094 /// This is what the old body scan cost. It stopped at the first line not
9095 /// indented under the note — a blank line is not — so a two-paragraph note
9096 /// came back as its first paragraph, and "go to note" framed half of it.
9097 /// Reading the span twig gives is both simpler and right.
9098 #[test]
9099 fn footnote_at_keeps_a_notes_second_paragraph() {
9100 let src = "Claim[^1].\n\n[^1]: first para.\n\n second para.\n\nAfter.\n";
9101 let mut d = Doc::from_source(src.to_string(), Format::Djot).unwrap();
9102 d.caret = 7;
9103 let f = d.footnote_at_caret().expect("a reference");
9104 assert_eq!(f.text.as_deref(), Some("first para.\n\n second para."));
9105 // And it stops there — `After.` is the next block, not more note.
9106 assert_eq!(
9107 &src[f.offset.unwrap()..f.end.unwrap()],
9108 f.text.as_deref().unwrap()
9109 );
9110 assert!(!f.text.as_deref().unwrap().contains("After"));
9111 }
9112
9113 #[test]
9114 fn footnote_at_bounds_a_note_whose_body_is_empty() {
9115 // `[^1]:` with nothing after it. The range is empty rather than
9116 // inverted, and still points inside the definition — which is what keeps
9117 // a frontend's row lookup from walking off into the block above.
9118 let src = "A claim[^1].\n\n[^1]:\n";
9119 let mut d = doc_with("fn_empty_body", src);
9120 d.caret = 9;
9121 let f = d.footnote_at_caret().expect("a reference");
9122 assert_eq!(f.text.as_deref(), Some(""));
9123 assert_eq!(f.offset, f.end, "an empty note is an empty range");
9124 assert!(f.offset.unwrap() >= src.find("[^1]:").unwrap());
9125 }
9126
9127 #[test]
9128 fn footnote_at_caret_ignores_a_caret_that_stands_in_no_reference() {
9129 let mut d = doc_with(
9130 "fn_at_caret_none",
9131 "A claim[^1] and more.\n\n[^1]: the note\n",
9132 );
9133 d.caret = 2; // in the prose
9134 assert_eq!(d.footnote_at_caret(), None);
9135 }
9136
9137 #[test]
9138 fn footnote_at_caret_is_not_a_link_query_and_vice_versa() {
9139 // The two are deliberately separate: a reference names a note in this
9140 // document, a link names somewhere to leave for, and answering one with
9141 // the other is what made a reference click do nothing at all.
9142 let mut d = doc_with("fn_vs_link", "a[^1] b [t](https://x.dev)\n\n[^1]: note\n");
9143 d.caret = 3; // the `1` of `[^1]`
9144 assert!(d.footnote_at_caret().is_some());
9145 assert_eq!(
9146 d.link_destination_at_caret(),
9147 None,
9148 "a reference is not a link"
9149 );
9150
9151 d.caret = 10; // inside the link's label
9152 assert_eq!(d.footnote_at_caret(), None, "a link is not a reference");
9153 assert_eq!(
9154 d.link_destination_at_caret().as_deref(),
9155 Some("https://x.dev")
9156 );
9157 }
9158
9159 #[test]
9160 fn footnote_at_caret_reports_an_undefined_reference_rather_than_nothing() {
9161 // A `[^99]` the document never defines is a real state — a note deleted
9162 // out from under its reference — and the label is what lets a frontend
9163 // say so. `None` here would be indistinguishable from "not on a
9164 // reference", which is the wrong thing to tell a reader.
9165 let mut d = doc_with("fn_undefined", "A claim[^99] and more.\n");
9166 d.caret = 9;
9167 let f = d
9168 .footnote_at_caret()
9169 .expect("the reference is still a reference");
9170 assert_eq!(f.label, "99");
9171 assert_eq!(f.text, None);
9172 assert_eq!(f.offset, None);
9173 }
9174
9175 #[test]
9176 fn footnote_at_caret_reads_a_word_label_and_a_multiline_note() {
9177 // Labels are not always numbers, and a note's body runs past its first
9178 // line — the indented continuation belongs to the note, so it comes back
9179 // with it (source bytes, verbatim, as documented).
9180 let src = "see[^note] here\n\n[^note]: first line\n second line\n";
9181 let mut d = doc_with("fn_word_label", src);
9182 d.caret = 6;
9183 let f = d
9184 .footnote_at_caret()
9185 .expect("the caret stands in a reference");
9186 assert_eq!(f.label, "note");
9187 assert_eq!(f.text.as_deref(), Some("first line\n second line"));
9188 }
9189
9190 #[test]
9191 fn footnote_at_answers_for_an_offset_the_caret_is_nowhere_near() {
9192 // The point of the offset form: a pointer hovering a reference asks what
9193 // note it names, and must not drag the caret along to ask.
9194 let mut d = doc_with("fn_at_off", "A claim[^1] and more.\n\n[^1]: the note\n");
9195 d.caret = 0;
9196 let f = d.footnote_at(9).expect("offset 9 stands in the reference");
9197 assert_eq!(f.label, "1");
9198 assert_eq!(f.text.as_deref(), Some("the note"));
9199 assert_eq!(d.caret, 0, "asking must not move the caret");
9200 assert_eq!(d.footnote_at(2), None, "offset 2 is prose");
9201 }
9202
9203 #[test]
9204 fn footnote_definition_at_caret_points_back_at_the_reference() {
9205 // The return leg. `[^1]` spans 7..11, so its label — the only byte of it
9206 // the caret can rest on — is at 9.
9207 let mut d = doc_with("fn_def", "A claim[^1] and more.\n\n[^1]: the note\n");
9208 d.caret = 30; // inside the note's body
9209 let f = d
9210 .footnote_definition_at_caret()
9211 .expect("the caret stands in a definition");
9212 assert_eq!(f.label, "1");
9213 assert_eq!(f.offset, Some(9));
9214 assert_eq!(&d.source[7..11], "[^1]");
9215 }
9216
9217 #[test]
9218 fn footnote_definition_at_covers_where_a_go_to_note_actually_lands() {
9219 // The two legs have to meet: wherever `footnote_at` sends the caret, the
9220 // definition query must answer for — otherwise arriving at a note leaves
9221 // the reader somewhere the way back isn't offered.
9222 let src = "A claim[^1] and more.\n\n[^1]: the note\n";
9223 let mut d = doc_with("fn_def_marker", src);
9224 let landed = d.footnote_at(9).unwrap().offset.unwrap();
9225 assert_eq!(
9226 d.footnote_definition_at(landed).and_then(|f| f.offset),
9227 Some(9),
9228 "the note a reference sends you to offers the way back"
9229 );
9230 }
9231
9232 #[test]
9233 fn footnote_definition_at_caret_ignores_prose_and_the_reference_itself() {
9234 // The two queries answer for disjoint places, which is what lets one
9235 // gesture mean "down to the note" in one and "back up" in the other
9236 // without either having to remember which way the reader is going.
9237 let mut d = doc_with("fn_def_none", "A claim[^1] and more.\n\n[^1]: the note\n");
9238 d.caret = 2; // prose
9239 assert_eq!(d.footnote_definition_at_caret(), None);
9240 d.caret = 9; // the reference
9241 assert_eq!(d.footnote_definition_at_caret(), None);
9242 assert!(
9243 d.footnote_at_caret().is_some(),
9244 "which is the reference's own query"
9245 );
9246 }
9247
9248 #[test]
9249 fn footnote_definition_at_caret_reports_an_orphan_note_rather_than_nothing() {
9250 // Nothing cites `[^2]`. Answering `None` would say "you are not in a
9251 // note", which is false and leaves a frontend unable to explain why the
9252 // way back is missing.
9253 let src = "A claim[^1].\n\n[^1]: cited\n\n[^2]: orphan\n";
9254 let mut d = doc_with("fn_def_orphan", src);
9255 d.caret = src.find("orphan").unwrap();
9256 let f = d
9257 .footnote_definition_at_caret()
9258 .expect("an orphan is still a definition");
9259 assert_eq!(f.label, "2");
9260 assert_eq!(f.offset, None);
9261 }
9262
9263 #[test]
9264 fn footnote_definition_at_caret_returns_to_the_first_of_repeated_references() {
9265 // One label, cited twice. The first is where the reader most likely came
9266 // from, and the only answer that doesn't depend on how they got here.
9267 let src = "One[^a] and two[^a].\n\n[^a]: the note\n";
9268 let mut d = doc_with("fn_def_repeat", src);
9269 d.caret = src.find("the note").unwrap();
9270 let f = d.footnote_definition_at_caret().expect("a definition");
9271 assert_eq!(
9272 f.offset,
9273 Some(5),
9274 "the first `[^a]`'s label, not the second's"
9275 );
9276 assert_eq!(&src[3..7], "[^a]");
9277 }
9278
9279 #[test]
9280 fn footnote_navigation_is_a_round_trip_through_placed_carets() {
9281 // Down and back up, each leg found from the document rather than from a
9282 // memory of the other — so it still works for a reader who scrolled to
9283 // the notes instead of jumping there.
9284 //
9285 // `place_caret` rather than assigning `caret`, because that is what a
9286 // frontend calls: it snaps to a real caret stop, and a jump that lands
9287 // on a byte the caret can't rest on would arrive somewhere the return
9288 // leg no longer answers for. `build_map` first, since snapping is a
9289 // no-op until the map exists — which is exactly how this went unnoticed
9290 // when the offsets pointed at the `[^` markers.
9291 let mut d = doc_with("fn_round", "A claim[^1] and more.\n\n[^1]: the note\n");
9292 d.build_map(None);
9293 d.place_caret(9, false);
9294 let down = d
9295 .footnote_at_caret()
9296 .expect("a reference")
9297 .offset
9298 .expect("a note");
9299 d.place_caret(down, false);
9300 let up = d
9301 .footnote_definition_at_caret()
9302 .expect("a definition")
9303 .offset
9304 .expect("a reference");
9305 d.place_caret(up, false);
9306 assert_eq!(d.caret, up, "the way back is a stop the caret can occupy");
9307 assert_eq!(
9308 d.footnote_at_caret().expect("back on the reference").label,
9309 "1"
9310 );
9311 }
9312
9313 #[test]
9314 fn insert_link_hands_the_destination_to_twig_raw() {
9315 // Escaping is twig's, and format-specific: Markdown ends a destination
9316 // at the first space and needs the `<…>` form, where djot would read
9317 // those angle brackets as part of the URL.
9318 let mut d = doc_with("link_space", "word\n");
9319 d.anchor = Some(0);
9320 d.caret = 4;
9321 d.insert_link("a b");
9322 assert_eq!(d.source, "[word](<a b>)\n");
9323 }
9324
9325 #[test]
9326 fn insert_link_reports_a_destination_no_format_can_carry() {
9327 let mut d = doc_with("link_bad", "word\n");
9328 d.anchor = Some(0);
9329 d.caret = 4;
9330 d.insert_link("a\nb");
9331 assert_eq!(d.source, "word\n"); // untouched, not quietly rewritten
9332 assert!(
9333 d.status.is_some(),
9334 "InvalidArgument should reach the status line"
9335 );
9336 assert!(!d.dirty);
9337 }
9338
9339 #[test]
9340 fn insert_link_works_in_wysiwyg_view() {
9341 let mut d = wysiwyg_doc("link_wys", "word here\n");
9342 d.anchor = Some(0);
9343 d.caret = 4;
9344 d.insert_link("http://x.dev");
9345 assert_eq!(d.source, "[word](http://x.dev) here\n");
9346 assert_eq!(d.selected_text(), Some("word"));
9347 // The map the caret has to keep riding is rebuilt each frame; motion
9348 // over the fresh one must still land on a real stop (the debug_assert).
9349 d.build_visual(80);
9350 d.move_right(false);
9351 d.move_left(false);
9352 }
9353
9354 #[test]
9355 fn click_maps_a_row_col_to_a_byte_offset() {
9356 let mut d = doc_with("click", "ab\ncd\n");
9357 d.click(1, 1, false); // row 1 ("cd"), col 1 -> the 'd'
9358 assert_eq!(d.caret, 4);
9359 }
9360
9361 // A pixel-hit-test placement (the GUI's `place_caret`) must land on a caret
9362 // stop just as the `(row, col)` click path does, so the caret can never come
9363 // to rest in the blank gap between two paragraphs — where it would draw in one
9364 // place and type in another.
9365 #[test]
9366 fn place_caret_snaps_out_of_the_blank_gap_between_paragraphs() {
9367 // "A\n\nB": offset 2 is the gap the paragraph break is drawn with, not a
9368 // caret stop (stops are 0,1,3,4).
9369 let mut d = wysiwyg_doc("place_gap", "A\n\nB");
9370 assert!(!d.vmap.is_stop(2), "offset 2 should be an unreachable gap");
9371 d.place_caret(2, false);
9372 assert!(d.vmap.is_stop(d.caret), "caret {} is not a stop", d.caret);
9373 assert_eq!(d.caret, 1, "should snap to the end of the paragraph above");
9374 }
9375
9376 #[test]
9377 fn place_caret_dragging_through_the_gap_keeps_selection_on_stops() {
9378 let mut d = wysiwyg_doc("place_gap_drag", "A\n\nB");
9379 d.place_caret(0, false); // anchor at the start of "A"
9380 d.place_caret(2, true); // drag into the gap
9381 assert!(d.vmap.is_stop(d.caret), "caret {} is not a stop", d.caret);
9382 let (s, e) = d.selection().expect("a selection");
9383 assert!(
9384 d.vmap.is_stop(s) && d.vmap.is_stop(e),
9385 "selection {s}..{e} off a stop"
9386 );
9387 }
9388
9389 #[test]
9390 fn place_caret_on_a_real_stop_is_left_untouched() {
9391 let mut d = wysiwyg_doc("place_stop", "A\n\nB");
9392 d.place_caret(3, false); // the start of "B" — a genuine stop
9393 assert_eq!(d.caret, 3);
9394 }
9395
9396 // An *empty paragraph* (two blank lines, an intentional blank line the user
9397 // opened) is a real caret stop, unlike the gap — a click into it must stay.
9398 #[test]
9399 fn place_caret_rests_in_an_empty_paragraph() {
9400 let mut d = wysiwyg_doc("place_empty_para", "A\n\n\n\nB");
9401 let empty = 3; // the navigable empty row's offset (stops: 0,1,3,5,6)
9402 assert!(d.vmap.is_stop(empty));
9403 d.place_caret(empty, false);
9404 assert_eq!(d.caret, empty);
9405 }
9406
9407 fn wysiwyg_doc(name: &str, body: &str) -> Doc {
9408 doc_in(View::Wysiwyg, name, body)
9409 }
9410
9411 /// How many list items the source actually parses into — the check that a
9412 /// marker Leaf wrote is a marker the format agrees is one.
9413 fn list_items(doc: &mut Doc) -> usize {
9414 doc.editor
9415 .nodes()
9416 .unwrap()
9417 .iter()
9418 .filter(|n| n.kind == Kind::ListItem || n.kind == Kind::TaskListItem)
9419 .count()
9420 }
9421
9422 /// A from-scratch, cache-free WYSIWYG map for `source` — the ground truth the
9423 /// incremental (`build_spliced` / `build_cached`) path must always match.
9424 fn reference_map(source: &str) -> crate::wysiwyg::VisualMap {
9425 reference_map_revealing(source, None)
9426 }
9427
9428 /// [`reference_map`] with a reveal line — the ground truth for the
9429 /// `MarkupMode::Full` builds, where the map is a function of the caret's
9430 /// line as well as the text.
9431 fn reference_map_revealing(
9432 source: &str,
9433 reveal: Option<Range<usize>>,
9434 ) -> crate::wysiwyg::VisualMap {
9435 // The same parse `Doc` uses. With twig's plain defaults instead, the two
9436 // sides disagree on what the *document* is before the renderer is even
9437 // reached — a bare `:word` is a text directive to one and prose to the
9438 // other — and the mismatch reads as a splice bug that isn't one.
9439 let mut ed =
9440 twig::Editor::new_ext(source.as_bytes(), Format::Markdown, parse_extensions()).unwrap();
9441 let nodes = ed.nodes().unwrap();
9442 crate::wysiwyg::build(
9443 &nodes,
9444 source,
9445 None,
9446 false,
9447 &std::collections::HashMap::new(),
9448 reveal,
9449 )
9450 }
9451
9452 fn maps_differ(a: &crate::wysiwyg::VisualMap, b: &crate::wysiwyg::VisualMap) -> bool {
9453 if a.rows.len() != b.rows.len() {
9454 return true;
9455 }
9456 for (ra, rb) in a.rows.iter().zip(&b.rows) {
9457 if ra.end_src != rb.end_src || ra.glyphs.len() != rb.glyphs.len() {
9458 return true;
9459 }
9460 for (ga, gb) in ra.glyphs.iter().zip(&rb.glyphs) {
9461 if ga.ch != gb.ch || ga.src != gb.src {
9462 return true;
9463 }
9464 }
9465 }
9466 false
9467 }
9468
9469 #[test]
9470 fn incremental_build_matches_a_fresh_build_across_edits() {
9471 // Every `Doc` edit rebuilds through `build_spliced` (the single-block
9472 // fast path, gated on twig's `dirty_range`) or falls back to
9473 // `build_cached`. After each edit the map must be byte-identical to a
9474 // from-scratch build — this is the correctness net under the splice.
9475 let docs = [
9476 "# Title\n\nThe quick brown fox jumps.\n\nAnother paragraph here.\n\n- a\n- b\n",
9477 "para one\n\n> quote **bold** text\n> continued line\n\ntail paragraph\n",
9478 "alpha\n\nbeta\n\ngamma\n\ndelta\n\nepsilon\n\nzeta\n",
9479 // A footnote definition is a root beside `doc`, merged back into the
9480 // top-level list by `wysiwyg::top_blocks`. The random edits below
9481 // make and unmake definitions as they go (a deleted `:` turns one
9482 // back into a paragraph, and vice versa), which is exactly the
9483 // structural churn the splice path has to notice and bail out of.
9484 "text[^1] here\n\n[^1]: the note\n\nmore text[^b]\n\n[^b]: second\n",
9485 ];
9486 // A deterministic mix: mostly single characters (which stay inside one
9487 // block → splice), plus edits that reshape structure (a paragraph break,
9488 // a heading marker, a code fence → fallback), so both paths are exercised.
9489 let inserts = ["x", "y", "\n\n", "#", "`", " ", "z"];
9490 for src in docs {
9491 let mut d = wysiwyg_doc("diff", src);
9492 d.build_visual_unwrapped();
9493 wysiwyg::assert_maps_eq(&d.vmap, &reference_map(&d.source), "initial");
9494
9495 for step in 0..60usize {
9496 let len = d.source.len();
9497 let raw = (step * 13 + 5) % (len + 1);
9498 let pos = (raw..=len).find(|&i| d.source.is_char_boundary(i)).unwrap();
9499 let pre = d.source.clone();
9500 let action;
9501 if step % 3 == 0 && pos < len {
9502 let end = (pos + 1..=len)
9503 .find(|&i| d.source.is_char_boundary(i))
9504 .unwrap();
9505 action = format!("delete [{pos},{end})");
9506 d.edit(pos, end, "");
9507 } else {
9508 let ins = inserts[step % inserts.len()];
9509 action = format!("insert {ins:?} @ {pos}");
9510 d.edit(pos, pos, ins);
9511 }
9512 d.build_visual_unwrapped();
9513 if maps_differ(&d.vmap, &reference_map(&d.source)) {
9514 panic!(
9515 "FIRST MISMATCH at step {step}: {action}\n pre = {pre:?}\n post = {:?}",
9516 d.source
9517 );
9518 }
9519 }
9520 }
9521 }
9522
9523 /// A frontend is handed [`Doc::vmap`] and may present it differently:
9524 /// leaf-ratatui splices blank filler rows under an oversized heading so the
9525 /// raster it paints there has somewhere to stand, and leaves them in the map
9526 /// because the caret and the mouse both read it between frames. The splice
9527 /// path addresses that map by *row index*, against the block layout the last
9528 /// build recorded — so handed a map with rows in it that no block owns, it
9529 /// laid the re-rendered block over one of the fillers and carried the rows
9530 /// the block really occupied into the suffix. One stranded copy of the
9531 /// edited line, and everything below it a row further down, per keystroke.
9532 ///
9533 /// A map that isn't the one the layout describes is a map this path can't
9534 /// patch, whoever changed it and for whatever reason. It rebuilds instead.
9535 #[test]
9536 fn an_edit_over_a_map_a_frontend_reshaped_rebuilds_it_whole() {
9537 let mut d = wysiwyg_doc("reshaped", "# Title\n\nThe quick brown fox jumps.\n");
9538 d.build_visual_unwrapped();
9539
9540 // Stand in for the heading filler rows: two blank rows past the heading
9541 // that no block accounts for. Cloning a real row keeps every field
9542 // plausible — it is the row *count* the splice can't survive.
9543 let filler = d.vmap.rows[0].clone();
9544 d.vmap.rows.insert(1, filler.clone());
9545 d.vmap.rows.insert(1, filler);
9546
9547 // An edit inside the last block: the single-block case the splice path
9548 // is for, and the one the frontend hits on every keystroke.
9549 let at = d.source.len() - 1;
9550 d.edit(at, at, "!");
9551 d.build_visual_unwrapped();
9552
9553 wysiwyg::assert_maps_eq(&d.vmap, &reference_map(&d.source), "after the edit");
9554 }
9555
9556 #[test]
9557 fn incremental_build_matches_a_fresh_build_under_full_reveal() {
9558 // The same correctness net as `incremental_build_matches_a_fresh_build_
9559 // across_edits`, under `MarkupMode::Full` — where the map depends on
9560 // the caret's *line* as well as the text, so the two caches have a new
9561 // way to be wrong. Both are exercised: the block cache can hand back
9562 // rows built for a line that is no longer the revealed one, and the
9563 // splice path can reuse a suffix that still has yesterday's line raw.
9564 //
9565 // Caret motion is interleaved with the edits deliberately, because a
9566 // caret that only ever moved with the edit would never cross a line
9567 // without also dirtying it — the case where a stale reveal survives.
9568 let docs = [
9569 "# Title\n\n*one* and **two**\n\n[lk](http://x) and `code`\n\n- a *b*\n",
9570 "para *em* one\n\n> quote **bold** text\n\ntail ~~del~~ paragraph\n",
9571 ];
9572 let inserts = ["x", "*", "\n\n", "#", "`", " ", "_"];
9573 for src in docs {
9574 let mut d = wysiwyg_doc("reveal_diff", src);
9575 d.set_markup_mode(MarkupMode::Full);
9576
9577 for step in 0..60usize {
9578 let len = d.source.len();
9579 let raw = (step * 13 + 5) % (len + 1);
9580 let pos = (raw..=len).find(|&i| d.source.is_char_boundary(i)).unwrap();
9581 let pre = d.source.clone();
9582 let action;
9583 if step % 3 == 0 && pos < len {
9584 let end = (pos + 1..=len)
9585 .find(|&i| d.source.is_char_boundary(i))
9586 .unwrap();
9587 action = format!("delete [{pos},{end})");
9588 d.edit(pos, end, "");
9589 } else {
9590 let ins = inserts[step % inserts.len()];
9591 action = format!("insert {ins:?} @ {pos}");
9592 d.edit(pos, pos, ins);
9593 }
9594 // Walk the caret somewhere else in the document, independently
9595 // of where the edit landed.
9596 let want = (step * 29 + 11) % (d.source.len() + 1);
9597 d.caret = (want..=d.source.len())
9598 .find(|&i| d.source.is_char_boundary(i))
9599 .unwrap();
9600 d.build_visual_unwrapped();
9601
9602 let want = reference_map_revealing(&d.source, d.reveal_line());
9603 if maps_differ(&d.vmap, &want) {
9604 panic!(
9605 "FIRST MISMATCH at step {step}: {action}, caret {}\n pre = {pre:?}\n post = {:?}",
9606 d.caret, d.source
9607 );
9608 }
9609 }
9610 }
9611 }
9612
9613 #[test]
9614 fn caret_motion_across_lines_rebuilds_only_under_full() {
9615 // The cache-key change has to earn its keep in both directions: `Full`
9616 // must rebuild when the caret changes line (or the reveal would never
9617 // move), and the hidden modes must *not* (or every arrow key would pay
9618 // for a feature they don't use). The existing `cache_motion` test pins
9619 // the second for the default mode; this pins the pair against a mode
9620 // change alone.
