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::wysiwyg::{self, MediaKind, MediaStop, VisualMap};
43
44/// Which view the body shows.
45#[derive(Clone, Copy, PartialEq, Eq, Debug)]
46pub enum View {
47 /// The raw document with a caret in source bytes.
48 Source,
49 /// Markup resolved to real styles, caret riding the rendered glyphs.
50 Wysiwyg,
51}
52
53/// How much of the source markup the WYSIWYG view exposes — a per-editor
54/// preference, orthogonal to [`View`]. Named for markup rather than for Markdown
55/// because leaf is grammar-agnostic: twig hands it Djot, HTML and XML on the same
56/// terms, and every rung below is about *delimiters*, whatever grammar spells
57/// them. The examples are Markdown only because that is what most documents are.
58///
59/// A single ladder over two underlying axes, because only three of their four
60/// combinations are coherent:
61///
62/// | | authoring off | authoring on |
63/// |---|---|---|
64/// | delimiters hidden | [`None`](Self::None) | [`Shortcuts`](Self::Shortcuts) |
65/// | caret line revealed | *incoherent* | [`Full`](Self::Full) |
66///
67/// The empty quadrant would show delimiters on the caret's line and then escape
68/// the ones you type — a surface that displays a syntax it refuses to accept.
69/// Someone who wants to read raw markup without authoring it has
70/// [`View::Source`], which is the better tool for it.
71///
72/// The two axes are read separately by the code that cares — see
73/// [`reveals_caret_line`](Self::reveals_caret_line) and
74/// [`authors`](Self::authors) — so neither behaviour has to know it's spelled
75/// as a ladder.
76#[derive(Clone, Copy, PartialEq, Eq, Debug, Default)]
77pub enum MarkupMode {
78 /// Delimiters stay hidden even on the caret's line, and typed syntax stays
79 /// literal — twig escapes anything that would open markup, so formatting
80 /// comes from commands (⌘b, the toolbar) instead of from spelling. The clean
81 /// reading surface for people who don't write markup by hand; the default,
82 /// and what Diaryx ships.
83 #[default]
84 None,
85 /// Delimiters stay hidden, but typing them authors real markup: `*x*`
86 /// becomes italic and the asterisks disappear into the styling
87 /// (Typora/Bear-shaped). For someone who knows the syntax but wants the
88 /// clean surface back once it has been applied.
89 Shortcuts,
90 /// The caret's line shows its raw markup while every other line renders
91 /// resolved (Obsidian live-preview-shaped), and typed syntax authors markup
92 /// — for people fluent in the document's grammar who want to see and edit
93 /// the delimiters they type.
94 Full,
95}
96
97impl MarkupMode {
98 /// Whether the rich view shows raw delimiters on the line holding the caret.
99 /// The rendering axis — read by [`Doc::reveal_line`] and threaded into the
100 /// WYSIWYG builder.
101 pub fn reveals_caret_line(self) -> bool {
102 matches!(self, MarkupMode::Full)
103 }
104
105 /// Whether typed markup characters author real formatting. The editing axis
106 /// — read by [`Doc::insert`], which escapes typed syntax when this is false.
107 pub fn authors(self) -> bool {
108 !matches!(self, MarkupMode::None)
109 }
110}
111
112/// How the WYSIWYG view treats a *soft break* — a bare newline inside a
113/// paragraph. An axis of its own, orthogonal to [`MarkupMode`] (which governs
114/// inline-markup delimiters) and to [`View`]: any reveal preference pairs with
115/// either flow. The renderer consults it when it lays a block's inline content
116/// into visual rows.
117#[derive(Clone, Copy, PartialEq, Eq, Debug, Default)]
118pub enum LineFlow {
119 /// A soft break folds into a space and the paragraph reflows to the
120 /// viewport width — flowing prose, where the source's line wrapping is
121 /// insignificant. The default, and what Diaryx ships.
122 #[default]
123 Fold,
124 /// A soft break renders as a line break exactly where it was written, so
125 /// the author's source line structure shows on screen unchanged — the mode
126 /// for people who lay out their prose deliberately (one sentence or clause
127 /// per line, semantic line breaks). The break is still a soft break in the
128 /// source; only its rendering changes.
129 Preserve,
130}
131
132/// What the file behind a document looks like right now, against the bytes leaf
133/// last read from it or wrote to it — the question a frontend asks before it
134/// saves (a `Changed` file plus a `dirty` document is an overwrite about to
135/// happen) or when its window regains focus. See [`Doc::disk_state`].
136#[derive(Clone, Copy, Debug, PartialEq, Eq)]
137pub enum DiskState {
138 /// The file holds exactly the bytes leaf last read or wrote.
139 Unchanged,
140 /// Someone else wrote the file since. Saving overwrites their work; see
141 /// [`Doc::reload`] for the other direction.
142 Changed,
143 /// The file is gone — deleted or renamed away. A save recreates it.
144 Missing,
145 /// There is a path, but the file couldn't be read (permissions, a directory
146 /// in the way): leaf can't tell, and won't guess.
147 Unreadable,
148 /// No file behind this document yet — see [`Doc::blank`]. Nothing can have
149 /// changed under a document that was never on disk.
150 Untitled,
151}
152
153/// The inline marks in force at a point in the document — what a toolbar
154/// lights up. A `Copy` bitset rather than a `HashSet`, because
155/// [`Doc::active_inline_marks`] is called on every frame that draws a toolbar
156/// and a set that allocates to answer "is Bold on?" is a set that shouldn't.
157#[derive(Clone, Copy, Debug, Default, PartialEq, Eq)]
158pub struct InlineMarks(u8);
159
160impl InlineMarks {
161 /// Every kind, in the order [`InlineMarks::iter`] yields them.
162 const ALL: [InlineKind; 8] = [
163 InlineKind::Strong,
164 InlineKind::Emph,
165 InlineKind::Verbatim,
166 InlineKind::Mark,
167 InlineKind::Superscript,
168 InlineKind::Subscript,
169 InlineKind::Insert,
170 InlineKind::Delete,
171 ];
172
173 pub const fn empty() -> Self {
174 InlineMarks(0)
175 }
176
177 /// Private: the set is an *answer*, and adding a mark to it doesn't mark
178 /// anything ([`Doc::toggle`] does that). `FromIterator` is the way in.
179 fn insert(&mut self, kind: InlineKind) {
180 self.0 |= Self::bit(kind);
181 }
182
183 /// Flip `kind` in the set — the sticky-marks toggle at a collapsed caret.
184 fn flip(&mut self, kind: InlineKind) {
185 self.0 ^= Self::bit(kind);
186 }
187
188 /// The symmetric difference: which marks differ between the two sets. Used
189 /// to resolve the marks already in force at the caret against the pending
190 /// delta — a bit set in the delta flips the base mark for the next keystroke.
191 fn xor(self, other: InlineMarks) -> InlineMarks {
192 InlineMarks(self.0 ^ other.0)
193 }
194
195 /// Whether `kind` is in force — the toolbar's "is Bold active?".
196 pub fn contains(self, kind: InlineKind) -> bool {
197 self.0 & Self::bit(kind) != 0
198 }
199
200 pub fn is_empty(self) -> bool {
201 self.0 == 0
202 }
203
204 /// The marks in force, for a frontend that renders whatever is on rather
205 /// than asking after a fixed list.
206 pub fn iter(self) -> impl Iterator<Item = InlineKind> {
207 Self::ALL.into_iter().filter(move |&k| self.contains(k))
208 }
209
210 fn bit(kind: InlineKind) -> u8 {
211 1 << match kind {
212 InlineKind::Strong => 0,
213 InlineKind::Emph => 1,
214 InlineKind::Verbatim => 2,
215 InlineKind::Mark => 3,
216 InlineKind::Superscript => 4,
217 InlineKind::Subscript => 5,
218 InlineKind::Insert => 6,
219 InlineKind::Delete => 7,
220 }
221 }
222}
223
224impl FromIterator<InlineKind> for InlineMarks {
225 fn from_iter<I: IntoIterator<Item = InlineKind>>(iter: I) -> Self {
226 let mut m = InlineMarks::empty();
227 for k in iter {
228 m.insert(k);
229 }
230 m
231 }
232}
233
234/// What kind of edit produced an undo group. Same-kind edits in a row coalesce
235/// into one undo step (a run of typed characters undoes together); `Other` never
236/// coalesces, so a paste, format toggle, or block change is always its own step.
237#[derive(Clone, Copy, PartialEq, Eq)]
238enum EditKind {
239 Insert,
240 Delete,
241 /// One step of an IME composition — see [`Doc::edit_composing`]. Its own kind
242 /// rather than `Insert`'s because a composition is not typing: each step
243 /// *replaces* the last (`か` → `かん` → `感`), so the run has to coalesce even
244 /// though no two steps insert the same bytes, and it must not fold into the
245 /// typed characters on either side of it.
246 Compose,
247 Other,
248}
249
250/// Which side of the caret a delete looks for an in-cell `<br>` break to swallow
251/// whole — see [`Doc::cell_break_at`]. `Backward` is Backspace (a break ending at
252/// the caret), `Forward` is Delete (one starting at it).