9621 let body = "*one* here\n\n*two* there\n";
9622
9623 let mut full = doc_in(View::Wysiwyg, "motion_full", body);
9624 full.set_markup_mode(MarkupMode::Full);
9625 caret_at(&mut full, "one");
9626 let before = full.revision();
9627 caret_at(&mut full, "two");
9628 assert_eq!(full.revision(), before, "motion is not an edit");
9629 assert!(
9630 drawn_rows(&full).iter().any(|r| r == "*two* there"),
9631 "the map followed the caret: {:?}",
9632 drawn_rows(&full)
9633 );
9634
9635 let mut hidden = doc_in(View::Wysiwyg, "motion_hidden", body);
9636 caret_at(&mut hidden, "one");
9637 let key = hidden.vmap_key.clone();
9638 caret_at(&mut hidden, "two");
9639 assert_eq!(
9640 hidden.vmap_key, key,
9641 "a hidden mode rebuilds nothing on motion"
9642 );
9643 }
9644
9645 #[test]
9646 fn wysiwyg_down_crosses_a_paragraph_boundary() {
9647 // Regression: the blank separator row used to share the previous
9648 // paragraph's end offset, so Down got pinned at the boundary (while Up
9649 // still crossed). Both directions must step through it symmetrically.
9650 //
9651 // It's now stepped *over* rather than onto: the blank line between two
9652 // paragraphs is the boundary being drawn, not a line of the document, so
9653 // one press of Down crosses it. The goal column survives the crossing —
9654 // col 3 at the end of "abc" is col 3 at the end of "def".
9655 let mut d = wysiwyg_doc("wys_down", "abc\n\ndef\n");
9656 d.caret = 3; // end of "abc" (row 0)
9657 d.move_down(false);
9658 assert_eq!(d.caret_pos().0, 2, "Down should reach the second paragraph");
9659 assert_eq!(d.caret, 8); // end of "def", col 3 kept
9660 d.move_up(false);
9661 assert_eq!(d.caret_pos().0, 0, "Up should come back symmetrically");
9662 assert_eq!(d.caret, 3);
9663 }
9664
9665 #[test]
9666 fn wysiwyg_up_and_down_are_inverse_across_paragraphs() {
9667 // The second Up and the second Down here run off the ends of the
9668 // document, which is no longer a place a press is swallowed: they carry
9669 // the caret to the start and the end of the text. The claim in the
9670 // middle — that a Down retraces the Up that crossed the paragraph gap —
9671 // is the one this test is for, and it is asserted where it is made.
9672 let mut d = wysiwyg_doc("wys_updown", "abc\n\ndef\n");
9673 d.caret = 5; // start of "def"
9674 let start = d.caret_pos();
9675 d.move_up(false);
9676 assert_eq!(d.caret_pos().0, 0, "Up reaches the first paragraph");
9677 d.move_up(false);
9678 assert_eq!(d.caret, 0, "a second Up runs on to the document's start");
9679 d.move_down(false);
9680 assert_eq!(d.caret_pos(), start, "Down retraces Up exactly");
9681 d.move_down(false);
9682 assert_eq!(d.caret, 8, "a second Down runs on to the document's end");
9683 }
9684
9685 #[test]
9686 fn wysiwyg_new_paragraph_shows_before_typing() {
9687 // Regression: two Enters at the end of a paragraph produced trailing
9688 // newlines with no AST node, so the caret appeared stuck on the old line
9689 // until a character was typed. It must ride down onto the new line now.
9690 let mut d = doc_with("wys_newpara", "abc\n");
9691 d.view = View::Wysiwyg;
9692 d.caret = 3;
9693 d.insert("\n");
9694 d.insert("\n"); // source is now "abc\n\n\n", caret at 5
9695 assert_eq!(d.source, "abc\n\n\n");
9696 d.build_visual(80);
9697 let (row, _) = d.caret_pos();
9698 assert!(
9699 row >= 2,
9700 "caret should have moved down to the new line, got row {row}"
9701 );
9702 assert!(
9703 d.vmap.num_rows() >= 3,
9704 "the blank lines should render as rows"
9705 );
9706 }
9707
9708 #[test]
9709 fn wysiwyg_enter_between_paragraphs_lands_on_an_empty_line() {
9710 // The reported bug: Enter at the end of a paragraph that has another
9711 // paragraph below put the caret at the *start of the next paragraph* —
9712 // the empty paragraph it opened had no row, so the caret snapped onto
9713 // "World". It must now sit on its own empty line, with a blank spacer
9714 // above it (the paragraph gap).
9715 let mut d = wysiwyg_doc("wys_gap_mid", "Hello\n\nWorld\n");
9716 d.caret = 5; // end of "Hello"
9717 d.newline();
9718 d.build_visual(80);
9719 let (row, col) = d.caret_pos();
9720 assert_eq!(col, 0, "caret should start an empty line, not sit in text");
9721 assert_eq!(
9722 d.vmap.row_width(row),
9723 0,
9724 "caret's row must be empty, not 'World'"
9725 );
9726 assert!(
9727 row >= 2,
9728 "a blank spacer row should sit above the caret, got row {row}"
9729 );
9730 // The row above the caret is a real (empty) gap, and "Hello" stays put.
9731 assert_eq!(
9732 d.vmap.row_width(row - 1),
9733 0,
9734 "the row above the caret is a gap"
9735 );
9736 let row0: String = d.vmap.rows[0].glyphs.iter().map(|g| g.ch).collect();
9737 assert_eq!(row0, "Hello", "the paragraph above the caret must not move");
9738 }
9739
9740 #[test]
9741 fn wysiwyg_enter_at_eof_shows_a_gap_before_typing() {
9742 // At the document end a single Enter must also show the paragraph gap —
9743 // a blank spacer row above the caret — so the layout already matches how
9744 // it will look once the new paragraph has text.
9745 let mut d = wysiwyg_doc("wys_gap_eof", "Hello");
9746 d.caret = 5; // end of "Hello", no trailing newline
9747 d.newline(); // source becomes "Hello\n\n"
9748 d.build_visual(80);
9749 let (row, col) = d.caret_pos();
9750 assert_eq!(col, 0);
9751 assert!(
9752 row >= 2,
9753 "caret should sit below a blank spacer, got row {row}"
9754 );
9755 assert_eq!(
9756 d.vmap.row_width(row - 1),
9757 0,
9758 "the row above the caret is a gap"
9759 );
9760 }
9761
9762 #[test]
9763 fn wysiwyg_typing_after_enter_does_not_shift_the_caret_row() {
9764 // The spacer is view-only: typing the new paragraph must not reflow the
9765 // caret onto a different row — the transient view already matched the
9766 // settled one.
9767 let mut d = wysiwyg_doc("wys_no_reflow", "Hello\n\nWorld\n");
9768 d.caret = 5;
9769 d.newline();
9770 d.build_visual(80);
9771 let before = d.caret_pos();
9772 d.insert("New");
9773 d.build_visual(80);
9774 let after = d.caret_pos();
9775 assert_eq!(
9776 after.0, before.0,
9777 "typing must not move the caret to another row ({before:?} -> {after:?})"
9778 );
9779 }
9780
9781 #[test]
9782 fn wysiwyg_hides_frontmatter_from_the_caret_and_copy() {
9783 let fm = "---\ntitle: hi\n---\n";
9784 let body = format!("{fm}# leaf\n\nbody\n");
9785 let mut d = wysiwyg_doc("wys_fm", &body);
9786 // Opening lifts the caret out of the now-hidden frontmatter.
9787 assert_eq!(
9788 d.caret,
9789 fm.len(),
9790 "caret should start at the first real block"
9791 );
9792 // Left at the content start can't step back into frontmatter.
9793 d.move_left(false);
9794 assert_eq!(d.caret, fm.len(), "left must not enter frontmatter");
9795 // Doc-start lands on the content floor, not offset 0.
9796 d.move_doc_start(false);
9797 assert_eq!(d.caret, fm.len());
9798 // Select-all + copy never include the frontmatter bytes.
9799 d.select_all();
9800 let sel = d.selected_text().unwrap().to_string();
9801 assert!(!sel.contains("title"), "copy leaked frontmatter: {sel:?}");
9802 assert!(
9803 sel.starts_with("# leaf"),
9804 "selection should begin at content: {sel:?}"
9805 );
9806 }
9807
9808 #[test]
9809 fn typing_in_a_frontmatter_only_document_lands_after_the_frontmatter() {
9810 // A fresh note is frontmatter and nothing else. With no rendered block
9811 // to floor the caret it opened at offset 0 — before the opening `---` —
9812 // so the first keystroke wrote itself in front of the metadata and the
9813 // file came out as `This---\ntitle: …`.
9814 let fm = "---\ntitle: 2026-08-29\nid: f8s32cd\n---\n";
9815 let mut d = wysiwyg_doc("wys_fm_only", fm);
9816 assert_eq!(d.caret, fm.len(), "caret must open past the frontmatter");
9817 // Nothing is rendered, so the caret draws at the origin of an empty view
9818 // — the same place an empty document puts it.
9819 assert_eq!(d.caret_pos(), (0, 0));
9820 d.insert("This");
9821 assert_eq!(d.source, format!("{fm}This"));
9822 }
9823
9824 /// `select_range` is the verb for a range a host already knows the bytes of,
9825 /// so it must not snap — and must still hold every invariant `place_caret`
9826 /// holds, the frontmatter floor above all.
9827 #[test]
9828 fn select_range_takes_the_range_as_given_but_still_floors_it() {
9829 let fm = "---\ntitle: foo\n---\n\n";
9830 let body = format!("{fm}body foo here\n");
9831 let mut d = wysiwyg_doc("wys_select_range", &body);
9832
9833 // The `foo` in the body: taken exactly, not snapped to a caret stop.
9834 let at = body.rfind("foo").unwrap();
9835 d.select_range(at, at + 3);
9836 assert_eq!(d.selection(), Some((at, at + 3)));
9837 assert_eq!(d.selected_text(), Some("foo"));
9838
9839 // The `foo` in the hidden frontmatter: below the floor, so both ends
9840 // come up to it rather than parking the caret in the metadata, where a
9841 // later keystroke would rewrite `title:`.
9842 let hidden = body.find("foo").unwrap();
9843 assert!(hidden < d.vmap.content_start);
9844 d.select_range(hidden, hidden + 3);
9845 assert!(
9846 d.caret >= d.vmap.content_start && d.anchor.unwrap() >= d.vmap.content_start,
9847 "a range under the floor must not leave the caret in the frontmatter"
9848 );
9849
9850 // Past the end, and mid-character, are both brought back to something
9851 // sliceable rather than panicking the next reader of the range.
9852 let multi = wysiwyg_doc("wys_select_range_utf8", "héllo\n");
9853 let mut d = multi;
9854 d.select_range(2, 9_999);
9855 assert_eq!(d.caret, d.source.len());
9856 assert!(d.source.is_char_boundary(d.anchor.unwrap()));
9857 assert!(d.source.is_char_boundary(d.caret));
9858 }
9859
9860 /// The bug `select_range` exists for: a match butting up against a hidden
9861 /// delimiter. `place_caret` snaps to the nearest *visible* stop, which is
9862 /// the one before the `**`.
9863 #[test]
9864 fn select_range_does_not_snap_off_a_hidden_delimiter() {
9865 let mut d = wysiwyg_doc("wys_select_range_bold", "a **needle** in it\n");
9866 let at = d.source.find("needle").unwrap();
9867 d.select_range(at, at + 6);
9868 assert_eq!(d.selected_text(), Some("needle"), "not \"needl\"");
9869 }
9870
9871 #[test]
9872 fn wysiwyg_backspace_at_content_start_leaves_frontmatter_intact() {
9873 // Backspace deletes `prev_boundary..caret` directly; at the first real
9874 // block that boundary is inside the hidden frontmatter, so it must be a
9875 // no-op rather than eating the closing `---`.
9876 let fm = "---\ntitle: hi\n---\n";
9877 let body = format!("{fm}leaf\n");
9878 let mut d = wysiwyg_doc("wys_fm_bs", &body);
9879 assert_eq!(d.caret, fm.len());
9880 d.backspace();
9881 assert_eq!(d.source, body, "backspace must not touch frontmatter");
9882 d.delete_word_back();
9883 assert_eq!(
9884 d.source, body,
9885 "word-delete must not touch frontmatter either"
9886 );
9887 }
9888
9889 #[test]
9890 fn wysiwyg_edits_inside_a_vis_directive_block_without_disturbing_its_fences() {
9891 // diaryx's `:::vis{.audience}` visibility block — any `:::name{.class}`
9892 // fenced div, really, since core parses these on for every document
9893 // now (`parse_extensions`). The container is a `directive` node, an
9894 // `is_block_container` kind like `block_quote`, so the caret works
9895 // inside its child paragraph exactly as it would inside a quote: typing
9896 // edits the paragraph, and the `:::vis{...}` / `:::` fences round-trip
9897 // untouched.
9898 let body = ":::vis{.public .family}\nhello\n:::\nafter\n";
9899 let mut d = wysiwyg_doc("wys_vis", body);
9900 d.caret = body.find("hello").unwrap() + "hello".len();
9901 d.insert("!");
9902 assert_eq!(
9903 d.source, ":::vis{.public .family}\nhello!\n:::\nafter\n",
9904 "typing inside the block edits its content in place"
9905 );
9906 assert!(
9907 d.source.contains(":::vis{.public .family}"),
9908 "opening fence survives"
9909 );
9910 assert!(d.source.contains(":::\nafter"), "closing fence survives");
9911 }
9912
9913 #[test]
9914 fn source_view_still_reaches_frontmatter() {
9915 // The metadata is only *hidden*, never lost: the source view edits and
9916 // selects it in full, and it's always preserved on save.
9917 let fm = "---\ntitle: hi\n---\n";
9918 let body = format!("{fm}# leaf\n");
9919 let mut d = doc_with("src_fm", &body);
9920 d.select_all();
9921 let sel = d.selected_text().unwrap();
9922 assert!(
9923 sel.contains("title"),
9924 "source view should select everything"
9925 );
9926 d.move_doc_start(false);
9927 assert_eq!(d.caret, 0, "source view can reach offset 0");
9928 }
9929
9930 const TABLE: &str = "| Name | Qty |\n|:-----|----:|\n| Pear | 3 |\n| Fig | 12 |\n";
9931
9932 #[test]
9933 fn wysiwyg_right_crosses_a_cell_border_without_stalling() {
9934 // The border and padding between two cells all share one source offset,
9935 // so a column-stepping caret would sit on `│` and then stall there
9936 // forever. Right must step: end of "Name" -> start of "Qty".
9937 let mut d = wysiwyg_doc("tbl_right", TABLE);
9938 d.caret = TABLE.find("Name").unwrap() + 4; // just after "Name"
9939 d.move_right(false);
9940 assert_eq!(
9941 d.caret,
9942 TABLE.find("Qty").unwrap(),
9943 "should land in the next cell"
9944 );
9945 let (r, c) = d.caret_pos();
9946 assert_eq!(d.vmap.rows[r].glyphs[c].ch, 'Q');
9947 }
9948
9949 #[test]
9950 fn wysiwyg_left_crosses_back_to_the_previous_cell() {
9951 let mut d = wysiwyg_doc("tbl_left", TABLE);
9952 d.caret = TABLE.find("Qty").unwrap();
9953 d.move_left(false);
9954 assert_eq!(
9955 d.caret,
9956 TABLE.find("Name").unwrap() + 4,
9957 "end of the previous cell"
9958 );
9959 }
9960
9961 #[test]
9962 fn wysiwyg_down_steps_over_a_table_rule() {
9963 // Between the header and the first body row sits a `├───┼───┤` rule.
9964 // It's drawn but holds no caret, so one Down must reach "Pear".
9965 let mut d = wysiwyg_doc("tbl_down", TABLE);
9966 d.caret = TABLE.find("Name").unwrap();
9967 d.move_down(false);
9968 assert_eq!(
9969 d.caret,
9970 TABLE.find("Pear").unwrap(),
9971 "one Down reaches the body row"
9972 );
9973 d.move_down(false);
9974 assert_eq!(d.caret, TABLE.find("Fig").unwrap());
9975 }
9976
9977 #[test]
9978 fn wysiwyg_tab_walks_the_cells_and_shift_tab_walks_back() {
9979 let mut d = wysiwyg_doc("tbl_tab", TABLE);
9980 d.caret = TABLE.find("Name").unwrap();
9981 // A hop lands with the destination cell's whole content selected, the
9982 // caret at its end — so typing replaces the cell like a form field.
9983 assert!(d.cell_hop(true));
9984 assert_eq!(
9985 d.selected_text(),
9986 Some("Qty"),
9987 "the target cell comes up selected"
9988 );
9989 assert_eq!(d.caret, TABLE.find("Qty").unwrap() + "Qty".len());
9990 assert!(d.cell_hop(true), "Tab wraps onto the next row's first cell");
9991 assert_eq!(d.selected_text(), Some("Pear"));
9992 assert!(d.cell_hop(false));
9993 assert_eq!(d.selected_text(), Some("Qty"));
9994 }
9995
9996 #[test]
9997 fn tab_outside_a_table_is_not_a_cell_hop() {
9998 // `cell_hop` reports false so the frontend can indent as usual.
9999 let mut d = wysiwyg_doc("tbl_none", "just a paragraph\n");
10000 d.caret = 4;
10001 assert!(!d.cell_hop(true));
10002 assert_eq!(d.caret, 4, "a refused hop leaves the caret alone");
10003 }
10004
10005 #[test]
10006 fn tab_at_the_last_cell_declines_rather_than_leaving_the_table() {
10007 let mut d = wysiwyg_doc("tbl_edge", TABLE);
10008 d.caret = TABLE.rfind("12").unwrap(); // the final cell
10009 assert!(!d.cell_hop(true), "no cell after the last one");
10010 d.caret = TABLE.find("Name").unwrap();
10011 assert!(!d.cell_hop(false), "no cell before the first one");
10012 }
10013
10014 #[test]
10015 fn wysiwyg_vertical_cell_motion_holds_the_column() {
10016 // Down/Up step to the cell above/below in the *same column*, not back to
10017 // the top-left the way a naive row/col motion over the picture would.
10018 let mut d = wysiwyg_doc("tbl_vert", TABLE);
10019 d.caret = TABLE.find("Qty").unwrap();
10020 // Each vertical hop selects the destination cell, holding the column.
10021 assert!(d.cell_move_vertical(true));
10022 assert_eq!(d.selected_text(), Some("3"), "Down holds column 1");
10023 assert!(d.cell_move_vertical(true));
10024 assert_eq!(d.selected_text(), Some("12"), "Down again, still column 1");
10025 assert!(!d.cell_move_vertical(true), "no row below the last");
10026 assert!(d.cell_move_vertical(false));
10027 assert_eq!(d.selected_text(), Some("3"), "Up holds column 1");
10028 assert!(d.cell_move_vertical(false));
10029 assert_eq!(d.selected_text(), Some("Qty"), "Up onto the header");
10030 assert!(!d.cell_move_vertical(false), "no row above the header");
10031 }
10032
10033 #[test]
10034 fn tab_off_the_last_cell_grows_a_row_and_enters_it() {
10035 let mut d = wysiwyg_doc("tbl_grow", TABLE);
10036 d.caret = TABLE.rfind("12").unwrap();
10037 let rows_before = d.source.matches('\n').count();
10038 assert!(d.cell_tab(true), "acts as a table key");
10039 assert_eq!(
10040 d.source.matches('\n').count(),
10041 rows_before + 1,
10042 "a fresh row was appended"
10043 );
10044 assert!(d.caret_in_table(), "the caret entered the new row");
10045 // The caret sits in the new row's first cell — past the old last cell.
10046 assert!(d.caret > TABLE.rfind("12").unwrap());
10047 }
10048
10049 #[test]
10050 fn return_in_a_table_drops_a_cell_and_grows_a_row_at_the_bottom() {
10051 let mut d = wysiwyg_doc("tbl_ret", TABLE);
10052 d.caret = TABLE.find("Name").unwrap();
10053 assert!(d.cell_return(), "acts as a table key");
10054 assert_eq!(
10055 d.selected_text(),
10056 Some("Pear"),
10057 "Return drops one cell, selecting it"
10058 );
10059 // From the last row, Return appends a row and enters it.
10060 d.caret = TABLE.rfind("Fig").unwrap();
10061 let rows_before = d.source.matches('\n').count();
10062 assert!(d.cell_return());
10063 assert_eq!(d.source.matches('\n').count(), rows_before + 1);
10064 assert!(d.caret_in_table());
10065 }
10066
10067 #[test]
10068 fn return_and_tab_outside_a_table_decline() {
10069 let mut d = wysiwyg_doc("tbl_decline", "just a paragraph\n");
10070 d.caret = 4;
10071 assert!(!d.cell_return(), "no table: the frontend inserts a newline");
10072 assert!(!d.cell_tab(true), "no table: the frontend indents");
10073 assert!(
10074 !d.cell_line_break(),
10075 "no table: the frontend breaks the line"
10076 );
10077 }
10078
10079 #[test]
10080 fn shift_return_inserts_an_in_cell_break_the_renderer_reads_as_a_line() {
10081 let mut d = wysiwyg_doc("tbl_break", TABLE);
10082 d.caret = TABLE.find("Pear").unwrap() + 4; // just after "Pear"
10083 assert!(d.cell_line_break(), "acts as a table key");
10084 assert!(
10085 d.source.contains("Pear<br>"),
10086 "spelled as an inline <br>: {}",
10087 d.source
10088 );
10089 assert!(d.caret_in_table(), "still in the cell, past the break");
10090 // The break renders as a real line: the "Pear" cell now draws two lines,
10091 // so the table's picture is one row taller than a single-line table.
10092 d.build_visual(80);
10093 let table = &d.vmap.tables[0];
10094 let cell = &table.grid[1].cells[0]; // first body row, first column
10095 assert!(
10096 cell.glyphs.iter().any(|g| g.ch == '\n'),
10097 "the cell carries the break as a newline glyph for the frontend to split"
10098 );
10099 }
10100
10101 #[test]
10102 fn shift_return_in_a_markdown_cell_leaves_a_semantic_hard_break_not_raw_html() {
10103 // twig promotes the in-cell `<br>` to a `hard_break`, so the break reads
10104 // back as structure — the whole point of routing through insert_line_break
10105 // instead of splicing raw `<br>` bytes.
10106 let mut d = wysiwyg_doc("tbl_break_semantic", TABLE);
10107 d.caret = TABLE.find("Pear").unwrap() + 4;
10108 assert!(d.cell_line_break());
10109 let kinds: Vec<Kind> = d
10110 .editor
10111 .nodes()
10112 .unwrap()
10113 .iter()
10114 .map(|n| n.kind.clone())
10115 .collect();
10116 assert!(kinds.contains(&Kind::HardBreak), "got {kinds:?}");
10117 assert!(
10118 !kinds.contains(&Kind::RawInline),
10119 "still raw HTML: {kinds:?}"
10120 );
10121 }
10122
10123 #[test]
10124 fn backspace_over_an_in_cell_break_deletes_the_whole_br_not_a_byte() {
10125 // The `<br>` draws as one newline glyph, so Backspace over it must take
10126 // all four bytes — a one-byte delete would strand a visible `<br` in the
10127 // cell (the reported bug).
10128 let mut d = wysiwyg_doc("tbl_break_bs", TABLE);
10129 d.caret = TABLE.find("Pear").unwrap() + 4;
10130 assert!(d.cell_line_break());
10131 assert!(d.source.contains("Pear<br>"), "precondition: {}", d.source);
10132 d.backspace(); // caret sits just past the break
10133 assert!(
10134 !d.source.contains("<br"),
10135 "no half-deleted <br left: {}",
10136 d.source
10137 );
10138 assert!(
10139 d.source.contains("| Pear |"),
10140 "the cell is back to one line: {}",
10141 d.source
10142 );
10143 }
10144
10145 #[test]
10146 fn delete_forward_over_an_in_cell_break_deletes_the_whole_br() {
10147 let mut d = wysiwyg_doc("tbl_break_del", TABLE);
10148 d.caret = TABLE.find("Pear").unwrap() + 4;
10149 assert!(d.cell_line_break());
10150 d.caret = TABLE.find("Pear").unwrap() + 4; // back onto the break's start
10151 d.delete_forward();
10152 assert!(
10153 !d.source.contains("<br"),
10154 "no half-deleted <br: {}",
10155 d.source
10156 );
10157 assert!(
10158 d.source.contains("| Pear |"),
10159 "cell back to one line: {}",
10160 d.source
10161 );
10162 }
10163
10164 #[test]
10165 fn shift_return_in_a_djot_cell_is_swallowed_and_leaves_the_row_intact() {
10166 // Djot has no idiomatic in-cell break, so twig refuses it. The gesture is
10167 // still consumed (a real newline would split the one-line row), but the
10168 // cell must be left exactly as it was — no non-idiomatic `<br>` spliced in.