253#[derive(Clone, Copy)]
254enum BreakEdge {
255 Backward,
256 Forward,
257}
258
259/// A re-spelling of one inline mark run, held ready in case the edit about to
260/// happen breaks it — see [`Doc::mark_edge_fix`] and [`Doc::repair_mark_edges`].
261/// Every offset in it is in the coordinates the document will have *after* the
262/// plain edit, since that is when it may be applied.
263struct MarkEdgeFix {
264 /// The run's kind, and an offset inside what was its content: together they
265 /// answer "did the plain edit actually break this mark?" — the question that
266 /// decides whether any of this is applied at all.
267 kind: InlineKind,
268 probe: usize,
269 /// The byte range to re-spell (the run's delimiters included) and its new
270 /// spelling, with the edge whitespace moved outside the delimiters.
271 start: usize,
272 end: usize,
273 text: String,
274 /// Where the caret belongs afterwards — the same place on screen it would
275 /// have had, which is now on the other side of a delimiter.
276 caret: usize,
277 /// The marks in force for text typed at that caret. The caret can land
278 /// outside a run it was inside, and the marks have to survive the move or
279 /// the toolbar goes dark mid-word.
280 want: InlineMarks,
281}
282
283/// The caret and selection at one moment — the part of a history step twig's
284/// `Change` cannot carry, because the caret is leaf's state and twig only knows
285/// about bytes. leaf serializes it into the opaque per-state blob twig now
286/// stores in its own undo history (see `record_caret`), so undo and redo hand
287/// back the caret that matches the source they restore.
288#[derive(Clone, Copy)]
289struct CaretState {
290 caret: usize,
291 anchor: Option<usize>,
292}
293
294impl CaretState {
295 /// Pack into the fixed 17-byte blob leaf hands twig: the caret as a u64,
296 /// then an anchor-present flag and the anchor. twig copies these bytes and
297 /// never reads them.
298 fn to_blob(self) -> [u8; 17] {
299 let mut b = [0u8; 17];
300 b[..8].copy_from_slice(&(self.caret as u64).to_le_bytes());
301 if let Some(a) = self.anchor {
302 b[8] = 1;
303 b[9..].copy_from_slice(&(a as u64).to_le_bytes());
304 }
305 b
306 }
307
308 /// Recover a state from twig's blob, or `None` when it is empty or the wrong
309 /// length — a state twig restored that never had a caret set on it, which
310 /// leaves the caller to fall back to the edit site.
311 fn from_blob(b: &[u8]) -> Option<Self> {
312 let b: &[u8; 17] = b.try_into().ok()?;
313 let caret = u64::from_le_bytes(b[..8].try_into().unwrap()) as usize;
314 let anchor = (b[8] != 0).then(|| u64::from_le_bytes(b[9..].try_into().unwrap()) as usize);
315 Some(CaretState { caret, anchor })
316 }
317}
318
319/// A footnote reference and the note it names — the answer to
320/// [`Doc::footnote_at`].
321///
322/// The two `Option`s move together: a reference whose definition is missing has
323/// neither a body to show nor a place to jump to, and one that resolved has
324/// both.
325#[derive(Clone, PartialEq, Eq, Debug)]
326pub struct FootnoteRef {
327 /// The reference's label — the `1` of `[^1]`, with neither the `^` that
328 /// spells it a footnote nor the brackets around it.
329 pub label: String,
330 /// The note's body as source bytes (see
331 /// [`wysiwyg::footnote_body_span`](crate::wysiwyg)), or `None` when the
332 /// document defines no `[^label]:` to read one from.
333 pub text: Option<String>,
334 /// Where the note's *body* starts, for a "go to note" that moves the caret
335 /// there. `None` alongside a `None` `text`.
336 ///
337 /// The body rather than the definition, because this is an offset to put a
338 /// caret on and the `[^1]:` marker is decoration the caret can't occupy —
339 /// aiming at the definition's first byte snaps to the nearest real stop,
340 /// which is up in the paragraph above the note. It is also simply where a
341 /// reader following a reference wants to land: at the note's first word,
342 /// ready to read or amend it.
343 pub offset: Option<usize>,
344 /// Where the note's body ends, exclusive — so a frontend can ask which
345 /// *rendered rows* the note occupies and draw those instead of [`text`](Self::text).
346 ///
347 /// The rows are the note with its markup resolved: `see *later*` reaches a
348 /// frontend as an italic run, not as asterisks. `text` is the source bytes
349 /// and stays the honest answer for anything that wants the note as written
350 /// (a search index, a copy); this pair of offsets is for anything that wants
351 /// it as *read*. `None` alongside a `None` `offset`.
352 pub end: Option<usize>,
353}
354
355/// A footnote definition and the reference that sends a reader to it — the
356/// answer to [`Doc::footnote_definition_at`], and the other half of the round
357/// trip [`FootnoteRef`] starts.
358///
359/// A note is a place a reader *arrives*, so the useful thing to know while
360/// standing in one is the way back. Without this the jump to a note is a
361/// one-way door: the definitions sit at the foot of the document, so returning
362/// by hand means scrolling back up and finding the sentence again.
363#[derive(Clone, PartialEq, Eq, Debug)]
364pub struct FootnoteDef {
365 /// The definition's label — the `1` of `[^1]: …`, marker and colon stripped,
366 /// spelled exactly as [`FootnoteRef::label`] spells the same footnote's.
367 pub label: String,
368 /// Where the reference's *label* is, for a "back to reference" that moves
369 /// the caret there. `None` for a note nothing refers to — an orphan, which
370 /// is worth being able to say rather than silently doing nothing.
371 ///
372 /// The label rather than the reference's first byte, for
373 /// [`FootnoteRef::offset`]'s reason: a reference's brackets are decoration
374 /// and its label is the only part of it the caret can rest on.
375 ///
376 /// The *first* reference, when a label is cited more than once: a repeated
377 /// citation has no one true home, and the first is both the one a reader
378 /// most likely came from and the only choice that doesn't depend on how
379 /// they got here.
380 pub offset: Option<usize>,
381}
382
383/// Where a locator lands — the answer to [`Doc::locate`].
384///
385/// A locator (the `v2` of a `chapter.dj#v2`) names a *place* rather than a
386/// document, and a place is a span rather than a point: a reader following one
387/// wants the caret at its first byte, and a reader merely *peeking* at one wants
388/// the block it covers drawn. Both are served by carrying the whole span, and
389/// only one of the two can be recovered from an offset alone.
390#[derive(Clone, PartialEq, Eq, Debug)]
391pub struct Landing {
392 /// The first byte of the block the locator names — where a caret goes.
393 pub start: usize,
394 /// One past its last byte, so a frontend can map the pair through
395 /// [`Doc::pos_for_offset`] to the rendered rows the block occupies and draw
396 /// those, the way a footnote peek draws a note ([`FootnoteRef::end`]).
397 pub end: usize,
398}
399
400/// A selection cited out of the source: the text itself, up to a requested
401/// number of characters either side, and the byte range it came from. See
402/// [`Doc::selection_quote`].
403///
404/// The prefix and suffix are what make the quote *re-findable*: the same text
405/// can occur twice, and a little of what surrounded it is how a later reader —
406/// or the same document after an edit — tells the occurrences apart. The Web
407/// Annotation model calls this a `TextQuoteSelector`; the shape is older than
408/// the name.
409#[derive(Debug, Clone, PartialEq, Eq)]
410pub struct Quote {
411 /// The selected source, verbatim.
412 pub exact: String,
413 /// What immediately preceded it — possibly empty, at the document's start.
414 pub prefix: String,
415 /// What immediately followed it — possibly empty, at the document's end.
416 pub suffix: String,
417 /// Byte offset in the source where the selection begins.
418 pub start: usize,
419 /// Byte offset where it ends (exclusive).
420 pub end: usize,
421}
422
423/// A host-painted range of the source — an annotation's footprint, a search
424/// hit, a reviewer's mark. Leaf renders it (a background wash behind the
425/// glyphs whose source falls inside it) and hands back the `id` when the
426/// reader activates it; what the range *means* is entirely the host's.
427///
428/// Ranges are source bytes, like the caret and the selection, so a host that
429/// anchors quotes against the source ([`Doc::selection_quote`] is the other
430/// half of that loop) can paint what it found without any coordinate
431/// conversion. A range that drifts off the text it meant is the host's to
432/// re-anchor; leaf draws what it is told.
433#[derive(Debug, Clone, PartialEq, Eq)]
434pub struct Highlight {
435 /// Byte offset in the source where the wash begins.
436 pub start: usize,
437 /// Byte offset where it ends (exclusive).
438 pub end: usize,
439 /// The host's name for it, handed back on activation. Opaque to leaf.
440 pub id: String,
441 /// A rendering hint the frontend maps — a `#RRGGBB` hex string, or
442 /// nothing for the theme's default wash.
443 pub color: Option<String>,
444 /// A margin glyph's name, or nothing for wash-only ink. A highlight with
445 /// a marker gets a small glyph in the margin beside its first line, and
446 /// the glyph — not the wash — is what activates it: the wash is ink, the
447 /// marker is the control, which is what lets a reader put a caret in (or
448 /// copy from) annotated text without a card leaping at them. The name is
449 /// opaque to leaf; an Apple frontend reads it as an SF Symbol, a web one
450 /// as a class.
451 pub marker: Option<String>,
452}
453
454pub struct Doc {
455 editor: Editor,
456 pub format: Format,
457 pub path: PathBuf,
458 /// Current source, refreshed from the editor after every successful edit.
459 pub source: String,
460 /// The caret, as a byte offset into `source` (always on a char boundary).
461 pub caret: usize,
462 /// The selection's fixed end, if a selection is active; the moving end is
463 /// the caret. `None` means no selection.
464 pub anchor: Option<usize>,
465 pub dirty: bool,
466 pub status: Option<String>,
467 pub view: View,
468 /// Whether the document refuses to change — a *reading* surface over the
469 /// same rendering, selection, and navigation the editor has.
470 ///
471 /// Enforced here rather than by each frontend hiding its input paths,
472 /// because every mutation funnels through three doors —
473 /// [`splice_exact`](Self::splice_exact), [`undo`](Self::undo),
474 /// [`redo`](Self::redo) — and three guarded doors are a guarantee where a
475 /// frontend's suppressed keyboard is a hope. A gated splice reports
476 /// exactly like a rolled-back one, a path every caller already handles.
477 read_only: bool,
478 /// The host-painted ranges, kept sorted by start — see [`Highlight`].
479 /// State like the selection rather than like the text: no edit history,
480 /// no dirty bit, redrawn from whatever the host last set.
481 highlights: Vec<Highlight>,
482 /// How much of the source markup the rich view exposes — a frontend preference (see
483 /// [`MarkupMode`]). Its two axes are read apart: the rendering one by
484 /// [`reveal_line`](Self::reveal_line), the editing one by
485 /// [`insert`](Self::insert).
486 markup_mode: MarkupMode,
487 /// Whether soft breaks fold into the reflowed paragraph or render where
488 /// they were written (see [`LineFlow`]) — an independent frontend
489 /// preference the WYSIWYG builder consults when it lays out a block.
490 line_flow: LineFlow,
491 /// The kind of the last edit, for coalescing: twig owns the undo *history*
492 /// (see `undo`/`redo`), but "what counts as one undo step" is a frontend-UX
493 /// call, so leaf decides when a run continues and tells twig to coalesce.
494 last_edit_kind: Option<EditKind>,
495 /// The inline marks the user has toggled *at a collapsed caret* with no
496 /// selection — "start typing bold here". Held as the XOR delta from the marks
497 /// already in force at [`pending_at`](Self::pending_at): a set bit means
498 /// "flip this kind for the next typed text", so it both turns a mark on where
499 /// none is (type into bold) and off where one already covers the caret (type
500 /// past the bold you're standing in). [`Doc::insert`] realises it onto the
501 /// freshly typed text and then clears it — a mark once realised is carried by
502 /// the caret sitting inside the run, not by this delta.
503 pending_marks: InlineMarks,
504 /// The caret offset [`pending_marks`](Self::pending_marks) applies to. The
505 /// delta is live only while the caret still stands here with no selection;
506 /// any motion or edit ([`move_to`](Self::move_to), a splice, a click) drops
507 /// it, so a toggled-but-never-typed format doesn't leak onto text elsewhere.
508 pending_at: Option<usize>,
509 /// The source as of the last open/save — `dirty` is `source != clean_source`,
510 /// so undoing back to the saved state correctly clears the modified flag.
511 clean_source: String,
512 /// A hash of the bytes leaf last read from `path` or wrote to it; `None`
513 /// while the document has no file behind it. [`Doc::disk_state`] compares
514 /// the file against this to catch an edit made *outside* leaf before a save
515 /// silently overwrites it — `clean_source` only knows what leaf itself did.
516 ///
517 /// A hash, not an mtime: mtime is the cheap answer and the wrong one — two
518 /// writes inside one filesystem timestamp tick are indistinguishable, a
519 /// clock that steps backwards (or a writer that restores an mtime) hides a
520 /// real change, and a `touch` invents one. The whole point of the watermark
521 /// is to not clobber someone's work, so it reads the bytes and compares what
522 /// is actually there. That costs a file read per question, which is why the
523 /// question is asked on a user event (focus, save) and not every frame.
524 disk_hash: Option<u64>,
525 /// The "sticky" display column vertical motion aims for, in the active
526 /// view's grid. Set on the first `move_up`/`move_down` of a run and
527 /// reused by every subsequent one in that run, so passing through a
528 /// shorter line doesn't permanently forget the original column. Any
529 /// horizontal motion or edit clears it.
530 ///
531 /// A column, not a character index: dropping down a line of `你好` onto one
532 /// of ASCII has to land under the glyph the caret was drawn beneath, which
533 /// is the only thing the user can see to aim by. Where the goal falls inside
534 /// a wide character on the target line, the mapping resolves it to that
535 /// character — the caret lands on it rather than between its cells.
536 goal_col: Option<usize>,
537 /// The rendered map for the WYSIWYG view; empty in the source view. Movement
538 /// and clicks read it to stay in visible space.
539 pub vmap: VisualMap,
540 /// Everything the map is built from, as one number: bumped whenever the
541 /// document's text changes, and never by a motion, a selection, or a save.
542 /// A frontend can hold work against it — see [`Doc::revision`].
543 revision: u64,
544 /// What `vmap` was built from, or `None` before the first build. The map is
545 /// a pure function of `(revision, wrap, reveal line)`, so when those haven't
546 /// moved, rebuilding it produces the identical map — see
547 /// [`Doc::build_visual`].
548 ///
549 /// The reveal line ([`Doc::reveal_line`]) is the caret's, and is `None` in
550 /// every mode but [`MarkupMode::Full`] — so outside that mode the key is
551 /// text and width alone, and a caret motion still rebuilds nothing.
552 vmap_key: Option<(u64, Option<usize>, Option<Range<usize>>)>,
553 /// Per-block row cache backing the incremental rebuild: when the text
554 /// changes, only the top-level blocks whose bytes moved are re-rendered and
555 /// the rest are reused shifted (see [`wysiwyg::BlockCache`]). Persists across
556 /// builds; a pure accelerator, so it's never read for correctness.
557 block_cache: wysiwyg::BlockCache,
558 /// How many visual rows each block image reserves, keyed by its destination —
559 /// set by the frontend through [`Doc::set_media_rows`] once it has decoded and
560 /// measured the pictures. Core does no image I/O, so this is the only way it
561 /// learns a picture's height; a destination not in the map reserves the bare
562 /// one-row placeholder. Threaded into the builder so [`wysiwyg::build_cached`]
563 /// sizes each placeholder, and folded into `vmap_key` so a height change
564 /// rebuilds the map.
565 media_rows: HashMap<String, usize>,
566
567 // View geometry the renderer stamps each frame, so mouse events can map a
568 // screen cell back to a byte offset.
569 pub scroll: usize,
570 pub body_origin: (u16, u16),
571 pub body_height: u16,
572 /// The caret as of the last frame drawn, or `None` before the first.
573 ///
574 /// Scrolling is the viewport's business, not the caret's: the view follows
575 /// the caret when the caret *moves*, but a wheel that doesn't touch the
576 /// caret has to be free to scroll away from it — otherwise the view is
577 /// pinned to the caret and stops dead at the edge of the document you can
578 /// see. Comparing against this is what tells the two apart, and it catches a
579 /// caret set by any route, including a frontend assigning the field itself.
580 pub drawn_caret: Option<usize>,
581}
582
583/// The Markdown extensions every leaf document is parsed with. `html_elements`
584/// and `directives` depart from twig's defaults. `html_elements` promotes
585/// embedded raw HTML (`<img>`, `<picture>`, `<source>`, …) into semantic AST
586/// nodes, so a picture becomes a real `image` node the frontends can frame and
587/// rasterize instead of opaque `raw_block` text. `directives` turns on generic
588/// `:::name{.class}` fenced-div containers (`directive` nodes), which a host
589/// app uses for its own semantics (diaryx's `:::vis{.audience}` visibility
590/// blocks) — core renders any directive as a plain tinted container, agnostic
591/// of `name`. Both flags are inert for non-Markdown formats, so it's safe to
592/// pass them unconditionally. Threading this through every constructor (not
593/// just `open`) keeps `from_source`, `blank`, and `reload` parsing the same
594/// document the same way — twig reparses with these same flags after each edit.
595fn parse_extensions() -> MarkdownExtensions {
596 MarkdownExtensions {
597 html_elements: true,
598 directives: true,
599 ..Default::default()
600 }
601}
602
603/// Build an editor over `bytes` in `format` with leaf's [`parse_extensions`],
604/// mapping twig's error into the `anyhow` context every constructor shares.
605fn new_editor(bytes: &[u8], format: Format) -> Result<Editor> {
606 Editor::new_ext(bytes, format, parse_extensions()).map_err(|e| anyhow!("twig parse: {e}"))
607}
608
609/// Does `format` spell a table as a **pipe table** — the one grid twig's table
610/// editor knows how to emit?
611///
612/// This is the single capability leaf still has to answer for itself, and the
613/// only hand-maintained format list left in this file. Every other gesture is
614/// [`Format::supports`], which is twig's own answer read across the C ABI — but
615/// twig deliberately leaves the table ops out of that query, because they read
616/// no `Syntax` table at all. They rewrite a grid that is already in the source
617/// and refuse on *position*, never on format. Handed a caret inside an HTML
618/// `<table>`, `table_insert_row` therefore re-emits the whole element as
619/// `| a | b |` and reports success — a real splice, a clean reparse, an honest
620/// `dirty` flag, and nothing downstream able to tell it from a good edit.