10169 let src = "| Name | Qty |\n|:-----|----:|\n| Pear | 3 |\n";
10170 let mut d = Doc::from_source(src.to_string(), Format::Djot).unwrap();
10171 d.caret = src.find("Pear").unwrap() + 4;
10172 assert!(d.caret_in_table(), "caret should be inside the djot table");
10173 assert!(
10174 d.cell_line_break(),
10175 "the key is consumed, not passed to the frontend"
10176 );
10177 assert_eq!(d.source, src, "the djot cell is left untouched");
10178 assert!(
10179 !d.source.contains("<br>"),
10180 "no non-idiomatic <br> spliced into djot"
10181 );
10182 assert!(
10183 d.status.is_some(),
10184 "the refusal is surfaced on the status line"
10185 );
10186 }
10187
10188 #[test]
10189 fn typing_in_a_cell_edits_that_cell() {
10190 // Editing comes free once offsets map correctly: the caret is a source
10191 // offset, so a normal splice lands inside the pipe table.
10192 let mut d = wysiwyg_doc("tbl_type", TABLE);
10193 d.caret = TABLE.find("Pear").unwrap() + 4;
10194 d.insert("s");
10195 assert!(d.source.contains("| Pears | 3 |"), "got {:?}", d.source);
10196 }
10197
10198 #[test]
10199 fn motion_and_delete_treat_an_emoji_as_one_character() {
10200 // 👨👩👧 is a single grapheme built from three emoji joined by ZWJ — 18
10201 // bytes, several codepoints. Right-arrow must clear it in one step, and
10202 // backspace must remove the whole cluster, not a stray joiner.
10203 let family = "👨👩👧";
10204 let mut d = doc_with("emoji", &format!("a{family}b\n"));
10205 d.caret = 1; // just after 'a', before the emoji
10206 d.move_right(false);
10207 assert_eq!(
10208 d.caret,
10209 1 + family.len(),
10210 "one step clears the whole cluster"
10211 );
10212 assert_eq!(&d.source[d.caret..d.caret + 1], "b");
10213
10214 d.backspace(); // delete the emoji as a unit
10215 assert_eq!(d.source, "ab\n");
10216 assert_eq!(d.caret, 1);
10217 }
10218
10219 #[test]
10220 fn motion_handles_a_combining_accent_as_one_character() {
10221 // "e" + U+0301 (combining acute) renders as one é.
10222 let mut d = doc_with("combining", "e\u{0301}x\n");
10223 d.caret = 0;
10224 d.move_right(false);
10225 assert_eq!(
10226 d.caret,
10227 "e\u{0301}".len(),
10228 "steps past base + combining mark"
10229 );
10230 }
10231
10232 #[test]
10233 fn undo_then_redo_round_trips_an_edit() {
10234 let mut d = doc_with("undo", "hello\n");
10235 d.caret = 5;
10236 d.insert("!");
10237 assert_eq!(d.source, "hello!\n");
10238 d.undo();
10239 assert_eq!(d.source, "hello\n");
10240 assert_eq!(d.caret, 5, "undo restores the caret");
10241 d.redo();
10242 assert_eq!(d.source, "hello!\n");
10243 }
10244
10245 #[test]
10246 fn a_run_of_typing_undoes_as_one_step() {
10247 let mut d = doc_with("coalesce", "\n");
10248 d.caret = 0;
10249 d.insert("a");
10250 d.insert("b");
10251 d.insert("c");
10252 assert_eq!(d.source, "abc\n");
10253 d.undo(); // the whole typed run, not just "c"
10254 assert_eq!(d.source, "\n");
10255 d.undo(); // nothing left — the run was one step
10256 assert_eq!(d.source, "\n");
10257 assert_eq!(d.status.as_deref(), Some("nothing to undo"));
10258 }
10259
10260 // ── IME composition ──────────────────────────────────────────────────────
10261
10262 #[test]
10263 fn a_composition_run_undoes_as_one_step() {
10264 let mut d = doc_with("compose", "\n");
10265 d.caret = 0;
10266 // What an IME does: each step replaces the last one's provisional bytes.
10267 d.edit_composing(0, 0, "k");
10268 d.edit_composing(0, 1, "か");
10269 d.edit_composing(0, 3, "かん");
10270 d.edit_composing(0, 6, "感"); // the commit
10271 d.end_composition();
10272 assert_eq!(d.source, "感\n");
10273 d.undo(); // the whole composition, not its last keystroke
10274 assert_eq!(d.source, "\n");
10275 assert_eq!(d.status.as_deref(), None, "the run was a single step");
10276 }
10277
10278 #[test]
10279 fn two_compositions_are_two_undo_steps() {
10280 let mut d = doc_with("compose_two", "\n");
10281 d.caret = 0;
10282 d.edit_composing(0, 0, "か");
10283 d.edit_composing(0, 3, "蚊");
10284 d.end_composition();
10285 d.edit_composing(3, 3, "き");
10286 d.edit_composing(3, 6, "木");
10287 d.end_composition();
10288 assert_eq!(d.source, "蚊木\n");
10289 d.undo();
10290 assert_eq!(d.source, "蚊\n", "only the second composition");
10291 d.undo();
10292 assert_eq!(d.source, "\n");
10293 }
10294
10295 #[test]
10296 fn a_composition_does_not_fold_into_the_typing_around_it() {
10297 let mut d = doc_with("compose_typing", "\n");
10298 d.caret = 0;
10299 d.insert("a");
10300 d.insert("b");
10301 d.edit_composing(2, 2, "か");
10302 d.edit_composing(2, 5, "蚊");
10303 d.end_composition();
10304 d.insert("c");
10305 assert_eq!(d.source, "ab蚊c\n");
10306 d.undo();
10307 assert_eq!(d.source, "ab蚊\n");
10308 d.undo();
10309 assert_eq!(d.source, "ab\n");
10310 d.undo();
10311 assert_eq!(d.source, "\n");
10312 }
10313
10314 #[test]
10315 fn ending_a_composition_that_never_began_leaves_a_typing_run_alone() {
10316 let mut d = doc_with("compose_spurious", "\n");
10317 d.caret = 0;
10318 d.insert("a");
10319 d.end_composition(); // an IME unmarking unprompted
10320 d.insert("b");
10321 assert_eq!(d.source, "ab\n");
10322 d.undo();
10323 assert_eq!(d.source, "\n", "still one typed run");
10324 }
10325
10326 // ── the clipboard's rich flavor ──────────────────────────────────────────
10327
10328 #[test]
10329 fn an_inline_selection_publishes_html_without_a_paragraph_wrapper() {
10330 let mut d = doc_with("sel_inline", "a **bold** c\n");
10331 d.anchor = Some(2);
10332 d.caret = 10; // `**bold**`, inside the paragraph
10333 assert_eq!(d.selection_html().as_deref(), Some("<strong>bold</strong>"));
10334 }
10335
10336 #[test]
10337 fn a_whole_block_selection_keeps_its_paragraph() {
10338 let mut d = doc_with("sel_block", "a **bold** c\n");
10339 d.anchor = Some(0);
10340 d.caret = 12; // the entire paragraph
10341 assert_eq!(
10342 d.selection_html().as_deref(),
10343 Some("<p>a <strong>bold</strong> c</p>")
10344 );
10345 }
10346
10347 #[test]
10348 fn a_multi_block_selection_keeps_its_structure() {
10349 let mut d = doc_with("sel_multi", "para\n\n- one\n- two\n");
10350 d.select_all();
10351 let html = d.selection_html().expect("renders");
10352 assert!(html.contains("<p>para</p>"), "{html:?}");
10353 assert!(html.contains("<li>one</li>"), "{html:?}");
10354 }
10355
10356 #[test]
10357 fn a_word_inside_a_heading_publishes_as_text_not_a_heading() {
10358 // The fragment `Head` is a paragraph standalone; the *document* says it
10359 // sits inside one block, so the wrapper is an artifact either way.
10360 let mut d = doc_with("sel_heading", "# Head line\n");
10361 d.anchor = Some(2);
10362 d.caret = 6;
10363 assert_eq!(d.selection_html().as_deref(), Some("Head"));
10364 }
10365
10366 #[test]
10367 fn no_selection_publishes_no_html() {
10368 let mut d = doc_with("sel_none", "a b\n");
10369 d.caret = 1;
10370 assert_eq!(d.selection_html(), None);
10371 }
10372
10373 #[test]
10374 fn pasting_html_converts_it_and_is_one_undo_step() {
10375 let mut d = doc_with("paste_html", "x\n");
10376 d.caret = 1;
10377 assert!(d.paste_html("<p>a <strong>b</strong> c</p>"));
10378 assert_eq!(d.source, "xa **b** c\n");
10379 d.undo();
10380 assert_eq!(d.source, "x\n", "the whole paste, in one step");
10381 }
10382
10383 #[test]
10384 fn pasting_html_replaces_the_selection() {
10385 let mut d = doc_with("paste_html_sel", "keep drop\n");
10386 d.anchor = Some(5);
10387 d.caret = 9;
10388 assert!(d.paste_html("<em>new</em>"));
10389 assert_eq!(d.source, "keep *new*\n");
10390 }
10391
10392 #[test]
10393 fn html_that_would_paste_garbage_declines_so_the_caller_falls_back() {
10394 let mut d = doc_with("paste_html_bad", "x\n");
10395 d.caret = 1;
10396 // twig builds no table from HTML; raw `<table>` in prose is worse than
10397 // the plain flavor the caller still holds.
10398 assert!(!d.paste_html("<table><tr><td>a</td></tr></table>"));
10399 assert_eq!(d.source, "x\n", "declined edits nothing");
10400 }
10401
10402 #[test]
10403 fn copy_then_paste_round_trips_through_the_html_flavor() {
10404 let mut d = doc_with("clip_round", "a **b** and [l](https://x.dev)\n");
10405 d.select_all();
10406 let html = d.selection_html().expect("renders");
10407 let mut into = doc_with("clip_round_dst", "\n");
10408 into.caret = 0;
10409 assert!(into.paste_html(&html));
10410 assert_eq!(into.source, "a **b** and [l](https://x.dev)\n");
10411 }
10412
10413 #[test]
10414 fn moving_the_caret_starts_a_new_undo_group() {
10415 let mut d = doc_with("break", "\n");
10416 d.caret = 0;
10417 d.insert("a");
10418 d.insert("b"); // "ab\n", caret at 2
10419 d.move_left(false); // breaks the run
10420 d.insert("X"); // "aXb\n"
10421 assert_eq!(d.source, "aXb\n");
10422 d.undo();
10423 assert_eq!(
10424 d.source, "ab\n",
10425 "first undo removes only the post-move insert"
10426 );
10427 d.undo();
10428 assert_eq!(d.source, "\n", "second undo removes the earlier run");
10429 }
10430
10431 #[test]
10432 fn undo_reverses_a_format_toggle() {
10433 let mut d = doc_with("fmt_undo", "a word b\n");
10434 d.anchor = Some(2);
10435 d.caret = 6;
10436 d.toggle(InlineKind::Strong);
10437 assert_eq!(d.source, "a **word** b\n");
10438 d.undo();
10439 assert_eq!(d.source, "a word b\n");
10440 }
10441
10442 #[test]
10443 fn undo_back_to_the_saved_state_clears_dirty() {
10444 let mut d = doc_with("dirty_undo", "hello\n");
10445 assert!(!d.dirty);
10446 d.caret = 5;
10447 d.insert("!");
10448 assert!(d.dirty);
10449 d.undo();
10450 assert!(
10451 !d.dirty,
10452 "undoing to the saved source is not a modification"
10453 );
10454 }
10455
10456 #[test]
10457 fn a_new_edit_invalidates_redo() {
10458 let mut d = doc_with("redo_inv", "\n");
10459 d.caret = 0;
10460 d.insert("a");
10461 d.undo();
10462 d.insert("b"); // diverges — the redo of "a" is now gone
10463 d.redo();
10464 assert_eq!(d.source, "b\n");
10465 }
10466
10467 #[test]
10468 fn can_undo_and_can_redo_follow_the_history_a_menu_would_enable_by() {
10469 let mut d = doc_with("can_undo", "hello\n");
10470 assert!(
10471 !d.can_undo() && !d.can_redo(),
10472 "a fresh document has no history"
10473 );
10474 d.caret = 5;
10475 d.insert("!");
10476 assert!(
10477 d.can_undo() && !d.can_redo(),
10478 "an edit is a step to take back"
10479 );
10480 d.undo();
10481 assert!(!d.can_undo() && d.can_redo(), "undone: only redo remains");
10482 d.redo();
10483 assert!(d.can_undo() && !d.can_redo(), "redone: back to undoable");
10484 d.undo();
10485 d.insert("?");
10486 assert!(
10487 d.can_undo() && !d.can_redo(),
10488 "a fresh edit ends the redo chain"
10489 );
10490 // A coalesced run over-counts steps — the bound is what a menu needs,
10491 // and it reconciles the moment twig reports the history empty.
10492 d.insert("a");
10493 d.insert("b");
10494 while d.can_undo() {
10495 d.undo();
10496 }
10497 assert_eq!(d.source, "hello\n");
10498 assert!(!d.can_undo());
10499 // A reading surface has nothing to undo, whatever the history holds.
10500 d.redo();
10501 d.set_read_only(true);
10502 assert!(!d.can_undo() && !d.can_redo());
10503 }
10504
10505 #[test]
10506 fn undo_on_empty_history_is_a_no_op() {
10507 let mut d = doc_with("undo_empty", "hi\n");
10508 d.undo();
10509 assert_eq!(d.source, "hi\n");
10510 assert_eq!(d.status.as_deref(), Some("nothing to undo"));
10511 }
10512
10513 #[test]
10514 fn a_one_character_paste_is_its_own_undo_step() {
10515 for view in [View::Source, View::Wysiwyg] {
10516 let mut d = doc_in(view, "paste_step", "ab\n");
10517 d.caret = 0;
10518 d.insert("x");
10519 d.insert("y"); // a run of typing
10520 d.paste("z"); // one character, but pasted — not part of that run
10521 assert_eq!(d.source, "xyzab\n");
10522 d.undo();
10523 assert_eq!(d.source, "xyab\n", "the paste undoes on its own");
10524 assert_eq!(d.caret, 2, "and hands back the caret it found");
10525 d.undo();
10526 assert_eq!(d.source, "ab\n", "the typed run is still one step under it");
10527 }
10528 }
10529
10530 #[test]
10531 fn the_same_character_typed_still_joins_the_run() {
10532 // The other half of the pair: `z` is a keystroke here and a paste above,
10533 // and the two undo differently. Nothing about the *string* says which —
10534 // which is why provenance has to come from the door the caller uses.
10535 for view in [View::Source, View::Wysiwyg] {
10536 let mut d = doc_in(view, "typed_run", "ab\n");
10537 d.caret = 0;
10538 d.insert("x");
10539 d.insert("y");
10540 d.insert("z");
10541 d.undo();
10542 assert_eq!(d.source, "ab\n", "one run, one step");
10543 }
10544 }
10545
10546 #[test]
10547 fn undo_restores_the_caret_to_where_it_was_not_to_the_edit_site() {
10548 for view in [View::Source, View::Wysiwyg] {
10549 let mut d = doc_in(view, "undo_caret", "hello world\n");
10550 d.caret = 11; // standing at the end of "world", away from the edit
10551 d.edit(0, 5, "goodbye");
10552 assert_eq!(d.source, "goodbye world\n");
10553 d.undo();
10554 assert_eq!(d.source, "hello world\n");
10555 // The undone edit ends at offset 5; the user was at 11.
10556 assert_eq!(d.caret, 11, "the caret comes back with the bytes");
10557 }
10558 }
10559
10560 #[test]
10561 fn undo_restores_the_selection_the_edit_replaced() {
10562 for view in [View::Source, View::Wysiwyg] {
10563 let mut d = doc_in(view, "undo_sel", "a word b\n");
10564 d.anchor = Some(2);
10565 d.caret = 6; // "word" selected
10566 d.insert("X");
10567 assert_eq!(d.source, "a X b\n");
10568 d.undo();
10569 assert_eq!(d.source, "a word b\n");
10570 assert_eq!(d.selection(), Some((2, 6)), "the selection comes back too");
10571 }
10572 }
10573
10574 #[test]
10575 fn redo_restores_the_caret_the_edit_left_behind() {
10576 for view in [View::Source, View::Wysiwyg] {
10577 let mut d = doc_in(view, "redo_caret", "hello world\n");
10578 d.caret = 11;
10579 d.edit(0, 5, "goodbye");
10580 assert_eq!(d.caret, 7, "the edit left the caret after its new text");
10581 d.undo();
10582 d.redo();
10583 assert_eq!(d.source, "goodbye world\n");
10584 assert_eq!(d.caret, 7, "redo puts it back where the edit had it");
10585 }
10586 }
10587
10588 #[test]
10589 fn undoing_a_typed_run_restores_the_caret_from_before_the_whole_run() {
10590 for view in [View::Source, View::Wysiwyg] {
10591 let mut d = doc_in(view, "run_caret", "hi\n");
10592 d.caret = 2;
10593 d.insert("a");
10594 d.insert("b");
10595 d.insert("c");
10596 assert_eq!(d.source, "hiabc\n");
10597 d.undo();
10598 assert_eq!(d.source, "hi\n");
10599 assert_eq!(d.caret, 2, "before the run, not before its last keystroke");
10600 d.redo();
10601 assert_eq!(d.caret, 5, "and redo restores the end of the whole run");
10602 }
10603 }
10604
10605 #[test]
10606 fn undo_restores_the_caret_across_a_format_toggle() {
10607 // A toggle reaches twig without going through `splice`, so it has to
10608 // record its own step — miss it and every stack depth below it is off by
10609 // one, and undo starts handing back another edit's caret.
10610 for view in [View::Source, View::Wysiwyg] {
10611 let mut d = doc_in(view, "fmt_caret", "a word b\n");
10612 d.caret = 8;
10613 d.anchor = Some(2);
10614 d.caret = 6;
10615 d.toggle(InlineKind::Strong);
10616 assert_eq!(d.source, "a **word** b\n");
10617 d.undo();
10618 assert_eq!(d.source, "a word b\n");
10619 assert_eq!(
10620 d.selection(),
10621 Some((2, 6)),
10622 "the toggled selection comes back"
10623 );
10624 }
10625 }
10626
10627 #[test]
10628 fn an_edit_after_an_undo_truncates_the_caret_history_with_twigs() {
10629 // The drift that would never announce itself: twig drops its redo stack
10630 // on any fresh edit, so a leaf redo entry that outlives it would restore
10631 // a caret from the timeline that edit abandoned.
10632 for view in [View::Source, View::Wysiwyg] {
10633 let mut d = doc_in(view, "redo_trunc", "hello world\n");
10634 d.caret = 11;
10635 d.edit(0, 5, "goodbye"); // step A, caret 11 → 7
10636 d.undo();
10637 assert_eq!(d.caret, 11);
10638 d.caret = 0;
10639 d.insert("X"); // diverges: A's redo is gone from twig
10640 assert_eq!(d.source, "Xhello world\n");
10641
10642 d.redo();
10643 assert_eq!(d.source, "Xhello world\n", "nothing to redo onto");
10644 assert_eq!(d.status.as_deref(), Some("nothing to redo"));
10645 d.undo();
10646 assert_eq!(d.source, "hello world\n");
10647 assert_eq!(
10648 d.caret, 0,
10649 "the surviving step's caret, not the dropped one"
10650 );
10651 }
10652 }
10653
10654 #[test]
10655 fn indent_and_outdent_move_the_caret_line_with_its_text() {
10656 for view in [View::Source, View::Wysiwyg] {
10657 let g = |m, f: fn(&mut Doc)| golden_in(view, "indent_line", m, f);
10658 assert_eq!(g("he|llo\n", |d| d.indent()), " he|llo\n");
10659 assert_eq!(g(" he|llo\n", |d| d.outdent()), "he|llo\n");
10660 // Indentation the caret is standing *in* collapses to the line start
10661 // rather than dragging the caret into the text.
10662 assert_eq!(g("| hello\n", |d| d.outdent()), "|hello\n");
10663 // A line with none to give back is left exactly as it was.
10664 assert_eq!(g("he|llo\n", |d| d.outdent()), "he|llo\n");
10665 // Less than a full level gives back what it has.
10666 assert_eq!(g(" he|llo\n", |d| d.outdent()), "he|llo\n");
10667 // A tab is one level however many spaces it isn't.
10668 assert_eq!(g("\the|llo\n", |d| d.outdent()), "he|llo\n");
10669 }
10670 }
10671
10672 #[test]
10673 fn one_indent_level_leaves_a_paragraph_a_paragraph() {
10674 // Why the level is two spaces and not the four both frontends type
10675 // today. Four is markdown's indented-code-block marker, so a Tab on a
10676 // paragraph would silently restyle it as code — a width that changes
10677 // what the document *means* isn't an indent. Pinned because the number
10678 // is the kind of thing a later list-aware pass would reach for.
10679 let mut d = doc_with("indent_kind", "hello\n");
10680 d.caret = 2;
10681 d.indent();
10682 assert_eq!(d.source, " hello\n");
10683 assert!(
10684 d.nodes().iter().any(|n| n.kind == Kind::Para),
10685 "still prose after a Tab"
10686 );
10687 assert!(!d.nodes().iter().any(|n| n.kind == Kind::CodeBlock));
10688
10689 // The four-space level this replaces, for contrast: same text, and twig
10690 // reparses the paragraph into a code block.
10691 let mut wide = doc_with("indent_kind_4", " hello\n");
10692 wide.build_visual(80);
10693 assert!(
10694 wide.nodes().iter().any(|n| n.kind == Kind::CodeBlock),
10695 "four spaces is a code block, not an indented paragraph"
10696 );
10697 }
10698
10699 #[test]
10700 fn indent_nests_a_list_item_under_its_parent() {
10701 // Tab indents a list item by its own marker width, landing its marker at
10702 // the parent's content column so twig reparses it as a nested list.
10703 for view in [View::Source, View::Wysiwyg] {
10704 let mut d = doc_in(view, "indent_nest", "- a\n- b\n");
10705 d.caret = 6; // on the second item
10706 d.indent();
10707 assert_eq!(d.source, "- a\n - b\n");
10708 let lists = d
10709 .nodes()
10710 .iter()
10711 .filter(|n| n.kind == Kind::BulletList)
10712 .count();
10713 assert_eq!(lists, 2, "the indented item is a nested list");
10714 }
10715 }
10716
10717 #[test]
10718 fn indent_nests_an_ordered_item_at_its_marker_width() {
10719 // An ordered marker `1. ` is three columns wide, so a two-space step
10720 // (which nests a bullet) leaves it flat. Regression: Tab must use the
10721 // marker width, three, so the item actually nests — and the source
10722 // renumbers so the sub-list restarts at 1 and the outer list resumes.
10723 for view in [View::Source, View::Wysiwyg] {
10724 let mut d = doc_in(view, "indent_ord", "1. a\n2. b\n3. c\n");
10725 d.caret = d.source.find('b').unwrap();
10726 d.indent();
10727 assert_eq!(d.source, "1. a\n 1. b\n2. c\n");
10728 let lists = d
10729 .nodes()
10730 .iter()
10731 .filter(|n| n.kind == Kind::OrderedList)
10732 .count();
10733 assert_eq!(lists, 2, "the indented item is a nested ordered list");
10734 }
10735 }
10736
10737 #[test]
10738 fn indent_leaves_a_lists_first_item_put() {
10739 // The first item of a list has no sibling above it to nest under, so Tab
10740 // is a no-op there — the marker stays at column zero rather than being
10741 // shoved into indentation twig can't read as a sub-list.