621///
622/// So the list is narrow on purpose. `Format` is `#[non_exhaustive]`, and the
623/// wildcard answers "no" for a format leaf has never heard of: a new twig
624/// language that *does* spell pipe tables loses its grid controls until this
625/// line is updated, which shows up as a missing button. The other default hands
626/// it to [`Doc::table_op`], which rewrites documents it cannot spell.
627fn spells_pipe_tables(format: Format) -> bool {
628 matches!(format, Format::Markdown | Format::Djot)
629}
630
631/// Which of leaf's authoring controls this document's format can actually
632/// spell — one flag per toolbar button, resolved once so a frontend can build
633/// its chrome instead of discovering each refusal on a click.
634///
635/// Every field but [`table`](Self::table) is `Format::supports` on the gesture
636/// the matching [`Doc`] method calls, so this record cannot drift from what the
637/// ops do; `table` is [`spells_pipe_tables`], the one answer twig doesn't
638/// export.
639///
640/// **The formats are ragged, and that is the point.** A single per-document
641/// boolean was enough while the two authorable formats were Markdown and djot
642/// and everything else spelled nothing. HTML is neither: it writes seven of the
643/// eight inline marks as a tag pair, plus `<code>`, `<hr>` and an in-cell
644/// `<br>`, and spells no heading marker, no line prefix, no fence, no task box,
645/// no link — because its versions of those have a different *shape*, not a
646/// different alphabet. So ⌘B works in an HTML document and ⌘1 does not, and no
647/// one flag can say that. Markdown and djot differ from each other too:
648/// `==mark==` is djot-only, and an in-cell `<br>` is Markdown-only.
649#[derive(Clone, Copy, Debug, Eq, PartialEq)]
650pub struct Capabilities {
651 /// ⌘B — `InlineKind::Strong`.
652 pub bold: bool,
653 /// ⌘I — `InlineKind::Emph`.
654 pub italic: bool,
655 /// Inline code — `InlineKind::Verbatim`.
656 pub code: bool,
657 /// Highlight — `InlineKind::Mark`. Djot spells it; Markdown does not.
658 pub mark: bool,
659 /// ⌘U — `InlineKind::Insert`, which every format that marks at all spells.
660 pub underline: bool,
661 /// Strikethrough — `InlineKind::Delete`.
662 pub strike: bool,
663 pub superscript: bool,
664 pub subscript: bool,
665 /// Heading levels and "make this a paragraph" — [`Doc::set_block`].
666 pub heading: bool,
667 pub blockquote: bool,
668 pub bullet_list: bool,
669 pub ordered_list: bool,
670 /// The checkbox controls: giving an item a box, and ticking one.
671 pub task: bool,
672 pub link: bool,
673 /// Covers [`Doc::insert_media`] too — see the note there on why the three
674 /// media kinds stand or fall together.
675 pub image: bool,
676 /// The horizontal-rule button. HTML spells this one (`<hr>`).
677 pub thematic_break: bool,
678 /// The footnote button — [`Doc::insert_footnote`]. Markdown and djot spell
679 /// the pair; HTML has no footnote of its own, so the button goes away rather
680 /// than writing brackets that would render as brackets.
681 pub footnote: bool,
682 /// Setting a fenced block's language — a control only ever offered with the
683 /// caret already in a fence.
684 pub code_language: bool,
685 /// The grid controls: insert/delete/move a row or column, set a column's
686 /// alignment. Pair with [`Doc::caret_in_table`], which asks the other
687 /// question — an HTML `<table>` holds the caret and still can't be edited.
688 pub table: bool,
689 /// Shift+Return inside a cell. Markdown and HTML spell it; djot has no
690 /// idiomatic in-cell break.
691 pub cell_line_break: bool,
692}
693
694impl Capabilities {
695 /// Resolve every flag for `format`. Pure and cheap — twig computes each from
696 /// a static table — but a frontend that wants to hold them can.
697 pub fn of(format: Format) -> Self {
698 let inline = |k| format.supports(Gesture::ToggleInline(k));
699 let container = |k| format.supports(Gesture::ToggleBlockContainer(k));
700 Self {
701 bold: inline(InlineKind::Strong),
702 italic: inline(InlineKind::Emph),
703 code: inline(InlineKind::Verbatim),
704 mark: inline(InlineKind::Mark),
705 underline: inline(InlineKind::Insert),
706 strike: inline(InlineKind::Delete),
707 superscript: inline(InlineKind::Superscript),
708 subscript: inline(InlineKind::Subscript),
709 heading: format.supports(Gesture::SetBlock),
710 blockquote: container(BlockContainerKind::BlockQuote),
711 bullet_list: container(BlockContainerKind::BulletList),
712 ordered_list: container(BlockContainerKind::OrderedList),
713 // Both halves of the checkbox story, and leaf offers no control that
714 // needs only one: the item gesture mints the box, the checked one
715 // ticks it, and a format spelling a `task_marker` spells both.
716 task: format.supports(Gesture::ToggleTaskItem)
717 && format.supports(Gesture::ToggleTaskChecked),
718 link: format.supports(Gesture::InsertLink),
719 image: format.supports(Gesture::InsertImage),
720 thematic_break: format.supports(Gesture::InsertThematicBreak),
721 footnote: format.supports(Gesture::InsertFootnote),
722 code_language: format.supports(Gesture::SetCodeLanguage),
723 table: spells_pipe_tables(format),
724 cell_line_break: format.supports(Gesture::InsertLineBreak),
725 }
726 }
727}
728
729impl Doc {
730 #[cfg(feature = "fs")]
731 pub fn open(path: PathBuf) -> Result<Self> {
732 let bytes = std::fs::read(&path).with_context(|| format!("reading {}", path.display()))?;
733 let format = detect_format(&path)?;
734 let editor = new_editor(&bytes, format)?;
735 let source = String::from_utf8(bytes).map_err(|_| anyhow!("document is not UTF-8"))?;
736 let disk_hash = Some(hash_bytes(source.as_bytes()));
737 // Store the document's *absolute* path. A relative one (`leaf README.md`)
738 // has an empty parent, so a frontend can't resolve a relative image
739 // destination (``) against the document's directory and the
740 // picture silently falls back to its text placeholder. `absolute` is
741 // purely lexical — it prefixes the current directory and normalizes, but
742 // reads nothing and resolves no symlinks — so `file_name` and save are
743 // unchanged; it only gives `path.parent()` something to join against.
744 let path = std::path::absolute(&path).unwrap_or(path);
745 Ok(Doc::from_parts(editor, format, path, source, disk_hash))
746 }
747
748 /// Build a document from an in-memory string, the format named explicitly —
749 /// the portable, filesystem-free counterpart to [`Doc::open`] (which reads a
750 /// path and sniffs the format from its extension). A wasm or FFI host, which
751 /// has no path to read, uses this: it hands over bytes it fetched however it
752 /// could, and later persists [`Doc::source`] however it can (a browser
753 /// download, `localStorage`, a backend `PUT`) and calls [`Doc::mark_saved`].
754 ///
755 /// No file backs the result, so it starts untitled ([`Doc::is_untitled`] is
756 /// true) exactly like a [`Doc::blank`] that has been given content.
757 pub fn from_source(source: String, format: Format) -> Result<Self> {
758 let editor = new_editor(source.as_bytes(), format)?;
759 Ok(Doc::from_parts(
760 editor,
761 format,
762 PathBuf::new(),
763 source,
764 None,
765 ))
766 }
767
768 /// An untitled, empty document — the `+` button and a `leaf` launched with
769 /// no file argument. Nothing on disk backs it until a [`Doc::save_as`].
770 ///
771 /// It is Markdown, because a format has to be chosen before a name exists to
772 /// read one from: `detect_format` reads the extension and an untitled
773 /// document has neither. Markdown is what leaf's own files are, what its
774 /// block markers are already written for (`insert_block_prefix`), and the
775 /// extension a Save As will overwhelmingly pick — a wrong guess here would
776 /// mean typing djot into a buffer parsing it as Markdown. Note that Save As
777 /// *doesn't* revisit this: see [`Doc::save_as`].
778 pub fn blank() -> Result<Self> {
779 let format = Format::Markdown;
780 let editor = new_editor(b"", format)?;
781 // An empty `path` is the untitled marker (`path` is a public `PathBuf`
782 // field two frontends already read; making it an `Option` to say this
783 // would break both). `is_untitled` is the question to ask, not the
784 // representation to copy.
785 Ok(Doc::from_parts(
786 editor,
787 format,
788 PathBuf::new(),
789 String::new(),
790 None,
791 ))
792 }
793
794 /// The fields every constructor agrees on, so `open` and `blank` can't drift
795 /// apart in the ones neither of them has an opinion about.
796 fn from_parts(
797 editor: Editor,
798 format: Format,
799 path: PathBuf,
800 source: String,
801 disk_hash: Option<u64>,
802 ) -> Self {
803 Doc {
804 editor,
805 format,
806 path,
807 disk_hash,
808 clean_source: source.clone(),
809 source,
810 caret: 0,
811 anchor: None,
812 dirty: false,
813 status: None,
814 read_only: false,
815 highlights: Vec::new(),
816 // leaf opens in the rich-text (WYSIWYG) view by default — the
817 // markup-resolved surface is leaf's differentiator. Frontends can
818 // still start in source view explicitly (e.g. a CLI flag), and ⌘e/⌥w
819 // toggles at runtime.