10742 for view in [View::Source, View::Wysiwyg] {
10743 let mut d = doc_in(view, "indent_first", "- a\n- b\n");
10744 d.caret = 1; // on the FIRST item
10745 d.indent();
10746 assert_eq!(d.source, "- a\n- b\n", "the first item doesn't nest");
10747 // The sibling below still nests, proving the guard is per-item.
10748 d.caret = d.source.find('b').unwrap();
10749 d.indent();
10750 assert_eq!(d.source, "- a\n - b\n");
10751 }
10752 }
10753
10754 #[test]
10755 fn hidden_mode_keeps_typed_markup_literal() {
10756 // The Diaryx default: typing `*hi*` gives the characters, not emphasis —
10757 // twig escapes what would open markup, so the source is `\*hi\*` and the
10758 // AST is a plain string. Formatting is the commands' job in this mode.
10759 let mut d = doc_in(View::Wysiwyg, "hidden_literal", "");
10760 d.insert("*hi*");
10761 assert_eq!(d.source, "\\*hi\\*");
10762 assert!(
10763 d.nodes()
10764 .iter()
10765 .all(|n| n.kind != Kind::Emph && n.kind != Kind::Strong)
10766 );
10767 }
10768
10769 #[test]
10770 fn hidden_mode_escapes_a_line_start_block_marker() {
10771 // A `#`/`-`/`>` at a line start would open a block, so Hidden mode keeps
10772 // it literal too — a Diaryx user's "# 1 idea" stays prose, not a heading.
10773 let mut d = doc_in(View::Wysiwyg, "hidden_block", "");
10774 d.insert("# hi");
10775 assert_eq!(d.source, "\\# hi");
10776 assert!(d.nodes().iter().all(|n| n.kind != Kind::Heading));
10777 }
10778
10779 #[test]
10780 fn authoring_modes_keep_typed_markup_live() {
10781 // Both authoring rungs of the ladder: typing `*hi*` really is emphasis
10782 // (no escape), the same as source view — escaping is `None`'s alone, and
10783 // it's the axis, not the reveal, that decides.
10784 for (view, mode) in [
10785 (View::Wysiwyg, MarkupMode::Shortcuts),
10786 (View::Wysiwyg, MarkupMode::Full),
10787 (View::Source, MarkupMode::None),
10788 ] {
10789 let mut d = doc_in(view, "live_markup", "");
10790 d.set_markup_mode(mode);
10791 d.insert("*hi*");
10792 assert_eq!(d.source, "*hi*", "{mode:?} in {view:?} types raw markup");
10793 }
10794 }
10795
10796 #[test]
10797 fn hidden_mode_overwrite_undoes_in_one_step() {
10798 // Typing over a selection escapes the replacement *and* stays a single
10799 // undo — the selection-delete and the literal insert fold together, so
10800 // one undo brings the whole selection back, like a plain overwrite.
10801 let mut d = doc_in(View::Wysiwyg, "hidden_overwrite", "a word b\n");
10802 d.anchor = Some(2);
10803 d.caret = 6; // "word"
10804 d.insert("*");
10805 assert_eq!(d.source, "a \\* b\n", "the replacement is escaped");
10806 d.undo();
10807 assert_eq!(d.source, "a word b\n");
10808 assert_eq!(d.selection(), Some((2, 6)), "one undo, selection restored");
10809 }
10810
10811 #[test]
10812 fn backspace_over_an_escaped_char_takes_the_hidden_backslash_too() {
10813 // Type `*` in Hidden mode → `\*` (drawn as one `*`); one Backspace clears
10814 // the whole visual character, never stranding the hidden `\`.
10815 let mut d = doc_in(View::Wysiwyg, "bsp_escape", "");
10816 d.insert("*");
10817 assert_eq!(d.source, "\\*");
10818 d.backspace();
10819 assert_eq!(d.source, "", "the escape backslash went with the *");
10820 // A *literal* backslash (source view, no escape) is an ordinary char.
10821 let mut s = doc_in(View::Source, "bsp_lit", "a\\b\n");
10822 s.caret = 3; // after `b`
10823 s.backspace();
10824 assert_eq!(s.source, "a\\\n", "only the b is deleted, the \\ stays");
10825 }
10826
10827 #[test]
10828 fn hidden_mode_leaves_structural_markup_alone() {
10829 // Enter continues a bullet list by writing a real `- ` marker (an
10830 // `insert_raw`, not the typing path), so Hidden mode's escaping never
10831 // touches it — the list keeps working.
10832 let mut d = doc_in(View::Wysiwyg, "hidden_struct", "- item\n");
10833 d.caret = 6;
10834 d.newline();
10835 d.insert("two");
10836 assert_eq!(d.source, "- item\n- two\n");
10837 }
10838
10839 #[test]
10840 fn markup_mode_defaults_to_none_and_round_trips() {
10841 // Diaryx's default is the clean `None` surface; a markup-fluent
10842 // frontend can climb the ladder, and the choice sticks.
10843 let mut d = doc_in(View::Wysiwyg, "markup_mode", "hi\n");
10844 assert_eq!(d.markup_mode(), MarkupMode::None, "None by default");
10845 for mode in [MarkupMode::Shortcuts, MarkupMode::Full, MarkupMode::None] {
10846 d.set_markup_mode(mode);
10847 assert_eq!(d.markup_mode(), mode);
10848 }
10849 }
10850
10851 #[test]
10852 fn full_mode_reveals_only_the_caret_line() {
10853 // The mode's whole claim: the caret's line shows its raw delimiters and
10854 // every other line stays resolved. Two paragraphs with identical markup
10855 // so the only difference between the rows is where the caret is.
10856 let mut d = doc_in(
10857 View::Wysiwyg,
10858 "reveal_caret_line",
10859 "*one* here\n\n*two* there\n",
10860 );
10861 d.set_markup_mode(MarkupMode::Full);
10862
10863 caret_at(&mut d, "one");
10864 let rows = drawn_rows(&d);
10865 assert!(
10866 rows.iter().any(|r| r == "*one* here"),
10867 "caret's line raw: {rows:?}"
10868 );
10869 assert!(
10870 rows.iter().any(|r| r == "two there"),
10871 "other line resolved: {rows:?}"
10872 );
10873
10874 // Move to the other paragraph: the reveal follows, and the line just
10875 // left goes back to being resolved.
10876 caret_at(&mut d, "two");
10877 let rows = drawn_rows(&d);
10878 assert!(
10879 rows.iter().any(|r| r == "*two* there"),
10880 "caret's line raw: {rows:?}"
10881 );
10882 assert!(
10883 rows.iter().any(|r| r == "one here"),
10884 "left line resolved: {rows:?}"
10885 );
10886 }
10887
10888 #[test]
10889 fn revealing_a_coloured_highlight_shows_the_emoji_that_spelled_it() {
10890 // The emoji is a delimiter, not content — so `MarkupMode::Full` owes it
10891 // the same treatment as an emphasis's `*`: hidden while the caret is
10892 // elsewhere, shown in full where the caret lands. That falls out of
10893 // `delims` reading the bytes between the mark's span and its content
10894 // span, which is exactly `==🔴 ` and `==`, rather than from a table
10895 // of spellings — so the no-space form `==🟢green==` reveals right too.
10896 let mut d = doc_in(
10897 View::Wysiwyg,
10898 "reveal_coloured_mark",
10899 "a ==🔴 red== one\n\nb ==plain== two\n",
10900 );
10901 d.set_markup_mode(MarkupMode::Full);
10902
10903 caret_at(&mut d, "red");
10904 let rows = drawn_rows(&d);
10905 assert!(
10906 rows.iter().any(|r| r == "a ==🔴 red== one"),
10907 "the caret's line shows the colour it was written with: {rows:?}"
10908 );
10909 assert!(
10910 rows.iter().any(|r| r == "b plain two"),
10911 "and every other line stays resolved: {rows:?}"
10912 );
10913
10914 // Away from it, the emoji goes back to being markup — the reader sees
10915 // the words and the wash.
10916 caret_at(&mut d, "two");
10917 let rows = drawn_rows(&d);
10918 assert!(
10919 rows.iter().any(|r| r == "a red one"),
10920 "resolved again: {rows:?}"
10921 );
10922 }
10923
10924 #[test]
10925 fn hidden_modes_never_reveal_wherever_the_caret_is() {
10926 // The two rungs below `Full` share a rendering: delimiters stay hidden
10927 // even under the caret. `Shortcuts` differing from `None` only in what
10928 // typing does is exactly the point of splitting the axes.
10929 for mode in [MarkupMode::None, MarkupMode::Shortcuts] {
10930 let mut d = doc_in(View::Wysiwyg, "reveal_hidden", "*one* here\n");
10931 d.set_markup_mode(mode);
10932 caret_at(&mut d, "one");
10933 let rows = drawn_rows(&d);
10934 assert!(
10935 rows.iter().any(|r| r == "one here"),
10936 "{mode:?} hides: {rows:?}"
10937 );
10938 assert!(
10939 !rows.iter().any(|r| r.contains('*')),
10940 "{mode:?} shows no `*`: {rows:?}"
10941 );
10942 }
10943 }
10944
10945 #[test]
10946 fn revealed_delimiters_are_the_authors_own_spelling() {
10947 // Delimiters are re-read from the source rather than synthesized per
10948 // kind, so a line comes back spelled the way it was written: `_em_` does
10949 // not turn into `*em*`, and a two-backtick fence keeps both backticks.
10950 let body = "_em_ and __st__ and ``lit ` tick`` and [lk](http://x) and ~~del~~\n";
10951 let mut d = doc_in(View::Wysiwyg, "reveal_spelling", body);
10952 d.set_markup_mode(MarkupMode::Full);
10953 caret_at(&mut d, "em");
10954 let rows = drawn_rows(&d);
10955 assert!(
10956 rows.iter().any(|r| r == body.trim_end()),
10957 "the revealed line is its own source: {rows:?}"
10958 );
10959 }
10960
10961 #[test]
10962 fn revealed_heading_shows_its_hashes() {
10963 // The `# ` marker is a block-level prefix, not an inline delimiter, so
10964 // it takes its own path — but it reveals on the same rule.
10965 let mut d = doc_in(View::Wysiwyg, "reveal_heading", "# Title\n\nbody\n");
10966 d.set_markup_mode(MarkupMode::Full);
10967
10968 caret_at(&mut d, "Title");
10969 assert!(
10970 drawn_rows(&d).iter().any(|r| r == "# Title"),
10971 "{:?}",
10972 drawn_rows(&d)
10973 );
10974
10975 caret_at(&mut d, "body");
10976 let rows = drawn_rows(&d);
10977 assert!(
10978 rows.iter().any(|r| r == "Title"),
10979 "hashes hidden again: {rows:?}"
10980 );
10981 }
10982
10983 #[test]
10984 fn revealed_delimiters_are_caret_stops() {
10985 // A delimiter that is drawn but can't be reached is worse than one
10986 // that's hidden: the mode exists so the markup can be *edited*. Every
10987 // revealed byte must be somewhere the caret can stand.
10988 let mut d = doc_in(View::Wysiwyg, "reveal_stops", "*em* x\n");
10989 d.set_markup_mode(MarkupMode::Full);
10990 caret_at(&mut d, "em");
10991 let opener = d.source.find('*').unwrap();
10992 assert!(d.vmap.is_stop(opener), "the opening `*` is a caret stop");
10993 assert!(
10994 d.vmap.is_stop(opener + 3),
10995 "the closing `*` is a caret stop"
10996 );
10997 }
10998
10999 #[test]
11000 fn setext_heading_reveals_nothing_across_its_newline() {
11001 // A setext heading's underline is on another line, so it is not the
11002 // caret line's to reveal — and emitting it would inject a `\n` glyph
11003 // that splits the row where the author wrote no break.
11004 let mut d = doc_in(View::Wysiwyg, "reveal_setext", "Title\n=====\n\nbody\n");
11005 d.set_markup_mode(MarkupMode::Full);
11006 caret_at(&mut d, "Title");
11007 let rows = drawn_rows(&d);
11008 assert!(
11009 rows.iter().any(|r| r == "Title"),
11010 "title renders alone: {rows:?}"
11011 );
11012 assert!(
11013 !rows.iter().any(|r| r.contains('=')),
11014 "no underline leaks in: {rows:?}"
11015 );
11016 }
11017
11018 #[test]
11019 fn markup_mode_axes_split_the_ladder() {
11020 // The two behaviours the ladder spells: `Shortcuts` is the middle rung
11021 // that authors markup but still hides it, and it's the only rung where
11022 // the two axes disagree.
11023 assert!(!MarkupMode::None.authors());
11024 assert!(!MarkupMode::None.reveals_caret_line());
11025 assert!(MarkupMode::Shortcuts.authors());
11026 assert!(!MarkupMode::Shortcuts.reveals_caret_line());
11027 assert!(MarkupMode::Full.authors());
11028 assert!(MarkupMode::Full.reveals_caret_line());
11029 }
11030
11031 #[test]
11032 fn indenting_an_empty_dash_item_under_text_dodges_the_setext_collapse() {
11033 // Tabbing an empty `- ` under a text line would spell `- hello\n - `,
11034 // which twig (correctly, per CommonMark — pandoc agrees) reparses as a
11035 // setext H2. leaf swaps the dash for a `*` so the item stays an empty
11036 // nested bullet and `hello` stays prose: the file round-trips instead of
11037 // hiding a heading the user never asked for.
11038 for view in [View::Source, View::Wysiwyg] {
11039 let mut d = doc_in(view, "setext_guard", "- hello\n- \n");
11040 d.caret = d.source.find("- \n").unwrap() + 2; // after the empty marker
11041 d.indent();
11042 assert_eq!(d.source, "- hello\n * \n");
11043 assert!(
11044 d.nodes().iter().all(|n| n.kind != Kind::Heading),
11045 "no heading"
11046 );
11047 // And it's genuinely a nested list, not a flat one.
11048 assert_eq!(
11049 d.nodes()
11050 .iter()
11051 .filter(|n| n.kind == Kind::BulletList)
11052 .count(),
11053 2
11054 );
11055 }
11056 }
11057
11058 #[test]
11059 fn indenting_a_dash_item_with_content_keeps_its_dash() {
11060 // With content, `- x` can't be a setext underline, so there's nothing to
11061 // dodge: the marker stays a dash and nests as an ordinary sub-bullet.
11062 let mut d = doc_in(View::Wysiwyg, "setext_ok", "- hello\n- x\n");
11063 d.caret = d.source.find('x').unwrap();
11064 d.indent();
11065 assert_eq!(d.source, "- hello\n - x\n");
11066 }
11067
11068 #[test]
11069 fn the_setext_swap_undoes_as_one_step_with_the_indent() {
11070 // The dash→`*` repair coalesces into the Tab, so a single undo restores
11071 // the whole pre-Tab state rather than stranding a half-collapsed doc.
11072 let mut d = doc_in(View::Wysiwyg, "setext_undo", "- hello\n- \n");
11073 d.caret = d.source.find("- \n").unwrap() + 2;
11074 d.indent();
11075 assert_eq!(d.source, "- hello\n * \n");
11076 d.undo();
11077 assert_eq!(d.source, "- hello\n- \n", "one undo, not two");
11078 }
11079
11080 #[test]
11081 fn indent_leaves_a_nested_lists_first_item_put_too() {
11082 // The guard is about siblings, not depth: the first item of an *inner*
11083 // list (already nested under `a`) still has nothing before it at its own
11084 // level, so Tab can't take it deeper.
11085 let mut d = doc_in(View::Wysiwyg, "indent_first_nested", "- a\n - b\n - c\n");
11086 d.caret = d.source.find('b').unwrap();
11087 d.indent();
11088 assert_eq!(d.source, "- a\n - b\n - c\n", "inner first item holds");
11089 // But `c` (a sibling of `b`) nests under `b`.
11090 d.caret = d.source.find('c').unwrap();
11091 d.indent();
11092 assert_eq!(d.source, "- a\n - b\n - c\n");
11093 }
11094
11095 #[test]
11096 fn backspace_at_a_nested_item_start_outdents_it() {
11097 // Backspace with the caret right after a nested item's marker gives back
11098 // one level of nesting, the mirror of Tab — and renumbers the flattened
11099 // ordered list back to a clean run.
11100 let mut d = doc_in(View::Wysiwyg, "bsp_outdent", "1. a\n 1. b\n2. c\n");
11101 d.caret = d.source.find('b').unwrap(); // start of the nested item's content
11102 d.backspace();
11103 assert_eq!(d.source, "1. a\n2. b\n3. c\n");
11104 }
11105
11106 #[test]
11107 fn backspace_at_a_top_level_item_start_strips_the_marker() {
11108 // At the outermost level there's no nesting left to give back, so the same
11109 // keystroke drops the bullet and leaves a plain paragraph.
11110 let mut d = doc_in(View::Wysiwyg, "bsp_strip", "- a\n- b\n");
11111 d.caret = d.source.find('b').unwrap(); // right after `- `
11112 d.backspace();
11113 assert_eq!(d.source, "- a\nb\n", "the marker is gone, the text stays");
11114 }
11115
11116 #[test]
11117 fn backspace_mid_item_still_deletes_a_character() {
11118 // The list behaviour is armed only at the item's content start; anywhere
11119 // else Backspace is the ordinary character delete.
11120 let mut d = doc_in(View::Wysiwyg, "bsp_mid", "- ab\n");
11121 d.caret = d.source.find('b').unwrap(); // between `a` and `b`
11122 d.backspace();
11123 assert_eq!(d.source, "- b\n");
11124 }
11125
11126 #[test]
11127 fn backspace_at_a_heading_start_strips_the_marker() {
11128 // The `# ` is markup the rich view hides, so Backspace over it takes the
11129 // whole marker and leaves a paragraph. Deleting a byte of it instead left
11130 // `#Title` — no longer a heading, with the hash now literal text the user
11131 // never typed and has to delete again.
11132 let mut d = doc_in(View::Wysiwyg, "bsp_head", "## Title\n");
11133 d.caret = d.source.find('T').unwrap(); // right after `## `
11134 d.backspace();
11135 assert_eq!(d.source, "Title\n");
11136 assert_eq!(
11137 d.caret, 0,
11138 "the caret stays with the text it was in front of"
11139 );
11140 }
11141
11142 #[test]
11143 fn backspace_at_a_heading_start_keeps_the_block_around_it() {
11144 // Only the heading's own marker goes — the quote (or list) it sits in is
11145 // untouched, exactly as un-heading it should be.
11146 let mut d = doc_in(View::Wysiwyg, "bsp_head_quote", "> # Title\n");
11147 d.caret = d.source.find('T').unwrap();
11148 d.backspace();
11149 assert_eq!(d.source, "> Title\n");
11150 }
11151
11152 #[test]
11153 fn backspace_at_a_heading_start_takes_its_closing_sequence_too() {
11154 // `# Title #`'s trailing hashes are hidden at the other end; leaving them
11155 // behind would surface the same stray hash the marker delete just avoided.
11156 let mut d = doc_in(View::Wysiwyg, "bsp_head_closed", "# Title #\n");
11157 d.caret = d.source.find('T').unwrap();
11158 d.backspace();
11159 assert_eq!(d.source, "Title\n");
11160 // And it's one edit: a single undo puts the whole heading back.
11161 d.undo();
11162 assert_eq!(d.source, "# Title #\n");
11163 }
11164
11165 #[test]
11166 fn backspace_mid_heading_still_deletes_a_character() {
11167 // The heading behaviour is armed only at the content's start; anywhere
11168 // else Backspace is the ordinary character delete.
11169 let mut d = doc_in(View::Wysiwyg, "bsp_head_mid", "# ab\n");
11170 d.caret = d.source.find('b').unwrap();
11171 d.backspace();
11172 assert_eq!(d.source, "# b\n");
11173 }
11174
11175 #[test]
11176 fn source_view_backspace_still_edits_the_heading_marker_literally() {
11177 // In source view the `# ` is text on the screen the user is deleting a
11178 // byte of, so it keeps its literal meaning — the same split the list
11179 // ladder and Enter draw between the two views.
11180 let mut d = doc_with("bsp_head_src", "# Title\n");
11181 d.caret = d.source.find('T').unwrap();
11182 d.backspace();
11183 assert_eq!(d.source, "#Title\n");
11184 }
11185
11186 #[test]
11187 fn outdent_unnests_an_ordered_item_in_one_press() {
11188 // Shift+Tab gives back exactly the marker width the indent added, so a
11189 // nested ordered item unnests in a single press, and the flattened list
11190 // renumbers back to a clean 1, 2, 3.
11191 let mut d = doc_with("outdent_ord", "1. a\n 2. b\n3. c\n");
11192 d.caret = d.source.find('b').unwrap();
11193 d.outdent();
11194 assert_eq!(d.source, "1. a\n2. b\n3. c\n");
11195 let lists = d
11196 .nodes()
11197 .iter()
11198 .filter(|n| n.kind == Kind::OrderedList)
11199 .count();
11200 assert_eq!(lists, 1, "back to one flat list");
11201 }
11202
11203 #[test]
11204 fn table_insert_row_adds_a_row_below_the_caret() {
11205 let mut d = doc_with("tbl_ins_row", "| a | b |\n| --- | --- |\n| 1 | 2 |\n");
11206 d.caret = d.source.find('1').unwrap(); // in the body row
11207 d.table_insert_row(true);
11208 assert_eq!(d.source, "| a | b |\n| --- | --- |\n| 1 | 2 |\n| | |\n");
11209 }
11210
11211 #[test]
11212 fn table_insert_and_delete_column_at_the_caret() {
11213 let mut d = doc_with("tbl_col", "| a | b |\n| --- | --- |\n| 1 | 2 |\n");
11214 d.caret = d.source.find('a').unwrap(); // column 0
11215 d.table_insert_column(true); // add a column to the right of `a`
11216 assert_eq!(
11217 d.source,
11218 "| a | | b |\n| --- | --- | --- |\n| 1 | | 2 |\n"
11219 );
11220 d.caret = d.source.find('b').unwrap(); // now the third column
11221 d.table_delete_column();
11222 assert_eq!(d.source, "| a | |\n| --- | --- |\n| 1 | |\n");
11223 }
11224
11225 // ── ragged formats ───────────────────────────────────────────────────────
11226 // No format spells every gesture. HTML writes the inline marks as a tag pair
11227 // and no heading, list, quote or link; Markdown spells three of the eight
11228 // marks; djot spells all eight and no in-cell break. leaf asks twig per
11229 // gesture (`Doc::supports`) and refuses at the door, rather than letting each
11230 // op discover the fact on its own — one of them didn't.
11231
11232 /// An HTML document in the rich view, ready for a gesture.
11233 fn html_doc(body: &str) -> Doc {
11234 let mut d = Doc::from_source(body.to_string(), Format::Html).unwrap();
11235 d.view = View::Wysiwyg;
11236 d.build_visual(80);
11237 d
11238 }
11239
11240 #[test]
11241 fn a_table_gesture_leaves_an_html_table_alone() {
11242 // The regression this guard exists for. twig's table editor consults no
11243 // `Syntax` table — it spells a grid, not a delimiter — so it rebuilt an
11244 // HTML `<table>` as a *pipe table* and reported success: the whole
11245 // element replaced by `| a | b |`, silently, on one press of a toolbar
11246 // button. Every grid op went the same way.
11247 let src = "<table><tr><td>a</td><td>b</td></tr><tr><td>c</td><td>d</td></tr></table>\n";
11248 // A table of named operations, which is what it looks like.
11249 #[allow(clippy::type_complexity)]
11250 let ops: [(&str, &dyn Fn(&mut Doc)); 7] = [
11251 ("insert row", &|d: &mut Doc| d.table_insert_row(true)),
11252 ("delete row", &|d: &mut Doc| d.table_delete_row()),
11253 ("insert column", &|d: &mut Doc| d.table_insert_column(true)),
11254 ("delete column", &|d: &mut Doc| d.table_delete_column()),
11255 ("align", &|d: &mut Doc| {
11256 d.table_set_alignment(Alignment::Right)
11257 }),
11258 ("move row", &|d: &mut Doc| d.table_move_row(true)),
11259 ("move column", &|d: &mut Doc| d.table_move_column(true)),
11260 ];
11261 for (name, op) in ops {
11262 let mut d = html_doc(src);
11263 d.caret = d.source.find('a').unwrap();
11264 assert!(d.caret_in_table(), "{name}: the caret really is in a table");
11265 op(&mut d);
11266 assert_eq!(d.source, src, "{name} rewrote an HTML table");
11267 assert!(
11268 !d.dirty,
11269 "{name} marked the document dirty without editing it"
11270 );
11271 assert!(d.status.is_some(), "{name} refused without saying why");
11272 }
11273 }
11274
11275 #[test]
11276 fn the_block_gestures_html_cannot_spell_are_refused_with_a_reason() {
11277 // A heading is a wrapping tag pair carrying its level in both ends, a
11278 // quote wraps a range rather than prefixing each line, a link's
11279 // destination lives in an attribute — different *shapes*, not a
11280 // different alphabet, so twig spells none of them and neither does leaf.
11281 let src = "<h1>Title</h1>\n<p>Hello world</p>\n<ul><li>one</li></ul>\n";
11282 // A table of named operations, which is what it looks like.
11283 #[allow(clippy::type_complexity)]
11284 let ops: [(&str, &dyn Fn(&mut Doc)); 9] = [
11285 ("heading", &|d: &mut Doc| d.toggle_heading(2)),
11286 ("paragraph", &|d: &mut Doc| {
11287 d.set_block(BlockKind::Paragraph)
11288 }),
11289 ("quote", &|d: &mut Doc| d.toggle_blockquote()),
11290 ("list", &|d: &mut Doc| d.toggle_list(false)),
11291 ("task item", &|d: &mut Doc| d.toggle_task_item()),
11292 ("task tick", &|d: &mut Doc| d.toggle_task_checked()),
11293 ("link", &|d: &mut Doc| d.insert_link("https://example.dev")),
11294 ("image", &|d: &mut Doc| d.insert_image("pic.png", "alt")),
11295 ("video", &|d: &mut Doc| {
11296 d.insert_media(MediaKind::Video, "clip.mp4", "")
11297 }),
11298 ];
11299 for (name, op) in ops {
11300 let mut d = html_doc(src);
11301 let at = d.source.find("Hello").unwrap();
11302 d.caret = at;
11303 d.anchor = Some(at + 5); // a selection, for the ops that want one
11304 op(&mut d);
11305 assert_eq!(d.source, src, "{name} edited an HTML document");
11306 assert!(
11307 !d.dirty,
11308 "{name} marked the document dirty without editing it"
11309 );
11310 let status = d.status.as_deref().unwrap_or("");
11311 assert!(
11312 status.contains("html"),
11313 "{name}: the refusal should name the format, got {status:?}"
11314 );
11315 }
11316 }
11317
11318 #[test]
11319 fn html_spells_the_inline_marks_and_the_rule() {
11320 // The other half, and why one per-document flag stopped being enough:
11321 // ⌘B in an HTML document writes `<strong>` — the tag the serializer
11322 // already emits and the parser reads straight back as the same mark —
11323 // and the rule button writes an `<hr>`. Refusing these on the old
11324 // "HTML is parse-only" reading would now be leaf's own limitation.
11325 let mut d = html_doc("<p>Hello world</p>\n");
11326 let at = d.source.find("world").unwrap();
11327 d.caret = at;
11328 d.anchor = Some(at + 5);
11329 d.toggle(InlineKind::Strong);
11330 assert_eq!(d.source, "<p>Hello <strong>world</strong></p>\n");
11331 assert!(d.dirty);
11332 assert_eq!(d.status, None, "a supported gesture reports nothing");
11333
11334 // And off again — the toggle reverses, which is the property that makes
11335 // authoring in HTML worth offering rather than a one-way trip.
11336 d.toggle(InlineKind::Strong);
11337 assert_eq!(d.source, "<p>Hello world</p>\n");
11338
11339 let mut d = html_doc("<p>Hello world</p>\n");
11340 d.caret = d.source.find("world").unwrap();
11341 d.insert_thematic_break();
11342 assert!(d.source.contains("<hr>"), "got {:?}", d.source);
11343 }
11344
11345 #[test]
11346 fn a_mark_the_format_cannot_spell_arms_nothing() {
11347 // `toggle` with a collapsed caret doesn't reach twig at all — it arms a
11348 // sticky mark for the next text typed. Guarding only the twig call
11349 // leaves that path live, promising a highlight the gesture will not
11350 // write and then swallowing the error inside `insert`.
11351 //
11352 // Markdown is still the case that carries this, and the reason is finer
11353 // than it was: since twig 3.3 leaf *parses* `==mark==` in Markdown, so a
11354 // highlight reads back and paints. What twig will not do is author one —
11355 // its Markdown syntax table spells the mark but marks it unauthorable,
11356 // because the reader on the other end may have the extension off. So the
11357 // button stays disabled over a Markdown document that renders highlights
11358 // perfectly well, and this test is what says so.
11359 let mut d = doc_with("mark", "Hello world\n");
11360 d.view = View::Wysiwyg;
11361 d.build_visual(80);
11362 d.caret = d.source.find("world").unwrap();
11363 d.toggle(InlineKind::Mark);
11364 assert!(d.pending_marks.is_empty(), "no mark should be armed");
11365 assert!(d.status.as_deref().unwrap_or("").contains("markdown"));
11366 d.insert("X");
11367 assert_eq!(d.source, "Hello Xworld\n");
11368 }
11369
11370 #[test]
11371 fn html_documents_still_take_typed_text() {
11372 // The guard covers *markup* gestures and must not touch plain editing:
11373 // twig's splicer is language-neutral, and typing into an HTML document
11374 // is the thing that does work today.
11375 let mut d = html_doc("<p>Hello world</p>\n");
11376 d.caret = d.source.find("world").unwrap();
11377 d.insert("big ");
11378 assert_eq!(d.source, "<p>Hello big world</p>\n");
11379 assert!(d.dirty);
11380 d.backspace();
11381 assert_eq!(d.source, "<p>Hello bigworld</p>\n");
11382 d.undo();
11383 d.undo();
11384 assert_eq!(d.source, "<p>Hello world</p>\n");
11385 }
11386
11387 #[test]
11388 fn authorable_is_the_coarse_question_and_capabilities_the_useful_one() {
11389 // `authorable` only separates "there is a door in" from "there is not",
11390 // and HTML is on the near side of that line — which is exactly why a
11391 // toolbar must not be built from it.
11392 let html = Doc::from_source("<p>x</p>\n".into(), Format::Html).unwrap();
11393 assert!(html.authorable());
11394 assert!(
11395 !Doc::from_source("<r>x</r>".into(), Format::Xml)
11396 .unwrap()
11397 .authorable()
11398 );
11399
11400 let caps = html.capabilities();
11401 assert!(caps.bold && caps.italic && caps.code && caps.mark);
11402 assert!(caps.thematic_break && caps.cell_line_break);
11403 assert!(!caps.heading && !caps.blockquote && !caps.bullet_list);
11404 assert!(!caps.task && !caps.link && !caps.image && !caps.code_language);
11405 // The one flag that isn't twig's answer: an HTML `<table>` is a grid
11406 // twig's table editor would happily re-emit as `| a | b |`.
11407 assert!(!caps.table);
11408
11409 // The two lightweight formats spell everything leaf offers — and still
11410 // differ from each other, which is the other half of why one boolean
11411 // can't serve.
11412 for fmt in [Format::Markdown, Format::Djot] {
11413 let caps = Capabilities::of(fmt);
11414 assert!(
11415 caps.heading && caps.blockquote && caps.ordered_list,
11416 "{fmt:?}"
11417 );
11418 assert!(
11419 caps.task && caps.link && caps.image && caps.table,
11420 "{fmt:?}"
11421 );
11422 }
11423 assert!(Capabilities::of(Format::Djot).mark);
11424 assert!(!Capabilities::of(Format::Markdown).mark);
11425 assert!(Capabilities::of(Format::Markdown).cell_line_break);
11426 assert!(!Capabilities::of(Format::Djot).cell_line_break);
11427
11428 // A parse-only format answers no to every one of them, so the coarse
11429 // predicate and the record agree there.
11430 let caps = Capabilities::of(Format::Xml);
11431 assert!(!caps.bold && !caps.heading && !caps.table && !caps.thematic_break);
11432 }
11433
11434 #[test]
11435 fn a_refused_gesture_says_so_where_twig_would_have_said_it() {
11436 // The guard exists to name the *document's* format rather than twig's
11437 // internals, so the message has to survive being one leaf writes itself.
11438 // Checked against the gesture twig also refuses, since that is the pair
11439 // most at risk of drifting apart.
11440 let mut d = html_doc("<p>Hello</p>\n");
11441 d.caret = d.source.find("Hello").unwrap();
11442 d.set_code_language("zig");
11443 assert_eq!(
11444 d.status.as_deref(),
11445 Some("code language: not supported in html")
11446 );
11447 assert!(!d.dirty);
11448 }
11449
11450 #[test]
11451 fn table_set_alignment_respells_the_delimiter() {
11452 let mut d = doc_with("tbl_align", "| a | b |\n| --- | --- |\n| 1 | 2 |\n");
11453 d.caret = d.source.find('b').unwrap();
11454 d.table_set_alignment(Alignment::Right);
11455 assert_eq!(d.source, "| a | b |\n| --- | ---: |\n| 1 | 2 |\n");
11456 }
11457
11458 #[test]
11459 fn each_empty_table_cell_has_its_own_editable_home() {
11460 // Regression: an empty cell has no twig content_span, so both cells of a
11461 // `| | |` row collapsed onto the row's start (before the first `│`).
11462 // Typing there inserted *before* the table (`hello| | |`); nav couldn't
11463 // tell the cells apart. Each empty cell must now have a distinct home
11464 // inside it.
11465 let mut d = wysiwyg_doc("tbl_empty", "| a | b |\n| --- | --- |\n| | |\n");
11466 let (c0, c1) = {
11467 let cells = &d.vmap.tables[0].grid[1].cells;
11468 (cells[0].start, cells[1].start)
11469 };
11470 assert!(
11471 c0 < c1,
11472 "the two empty cells have distinct homes: {c0} < {c1}"
11473 );
11474 d.caret = c0;
11475 d.insert("x");
11476 assert_eq!(
11477 d.source, "| a | b |\n| --- | --- |\n| x | |\n",
11478 "typed inside the cell"
11479 );
11480 }
11481
11482 #[test]
11483 fn arrows_step_into_each_empty_table_cell() {
11484 let mut d = wysiwyg_doc("tbl_empty_nav", "| a | b |\n| --- | --- |\n| | |\n");
11485 let (c0, c1) = {
11486 let cells = &d.vmap.tables[0].grid[1].cells;
11487 (cells[0].start, cells[1].start)
11488 };
11489 d.caret = d.source.find('b').unwrap(); // in the header's second cell
11490 let mut seen = std::collections::HashSet::new();
11491 for _ in 0..6 {
11492 d.move_right(false);
11493 seen.insert(d.caret);
11494 }
11495 assert!(
11496 seen.contains(&c0),
11497 "right arrow reaches the first empty cell"
11498 );
11499 assert!(
11500 seen.contains(&c1),
11501 "right arrow reaches the second empty cell"
11502 );
11503 }
11504
11505 #[test]
11506 fn table_op_off_a_table_is_a_no_op_with_a_status() {
11507 let mut d = doc_with("tbl_none", "just text\n");
11508 d.caret = 3;
11509 d.table_insert_row(true);
11510 assert_eq!(d.source, "just text\n", "nothing changed");
11511 assert!(d.status.is_some(), "a status explains why");
11512 assert!(!d.caret_in_table());
11513 }
11514
11515 #[test]
11516 fn enter_in_an_ordered_list_renumbers_the_following_items() {
11517 // Inserting an item mid-list left the source markers stale (`1. 2. 2. 3.`);
11518 // the renumber pass keeps them sequential, matching what the view draws.
11519 let mut d = wysiwyg_doc("enter_renumber", "1. a\n2. b\n3. c\n");
11520 d.caret = d.source.find('a').unwrap() + 1; // end of item a
11521 d.newline();
11522 d.insert("x");
11523 assert_eq!(d.source, "1. a\n2. x\n3. b\n4. c\n");
11524 }
11525
11526 #[test]
11527 fn outdent_with_nothing_to_give_back_records_no_undo_step() {
11528 for view in [View::Source, View::Wysiwyg] {
11529 let mut d = doc_in(view, "outdent_noop", "hello\n");
11530 d.caret = 2;
11531 d.outdent();
11532 assert_eq!(d.source, "hello\n");
11533 assert!(!d.dirty, "a no-op is not a modification");
11534 d.undo();
11535 assert_eq!(
11536 d.status.as_deref(),
11537 Some("nothing to undo"),
11538 "spends no undo step"
11539 );
11540 assert_eq!(d.source, "hello\n");
11541 }
11542 }
11543
11544 #[test]
11545 fn indent_shifts_every_selected_line_and_keeps_them_selected() {
11546 for view in [View::Source, View::Wysiwyg] {
11547 let mut d = doc_in(view, "indent_sel", "one\n\ntwo\n");
11548 d.anchor = Some(0);
11549 d.caret = 7; // through "two"
11550 d.indent();
11551 assert_eq!(
11552 d.source, " one\n\n two\n",
11553 "the blank line keeps no trailing pad"
11554 );
11555 // Selected, so a second Tab lands on the same lines rather than on
11556 // whatever the shifted offsets now cover.
11557 assert_eq!(d.selection(), Some((0, 12)));
11558 d.indent();
11559 assert_eq!(d.source, " one\n\n two\n");
11560 }
11561 }
11562
11563 #[test]
11564 fn outdent_takes_what_each_line_has_and_leaves_the_rest_alone() {
11565 for view in [View::Source, View::Wysiwyg] {
11566 let mut d = doc_in(view, "outdent_sel", " two\n one\nnone\n");
11567 d.anchor = Some(0);
11568 d.caret = 15;
11569 d.outdent();
11570 assert_eq!(d.source, "two\none\nnone\n");
11571 }
11572 }
11573
11574 #[test]
11575 fn a_tab_undoes_as_one_step_however_many_lines_it_moved() {
11576 for view in [View::Source, View::Wysiwyg] {
11577 let mut d = doc_in(view, "indent_undo", "one\n\ntwo\n");
11578 d.anchor = Some(0);
11579 d.caret = 7;
11580 d.indent();
11581 assert_eq!(d.source, " one\n\n two\n");
11582 d.undo();
11583 assert_eq!(d.source, "one\n\ntwo\n", "one step, not one per line");
11584 assert_eq!(
11585 d.selection(),
11586 Some((0, 7)),
11587 "with the selection it was aimed at"
11588 );
11589 d.redo();
11590 assert_eq!(d.source, " one\n\n two\n");
11591 assert_eq!(
11592 d.selection(),
11593 Some((0, 12)),
11594 "redo replays the caret the indent placed, not the one splice left"
11595 );
11596 }
11597 }
11598
11599 #[test]
11600 fn vertical_motion_keeps_the_column() {
11601 let mut d = doc_with("move", "abcd\nef\n");
11602 d.caret = 3; // "abc|d" on row 0, col 3
11603 d.move_down(false); // row 1 "ef" only has cols 0..2 -> clamps to end
11604 assert_eq!(d.caret, 7); // just after "ef"
11605 }
11606
11607 // ── goal column ──────────────────────────────────────────────────────────
11608
11609 #[test]
11610 fn vertical_motion_goal_column_survives_a_short_line() {
11611 // Regression: re-deriving the column from the clamped position on
11612 // every step permanently forgets it once a short line clamps it.
11613 // Down through "xy" (2 cols) and into "ghijkl" must return to col 4.
11614 let g = |m, f: fn(&mut Doc)| golden("goalcol", m, f);
11615 assert_eq!(
11616 g("abcd|ef\nxy\nghijkl\n", |d| {
11617 d.move_down(false); // clamps to end of "xy"
11618 d.move_down(false); // restores col 4 on the long line
11619 }),
11620 "abcdef\nxy\nghij|kl\n"
11621 );
11622 }
11623
11624 #[test]
11625 fn goal_column_state_is_set_by_vertical_motion_and_cleared_by_horizontal() {
11626 let mut d = doc_with("goalcol_state", "abcdef\nxy\nghijkl\n");
11627 assert_eq!(d.goal_col, None);
11628 d.caret = 4; // row 0, col 4
11629 d.move_down(false); // clamps into "xy"; goal stays the original col
11630 assert_eq!(d.goal_col, Some(4));
11631 assert_eq!(d.caret_pos(), (1, 2));
11632
11633 // A horizontal motion drops the goal column...
11634 d.move_left(false);
11635 assert_eq!(d.goal_col, None);
11636
11637 // ...so the next vertical motion picks up the *new* column (1), not
11638 // the stale one (4).
11639 d.move_down(false);
11640 assert_eq!(d.goal_col, Some(1));
11641 assert_eq!(d.caret_pos(), (2, 1));
11642 }
11643
11644 #[test]
11645 fn editing_clears_the_goal_column() {
11646 let mut d = doc_with("goalcol_edit", "abcdef\nxy\nghijkl\n");
11647 d.caret = 4;
11648 d.move_down(false);
11649 assert_eq!(d.goal_col, Some(4));
11650 d.insert("Z");
11651 assert_eq!(d.goal_col, None);
11652 }
11653
11654 #[test]
11655 fn vertical_motion_on_an_empty_document_is_a_no_op() {
11656 let mut d = doc_with("empty_vert", "");
11657 d.move_down(false);
11658 assert_eq!(d.caret, 0);
11659 d.move_up(false);
11660 assert_eq!(d.caret, 0);
11661 }
11662
11663 // ── the document's edges ─────────────────────────────────────────────────
11664
11665 #[test]
11666 fn vertical_motion_at_the_document_edges_runs_to_them_in_both_views() {
11667 // The reproduction, and the disagreement: Down on the last line ran to
11668 // the end of the document in the source view — by accident, an
11669 // out-of-range row clamping to the end of the string — and did nothing
11670 // whatever in the view leaf opens in. One rule now, in both.
11671 for (view, tag) in VIEWS {
11672 let mut d = doc_in(view, &format!("edge_{tag}"), "abc");
11673 d.caret = 1;
11674 d.move_down(false);
11675 assert_eq!(d.caret, 3, "{tag}: Down on the last line runs to the end");
11676 d.move_up(false);
11677 assert_eq!(d.caret, 0, "{tag}: Up on the first line runs to the start");
11678 }
11679 }
11680
11681 #[test]
11682 fn vertical_motion_at_the_edges_carries_the_column_across_the_lines_between() {
11683 // Down off the bottom is a motion like any other, so it latches a goal
11684 // column — and Up comes back to the column the caret left, not to the
11685 // one the document's end happened to be in.
11686 for (view, tag) in VIEWS {
11687 let gap = if view == View::Source { "\n" } else { "\n\n" };
11688 let src = format!("abcdef{gap}ghijkl");
11689 let mut d = doc_in(view, &format!("edge_goal_{tag}"), &src);
11690 d.caret = 2; // row 0, col 2
11691 d.move_down(false);
11692 assert_eq!(d.caret_pos().1, 2, "{tag}: Down keeps the column");
11693 d.move_down(false);
11694 assert_eq!(
11695 d.caret,
11696 src.len(),
11697 "{tag}: Down off the bottom reaches the end"
11698 );
11699 d.move_up(false);
11700 assert_eq!(
11701 d.caret_pos().1,
11702 2,
11703 "{tag}: Up returns to the column Down left"
11704 );
11705 }
11706 }
11707
11708 #[test]
11709 fn vertical_motion_with_nowhere_to_go_latches_no_goal_column() {
11710 // `goal_col.get_or_insert` ran *before* the early return at row 0, so an
11711 // Up that did nothing still armed a goal column, and the next Down aimed
11712 // at a column the caret had never been in.
11713 for (view, tag) in VIEWS {
11714 let mut d = doc_in(view, &format!("noop_goal_{tag}"), "abc\n\ndef");
11715 d.caret = 0;
11716 d.move_up(false);
11717 assert_eq!(d.caret, 0, "{tag}: already at the start");
11718 assert_eq!(d.goal_col, None, "{tag}: a no-op Up latched a goal column");
11719
11720 d.caret = d.source.len();
11721 d.move_down(false);
11722 assert_eq!(d.caret, d.source.len(), "{tag}: already at the end");
11723 assert_eq!(
11724 d.goal_col, None,
11725 "{tag}: a no-op Down latched a goal column"
11726 );
11727 }
11728 }
11729
11730 // ── soft wrap ────────────────────────────────────────────────────────────
11731 // Every other test here builds the map at 80 columns, where no fixture is
11732 // long enough to fold. A wrap is where one offset belongs to two rows at
11733 // once, and it broke everything that asks the caret what row it is on.
11734
11735 /// The wrapped fixture these cases share, folded at 12 columns into
11736 /// `one two ` / `three four ` / `five six ` / `seven eight`.
11737 fn wrapped_doc(name: &str) -> Doc {
11738 let mut d = wysiwyg_doc(name, "one two three four five six seven eight");
11739 d.build_visual(12);
11740 d
11741 }
11742
11743 #[test]
11744 fn home_and_end_work_from_a_wrapped_row() {
11745 // The reproduction: offset 19 is the `f` of "five", the first character
11746 // of the third row — and also the offset the second row ends at. It
11747 // resolved to the *second* row, so End aimed at a place the caret was
11748 // already in and did nothing, while Home walked backwards onto a row the
11749 // caret had left.
11750 let mut d = wrapped_doc("wrap_home_end");
11751 d.caret = 19;
11752 assert_eq!(
11753 d.caret_pos(),
11754 (2, 0),
11755 "the wrap boundary opens the third row"
11756 );
11757 d.move_end(false);
11758 assert_eq!(d.caret, 27, "End stalled at the wrap boundary");
11759 d.move_home(false);
11760 assert_eq!(d.caret, 19, "Home left the row the caret was on");
11761 }
11762
11763 #[test]
11764 fn end_of_a_wrapped_row_stays_put_when_pressed_again() {
11765 // The row's end is the last offset that is only ever its own: the offset
11766 // past it opens the row below, and aiming there would send a second
11767 // press on to *that* row's end, and a third to the next — End walking
11768 // down the paragraph rather than sitting where it landed.
11769 let mut d = wrapped_doc("wrap_end_twice");
11770 d.caret = 12; // inside "three", on the second row
11771 d.move_end(false);
11772 assert_eq!(
11773 d.caret, 18,
11774 "the end of `three four`, before the space the wrap ate"
11775 );
11776 assert_eq!(d.caret_pos(), (1, 10), "drawn on the row it is the end of");
11777 d.move_end(false);
11778 assert_eq!(d.caret, 18, "a second End moved the caret");
11779 d.move_home(false);
11780 assert_eq!(d.caret, 8, "Home takes the row's own start");
11781 }
11782
11783 #[test]
11784 fn vertical_motion_crosses_a_soft_wrap() {
11785 // Down aimed at the row below's column 0, an offset that resolved *up*
11786 // to the row above's end — so it landed on the offset it already had and
11787 // the caret could never leave a paragraph's first row.
11788 let mut d = wrapped_doc("wrap_down");
11789 d.caret = 0;
11790 for (want, row) in [(8, 1), (19, 2), (28, 3), (39, 3)] {
11791 d.move_down(false);
11792 assert_eq!(d.caret, want, "Down stalled");
11793 assert_eq!(d.caret_pos().0, row, "Down landed on the wrong row");
11794 }
11795 d.move_down(false);
11796 assert_eq!(d.caret, 39, "the last row's Down runs to the end and stops");
11797
11798 // ...and back up, one row per press. The goal column is the end of the
11799 // last row, past every other row's width, so each press clamps to the
11800 // row's own last offset rather than to the one that opens the next.
11801 let mut d = wrapped_doc("wrap_up");
11802 d.caret = 39;
11803 for (want, pos) in [(27, (2, 8)), (18, (1, 10)), (7, (0, 7)), (0, (0, 0))] {
11804 d.move_up(false);
11805 assert_eq!(d.caret, want, "Up stalled");
11806 assert_eq!(d.caret_pos(), pos, "Up landed on the wrong row");
11807 }
11808 }
11809
11810 #[test]
11811 fn a_kill_on_a_wrapped_row_stops_at_the_row() {
11812 // The kills take the same line Home and End do, so in WYSIWYG they take
11813 // the visual row — and a soft wrap has no newline in it to delete, so
11814 // nothing is joined by reaching the end of one.