820 view: View::Wysiwyg,
821 // `None` by default — the clean surface Diaryx ships, with typed
822 // syntax kept literal; a markup-fluent frontend can climb the
823 // ladder to `Shortcuts` or `Full`.
824 markup_mode: MarkupMode::default(),
825 // Fold by default — flowing prose that reflows to the viewport, the
826 // behaviour every frontend had before this preference existed.
827 line_flow: LineFlow::default(),
828 last_edit_kind: None,
829 pending_marks: InlineMarks::empty(),
830 pending_at: None,
831 goal_col: None,
832 vmap: VisualMap::default(),
833 revision: 0,
834 // No map yet — the first `build_visual` always builds.
835 vmap_key: None,
836 block_cache: wysiwyg::BlockCache::default(),
837 media_rows: HashMap::new(),
838 scroll: 0,
839 body_origin: (0, 0),
840 body_height: 0,
841 drawn_caret: None,
842 }
843 }
844
845 /// Whether this document has no file behind it yet — a [`Doc::blank`] that
846 /// has never been saved. The question a ⌘S handler asks to know it should
847 /// open a Save As picker instead ([`Doc::save`] won't guess a name), and the
848 /// header asks to know the name it shows is a placeholder.
849 pub fn is_untitled(&self) -> bool {
850 self.path.as_os_str().is_empty()
851 }
852
853 pub fn toggle_view(&mut self) {
854 self.view = match self.view {
855 View::Source => View::Wysiwyg,
856 View::Wysiwyg => View::Source,
857 };
858 self.scroll = 0;
859 self.status = None;
860 // Entering WYSIWYG, the caret may be sitting in now-hidden frontmatter;
861 // lift it to the first rendered offset.
862 self.clamp_caret();
863 }
864
865 /// The current markup-exposure preference (see [`MarkupMode`]).
866 pub fn markup_mode(&self) -> MarkupMode {
867 self.markup_mode
868 }
869
870 /// Set the markup-exposure preference. Both of its axes take effect at
871 /// once: the editing one on the next [`insert`](Self::insert), and the
872 /// rendering one on the next build — which is why this drops the cached
873 /// visual map and the per-block render cache, exactly as
874 /// [`set_line_flow`](Self::set_line_flow) does.
875 pub fn set_markup_mode(&mut self, mode: MarkupMode) {
876 if self.markup_mode == mode {
877 return;
878 }
879 self.markup_mode = mode;
880 // Neither cache is keyed on the mode, and moving between `Full` and the
881 // hidden modes changes every row the caret's line renders to — so
882 // invalidate both explicitly.
883 self.vmap_key = None;
884 self.block_cache = wysiwyg::BlockCache::default();
885 }
886
887 /// The source byte range of the line the caret sits on, when that line
888 /// should render its raw delimiters — `None` in every mode and view that
889 /// hides them, which is what the builder reads as "reveal nothing".
890 ///
891 /// A *source* line (newline to newline), not a visual row: a wrapped
892 /// paragraph and a `LineFlow::Preserve` soft break both split one source
893 /// line across several rows, and revealing half a delimiter pair because the
894 /// other half wrapped would be worse than revealing neither. The range
895 /// excludes the terminating newline and is empty-but-present on a blank
896 /// line, which reveals nothing but still keys the caches correctly.
897 ///
898 /// Only in [`View::Wysiwyg`]: source view already shows every byte, so
899 /// there is nothing there to reveal.
900 pub(crate) fn reveal_line(&self) -> Option<Range<usize>> {
901 if !self.markup_mode.reveals_caret_line() || self.view != View::Wysiwyg {
902 return None;
903 }
904 Some(source_line_range(&self.source, self.caret))
905 }
906
907 /// The current soft-break flow preference (see [`LineFlow`]).
908 pub fn line_flow(&self) -> LineFlow {
909 self.line_flow
910 }
911
912 /// Set the soft-break flow preference. The mode changes how every block lays
913 /// out, so a change drops the cached visual map and the per-block render
914 /// cache, forcing the next [`build_visual`] to rebuild under the new flow.
915 ///
916 /// [`build_visual`]: Self::build_visual
917 pub fn set_line_flow(&mut self, mode: LineFlow) {
918 if self.line_flow == mode {
919 return;
920 }
921 self.line_flow = mode;
922 // Both caches are keyed on `(revision, wrap)`, neither of which moved —
923 // so invalidate them explicitly, or the next build would reuse rows laid
924 // out under the old flow.
925 self.vmap_key = None;
926 self.block_cache = wysiwyg::BlockCache::default();
927 }
928
929 pub fn view_name(&self) -> &'static str {
930 match self.view {
931 View::Source => "source",
932 View::Wysiwyg => "wysiwyg",
933 }
934 }
935
936 /// Rebuild the WYSIWYG visual map for the current tree at `width` columns
937 /// (called by the renderer each frame it's in the WYSIWYG view).
938 /// Build the WYSIWYG map, wrapped at `width` display columns.
939 ///
940 /// Cheap to call every frame, which is what both frontends do: the map is a
941 /// pure function of the document and the wrap width, so a call that would
942 /// rebuild the same map returns the one already built. Only an edit (or a
943 /// resize) pays.
944 ///
945 /// That isn't a micro-optimisation. A frontend repaints for reasons that have
946 /// nothing to do with the text — a blinking caret, a scroll, a focus change —
947 /// and rebuilding here is O(document): 23 ms on a 1 MB file, of which 5 ms is
948 /// marshalling twig's AST across the C ABI. Paid twice a second by the GUI's
949 /// blink timer, that was 14% of a core spent redrawing an unchanged document.
950 /// (`cargo run --release -p leaf-core --example bench` for the numbers.)
951 pub fn build_visual(&mut self, width: usize) {
952 self.build_map(Some(width));
953 }
954
955 /// Build the WYSIWYG map with each block as a single unwrapped row — for a
956 /// frontend (the GUI) that wraps at its own proportional pixel width rather
957 /// than a fixed character column.
958 pub fn build_visual_unwrapped(&mut self) {
959 self.build_map(None);
960 }
961
962 /// Tell the model how many visual rows each block image should reserve, keyed
963 /// by the image's destination. A terminal frontend calls this once it has
964 /// decoded and measured its pictures — core does no image I/O, so this is the
965 /// only way it learns a height — and the next [`Doc::build_visual`] lays each
966 /// placeholder out that tall (the label row plus blank filler rows the
967 /// frontend paints the raster over). A destination left out of the map falls
968 /// back to the bare one-row placeholder, which is also what a frontend that
969 /// can't draw pictures (or lays them out in its own units, like the GUI) gets
970 /// by never calling this.
971 ///
972 /// Cheap to call every frame with the same map: only a *change* invalidates
973 /// the built map (and the block-row cache, since a height isn't part of a
974 /// block's bytes and so wouldn't otherwise re-render it). Steady state is a
975 /// no-op, so a frontend can just hand over its current measurements each frame.
976 pub fn set_media_rows(&mut self, rows: HashMap<String, usize>) {
977 if self.media_rows == rows {
978 return;
979 }
980 self.media_rows = rows;
981 // A height lives outside the block's source bytes, so the content-keyed
982 // block cache would hand back the old-height rows on a hit. Drop it (and
983 // the splice layout it carries) so the next build re-renders every block
984 // at the new heights, and force that build by clearing the map key.
985 self.block_cache = wysiwyg::BlockCache::default();
986 self.vmap_key = None;
987 }
988
989 /// The revision the document's text is at — bumped by every edit, undo,
990 /// redo, and reload, and by nothing else. A frontend caches against this to
991 /// tell a repaint that needs new work from one that doesn't.
992 ///
993 /// It counts *edits*, not distinct texts: typing `x` and deleting it again
994 /// lands on the same text two revisions later. Work is only ever rebuilt
995 /// needlessly, never wrongly reused.
996 pub fn revision(&self) -> u64 {
997 self.revision
998 }
999
1000 /// The map, built at most once per `(revision, wrap)`. `clamp_caret` still
1001 /// runs on every call: the caret moves without the document changing, and
1002 /// keeping it on a legal stop is this function's job either way.
1003 fn build_map(&mut self, wrap: Option<usize>) {
1004 // Under `MarkupMode::Full` the map is a function of the caret's *line*
1005 // as well as the text, so the line joins the key: moving within a line
1006 // still reuses the map, and crossing into another one rebuilds it. In
1007 // every other mode `reveal_line` is `None` and the key is what it was,
1008 // so caret motion goes on costing nothing.
1009 let reveal = self.reveal_line();
1010 let key = (self.revision, wrap, reveal.clone());
1011 if self.vmap_key.as_ref() != Some(&key) {
1012 // Enumerate the top-level blocks cheaply — no whole-arena marshal.
1013 // A subtree is pulled only for the block(s) that actually changed, so
1014 // the FFI marshal shrinks from O(document) to O(edited block).
1015 let top = self.top_blocks();
1016
1017 // Fast path: when twig reports a dirty byte range, try to patch the
1018 // previous map in place — a single-block edit moves the prefix,
1019 // shifts the suffix, and re-renders only one block. `build_spliced`
1020 // returns `None` (and we fall back to the always-correct full rebuild)
1021 // whenever the edit reshaped the block structure, hit a table, or
1022 // there's no previous map to patch.
1023 // Preserve soft breaks as written when the flow preference asks for
1024 // it — the builder renders each as its own visual row instead of
1025 // folding it into the reflowed paragraph.
1026 let preserve_soft = self.line_flow == LineFlow::Preserve;
1027 let spliced = match self.editor.dirty_range() {
1028 Some(dirty) => {
1029 let prev = std::mem::take(&mut self.vmap);
1030 let source = &self.source;
1031 let cache = &mut self.block_cache;
1032 let media_rows = &self.media_rows;
1033 let editor = &mut self.editor;
1034 wysiwyg::build_spliced(
1035 prev,
1036 source,
1037 wrap,
1038 preserve_soft,
1039 &top,
1040 dirty,
1041 media_rows,
1042 reveal.clone(),
1043 cache,
1044 |id| editor.subtree(NodeId(id)).unwrap_or_default(),
1045 )
1046 }
1047 None => None,
1048 };
1049 self.vmap = spliced.unwrap_or_else(|| {
1050 let source = &self.source;
1051 let cache = &mut self.block_cache;
1052 let media_rows = &self.media_rows;
1053 let editor = &mut self.editor;
1054 wysiwyg::build_cached(
1055 &top,
1056 source,
1057 wrap,
1058 preserve_soft,
1059 media_rows,
1060 reveal,
1061 cache,
1062 |id| editor.subtree(NodeId(id)).unwrap_or_default(),
1063 )
1064 });
1065 // Acknowledge the dirty range so the next edit's range starts fresh.
1066 self.editor.clear_dirty();
1067 self.vmap_key = Some(key);
1068 }
1069 self.clamp_caret();
1070 }
1071
1072 fn nodes(&mut self) -> Vec<FlatNode> {
1073 self.editor.nodes().unwrap_or_default()
1074 }
1075
1076 /// The document's top-level blocks for the incremental render. See
1077 /// [`wysiwyg::top_blocks`] for why this isn't simply `child_spans(None)`.
1078 fn top_blocks(&mut self) -> Vec<QueryMatch> {
1079 wysiwyg::top_blocks(&mut self.editor)
1080 }
1081
1082 pub fn format_name(&self) -> &'static str {
1083 // `Format` is `#[non_exhaustive]` as of twig 3.0, so the wildcard is
1084 // required. It also covers `Asciidoc`, which twig parses but cannot
1085 // serialize — leaf never opens a document in it (see `Doc::open`).
1086 match self.format {
1087 Format::Djot => "djot",
1088 Format::Markdown => "markdown",
1089 Format::Xml => "xml",
1090 Format::Html => "html",
1091 _ => "unknown",
1092 }
1093 }
1094
1095 /// Whether this document's format offers *any* door in — `false` only for a
1096 /// wholly parse-only format (XML, AsciiDoc), where every gesture refuses and
1097 /// a frontend may as well open the file read-only.
1098 ///
1099 /// This is a much weaker claim than the name suggests, and driving per-button
1100 /// state from it is exactly the mistake to avoid: HTML answers `true` because
1101 /// it spells the inline marks with a tag pair (`<strong>`, `<em>`, `<code>`)
1102 /// while a heading, a quote, a list, a task box, a link and a code fence all
1103 /// remain unspellable there. Ask [`capabilities`](Self::capabilities) — or
1104 /// [`supports`](Self::supports) — per control.
1105 pub fn authorable(&self) -> bool {
1106 self.format.is_authorable()
1107 }
1108
1109 /// Whether this document's format can spell `gesture`, which is twig's own
1110 /// answer rather than a copy of it: `Format::supports` reads the same
1111 /// `Syntax` table the `Editor` method consults before refusing.
1112 ///
1113 /// It is a fact about the *format*, not about the caret. `true` does not
1114 /// promise the gesture succeeds where it is standing — a link over a table
1115 /// border still fails — only that it will not fail with
1116 /// `UnsupportedFormat`. Gray out on `false`; don't read `true` as "this
1117 /// will work here".
1118 pub fn supports(&self, gesture: Gesture) -> bool {
1119 self.format.supports(gesture)
1120 }
1121
1122 /// Every control's enabled state in one read — what a toolbar builds itself
1123 /// from when a document opens or its format changes. See [`Capabilities`].
1124 pub fn capabilities(&self) -> Capabilities {
1125 Capabilities::of(self.format)
1126 }
1127
1128 /// Refuse a gesture this document's format cannot spell, saying so in the
1129 /// status line. `true` means the caller must return without calling twig.
1130 ///
1131 /// Most of these refusals duplicate one twig would make anyway, and they are
1132 /// made here regardless because a message naming the *document's* format
1133 /// reads better than one naming twig's internals. Two of them are not
1134 /// duplicates and are the reason this is a guard rather than an error
1135 /// translation:
1136 ///
1137 /// - The table family (see [`table_op`](Self::table_op)) consults no
1138 /// `Syntax` table, so twig does not refuse it at all.
1139 /// - [`toggle`](Self::toggle) at a collapsed caret never reaches twig — it
1140 /// arms a sticky mark for text not yet typed, which is a promise `insert`
1141 /// could not keep.
1142 fn refuse_unsupported(&mut self, what: &str, gesture: Gesture) -> bool {
1143 self.refuse_unless(what, self.supports(gesture))
1144 }
1145
1146 /// [`refuse_unsupported`](Self::refuse_unsupported) against a capability leaf
1147 /// answers itself — today only [`spells_pipe_tables`].
1148 fn refuse_unless(&mut self, what: &str, supported: bool) -> bool {
1149 if supported {
1150 return false;
1151 }
1152 self.status = Some(format!("{what}: not supported in {}", self.format_name()));
1153 true
1154 }
1155
1156 /// The name to show for this document. An untitled one has no file to name
1157 /// it, and both frontends put this straight on screen — an empty path
1158 /// renders as an empty header, so it says so instead.
1159 pub fn file_name(&self) -> String {
1160 if self.is_untitled() {
1161 return "untitled".into();
1162 }
1163 self.path
1164 .file_name()
1165 .map(|s| s.to_string_lossy().into_owned())
1166 .unwrap_or_else(|| self.path.display().to_string())
1167 }
1168
1169 /// The selection as an ordered `[start, end)` byte range, or `None` when the
1170 /// caret and anchor coincide (an empty selection is no selection).
1171 pub fn selection(&self) -> Option<(usize, usize)> {
1172 self.anchor
1173 .map(|a| (a.min(self.caret), a.max(self.caret)))
1174 .filter(|(s, e)| s != e)
1175 }
1176
1177 /// The selected text, or `None` when there's no selection — the source
1178 /// slice a copy/cut hands to the system clipboard.
1179 pub fn selected_text(&self) -> Option<&str> {
1180 self.selection().map(|(s, e)| &self.source[s..e])
1181 }
1182
1183 /// The selection as a quote with a little of what surrounds it — the shape
1184 /// a host that cites, annotates, or searches for a passage wants, cut from
1185 /// the **source** rather than from anything rendered, so the quote is
1186 /// findable in the document again by plain string search.
1187 ///
1188 /// `context` is a count of characters (not bytes) on each side, clipped at
1189 /// the document's edges; the slices land on char boundaries by
1190 /// construction. `None` when nothing is selected.
1191 pub fn selection_quote(&self, context: usize) -> Option<Quote> {
1192 let (start, end) = self.selection()?;
1193 let mut before = start;
1194 for _ in 0..context {
1195 match self.source[..before].chars().next_back() {
1196 Some(c) => before -= c.len_utf8(),
1197 None => break,
1198 }
1199 }
1200 let mut after = end;
1201 for _ in 0..context {
1202 match self.source[after..].chars().next() {
1203 Some(c) => after += c.len_utf8(),
1204 None => break,
1205 }
1206 }
1207 Some(Quote {
1208 exact: self.source[start..end].to_string(),
1209 prefix: self.source[before..start].to_string(),
1210 suffix: self.source[end..after].to_string(),
1211 start,
1212 end,
1213 })
1214 }
1215
1216 /// Whether the document refuses to change — see the field.
1217 pub fn read_only(&self) -> bool {
1218 self.read_only
1219 }
1220
1221 /// Turn the read-only gate on or off. A frontend preference like
1222 /// [`set_markup_mode`](Self::set_markup_mode): nothing about the document
1223 /// itself changes, only what may be done to it from here on.
1224 pub fn set_read_only(&mut self, on: bool) {
1225 self.read_only = on;
1226 }
1227
1228 /// The host-painted ranges, sorted by start — see [`Highlight`].
1229 pub fn highlights(&self) -> &[Highlight] {
1230 &self.highlights
1231 }
1232
1233 /// Replace the host-painted ranges wholesale. The whole set each time,
1234 /// rather than add/remove verbs: the host owns the list (it derives it
1235 /// from its own state — annotations, search hits), and a replace can
1236 /// never leave the two disagreeing about what should be on screen.