11815 let mut d = wrapped_doc("wrap_kill");
11816 d.caret = 19; // the `f` of "five", opening the third row
11817 d.delete_to_line_end();
11818 // The space the wrap ate goes with the row it was drawn on: sparing it
11819 // would leave "four seven", two spaces where the row had been.
11820 assert_eq!(d.source, "one two three four seven eight");
11821
11822 // Backwards from the row's last caret position — which is *before* that
11823 // space, so this one survives, being on the far side of the caret.
11824 let mut d = wrapped_doc("wrap_kill_back");
11825 d.caret = 27;
11826 d.delete_to_line_start();
11827 assert_eq!(d.source, "one two three four seven eight");
11828 }
11829
11830 // ── document start / end ────────────────────────────────────────────────
11831
11832 #[test]
11833 fn move_doc_start_and_end_jump_to_the_edges() {
11834 let g = |m, f: fn(&mut Doc)| golden("doc_edges", m, f);
11835 assert_eq!(
11836 g("hello\nwor|ld\n", |d| d.move_doc_start(false)),
11837 "|hello\nworld\n"
11838 );
11839 assert_eq!(
11840 g("hel|lo\nworld\n", |d| d.move_doc_end(false)),
11841 "hello\nworld\n|"
11842 );
11843 // Already at the edge: a no-op.
11844 assert_eq!(g("|hello\n", |d| d.move_doc_start(false)), "|hello\n");
11845 assert_eq!(g("hello|\n", |d| d.move_doc_end(false)), "hello\n|");
11846 }
11847
11848 #[test]
11849 fn move_doc_start_and_end_extend_the_selection() {
11850 assert_eq!(
11851 golden("doc_edges_ext_end", "hello wor|ld\n", |d| d
11852 .move_doc_end(true)),
11853 "hello wor[ld\n|]"
11854 );
11855 assert_eq!(
11856 golden("doc_edges_ext_start", "hello wor|ld\n", |d| d
11857 .move_doc_start(true)),
11858 "[|hello wor]ld\n"
11859 );
11860 }
11861
11862 #[test]
11863 fn move_doc_start_and_end_on_an_empty_document_are_a_no_op() {
11864 let mut d = doc_with("empty_edges", "");
11865 d.move_doc_end(false);
11866 assert_eq!(d.caret, 0);
11867 d.move_doc_start(false);
11868 assert_eq!(d.caret, 0);
11869 }
11870
11871 // ── arrow collapses an active selection ─────────────────────────────────
11872
11873 #[test]
11874 fn arrow_collapses_selection_to_its_near_edge() {
11875 let mut d = doc_with("collapse", "hello world\n");
11876
11877 // Forward selection (anchor before caret): Right -> end, Left -> start.
11878 d.anchor = Some(2);
11879 d.caret = 7;
11880 d.move_right(false);
11881 assert_eq!((d.caret, d.anchor), (7, None));
11882
11883 d.anchor = Some(2);
11884 d.caret = 7;
11885 d.move_left(false);
11886 assert_eq!((d.caret, d.anchor), (2, None));
11887
11888 // Backward selection (anchor after caret): edges are the same
11889 // regardless of which end the caret started on.
11890 d.anchor = Some(7);
11891 d.caret = 2;
11892 d.move_right(false);
11893 assert_eq!((d.caret, d.anchor), (7, None));
11894
11895 d.anchor = Some(7);
11896 d.caret = 2;
11897 d.move_left(false);
11898 assert_eq!((d.caret, d.anchor), (2, None));
11899 }
11900
11901 #[test]
11902 fn arrow_with_extend_keeps_growing_the_selection() {
11903 let mut d = doc_with("collapse_extend", "hello world\n");
11904 d.anchor = Some(2);
11905 d.caret = 7;
11906 d.move_right(true); // extend: no collapse, caret steps one further
11907 assert_eq!((d.caret, d.anchor), (8, Some(2)));
11908 }
11909
11910 #[test]
11911 fn arrow_without_a_selection_moves_one_character_as_before() {
11912 let mut d = doc_with("no_collapse", "hello\n");
11913 d.caret = 2;
11914 d.move_right(false);
11915 assert_eq!(d.caret, 3);
11916 d.move_left(false);
11917 assert_eq!(d.caret, 2);
11918 }
11919
11920 /// Press Right until it stops, collecting the offsets walked through. Every
11921 /// caret bug in the WYSIWYG view shows up here as a walk that ends early:
11922 /// two stops sharing one source offset can't be moved between, so the caret
11923 /// stalls on the first of them and the walk never reaches the rest.
11924 fn walk_right(d: &mut Doc) -> Vec<usize> {
11925 let mut seen = vec![d.caret];
11926 for _ in 0..2000 {
11927 let before = d.caret;
11928 d.move_right(false);
11929 if d.caret == before {
11930 break;
11931 }
11932 seen.push(d.caret);
11933 }
11934 seen
11935 }
11936
11937 #[test]
11938 fn the_caret_crosses_a_soft_break() {
11939 // A newline inside a paragraph is a `soft_break`, which twig gives no
11940 // span of its own — the space it renders as used to borrow the offset of
11941 // the character before it, and a caret can't move without changing
11942 // offset. Right must walk clean off the end of the first line.
11943 let mut d = wysiwyg_doc("soft_break_walk", "one two\nthree four\n");
11944 d.caret = 0;
11945 let seen = walk_right(&mut d);
11946 assert_eq!(seen, (0..=18).collect::<Vec<_>>(), "walk stalled: {seen:?}");
11947 }
11948
11949 #[test]
11950 fn line_flow_preserve_resplits_the_map_and_defaults_to_fold() {
11951 // The paragraph holds one soft break. Folded (the default) it lays out as
11952 // a single reflowed row; Preserve re-lays it as a row per source line.
11953 // The setter must invalidate the cached map for the change to show, and
11954 // again on the way back — so a round trip returns to the folded layout.
11955 let mut d = wysiwyg_doc("line_flow", "one two\nthree four\n");
11956 assert_eq!(d.line_flow(), LineFlow::Fold, "fold is the default");
11957 d.build_visual(80);
11958 assert_eq!(d.vmap.num_rows(), 1, "fold: one flowing row");
11959
11960 d.set_line_flow(LineFlow::Preserve);
11961 d.build_visual(80);
11962 assert_eq!(d.vmap.num_rows(), 2, "preserve: a row per source line");
11963
11964 d.set_line_flow(LineFlow::Fold);
11965 d.build_visual(80);
11966 assert_eq!(d.vmap.num_rows(), 1, "fold again: back to one row");
11967 }
11968
11969 #[test]
11970 fn the_caret_still_crosses_a_preserved_soft_break() {
11971 // Preserve renders the soft break as a row boundary rather than a space,
11972 // but the caret must still reach every offset — the break's own offset is
11973 // the first row's end stop, so Right walks clean off the end of line one
11974 // onto line two, exactly as it does when the break is folded.
11975 let mut d = wysiwyg_doc("preserve_walk", "one two\nthree four\n");
11976 d.set_line_flow(LineFlow::Preserve);
11977 d.build_visual(80);
11978 d.caret = 0;
11979 let seen = walk_right(&mut d);
11980 assert_eq!(seen, (0..=18).collect::<Vec<_>>(), "walk stalled: {seen:?}");
11981 }
11982
11983 #[test]
11984 fn the_caret_walks_a_code_block() {
11985 // Every glyph of a code block used to map to the block's start, so the
11986 // whole block was a single offset and the caret couldn't move inside it.
11987 let src = "```rust\nlet x = 1;\nfn f() {}\n```\n";
11988 let mut d = wysiwyg_doc("code_walk", src);
11989 d.caret = 0;
11990 let seen = walk_right(&mut d);
11991 // The fences are markup: hidden, and no caret stop. The code between
11992 // them is reached a character at a time.
11993 let code = src.find("let").unwrap()..src.find("\n```").unwrap();
11994 for off in code.clone() {
11995 assert!(seen.contains(&off), "offset {off} unreachable: {seen:?}");
11996 }
11997 assert!(seen.contains(&code.end), "no stop after the last line");
11998 }
11999
12000 #[test]
12001 fn the_caret_walks_an_indented_code_block() {
12002 // An indented block's text has the four-space indent stripped, so it
12003 // isn't a verbatim slice and its lines have to be re-found. The caret
12004 // lands on the code, never in the indent.
12005 let src = " indented\n code\n";
12006 let mut d = wysiwyg_doc("indent_code_walk", src);
12007 d.caret = 0;
12008 let seen = walk_right(&mut d);
12009 assert!(seen.contains(&src.find("indented").unwrap()));
12010 assert!(seen.contains(&src.find("code").unwrap()));
12011 assert!(
12012 !seen.contains(&0) || seen[0] == 0,
12013 "the caret starts where it was put"
12014 );
12015 // Nothing in the stripped indent is a stop.
12016 for off in [1, 2, 3] {
12017 assert!(!seen.contains(&off), "landed in the indent at {off}");
12018 }
12019 }
12020
12021 #[test]
12022 fn the_caret_leaves_a_tight_heading() {
12023 // "# H" with text directly under it: the heading row's end and the
12024 // separator row's end are the same offset. Right used to find the
12025 // separator's copy, set the caret to where it already was, and stop.
12026 let mut d = wysiwyg_doc("tight_heading_walk", "# H\ntext\n");
12027 d.caret = 2; // the "H"
12028 let seen = walk_right(&mut d);
12029 assert!(
12030 seen.len() > 2,
12031 "Right stalled at the heading's end: {seen:?}"
12032 );
12033 assert!(
12034 seen.contains(&8),
12035 "never reached the end of \"text\": {seen:?}"
12036 );
12037 }
12038
12039 #[test]
12040 fn the_caret_skips_the_gap_between_two_paragraphs() {
12041 // The blank line between two paragraphs is the boundary itself. The
12042 // caret used to be able to sit on it, and typing there landed in the
12043 // previous paragraph — "A\n\nB" became "A\nx\nB", one paragraph with a
12044 // soft break, so the text visibly snapped back up.
12045 let mut d = wysiwyg_doc("gap_skip", "A\n\nB\n");
12046 d.caret = 1; // the end of "A"
12047 d.move_right(false);
12048 assert_eq!(d.caret, 3, "Right stopped in the gap");
12049 d.insert("x");
12050 assert_eq!(d.source, "A\n\nxB\n", "typing landed outside B");
12051 }
12052
12053 #[test]
12054 fn down_from_a_paragraph_lands_on_the_next_one() {
12055 let mut d = wysiwyg_doc("gap_down", "A\n\nB\n");
12056 d.caret = 0;
12057 d.move_down(false);
12058 assert_eq!(d.caret, 3, "Down stopped in the gap");
12059 }
12060
12061 #[test]
12062 fn clicking_the_gap_lands_on_real_text() {
12063 // A click can still *reach* the gap — it's drawn, so it's clickable.
12064 // It has to resolve to somewhere the caret can be.
12065 let mut d = wysiwyg_doc("gap_click", "A\n\nB\n");
12066 d.click(1, 0, false); // the gap row
12067 assert!(
12068 d.caret == 1 || d.caret == 3,
12069 "click left the caret in the gap at {}",
12070 d.caret
12071 );
12072 d.insert("x");
12073 // Either edge of the boundary is a fair place to land; inside it isn't.
12074 assert!(
12075 d.source == "Ax\n\nB\n" || d.source == "A\n\nxB\n",
12076 "click in the gap typed into the boundary: {:?}",
12077 d.source
12078 );
12079 }
12080
12081 #[test]
12082 fn enter_opens_an_empty_paragraph_the_caret_can_type_into() {
12083 // Enter inserts a paragraph break, which leaves a blank line spare on
12084 // either side of a new one. That middle line is a real empty paragraph:
12085 // the caret lands there, and typing makes a paragraph rather than
12086 // extending a neighbour.
12087 let mut d = wysiwyg_doc("gap_enter", "A\n\nB\n");
12088 d.caret = 1;
12089 d.newline();
12090 assert_eq!(d.source, "A\n\n\n\nB\n");
12091 d.build_visual(80);
12092 let (row, _) = d.caret_pos();
12093 assert!(
12094 d.vmap.row_is_navigable(row),
12095 "the caret landed on a gap row"
12096 );
12097 d.insert("x");
12098 assert_eq!(
12099 d.source, "A\n\nx\n\nB\n",
12100 "the new paragraph merged into a neighbour"
12101 );
12102 }
12103
12104 #[test]
12105 fn enter_at_the_end_of_the_document_opens_a_paragraph_too() {
12106 let mut d = wysiwyg_doc("gap_eof", "A\n");
12107 d.caret = 1;
12108 d.newline();
12109 d.build_visual(80);
12110 let (row, _) = d.caret_pos();
12111 assert!(
12112 d.vmap.row_is_navigable(row),
12113 "the caret landed on a gap row"
12114 );
12115 d.insert("x");
12116 assert!(
12117 d.source.starts_with("A\n\n") && d.source.contains('x'),
12118 "typing at the end merged into A: {:?}",
12119 d.source
12120 );
12121 }
12122
12123 #[test]
12124 fn triple_click_selects_a_paragraph_across_its_soft_breaks() {
12125 // A paragraph broken over two source lines is one paragraph. Selecting
12126 // it must not stop at the newline inside it — that newline is markup the
12127 // rich-text view exists to hide.
12128 let src = "one two\nthree four\n\nnext\n";
12129 let mut d = wysiwyg_doc("triple_para", src);
12130 d.select_block_at(2);
12131 assert_eq!(
12132 d.selected_text(),
12133 Some("one two\nthree four"),
12134 "stopped at the soft break"
12135 );
12136 }
12137
12138 #[test]
12139 fn the_wheel_can_scroll_away_from_a_caret_that_stays_put() {
12140 // The reader scrolls down past the caret's row. Nothing moved the
12141 // caret, so the view must stay where it was put — the old code revealed
12142 // the caret every frame, which dragged the view straight back and made
12143 // the document unscrollable past the caret.
12144 let mut d = wysiwyg_doc("scroll_free", "a\n\nb\n\nc\n\nd\n\ne\n");
12145 d.caret = 0;
12146 d.follow_caret(0, 3, 9); // first frame: the caret is at the top
12147 d.scroll = 4; // the wheel
12148 d.follow_caret(0, 3, 9);
12149 assert_eq!(
12150 d.scroll, 4,
12151 "the wheel was overruled by a caret that never moved"
12152 );
12153 }
12154
12155 #[test]
12156 fn moving_the_caret_brings_the_view_back_to_it() {
12157 let mut d = wysiwyg_doc("scroll_follow", "a\n\nb\n\nc\n\nd\n\ne\n");
12158 d.caret = 0;
12159 d.follow_caret(0, 3, 9);
12160 d.scroll = 6; // scrolled away
12161 d.move_right(false); // ...and now the caret moves
12162 let (row, _) = d.caret_pos();
12163 d.follow_caret(row, 3, 9);
12164 assert!(
12165 d.scroll <= row && row < d.scroll + 3,
12166 "caret row {row} off screen at scroll {}",
12167 d.scroll
12168 );
12169 }
12170
12171 #[test]
12172 fn scrolling_stops_at_the_last_row() {
12173 let mut d = wysiwyg_doc("scroll_clamp", "a\n\nb\n");
12174 d.caret = 0;
12175 d.follow_caret(0, 3, 3); // a first frame, so the caret isn't "new"
12176 d.scroll = 999; // the wheel, spun hard
12177 d.follow_caret(0, 3, 3);
12178 assert_eq!(d.scroll, 2, "scrolled into the void past the document");
12179 }
12180
12181 #[test]
12182 fn every_cell_of_a_wide_table_is_reachable() {
12183 // A table whose cells are far wider than the surface: the columns are
12184 // cut to fit and the text wraps inside them, so no cell hangs off the
12185 // right edge where the caret can never go.
12186 let src = "| Ingredient | Notes |\n|---|---|\n\
12187 | flour milled coarse | sift it twice before folding it in |\n";
12188 let mut d = wysiwyg_doc("wide_table_walk", src);
12189 d.build_visual(30);
12190 d.caret = 0;
12191 let seen = walk_right(&mut d);
12192 for word in ["Ingredient", "Notes", "coarse", "folding"] {
12193 let at = src.find(word).unwrap();
12194 assert!(seen.contains(&at), "{word:?} at {at} unreachable: {seen:?}");
12195 }
12196 }
12197
12198 // ── view parity ──────────────────────────────────────────────────────────
12199 // `doc_with` pins the source view, so everything above tests a view users
12200 // never start in — `Doc::open` opens in WYSIWYG. These run the motion and
12201 // deletion golden cases through *both*, plus the WYSIWYG cases the two
12202 // can't share: where the source carries markup the rendered text is a
12203 // different string, and the views agreeing would itself be the bug.
12204
12205 const VIEWS: [(View, &str); 2] = [(View::Source, "source"), (View::Wysiwyg, "wysiwyg")];
12206
12207 /// Run `action` in both views on one `|`-marked fixture and assert they
12208 /// agree. Plain prose only: with no markup to hide, WYSIWYG renders the
12209 /// source verbatim, so the two views are looking at the same text and any
12210 /// disagreement is one of them having lost the plot.
12211 fn both_views(name: &str, marked: &str, action: fn(&mut Doc)) -> String {
12212 let (src, caret) = parse_caret(marked);
12213 let run = |view: View, tag: &str| {
12214 let mut d = doc_in(view, &format!("{name}_{tag}"), &src);
12215 d.caret = caret;
12216 action(&mut d);
12217 render_caret(&d)
12218 };
12219 let source = run(VIEWS[0].0, VIEWS[0].1);
12220 let wysiwyg = run(VIEWS[1].0, VIEWS[1].1);
12221 assert_eq!(source, wysiwyg, "the views disagree on {marked:?}");
12222 source
12223 }
12224
12225 #[test]
12226 fn word_motion_agrees_across_the_views_on_plain_prose() {
12227 let g = both_views;
12228 assert_eq!(
12229 g("par_wl", "hello wor|ld", |d| d.move_word_left(false)),
12230 "hello |world"
12231 );
12232 assert_eq!(
12233 g("par_wl2", "hello| world", |d| d.move_word_left(false)),
12234 "|hello world"
12235 );
12236 assert_eq!(
12237 g("par_wr", "hel|lo world", |d| d.move_word_right(false)),
12238 "hello| world"
12239 );
12240 assert_eq!(
12241 g("par_wr2", "hello| world", |d| d.move_word_right(false)),
12242 "hello world|"
12243 );
12244 assert_eq!(
12245 g("par_punct", "|foo.bar", |d| d.move_word_right(false)),
12246 "foo|.bar"
12247 );
12248 assert_eq!(
12249 g("par_ext", "hello |world", |d| d.move_word_right(true)),
12250 "hello [world|]"
12251 );
12252 }
12253
12254 #[test]
12255 fn word_deletion_agrees_across_the_views_on_plain_prose() {
12256 let g = both_views;
12257 assert_eq!(
12258 g("par_db", "hello world|", |d| d.delete_word_back()),
12259 "hello |"
12260 );
12261 assert_eq!(
12262 g("par_df", "hello |world", |d| d.delete_word_forward()),
12263 "hello |"
12264 );
12265 assert_eq!(
12266 g("par_db2", "foo |bar baz", |d| d.delete_word_back()),
12267 "|bar baz"
12268 );
12269 assert_eq!(g("par_utf8", "café |ok", |d| d.delete_word_back()), "|ok");
12270 }
12271
12272 #[test]
12273 fn character_motion_and_deletion_agree_across_the_views_on_plain_prose() {
12274 let g = both_views;
12275 assert_eq!(g("par_r", "he|llo", |d| d.move_right(false)), "hel|lo");
12276 assert_eq!(g("par_l", "he|llo", |d| d.move_left(false)), "h|ello");
12277 assert_eq!(g("par_bs", "hel|lo", |d| d.backspace()), "he|lo");
12278 assert_eq!(g("par_del", "hel|lo", |d| d.delete_forward()), "hel|o");
12279 }
12280
12281 #[test]
12282 fn wysiwyg_motion_steps_a_grapheme_cluster_the_way_the_source_view_does() {
12283 // The reproduction: the stop table was built one stop per `char`, so
12284 // Right parked the caret 4 bytes into a ZWJ sequence — a place the
12285 // source view, which steps by grapheme, can't reach and backspace can't
12286 // survive. The two views must land on the same offset.
12287 let family = "👨👩👧"; // three emoji strung together with joiners: one cluster
12288 for (view, tag) in VIEWS {
12289 let mut d = doc_in(view, &format!("cluster_{tag}"), &format!("a{family}b\n"));
12290 d.caret = 1;
12291 d.move_right(false);
12292 assert_eq!(d.caret, 1 + family.len(), "{tag} parked inside the cluster");
12293
12294 // ...and the edit that used to sever a joiner off the front of it.
12295 d.backspace();
12296 assert_eq!(d.source, "ab\n", "{tag} split the cluster");
12297 assert_eq!(d.caret, 1);
12298 }
12299 }
12300
12301 #[test]
12302 fn wysiwyg_motion_treats_a_combining_accent_as_one_character() {
12303 for (view, tag) in VIEWS {
12304 let mut d = doc_in(view, &format!("combining_{tag}"), "e\u{0301}x\n");
12305 d.caret = 0;
12306 d.move_right(false);
12307 assert_eq!(
12308 d.caret,
12309 "e\u{0301}".len(),
12310 "{tag} stopped on the combining mark"
12311 );
12312 }
12313 }
12314
12315 #[test]
12316 fn no_wysiwyg_motion_can_park_the_caret_inside_a_cluster() {
12317 // The general form: whatever route the caret takes through a document
12318 // full of clusters, it never lands between the codepoints of one — so no
12319 // motion-then-backspace sequence can leave a dangling joiner behind.
12320 use unicode_segmentation::UnicodeSegmentation;
12321
12322 let src = "a👨👩👧b e\u{0301}mo👨👩👧ji\n\nnext 👩🚀 line\n";
12323 let mut d = wysiwyg_doc("cluster_walk", src);
12324 d.caret = 0;
12325 let boundaries: Vec<usize> = src
12326 .grapheme_indices(true)
12327 .map(|(i, _)| i)
12328 .chain(std::iter::once(src.len()))
12329 .collect();
12330 for off in walk_right(&mut d) {
12331 assert!(
12332 boundaries.contains(&off),
12333 "Right stopped at {off}, inside a grapheme cluster"
12334 );
12335 }
12336 }
12337
12338 #[test]
12339 fn wysiwyg_word_motion_stays_out_of_hidden_delimiters() {
12340 // The reproduction: ⌥→ from inside the opening `**` computed its
12341 // boundary over the raw source and landed on byte 8 — inside the
12342 // *closing* `**`, which `caret_pos` draws at column 6, immediately after
12343 // "bold". The caret drew past the bold word and sat inside it.
12344 let mut d = wysiwyg_doc("wys_word_delim", "a **bold** c\n");
12345 d.caret = 2;
12346 d.move_word_right(false);
12347 assert!(
12348 d.vmap.is_stop(d.caret),
12349 "landed at {}, not a caret stop",
12350 d.caret
12351 );
12352 assert_eq!(d.caret, 10, "should land on the space after \"bold\"");
12353 // The rendered row is "a bold c": column 6 is the space just past "bold",
12354 // and now the caret is really there rather than only drawn there.
12355 assert_eq!(d.caret_pos(), (0, 6));
12356
12357 // ...and back again: ⌥← returns to the "b", not into the opening `**`.
12358 d.move_word_left(false);
12359 assert_eq!(d.caret, 4);
12360 assert_eq!(d.caret_pos(), (0, 2));
12361 }
12362
12363 #[test]
12364 fn wysiwyg_word_delete_takes_the_markup_with_the_word() {
12365 // The reproduction: ⌥⌫ from after "bold" walked the raw source, stopped
12366 // inside the closing `**`, and left "a ** c\n" — delimiters with no
12367 // opener. Glyph space covers the word alone, which would leave
12368 // "a **** c": markup wrapped around nothing. The word and the styling
12369 // that was only ever the word's go together.