1237 pub fn set_highlights(&mut self, mut highlights: Vec<Highlight>) {
1238 highlights.retain(|h| h.start < h.end);
1239 highlights.sort_by_key(|h| (h.start, h.end));
1240 self.highlights = highlights;
1241 }
1242
1243 /// The highlight covering source `offset`, if one does — first by start
1244 /// when several overlap, which makes overlapping washes resolvable rather
1245 /// than undefined. What a frontend asks when the reader activates a spot.
1246 pub fn highlight_at(&self, offset: usize) -> Option<&Highlight> {
1247 self.highlights
1248 .iter()
1249 .find(|h| h.start <= offset && offset < h.end)
1250 }
1251
1252 /// The AST breadcrumb at the caret (root → deepest), e.g.
1253 /// `doc › para › strong`. Read live from twig via `ancestors_at`.
1254 pub fn breadcrumb(&mut self) -> String {
1255 match self.editor.ancestors_at(self.caret) {
1256 Ok(chain) => chain
1257 .iter()
1258 .map(|m| m.kind.as_str())
1259 .collect::<Vec<_>>()
1260 .join(" › "),
1261 Err(_) => String::new(),
1262 }
1263 }
1264
1265 // ── editing ──────────────────────────────────────────────────────────────
1266
1267 /// Replace the byte range `[start, end)` with `text`, re-anchoring the caret
1268 /// after it. The public form of the internal splice — a pixel frontend that
1269 /// hit-tests to a byte offset (or an IME that hands back an explicit range)
1270 /// edits through this, the same twig `edit_range` the caret ops use.
1271 pub fn edit(&mut self, start: usize, end: usize, text: &str) {
1272 self.splice(start, end, text, EditKind::Other);
1273 }
1274
1275 /// Insert typed `text` at the caret, replacing the selection if there is one.
1276 /// A single typed character coalesces with the run of typing before it; a
1277 /// newline or a multi-character insert is its own undo step.
1278 ///
1279 /// Typed input only — clipboard text goes through [`paste`](Self::paste).
1280 pub fn insert(&mut self, text: &str) {
1281 // Typing against a block picture would dissolve it — see
1282 // `open_paragraph_at_block_media`. Give the text a paragraph first, so
1283 // what the caret was standing beside stays a picture.
1284 self.open_paragraph_at_block_media(text);
1285 // Armed sticky marks (⌘b with no selection) turn the next typed text
1286 // bold/italic/… and then retire — see `insert_with_marks`. Whitespace is
1287 // the exception: it takes no mark of its own and keeps the delta armed
1288 // for the character behind it — see `insert_space_with_marks`.
1289 let pending = self.pending_here();
1290 if !pending.is_empty() && self.selection().is_none() && !text.is_empty() {
1291 if text.trim().is_empty() {
1292 self.insert_space_with_marks(self.caret, text, pending);
1293 } else {
1294 self.insert_with_marks(self.caret, text, pending);
1295 }
1296 return;
1297 }
1298 // `MarkupMode::None`: typed syntax stays literal — twig escapes
1299 // anything that would open markup, so a Diaryx user never mints
1300 // formatting by keyboard (it comes from commands instead). The other two
1301 // rungs of the ladder author markup from what you type, which is the
1302 // whole difference between them and this one. Only in the rendered view
1303 // (source view is for typing raw markup) and only where the format has a
1304 // literal spelling at all: escaping is a backslash before a byte from the
1305 // format's own alphabet, and a format with no such alphabet (HTML escapes
1306 // with entities, XML spells nothing) would have `\&` written into it,
1307 // which is two literal characters and not an escape. Marks (⌘b) still
1308 // format — that path returned above; and leaf's own structural inserts go
1309 // through `insert_raw`, never here, so a list marker or quote gutter is
1310 // written as the markup it is.
1311 if !self.markup_mode.authors()
1312 && self.view == View::Wysiwyg
1313 && !text.is_empty()
1314 && self.supports(Gesture::InsertLiteral)
1315 {
1316 self.insert_literal_typed(text);
1317 return;
1318 }
1319 self.insert_raw(text);
1320 }
1321
1322 /// Insert `text` verbatim at the caret (replacing any selection) — the plain
1323 /// path with no Hidden-mode literal escaping. leaf's own structural inserts
1324 /// (a list marker, a quote gutter, an in-cell `<br>`) call this: they ARE
1325 /// markup by design and must not be escaped.
1326 fn insert_raw(&mut self, text: &str) {
1327 let (s, e) = self.selection().unwrap_or((self.caret, self.caret));
1328 self.splice(s, e, text, typed_edit_kind(text));
1329 }
1330
1331 /// Open a paragraph for text about to be inserted at one of a block media's
1332 /// two caret stops, and leave the caret standing in it.
1333 ///
1334 /// A block image is a paragraph whose entire content is the picture, and the
1335 /// caret's only homes on it are in front of it and just past it (see
1336 /// [`VisualMap::block_media_stop`]). Text inserted at either offset joins
1337 /// *that* paragraph — and a paragraph holding anything besides the image is
1338 /// no longer a block image but a line of text with an inline one in it. The
1339 /// frontend that was painting a photo there paints a text run instead; the
1340 /// picture is still in the file, and nothing said a word. Those two offsets
1341 /// are also exactly where a click on the picture lands, so the whole accident
1342 /// is one tap and one keystroke.
1343 ///
1344 /// So the break goes in first and the text lands in the new empty paragraph —
1345 /// what pressing Return before typing would have done, which is a habit no
1346 /// one should have to learn from losing a photo. A no-op everywhere else, and
1347 /// over a selection (which is replaced, not joined into).
1348 ///
1349 /// A picture inside a quote or a list leaves its container, because `\n\n`
1350 /// ends the block. The alternative is worse: the `\n> ` / next-item
1351 /// continuation [`newline`](Self::newline) writes stays in the same
1352 /// *paragraph*, which is the thing being prevented.
1353 ///
1354 /// Only in the rendered view. Source view is for typing raw markup, where
1355 /// putting a character against an image is exactly what it looks like.
1356 fn open_paragraph_at_block_media(&mut self, text: &str) {
1357 if self.view != View::Wysiwyg || text.is_empty() || text == "\n" {
1358 return;
1359 }
1360 if self.selection().is_some() {
1361 return;
1362 }
1363 // The map may be a revision behind (nothing has drawn since the last
1364 // edit), and this asks it about offsets — a stale answer would splice a
1365 // break into the wrong place. Free when it is already current, which it
1366 // is whenever a frontend drew a frame between keystrokes.
1367 self.rebuild_map();
1368 let at = self.caret;
1369 let Some((side, _)) = self.vmap.block_media_stop(at) else {
1370 return;
1371 };
1372 if !self.splice(at, at, "\n\n", EditKind::Other) {
1373 return;
1374 }
1375 // The break is part of the keystroke, not an edit of its own: leave the
1376 // run marked as typing so the character about to arrive folds into it and
1377 // one undo puts the document back the way it was found. (A paste, or a
1378 // multi-character insert, is `EditKind::Other` and stays its own step —
1379 // as it would have been anywhere else in the document.)
1380 self.last_edit_kind = Some(EditKind::Insert);
1381 if side == MediaStop::Before {
1382 // The break went in above the picture and the caret rode to the end
1383 // of it — which is still hard against the picture. Step back onto the
1384 // blank line it opened, so the text lands above rather than in front.
1385 self.caret = at;
1386 }
1387 }
1388
1389 /// A delete key pressed at one of a block picture's two caret stops, handled
1390 /// as the picture being an *atom* rather than a run of bytes. Returns whether
1391 /// the key was consumed.
1392 ///
1393 /// The caret rests in front of a block image and just past it, never inside
1394 /// its markup — which the rendered view doesn't show. So the byte a delete
1395 /// key nominally takes there is one the writer cannot see, and taking it
1396 /// leaves the picture as broken markup rather than as anything anyone asked
1397 /// for: Backspace at the stop past `` removes the closing paren, and
1398 /// a photo becomes the literal text `
1401 /// prevents from the typing side, and it cost this repository's own test vault
1402 /// a photo before it was found.
1403 ///
1404 /// So the key aimed *at* the picture deletes the picture, whole — Backspace
1405 /// when it is behind the caret, Delete when it is in front — which is what
1406 /// every editor does with an embed, and one undo away. The key aimed *away*
1407 /// from it would otherwise delete the paragraph break and merge a neighbour
1408 /// into the picture's own paragraph, which dissolves it just as surely; it
1409 /// steps the caret over the boundary instead and leaves the
1410 /// next press to delete in the block it has reached — the same "first press
1411 /// steps out of the atom, second press deletes" every delete key here gets,
1412 /// word-deletes included (⌥⌫ in front of a picture is aimed at the prose
1413 /// above, and reaches it on the second press rather than taking the break and
1414 /// the picture with it on the first).
1415 fn delete_around_block_media(&mut self, forward: bool) -> bool {
1416 // The map answers about offsets, so it has to be this revision's — see
1417 // the same call in `open_paragraph_at_block_media`.