12370 let mut d = wysiwyg_doc("wys_word_del_back", "a **bold** c\n");
12371 d.caret = 10;
12372 d.delete_word_back();
12373 assert_eq!(d.source, "a c\n");
12374 assert_eq!(d.caret, 2);
12375
12376 let mut d = wysiwyg_doc("wys_word_del_fwd", "a **bold** c\n");
12377 d.caret = 4; // the "b"
12378 d.delete_word_forward();
12379 assert_eq!(d.source, "a c\n");
12380 }
12381
12382 #[test]
12383 fn wysiwyg_word_delete_empties_a_nested_mark_and_a_code_span_too() {
12384 let src = "a ***bold*** c\n";
12385 let mut d = wysiwyg_doc("wys_word_del_nest", src);
12386 d.caret = src.find(" c").unwrap();
12387 d.delete_word_back();
12388 assert_eq!(
12389 d.source, "a c\n",
12390 "the emph inside the strong empties it too"
12391 );
12392
12393 let src = "a `code` c\n";
12394 let mut d = wysiwyg_doc("wys_word_del_code", src);
12395 d.caret = src.find(" c").unwrap();
12396 d.delete_word_back();
12397 assert_eq!(d.source, "a c\n");
12398 }
12399
12400 #[test]
12401 fn wysiwyg_word_delete_keeps_a_mark_that_still_has_text() {
12402 // Only an *emptied* node goes. Take one word of two and the `**` still
12403 // has a job to do — over the word that's left, with the space the delete
12404 // pushed against the opening delimiter moved out in front of it, or the
12405 // run would be no run at all (`** words**` is literal asterisks — see
12406 // the mark-edge rule on `splice`).
12407 let src = "a **two words** c\n";
12408 let mut d = wysiwyg_doc("wys_word_del_partial", src);
12409 d.caret = src.find(" words").unwrap();
12410 d.delete_word_back();
12411 assert_eq!(d.source, "a **words** c\n");
12412 }
12413
12414 #[test]
12415 fn source_view_word_motion_still_walks_the_markup() {
12416 // The other half of the decision: in the source view the `**` are
12417 // characters like any other — they're on the screen, so word motion has
12418 // to stop at them and a word-delete has to leave them behind. Only
12419 // WYSIWYG hides them, so only WYSIWYG steps over them.
12420 let g = |n, m, f: fn(&mut Doc)| golden(n, m, f);
12421 assert_eq!(
12422 g("src_word_motion", "a |**bold** c\n", |d| d
12423 .move_word_right(false)),
12424 "a **bold|** c\n"
12425 );
12426 // The same caret as the WYSIWYG reproduction, and the opposite outcome:
12427 // here "a ** c\n" is right, because `bold**` is what's to the left of it.
12428 assert_eq!(
12429 g("src_word_del", "a **bold**| c\n", |d| d.delete_word_back()),
12430 "a **| c\n"
12431 );
12432 }
12433
12434 #[test]
12435 fn every_wysiwyg_motion_lands_on_a_caret_stop() {
12436 // The single invariant both bugs violated: the caret draws and edits at
12437 // the same place only when it's on a stop. `debug_assert_on_a_stop`
12438 // makes the same claim in-place; this pins it from the outside, over a
12439 // document with every kind of thing the map has to be careful about.
12440 // At two widths: the wide one every other test builds at, where no
12441 // fixture folds, and one narrow enough that they all do. A soft wrap is
12442 // where an offset stops being on exactly one row, and testing only the
12443 // width that never wraps is how the caret came to be pinned at the first
12444 // one Down reached.
12445 let src = "# Title\n\na **bold** e\u{0301}mo👨👩👧ji `x` c\n\n\
12446 - item one\n\n| A | B |\n|---|---|\n| x | y |\n";
12447 // A table of named operations, which is what it looks like.
12448 #[allow(clippy::type_complexity)]
12449 let motions: [(&str, fn(&mut Doc)); 8] = [
12450 ("right", |d| d.move_right(false)),
12451 ("left", |d| d.move_left(false)),
12452 ("word_right", |d| d.move_word_right(false)),
12453 ("word_left", |d| d.move_word_left(false)),
12454 ("down", |d| d.move_down(false)),
12455 ("up", |d| d.move_up(false)),
12456 ("home", |d| d.move_home(false)),
12457 ("end", |d| d.move_end(false)),
12458 ];
12459 for width in [80, 12] {
12460 let mut d = wysiwyg_doc("stop_invariant", src);
12461 d.build_visual(width);
12462 let stops: Vec<usize> = (0..=src.len()).filter(|&o| d.vmap.is_stop(o)).collect();
12463 assert!(stops.len() > 20, "fixture should have plenty of stops");
12464 for start in stops {
12465 for (name, motion) in &motions {
12466 d.caret = start;
12467 d.anchor = None;
12468 motion(&mut d);
12469 assert!(
12470 d.vmap.is_stop(d.caret),
12471 "{name} from {start} at width {width} landed at {} — not a caret stop",
12472 d.caret
12473 );
12474 }
12475 }
12476 }
12477 }
12478
12479 #[test]
12480 fn no_wysiwyg_motion_is_a_dead_end() {
12481 // Down held to the bottom of a document reaches the bottom, and Up held
12482 // to the top reaches the top — from anywhere, at a width that wraps. The
12483 // invariant above says a motion lands somewhere legal; this one says it
12484 // gets somewhere at all, which is what a caret pinned at a wrap boundary
12485 // was quietly failing to do while every assertion around it held.
12486 let src = "# Title\n\none two three four five six seven eight nine ten\n\n\
12487 - item one two three four five\n\nlast\n";
12488 for width in [80, 12] {
12489 let mut d = wysiwyg_doc("no_dead_end", src);
12490 d.build_visual(width);
12491 let stops: Vec<usize> = (0..=src.len()).filter(|&o| d.vmap.is_stop(o)).collect();
12492 let (first, last) = (stops[0], stops[stops.len() - 1]);
12493 for &start in &stops {
12494 for (name, motion, want) in [
12495 (
12496 "down",
12497 (|d: &mut Doc| d.move_down(false)) as fn(&mut Doc),
12498 last,
12499 ),
12500 ("up", |d: &mut Doc| d.move_up(false), first),
12501 ] {
12502 d.caret = start;
12503 d.anchor = None;
12504 d.goal_col = None;
12505 // Every row, plus the presses the edges take, plus slack.
12506 for _ in 0..d.vmap.num_rows() + 4 {
12507 motion(&mut d);
12508 }
12509 assert_eq!(
12510 d.caret, want,
12511 "{name} held from {start} at width {width} never arrived"
12512 );
12513 }
12514 }
12515 }
12516 }
12517 // ── display columns ──────────────────────────────────────────────────────
12518 // A `col` is a terminal cell, not a character. The two are the same number
12519 // for the ASCII the fixtures above are written in, which is how they came
12520 // apart in the first place: `你` is one character drawn in two cells, so a
12521 // column counted in characters names a cell the text isn't in — one earlier
12522 // for every wide character to its left.
12523
12524 #[test]
12525 fn a_wide_character_is_two_columns_wide() {
12526 // The reproduction: `你` is one char and two cells, so the caret just
12527 // past it drew at column 1 — inside the character it had already left.
12528 for (view, tag) in VIEWS {
12529 let mut d = doc_in(view, &format!("wide_col_{tag}"), "你好\n");
12530 d.caret = "你".len();
12531 assert_eq!(d.caret_pos(), (0, 2), "{tag}: caret drew inside 你");
12532 d.caret = "你好".len();
12533 assert_eq!(d.caret_pos(), (0, 4), "{tag}");
12534 }
12535 }
12536
12537 #[test]
12538 fn a_cluster_is_as_wide_as_it_is_drawn_not_as_its_codepoints_measure() {
12539 // `👨👩👧` is five codepoints — two-cell, joiner, two-cell, joiner,
12540 // two-cell — measuring six cells one at a time, but the character they
12541 // spell is drawn in two. Width belongs to the cluster, not the glyph,
12542 // and the frontends measure it the same way.
12543 let family = "👨👩👧";
12544 for (view, tag) in VIEWS {
12545 let src = format!("a{family}b\n");
12546 let mut d = doc_in(view, &format!("wide_cluster_{tag}"), &src);
12547 d.caret = 1 + family.len();
12548 assert_eq!(
12549 d.caret_pos(),
12550 (0, 3),
12551 "{tag}: 'a' is one cell, the family two"
12552 );
12553 }
12554 }
12555
12556 #[test]
12557 fn both_cells_of_a_wide_character_mean_the_character() {
12558 // Clicking the far half of `好` is still clicking `好`: half a character
12559 // is not a place the caret can be, so it comes to rest at the
12560 // character's start — the column it would have been drawn at anyway.
12561 for (view, tag) in VIEWS {
12562 let mut d = doc_in(view, &format!("wide_click_{tag}"), "你好\n");
12563 for col in [2, 3] {
12564 d.caret = 0;
12565 d.click(0, col, false);
12566 assert_eq!(d.caret, "你".len(), "{tag}: click at col {col}");
12567 assert_eq!(d.caret_pos(), (0, 2), "{tag}: click at col {col}");
12568 }
12569 // Past the last cell is the line's end, as it is for ASCII.
12570 d.click(0, 9, false);
12571 assert_eq!(d.caret, "你好".len(), "{tag}: click past the end");
12572 }
12573 }
12574
12575 #[test]
12576 fn every_offset_survives_the_trip_out_to_a_column_and_back() {
12577 // The mapping is only a mapping if it inverts: the cell the caret is
12578 // drawn in has to be the cell that brings it back to the same offset.
12579 // Over a fixture where a character may be one cell or two, and one
12580 // codepoint or five.
12581 use unicode_segmentation::UnicodeSegmentation;
12582
12583 let src = "ab 你好 c\n\n👨👩👧 e\u{0301}x 漢字\n\nplain ascii\n";
12584
12585 let mut d = doc_in(View::Source, "roundtrip_source", src);
12586 // Every offset the source view's caret can occupy: it steps by grapheme
12587 // cluster, so those are its boundaries.
12588 for (off, _) in src
12589 .grapheme_indices(true)
12590 .chain(std::iter::once((src.len(), "")))
12591 {
12592 d.caret = off;
12593 let (row, col) = d.caret_pos();
12594 d.click(row, col, false);
12595 assert_eq!(d.caret, off, "source: {off} → ({row}, {col}) → {}", d.caret);
12596 }
12597
12598 // And in WYSIWYG, where the offsets the caret can occupy are the map's
12599 // stops rather than every boundary.
12600 let mut d = doc_in(View::Wysiwyg, "roundtrip_wysiwyg", src);
12601 let stops: Vec<usize> = (0..=src.len()).filter(|&o| d.vmap.is_stop(o)).collect();
12602 assert!(stops.len() > 20, "fixture should have plenty of stops");
12603 for off in stops {
12604 d.caret = off;
12605 let (row, col) = d.caret_pos();
12606 d.click(row, col, false);
12607 assert_eq!(
12608 d.caret, off,
12609 "wysiwyg: {off} → ({row}, {col}) → {}",
12610 d.caret
12611 );
12612 }
12613 }
12614
12615 #[test]
12616 fn vertical_motion_aims_at_a_column_the_reader_can_see() {
12617 // Down from under `世` lands under the glyph in that cell, not two
12618 // characters further along the line. The goal is a column, so a line of
12619 // wide characters and a line of ASCII line up the way they're drawn.
12620 //
12621 // The gap differs by view: a bare newline inside a paragraph is a soft
12622 // break, which WYSIWYG draws as a space on a single row. The views share
12623 // a grid only where the source's lines are the renderer's rows too.
12624 for (view, tag) in VIEWS {
12625 let gap = if view == View::Source { "\n" } else { "\n\n" };
12626 let src = format!("你好世{gap}abcdef\n");
12627 let mut d = doc_in(view, &format!("goal_wide_{tag}"), &src);
12628 d.caret = "你好".len();
12629 assert_eq!(d.caret_pos().1, 4, "{tag}: `世` is drawn at column 4");
12630 d.move_down(false);
12631 assert_eq!(d.caret_pos().1, 4, "{tag}: goal column lost");
12632 assert!(
12633 d.source[d.caret..].starts_with('e'),
12634 "{tag}: landed on the wrong glyph"
12635 );
12636 }
12637 }
12638
12639 #[test]
12640 fn a_goal_column_landing_inside_a_wide_character_lands_on_it() {
12641 // Down from column 3 onto `你好`, whose characters start at columns 0
12642 // and 2: column 3 is the *second* cell of `好`. There is nowhere to be
12643 // between the cells of one character, so the caret rests on it — and on
12644 // its start, which is the only offset there that is a caret stop.
12645 for (view, tag) in VIEWS {
12646 let gap = if view == View::Source { "\n" } else { "\n\n" };
12647 let src = format!("abcdef{gap}你好\n");
12648 let mut d = doc_in(view, &format!("goal_inside_{tag}"), &src);
12649 let line = src.find('你').unwrap();
12650 d.caret = 3;
12651 d.move_down(false);
12652 assert_eq!(d.caret, line + "你".len(), "{tag}: landed off `好`'s start");
12653 assert_eq!(d.caret_pos().1, 2, "{tag}: drew between `好`'s cells");
12654 }
12655 }
12656
12657 #[test]
12658 fn a_caret_in_a_table_cell_of_wide_text_draws_where_the_text_is() {
12659 // The column the cell's text is laid out in is measured in cells, so the
12660 // caret walking that text has to be too — the two agreeing is the whole
12661 // point of the grid staying square.
12662 let mut d = wysiwyg_doc("table_wide", "| A | B |\n|---|---|\n| 你好 | y |\n");
12663 let at = d.source.find("你").unwrap();
12664 d.caret = at;
12665 let (row, col) = d.caret_pos();
12666 // `│ ` opens the row, so the cell's text starts at column 2; `好` is two
12667 // cells further along.
12668 assert_eq!(col, 2, "the cell's first character");
12669 d.move_right(false);
12670 assert_eq!(
12671 d.caret_pos(),
12672 (row, 4),
12673 "`好` is drawn past `你`'s two cells"
12674 );
12675 assert_eq!(d.caret, at + "你".len());
12676 }
12677
12678 // ── active inline marks ───────────────────────────────────────────────────
12679
12680 /// The marks at a `|`-marked fixture's caret, in `InlineMarks::iter` order.
12681 fn marks(view: View, name: &str, marked: &str) -> Vec<InlineKind> {
12682 let (src, caret) = parse_caret(marked);
12683 let mut d = doc_in(view, name, &src);
12684 d.caret = caret;
12685 d.active_inline_marks().iter().collect()
12686 }
12687
12688 /// The marks over the selection `[start, end)`.
12689 fn marks_over(view: View, name: &str, src: &str, start: usize, end: usize) -> Vec<InlineKind> {
12690 let mut d = doc_in(view, name, src);
12691 d.anchor = Some(start);
12692 d.caret = end;
12693 d.active_inline_marks().iter().collect()
12694 }
12695
12696 #[test]
12697 fn a_caret_in_a_mark_reports_it() {
12698 for (view, tag) in VIEWS {
12699 let m = |marked| marks(view, &format!("marks_in_{tag}"), marked);
12700 assert_eq!(m("a **bo|ld** b"), [InlineKind::Strong], "{tag}");
12701 assert_eq!(m("a *it|alic* b"), [InlineKind::Emph], "{tag}");
12702 assert_eq!(m("a `co|de` b"), [InlineKind::Verbatim], "{tag}");
12703 // Plain text under no mark lights nothing — the toolbar's resting state.
12704 assert_eq!(m("a| **bold** b"), [], "{tag}");
12705 assert!(m("plain t|ext").is_empty(), "{tag}");
12706 }
12707 }
12708
12709 #[test]
12710 fn nested_marks_all_report() {
12711 // Bold *and* italic: a toolbar lights both buttons, so the set has both —
12712 // the ancestor chain is a chain, and every mark on it is in force.
12713 for (view, tag) in VIEWS {
12714 assert_eq!(
12715 marks(
12716 view,
12717 &format!("marks_nested_{tag}"),
12718 "**bold and *bo|th*** end"
12719 ),
12720 [InlineKind::Strong, InlineKind::Emph],
12721 "{tag}"
12722 );
12723 }
12724 }
12725
12726 #[test]
12727 fn the_caret_at_a_marks_edge_reports_it_where_typing_would_extend_it() {
12728 // The offsets a WYSIWYG caret actually reaches at a bold run's edges are
12729 // the first byte of its text and the byte after its last — both inside
12730 // the mark's span, both places typing lands inside the bold. The offset
12731 // past the closing delimiter is the next text, and reports nothing.
12732 let src = "a **bold** b";
12733 let inner_start = src.find("bold").unwrap(); // 4
12734 let inner_end = inner_start + "bold".len(); // 8, on the closing `**`
12735 for (view, tag) in VIEWS {
12736 let mut d = doc_in(view, &format!("marks_edge_{tag}"), src);
12737 for off in [2, 3, inner_start, inner_end, 9] {
12738 d.caret = off;
12739 assert!(
12740 d.active_inline_marks().contains(InlineKind::Strong),
12741 "{tag}: offset {off} is inside the strong span"
12742 );
12743 }
12744 for off in [0, 1, 10, 11, 12] {
12745 d.caret = off;
12746 assert!(
12747 !d.active_inline_marks().contains(InlineKind::Strong),
12748 "{tag}: offset {off} is outside the strong run"
12749 );
12750 }
12751 }
12752 }
12753
12754 #[test]
12755 fn a_mark_ends_the_same_way_at_the_end_of_the_buffer_as_in_the_middle() {
12756 // Regression: twig resolves an offset that is one node's end and the
12757 // next one's start to the node that *starts* there, so `**bold**|\n`
12758 // isn't bold. With nothing following there's no tie to break and the
12759 // chain still ended at the mark, which made a trailing `\n` — not the
12760 // text — decide whether the caret after a bold word reported bold. It's
12761 // the offset past the mark either way, and typing there is plain either
12762 // way. A blank document typed into is exactly this shape.
12763 for (view, tag) in VIEWS {
12764 let m = |name: String, marked| marks(view, &name, marked);
12765 assert_eq!(
12766 m(format!("marks_eob_{tag}"), "**bold**|"),
12767 [],
12768 "{tag}: no trailing newline"
12769 );
12770 assert_eq!(
12771 m(format!("marks_eol_{tag}"), "**bold**|\n"),
12772 [],
12773 "{tag}: with one"
12774 );
12775 // And the last offset that *is* in the mark still is.
12776 assert_eq!(
12777 m(format!("marks_eob_in_{tag}"), "**bold*|*"),
12778 [InlineKind::Strong],
12779 "{tag}"
12780 );
12781 }
12782 }
12783
12784 #[test]
12785 fn a_selection_reports_a_mark_only_when_it_covers_the_whole_thing() {
12786 let src = "a **bold** b";
12787 let (b, d_) = (src.find("bold").unwrap(), src.find("bold").unwrap() + 4);
12788 for (view, tag) in VIEWS {
12789 let m = |s, e| marks_over(view, &format!("marks_sel_{tag}"), src, s, e);
12790 // The whole bold word, and a slice of it.
12791 assert_eq!(m(b, d_), [InlineKind::Strong], "{tag}: the whole word");
12792 assert_eq!(m(b + 1, d_ - 1), [InlineKind::Strong], "{tag}: a slice");
12793 // Ending exactly at the closing delimiter's start is still all-bold:
12794 // an exclusive end sits *past* the last selected character, so the
12795 // question is asked of the character, not the boundary.
12796 assert_eq!(
12797 m(b, d_ + 2),
12798 [InlineKind::Strong],
12799 "{tag}: through the close"
12800 );
12801 // Half in, half out: Bold lit here would claim a press turns it off.
12802 assert_eq!(m(0, d_), [], "{tag}: leading plain text");
12803 assert_eq!(m(b, src.len()), [], "{tag}: trailing plain text");
12804 }
12805 }
12806
12807 #[test]
12808 fn a_selection_across_two_runs_of_the_same_mark_reports_nothing() {
12809 // Both ends are bold, but the space between them isn't — two runs are two
12810 // nodes, which is exactly what the node id catches and a kind-only
12811 // comparison would not.
12812 let src = "**one** **two**";
12813 for (view, tag) in VIEWS {
12814 let m = marks_over(view, &format!("marks_runs_{tag}"), src, 2, 13);
12815 assert_eq!(m, [], "{tag}: `one** **two` is not all bold");
12816 }
12817 }
12818
12819 #[test]
12820 fn marks_read_the_document_as_it_is_edited() {
12821 // The point of asking twig every frame instead of caching: the answer has
12822 // to follow the toggle that changed it.
12823 let mut d = wysiwyg_doc("marks_live", "one two\n");
12824 d.anchor = Some(0);
12825 d.caret = 3;
12826 assert!(d.active_inline_marks().is_empty(), "plain to start");
12827 d.toggle(InlineKind::Strong);
12828 assert_eq!(d.source, "**one** two\n");
12829 // `toggle` leaves the bolded text selected, so the button it lit stays lit.
12830 assert!(d.active_inline_marks().contains(InlineKind::Strong));
12831 d.toggle(InlineKind::Strong);
12832 assert!(d.active_inline_marks().is_empty(), "and off again");
12833 }
12834
12835 #[test]
12836 fn a_link_is_not_an_inline_mark() {
12837 // `link`/`str` are inline nodes, but nothing on the inline toolbar
12838 // toggles them — a set with a "link mark" in it would have no button.
12839 for (view, tag) in VIEWS {
12840 assert_eq!(
12841 marks(view, &format!("marks_link_{tag}"), "a [te|xt](u) b"),
12842 [],
12843 "{tag}"
12844 );
12845 }
12846 }
12847
12848 // ── blank documents ───────────────────────────────────────────────────────
12849
12850 #[test]
12851 fn a_blank_document_is_untitled_empty_and_markdown() {
12852 let mut d = Doc::blank().unwrap();
12853 assert!(d.is_untitled());
12854 assert_eq!(d.path, PathBuf::new());
12855 assert_eq!(
12856 d.file_name(),
12857 "untitled",
12858 "the header has to show something"
12859 );
12860 assert_eq!(d.format_name(), "markdown");
12861 assert_eq!(d.source, "");
12862 assert!(!d.dirty, "nothing typed yet is nothing to lose");
12863 assert_eq!(d.disk_state(), DiskState::Untitled);
12864 // And it's a document you can be in: the default view renders it.
12865 d.build_visual(80);
12866 assert_eq!(d.caret, 0);
12867 }
12868
12869 #[test]
12870 fn saving_an_untitled_document_asks_for_a_name_instead_of_writing() {
12871 let mut d = Doc::blank().unwrap();
12872 d.insert("hello");
12873 assert!(d.dirty);
12874 d.save();
12875 assert_eq!(d.status.as_deref(), Some("untitled — save as…"));
12876 assert!(d.dirty, "it must not come away believing it saved");
12877 assert!(d.is_untitled(), "and it still has no file");
12878 }
12879
12880 #[test]
12881 fn a_blank_document_becomes_a_real_one_at_the_first_save_as() {
12882 let p = temp_path("blank_save_as");
12883 let mut d = Doc::blank().unwrap();
12884 // Plain text — a blank doc opens in Hidden mode, where a typed `#` would
12885 // be kept literal (`\#`); this test is about save-as, not escaping (which
12886 // has its own test), so it types nothing that escaping would touch.
12887 d.insert("hi");
12888 d.save_as(p.clone());
12889 assert_eq!(std::fs::read_to_string(&p).unwrap(), "hi");
12890 assert!(!d.is_untitled());
12891 assert!(!d.dirty);
12892 assert_eq!(d.file_name(), p.file_name().unwrap().to_string_lossy());
12893 assert_eq!(
12894 d.disk_state(),
12895 DiskState::Unchanged,
12896 "the watermark is stamped"
12897 );
12898 // And ⌘S is a plain save from here on.
12899 d.insert("!");
12900 d.save();
12901 assert_eq!(std::fs::read_to_string(&p).unwrap(), "hi!");
12902 let _ = std::fs::remove_file(&p);
12903 }
12904
12905 // ── a file that isn't there yet ───────────────────────────────────────────
12906
12907 /// A unique path in the temp dir with the given extension, guaranteed not to
12908 /// exist — what `leaf notes.md` is handed when the file has never been made.
12909 fn missing_path(name: &str, ext: &str) -> PathBuf {
12910 static SEQ: std::sync::atomic::AtomicUsize = std::sync::atomic::AtomicUsize::new(0);
12911 let seq = SEQ.fetch_add(1, std::sync::atomic::Ordering::Relaxed);
12912 let mut p = std::env::temp_dir();
12913 p.push(format!("leaf_test_new_{name}_{seq}.{ext}"));
12914 let _ = std::fs::remove_file(&p);
12915 p
12916 }
12917
12918 #[test]
12919 fn a_file_that_doesnt_exist_opens_as_an_empty_named_document() {
12920 let p = missing_path("named", "md");
12921 let mut d = Doc::open_or_create(p.clone()).unwrap();
12922
12923 assert_eq!(d.source, "", "nothing was read, so there's nothing in it");
12924 assert!(!d.dirty, "an untouched new buffer has nothing to lose");
12925 assert!(
12926 !d.is_untitled(),
12927 "it has the name the user asked for — ^S must not detour to Save As"
12928 );
12929 assert_eq!(d.file_name(), p.file_name().unwrap().to_str().unwrap());
12930 assert!(d.path.is_absolute(), "the same absolute path `open` stores");
12931 assert!(!p.exists(), "and opening it wrote nothing");
12932 // And it's a document you can be in.
12933 d.build_visual(80);
12934 assert_eq!(d.caret, 0);
12935 }
12936
12937 #[test]
12938 fn a_new_file_is_created_by_its_first_save() {
12939 let p = missing_path("first_save", "md");
12940 let mut d = Doc::open_or_create(p.clone()).unwrap();
12941 d.insert("hello\n");
12942 assert!(d.dirty);
12943 d.save();
12944
12945 assert_eq!(
12946 std::fs::read_to_string(&p).unwrap(),
12947 "hello\n",
12948 "a plain ^S wrote it — no Save As, no name to invent"
12949 );
12950 assert!(!d.dirty);
12951 assert_eq!(d.disk_state(), DiskState::Unchanged);
12952 let _ = std::fs::remove_file(&p);
12953 }
12954
12955 #[test]
12956 fn a_new_file_takes_its_format_from_the_extension() {
12957 // The one thing `blank` can't do: with no name it has to assume Markdown,
12958 // and typing djot into a Markdown parse is the wrong buffer.
12959 let dj = missing_path("format", "dj");
12960 assert_eq!(Doc::open_or_create(dj).unwrap().format_name(), "djot");
12961 let md = missing_path("format", "md");
12962 assert_eq!(Doc::open_or_create(md).unwrap().format_name(), "markdown");
12963 }
12964
12965 #[test]
12966 fn a_new_file_reports_itself_missing_until_it_is_saved() {
12967 // Not `Untitled` — that's the answer for a document with no path, and it
12968 // would tell a frontend there is nothing a save could collide with. Here
12969 // there is a path, and the file simply isn't at it yet.
12970 let p = missing_path("disk_state", "md");
12971 let mut d = Doc::open_or_create(p.clone()).unwrap();
12972 assert_eq!(d.disk_state(), DiskState::Missing);
12973
12974 // Somebody else creates it while the buffer is open: that's an overwrite
12975 // the frontend has to be able to prompt about, exactly as for an opened
12976 // file. Their bytes, not ours, so `Changed`.
12977 std::fs::write(&p, "theirs\n").unwrap();
12978 assert_eq!(d.disk_state(), DiskState::Changed);
12979
12980 // Saving makes the file ours and re-stamps the watermark.
12981 d.insert("ours\n");
12982 d.save();
12983 assert_eq!(d.disk_state(), DiskState::Unchanged);
12984 assert_eq!(std::fs::read_to_string(&p).unwrap(), "ours\n");
12985 let _ = std::fs::remove_file(&p);
12986 }
12987
12988 #[test]
12989 fn open_or_create_still_opens_a_file_that_is_there() {
12990 let d = doc_with("open_or_create_existing", "body\n");
12991 let reopened = Doc::open_or_create(d.path.clone()).unwrap();
12992 assert_eq!(reopened.source, "body\n");
12993 assert_eq!(reopened.disk_state(), DiskState::Unchanged);
12994 }
12995
12996 #[test]
12997 fn a_missing_file_with_no_readable_extension_is_still_an_error() {
12998 // A mistyped flag or a stray argument must not become a buffer promising
12999 // to save somewhere — the same refusal `open` gives a real file.
13000 let mut p = std::env::temp_dir();
13001 p.push("leaf_test_new_bad_ext.wat");
13002 assert!(Doc::open_or_create(p).is_err());
13003 let mut none = std::env::temp_dir();
13004 none.push("leaf_test_new_no_ext");
13005 assert!(Doc::open_or_create(none).is_err());
13006 }
13007
13008 #[test]
13009 fn a_new_file_in_a_directory_that_doesnt_exist_opens_but_wont_save() {
13010 // Opening reads nothing, so there is nothing to fail on yet; the write is
13011 // where it fails, and it says so rather than claiming a save.
13012 let p = std::env::temp_dir().join("leaf_test_no_such_dir_c41/doc.md");
13013 let mut d = Doc::open_or_create(p).unwrap();
13014 d.insert("x");
13015 d.save();
13016 assert!(
13017 d.status.as_deref().unwrap().starts_with("save failed:"),
13018 "got {:?}",
13019 d.status
13020 );
13021 assert!(d.dirty, "it must not come away believing it saved");
13022 }
13023
13024 // ── save as ───────────────────────────────────────────────────────────────
13025
13026 /// A unique path in the temp dir that no fixture wrote — a Save As target.
13027 fn temp_path(name: &str) -> PathBuf {
13028 static SEQ: std::sync::atomic::AtomicUsize = std::sync::atomic::AtomicUsize::new(0);
13029 let seq = SEQ.fetch_add(1, std::sync::atomic::Ordering::Relaxed);
13030 let mut p = std::env::temp_dir();
13031 p.push(format!("leaf_test_target_{name}_{seq}.md"));
13032 let _ = std::fs::remove_file(&p);
13033 p
13034 }
13035
13036 #[test]
13037 fn save_as_moves_the_document_and_leaves_the_old_file_alone() {
13038 let mut d = doc_with("save_as_move", "original\n");
13039 let old = d.path.clone();
13040 let new = temp_path("save_as_move");
13041 d.insert("edited: ");
13042 d.save_as(new.clone());
13043
13044 assert_eq!(std::fs::read_to_string(&new).unwrap(), "edited: original\n");
13045 assert_eq!(
13046 std::fs::read_to_string(&old).unwrap(),
13047 "original\n",
13048 "Save As doesn't touch the file it came from"
13049 );
13050 assert_eq!(d.path, new, "the document moved");
13051 assert!(!d.dirty);
13052 assert_eq!(
13053 d.status.as_deref(),
13054 Some(&*format!("saved {}", d.file_name()))
13055 );
13056
13057 // Every later save follows it, which is the whole difference from a copy.
13058 d.caret = 0;
13059 d.insert("re-");
13060 d.save();
13061 assert_eq!(
13062 std::fs::read_to_string(&new).unwrap(),
13063 "re-edited: original\n"
13064 );
13065 assert_eq!(std::fs::read_to_string(&old).unwrap(), "original\n");
13066 let _ = std::fs::remove_file(&new);
13067 }
13068
13069 #[test]
13070 fn save_as_overwrites_an_existing_target() {
13071 // The picker already asked; asking again down here is the same question
13072 // twice, and the second one has no way to be answered.
13073 let new = temp_path("save_as_over");
13074 std::fs::write(&new, "theirs\n").unwrap();
13075 let mut d = doc_with("save_as_over", "ours\n");
13076 d.save_as(new.clone());
13077 assert_eq!(std::fs::read_to_string(&new).unwrap(), "ours\n");
13078 let _ = std::fs::remove_file(&new);
13079 }
13080
13081 #[test]
13082 fn a_save_as_that_fails_leaves_the_document_where_it_was() {
13083 let mut d = doc_with("save_as_fail", "body\n");
13084 let old = d.path.clone();
13085 d.insert("x");
13086 // A directory that doesn't exist: the write can't land.
13087 let bad = std::env::temp_dir().join("leaf_test_no_such_dir_9f2/doc.md");
13088 d.save_as(bad);
13089
13090 assert_eq!(
13091 d.path, old,
13092 "the document must not move to a file that isn't there"
13093 );
13094 assert!(d.dirty, "and must not believe it saved");
13095 assert!(
13096 d.status.as_deref().unwrap().starts_with("save failed:"),
13097 "the same failure a plain save reports, got {:?}",
13098 d.status
13099 );
13100 // The original is still the document's file, and still saveable.
13101 d.save();
13102 assert_eq!(std::fs::read_to_string(&old).unwrap(), "xbody\n");
13103 assert!(!d.dirty);
13104 }
13105
13106 #[test]
13107 fn save_as_renames_without_reparsing_the_format() {
13108 // `.dj` on the name doesn't make the buffer djot: it was parsed as
13109 // Markdown and still is, and saying otherwise would be a conversion the
13110 // user never asked for (and an undo history thrown away to do it).
13111 let mut d = doc_with("save_as_format", "**b**\n");
13112 let mut new = temp_path("save_as_format");
13113 new.set_extension("dj");
13114 d.save_as(new.clone());
13115 assert_eq!(d.format_name(), "markdown");
13116 let _ = std::fs::remove_file(&new);
13117 }
13118
13119 // ── external change / reload ──────────────────────────────────────────────
13120
13121 #[test]
13122 fn an_untouched_file_reports_unchanged() {
13123 let mut d = doc_with("disk_clean", "body\n");
13124 assert_eq!(d.disk_state(), DiskState::Unchanged);
13125 // Editing the buffer is not editing the file.
13126 d.insert("x");
13127 assert_eq!(d.disk_state(), DiskState::Unchanged);
13128 assert!(d.dirty);
13129 // Saving re-stamps the watermark rather than reporting our own bytes back.
13130 d.save();
13131 assert_eq!(d.disk_state(), DiskState::Unchanged);
13132 }
13133
13134 #[test]
13135 fn a_file_written_underneath_reports_changed() {
13136 let mut d = doc_with("disk_changed", "body\n");
13137 std::fs::write(&d.path, "someone else\n").unwrap();
13138 assert_eq!(d.disk_state(), DiskState::Changed);
13139 // Dirty *and* changed is the clobber: both halves are readable, and
13140 // leaf-core takes neither side.
13141 d.insert("x");
13142 assert!(d.dirty && d.disk_state() == DiskState::Changed);
13143 // Saving anyway is allowed — the frontend asked, or chose not to.
13144 d.save();
13145 assert_eq!(std::fs::read_to_string(&d.path).unwrap(), "xbody\n");
13146 assert_eq!(d.disk_state(), DiskState::Unchanged);
13147 }
13148
13149 #[test]
13150 fn a_file_rewritten_with_the_same_bytes_is_unchanged() {
13151 // The hash is what makes this honest: the file was written (a fresh
13152 // mtime), and nothing about the document is stale.
13153 let d = doc_with("disk_same_bytes", "body\n");
13154 std::fs::write(&d.path, "body\n").unwrap();
13155 assert_eq!(d.disk_state(), DiskState::Unchanged);
13156 }
13157
13158 #[test]
13159 fn a_deleted_file_reports_missing() {
13160 let mut d = doc_with("disk_missing", "body\n");
13161 std::fs::remove_file(&d.path).unwrap();
13162 assert_eq!(d.disk_state(), DiskState::Missing);
13163 // A save recreates it, and the document is whole again.
13164 d.save();
13165 assert_eq!(d.disk_state(), DiskState::Unchanged);
13166 assert_eq!(std::fs::read_to_string(&d.path).unwrap(), "body\n");
13167 }
13168
13169 #[test]
13170 fn reload_replaces_the_document_with_the_file() {
13171 for (view, tag) in VIEWS {
13172 let mut d = doc_in(view, &format!("reload_{tag}"), "one\n\ntwo\n");
13173 d.insert("edited ");
13174 assert!(d.dirty);
13175 std::fs::write(&d.path, "one\n\ntwo\n\nthree\n").unwrap();
13176 d.reload();
13177
13178 assert_eq!(d.source, "one\n\ntwo\n\nthree\n", "{tag}");
13179 assert!(!d.dirty, "{tag}: the file is what we have");
13180 assert_eq!(d.disk_state(), DiskState::Unchanged, "{tag}");
13181 assert_eq!(
13182 d.status.as_deref(),
13183 Some(&*format!("reloaded {}", d.file_name()))
13184 );
13185 // The reloaded tree is live, not the old parse.
13186 d.caret = d.source.find("three").unwrap();
13187 assert_eq!(d.breadcrumb(), "doc › para › str", "{tag}");
13188 }
13189 }
13190
13191 #[test]
13192 fn reload_clamps_the_caret_and_drops_the_selection() {
13193 let mut d = doc_with("reload_caret", "a long first line\n");
13194 d.caret = 12;
13195 d.anchor = Some(4);
13196 std::fs::write(&d.path, "short\n").unwrap();
13197 d.reload();
13198 assert_eq!(d.caret, d.source.len(), "clamped into the shorter file");
13199 assert_eq!(
13200 d.anchor, None,
13201 "a selection over bytes that changed is a lie"
13202 );
13203 assert!(d.selection().is_none());
13204
13205 // A caret the file still has room for stays put.
13206 let mut d = doc_with("reload_caret_keep", "one\n\ntwo\n");
13207 d.caret = 2;
13208 std::fs::write(&d.path, "one\n\ntwo\n\nthree\n").unwrap();
13209 d.reload();
13210 assert_eq!(d.caret, 2);
13211 }
13212
13213 /// A silent reload is something that happened *to* a reader — a formatter,
13214 /// a `git checkout` — so it has to be undoable like anything else that
13215 /// changes the document, and undoable as one step rather than as however
13216 /// many the file happens to differ by.
13217 #[test]
13218 fn reload_is_one_undo_step_and_keeps_the_history_under_it() {
13219 let mut d = doc_with("reload_undo", "body\n");
13220 d.insert("x");
13221 assert_eq!(d.source, "xbody\n");
13222 std::fs::write(&d.path, "replaced\n").unwrap();
13223 d.reload();
13224 assert_eq!(d.source, "replaced\n");
13225 assert!(!d.dirty, "a reload lands clean");
13226
13227 // One ^Z takes the whole swap off, and hands back the unsaved work it
13228 // replaced — which is unsaved again, because the file no longer says it.
13229 d.undo();
13230 assert_eq!(d.source, "xbody\n", "the reload comes off in one step");
13231 assert!(d.dirty, "and what it comes back to is unsaved");
13232 // …and the history under it is still there.
13233 d.undo();
13234 assert_eq!(
13235 d.source, "body\n",
13236 "the typing before the reload undoes too"
13237 );
13238 // Redo walks back up through the reload.
13239 d.redo();
13240 d.redo();
13241 assert_eq!(d.source, "replaced\n");
13242 }
13243
13244 /// A file rewritten with the bytes it already had is not an edit, so it
13245 /// must not leave an undo step behind for something nobody did.
13246 #[test]
13247 fn reloading_identical_bytes_pushes_no_undo_step() {
13248 let mut d = doc_with("reload_same", "body\n");
13249 d.insert("x");
13250 std::fs::write(&d.path, "xbody\n").unwrap();
13251 d.reload();
13252 assert_eq!(d.source, "xbody\n");
13253 assert!(!d.dirty, "the file now says what the buffer does");
13254 d.undo();
13255 assert_eq!(
13256 d.source, "body\n",
13257 "one step back is the typing, not a no-op"
13258 );
13259 }
13260
13261 #[test]
13262 fn a_reload_that_cant_read_leaves_the_document_alone() {
13263 let mut d = doc_with("reload_gone", "body\n");
13264 d.insert("x");
13265 std::fs::remove_file(&d.path).unwrap();
13266 d.reload();
13267 assert_eq!(d.source, "xbody\n", "the unsaved work is still here");
13268 assert!(d.dirty);
13269 assert!(
13270 d.status.as_deref().unwrap().starts_with("reload failed:"),
13271 "{:?}",
13272 d.status
13273 );
13274
13275 // And an untitled document has nothing to reload from.
13276 let mut d = Doc::blank().unwrap();
13277 d.insert("typed");
13278 d.reload();
13279 assert_eq!(d.source, "typed");
13280 assert_eq!(d.status.as_deref(), Some("no file to reload"));
13281 }
13282
13283 #[test]
13284 fn a_read_only_document_refuses_every_door() {
13285 let mut d = doc_with("readonly", "one two three\n");
13286 d.insert("x");
13287 assert!(d.dirty, "writable first, so the undo step exists");
13288 d.set_read_only(true);
13289 let before = d.source.clone();
13290 d.insert("y");
13291 d.backspace();
13292 d.undo();
13293 d.redo();
13294 assert_eq!(d.source, before, "no door moved a byte");
13295 d.set_read_only(false);
13296 d.undo();
13297 assert_ne!(d.source, before, "off again, the same doors work");
13298 }
13299
13300 /// The doors that go to twig's own verbs rather than through the splice.
13301 /// Typed text in the rendered view under the default markup mode is the
13302 /// everyday one — it is what a keystroke in leaf-web or the Apple views
13303 /// becomes — and it walked straight past the gate.
13304 #[test]
13305 fn a_read_only_document_refuses_the_doors_around_the_splice() {
13306 let mut d = wysiwyg_doc(
13307 "readonly-doors",
13308 "one two three\n\n| a | b |\n|---|---|\n| c | d |\n",
13309 );
13310 d.set_markup_mode(MarkupMode::None);
13311 d.set_read_only(true);
13312 let before = d.source.clone();
13313 d.place_caret(3, false);
13314 d.insert("y");
13315 d.insert_link("https://example.com");
13316 d.insert_image("a.png", "alt");
13317 d.insert_thematic_break();
13318 d.insert_footnote();
13319 d.place_caret(0, false);
13320 d.place_caret(3, true);
13321 d.toggle(InlineKind::Strong);
13322 d.toggle_heading(2);
13323 d.set_block(BlockKind::Paragraph);
13324 d.toggle_list(false);
13325 d.toggle_blockquote();
13326 d.toggle_task_item();
13327 d.newline();
13328 d.indent();
13329 d.set_code_language("rust");
13330 let in_cell = d.source.find("| c").unwrap() + 2;
13331 d.place_caret(in_cell, false);
13332 assert!(d.caret_in_table(), "the caret is in the grid");
13333 assert!(!d.cell_line_break(), "the cell break reports the refusal");
13334 assert_eq!(d.source, before, "no door moved a byte");
13335 assert!(!d.dirty, "nothing to save");
13336 d.set_read_only(false);
13337 d.place_caret(3, false);
13338 d.insert("y");
13339 assert_ne!(d.source, before, "off again, the same doors work");
13340 }
13341
13342 #[test]
13343 fn a_selection_quote_carries_its_context_on_char_boundaries() {
13344 let mut d = doc_with("quote", "before 你好 exact 世界 after\n");
13345 let start = d.source.find("exact").unwrap();
13346 d.place_caret(start, false);
13347 d.place_caret(start + "exact".len(), true);
13348 let q = d.selection_quote(3).unwrap();
13349 assert_eq!(q.exact, "exact");
13350 assert_eq!(
13351 q.prefix, "你好 ",
13352 "chars, not bytes — the multibyte pair counts as two"
13353 );
13354 assert_eq!(q.suffix, " 世界");
13355 assert_eq!(&d.source[q.start..q.end], "exact");
13356 // At the edges the context clips rather than erring.
13357 d.place_caret(0, false);
13358 d.place_caret(6, true);
13359 let q = d.selection_quote(40).unwrap();
13360 assert_eq!(q.prefix, "");
13361 assert_eq!(q.exact, "before");
13362 // No selection is no quote.
13363 d.place_caret(0, false);
13364 assert!(d.selection_quote(3).is_none());
13365 }
13366
13367 #[test]
13368 fn highlights_are_kept_sorted_and_answer_point_queries() {
13369 let mut d = doc_with("hl", "one two three\n");
13370 d.set_highlights(vec![
13371 Highlight {
13372 start: 8,
13373 end: 13,
13374 id: "b".into(),
13375 color: None,
13376 marker: None,
13377 },
13378 Highlight {
13379 start: 0,
13380 end: 3,
13381 id: "a".into(),
13382 color: Some("#ffe066".into()),
13383 marker: None,
13384 },
13385 Highlight {
13386 start: 5,
13387 end: 5,
13388 id: "empty".into(),
13389 color: None,
13390 marker: None,
13391 },
13392 ]);
13393 assert_eq!(
13394 d.highlights()
13395 .iter()
13396 .map(|h| h.id.as_str())
13397 .collect::<Vec<_>>(),
13398 ["a", "b"],
13399 "sorted by start, the empty range dropped"
13400 );
13401 assert_eq!(d.highlight_at(1).map(|h| h.id.as_str()), Some("a"));
13402 assert_eq!(d.highlight_at(3), None, "end is exclusive");
13403 assert_eq!(d.highlight_at(8).map(|h| h.id.as_str()), Some("b"));
13404 d.set_highlights(Vec::new());
13405 assert!(d.highlights().is_empty(), "a replace is a replace");
13406 }
13407
13408 /// `Highlight::covering` and the cursor over it are what both painters ask
13409 /// per glyph, so they have to answer the same as the scan they replaced —
13410 /// including in the gaps, which is where most glyphs are.
13411 #[test]
13412 fn covering_answers_from_a_sorted_list_without_scanning_all_of_it() {
13413 let hl = |start: usize, end: usize, id: &str| Highlight {
13414 start,
13415 end,
13416 id: id.into(),
13417 color: None,
13418 marker: None,
13419 };
13420 // Disjoint, as search hits are: in a range, in a gap, and past the end.
13421 let hits: Vec<Highlight> = (0..20).map(|i| hl(i * 10, i * 10 + 3, "hit")).collect();
13422 assert_eq!(Highlight::covering(&hits, 0).map(|h| h.start), Some(0));
13423 assert_eq!(Highlight::covering(&hits, 102).map(|h| h.start), Some(100));
13424 assert_eq!(
13425 Highlight::covering(&hits, 105),
13426 None,
13427 "a gap covers nothing"
13428 );
13429 assert_eq!(Highlight::covering(&hits, 103), None, "end is exclusive");
13430 assert_eq!(Highlight::covering(&hits, 9_999), None);
13431 assert_eq!(Highlight::covering(&[], 0), None);
13432
13433 // Nested: first by start, so a hit inside an annotation still resolves
13434 // to the annotation — and the range that stops short doesn't mask it.
13435 let nested = vec![hl(0, 20, "outer"), hl(5, 10, "inner")];
13436 assert_eq!(
13437 Highlight::covering(&nested, 7).map(|h| h.id.as_str()),
13438 Some("outer")
13439 );
13440 assert_eq!(
13441 Highlight::covering(&nested, 15).map(|h| h.id.as_str()),
13442 Some("outer")
13443 );
13444 }
13445
13446 /// The cursor is an optimisation, so the only thing worth asserting is that
13447 /// it is not also a change of answer — at every offset, over a list with a
13448 /// nest in it, walked forwards and then backwards.
13449 #[test]
13450 fn the_highlight_cursor_answers_exactly_what_a_fresh_scan_would() {
13451 let hl = |start: usize, end: usize, id: &str| Highlight {
13452 start,
13453 end,
13454 id: id.into(),
13455 color: None,
13456 marker: None,
13457 };
13458 let mut list = vec![
13459 hl(0, 20, "outer"),
13460 hl(5, 10, "inner"),
13461 hl(30, 33, "hit"),
13462 hl(40, 43, "hit"),
13463 ];
13464 list.sort_by_key(|h| (h.start, h.end));
13465
13466 let mut cursor = HighlightCursor::new(&list);
13467 for offset in 0..50 {
13468 assert_eq!(
13469 cursor.at(offset).map(|h| h.id.as_str()),
13470 Highlight::covering(&list, offset).map(|h| h.id.as_str()),
13471 "cursor disagrees at {offset}"
13472 );
13473 }
13474 // Backwards: the cursor re-seats rather than answering from where it
13475 // had got to, so a painter that revisits a row is still told the truth.
13476 for offset in (0..50).rev() {
13477 assert_eq!(
13478 cursor.at(offset).map(|h| h.id.as_str()),
13479 Highlight::covering(&list, offset).map(|h| h.id.as_str()),
13480 "cursor disagrees walking back at {offset}"
13481 );
13482 }
13483 }
13484}