1418 self.rebuild_map();
1419 let Some((side, span)) = self.vmap.block_media_stop(self.caret) else {
1420 return false;
1421 };
1422 let aimed_at_it = side
1423 == if forward {
1424 MediaStop::Before
1425 } else {
1426 MediaStop::After
1427 };
1428 if !aimed_at_it {
1429 let over = if forward {
1430 self.vmap.stop_after(self.caret)
1431 } else {
1432 self.vmap.stop_before(self.caret)
1433 };
1434 if let Some(off) = over.filter(|&o| o >= self.caret_floor()) {
1435 self.caret = off;
1436 self.anchor = None;
1437 self.goal_col = None;
1438 }
1439 return true;
1440 }
1441 // Take the break that held the picture apart from its neighbour with it,
1442 // so the delete doesn't leave a blank paragraph standing where the
1443 // picture was. The last arm is a picture that is the whole document.
1444 let (from, to) = if self.source[..span.start].ends_with("\n\n") {
1445 (span.start - 2, span.end)
1446 } else if self.source[span.end..].starts_with("\n\n") {
1447 (span.start, span.end + 2)
1448 } else {
1449 (span.start, span.end)
1450 };
1451 self.splice(from.max(self.caret_floor()), to, "", EditKind::Other);
1452 true
1453 }
1454
1455 /// The Hidden-mode typing path: replace any selection, then insert `text`
1456 /// escaped so it stays literal. When it replaces a selection the two edits
1457 /// fold into one undo step, so an overwrite undoes atomically (and restores
1458 /// the selection) exactly as a plain one does.
1459 fn insert_literal_typed(&mut self, text: &str) {
1460 let kind = typed_edit_kind(text);
1461 match self.selection() {
1462 Some((s, e)) => {
1463 if !self.splice(s, e, "", EditKind::Other) {
1464 return;
1465 }
1466 // Typing over a whole marked run takes its delimiters with it
1467 // (the empty content couldn't hold them — see
1468 // `repair_mark_edges`) and leaves its marks armed at the caret.
1469 // The text taking the run's place inherits them, exactly as it
1470 // would have by landing inside a run that survived.
1471 let pending = self.pending_here();
1472 if !pending.is_empty() && !text.trim().is_empty() {
1473 self.insert_with_marks(self.caret, text, pending);
1474 return;
1475 }
1476 self.insert_literal_at(self.caret, text, kind, true);
1477 }
1478 None => {
1479 self.insert_literal_at(self.caret, text, kind, false);
1480 }
1481 }
1482 }
1483
1484 /// The sticky-mark delta that is live right now: the marks armed by [`toggle`]
1485 /// at a collapsed caret, but only while the caret still stands where they
1486 /// were armed and nothing is selected. Empty otherwise, so a stale delta
1487 /// never styles text it wasn't meant for.
1488 fn pending_here(&self) -> InlineMarks {
1489 if self.anchor.is_none() && self.pending_at == Some(self.caret) {
1490 self.pending_marks
1491 } else {
1492 InlineMarks::empty()
1493 }
1494 }
1495
1496 /// Drop the armed sticky marks — any caret motion, selection, or edit does
1497 /// this, so "start bold here" only ever applies at the exact spot it was
1498 /// asked for.
1499 fn clear_pending(&mut self) {
1500 self.pending_marks = InlineMarks::empty();
1501 self.pending_at = None;
1502 }
1503
1504 /// Insert `text` at `at` carrying the armed sticky `marks`: a mark not yet in
1505 /// force is wrapped around the freshly typed text; a mark the caret already
1506 /// stands inside is *shed* — the text is inserted past the run's end so it
1507 /// lands unmarked ("type normally again"). The caret comes to rest inside any
1508 /// added runs, so continued typing inherits the marks with no re-wrapping,
1509 /// and the delta is cleared: the marks now live in the document, not here.
1510 fn insert_with_marks(&mut self, at: usize, text: &str, marks: InlineMarks) {
1511 let base = self.mark_spans_at(at);
1512 let base_set: InlineMarks = base.iter().map(|(k, _)| *k).collect();
1513 // Nothing to shed, and a run of exactly these marks standing just behind
1514 // the caret: carry on writing *that* run rather than opening a second
1515 // one beside it.
1516 if base_set.is_empty() && self.rejoin_run(at, text, marks) {
1517 return;
1518 }
1519 // Shed the marks we're turning off: step the insertion point past the
1520 // end of each run the caret sits in, so the new text falls outside it.
1521 let mut ins_at = at;
1522 for (kind, span) in &base {
1523 if marks.contains(*kind) {
1524 ins_at = ins_at.max(span.end);
1525 }
1526 }
1527 if !self.splice_exact(ins_at, ins_at, text, EditKind::Other) {
1528 return;
1529 }
1530 // The plain splice inserted exactly `text` at `ins_at`; that byte range
1531 // is the content every added mark wraps.
1532 let (mut cs, mut ce) = (ins_at, ins_at + text.len());
1533 for kind in marks.iter() {
1534 if !base_set.contains(kind) {
1535 let (ncs, nce) = self.wrap_span(cs, ce, kind);
1536 cs = ncs;
1537 ce = nce;
1538 }
1539 }
1540 self.caret = ce.min(self.source.len());
1541 self.anchor = None;
1542 self.last_edit_kind = None;
1543 // Realised: the marks are in the document now, and the caret sits inside
1544 // them, so there is no delta left to carry. Arm nothing, but remember the
1545 // spot so a *further* toggle before typing starts a clean delta here.
1546 self.pending_marks = InlineMarks::empty();
1547 self.pending_at = Some(self.caret);
1548 self.clamp_caret();
1549 self.record_caret();
1550 }
1551
1552 /// Carry on the marked run just behind `at` — moving its closing delimiters
1553 /// out past the new text — instead of opening a second run of the same marks
1554 /// beside it. Returns whether it did.
1555 ///
1556 /// This is the far half of the mark-edge rule (see [`splice`](Self::splice)).
1557 /// A space typed after a bold word steps the caret out of the run, because
1558 /// `**bold **` is not bold; the next character has to step back *in*, or the
1559 /// writer who typed one bold phrase is left with `**bold** **and**` — two
1560 /// runs that read the same to a reader but spell the file in a way nobody
1561 /// wrote. Only whitespace may stand in the gap (a run doesn't reach across
1562 /// words it isn't marking), and the marks behind it must be exactly the ones
1563 /// armed — a run of *some* other kind is a neighbour, not this phrase.
1564 fn rejoin_run(&mut self, at: usize, text: &str, marks: InlineMarks) -> bool {
1565 if text.is_empty() || text.trim() != text {
1566 return false;
1567 }
1568 let gap_at = self.source[..at].trim_end_matches([' ', '\t']).len();
1569 // Walk in through the delimiters stacked at that point, innermost last:
1570 // `***both*** ` closes two runs with one `***`, and rejoining means
1571 // getting behind all of them.
1572 let (mut cut, mut kinds) = (gap_at, InlineMarks::empty());
1573 while let Some((kind, content_end)) = self
1574 .editor
1575 .ancestors_at(prev_boundary(&self.source, cut))
1576 .unwrap_or_default()
1577 .into_iter()
1578 .filter(|m| m.span.end == cut)
1579 .find_map(|m| Some((inline_kind(&m.kind)?, m.content_span.clone()?.end)))
1580 {
1581 if content_end >= cut {
1582 break; // a mark with no closing delimiter to step behind
1583 }
1584 kinds.insert(kind);
1585 cut = content_end;
1586 }
1587 if cut == gap_at || kinds != marks {
1588 return false;
1589 }
1590 // Re-spell the tail: the gap, then the new text, then the delimiters that
1591 // used to close in front of them — read out of the document rather than
1592 // written from a table, so whatever twig spells them with is what moves.
1593 let tail = format!(
1594 "{}{text}{}",
1595 &self.source[gap_at..at],
1596 &self.source[cut..gap_at]
1597 );
1598 if !self.splice_exact(cut, at, &tail, EditKind::Other) {
1599 return false;
1600 }
1601 self.caret = (cut + (at - gap_at) + text.len()).min(self.source.len());
1602 self.anchor = None;
1603 self.last_edit_kind = None;
1604 self.pending_marks = InlineMarks::empty();
1605 self.pending_at = Some(self.caret);
1606 self.clamp_caret();
1607 self.record_caret();
1608 true
1609 }
1610
1611 /// Insert typed whitespace at a caret with sticky marks armed. Whitespace is
1612 /// never itself wrapped: a mark around a space draws nothing a reader can
1613 /// see, and in Markdown and Djot it draws its own delimiters instead
1614 /// (`** **`). So the space goes in unmarked — outside any run the armed
1615 /// marks are shedding — and the marks stay armed for the character after it,
1616 /// which rejoins the run (see [`rejoin_run`](Self::rejoin_run)).
1617 fn insert_space_with_marks(&mut self, at: usize, text: &str, marks: InlineMarks) {
1618 let base = self.mark_spans_at(at);
1619 // What the *next* character carries: the armed delta resolved against the
1620 // marks in force here, which the space must not quietly drop.
1621 let want = base
1622 .iter()
1623 .map(|(k, _)| *k)
1624 .collect::<InlineMarks>()
1625 .xor(marks);
1626 let mut ins_at = at;
1627 for (kind, span) in &base {
1628 if marks.contains(*kind) {
1629 ins_at = ins_at.max(span.end);
1630 }
1631 }
1632 if !self.splice(ins_at, ins_at, text, typed_edit_kind(text)) {
1633 return;
1634 }
1635 self.rearm(want);
1636 self.record_caret();
1637 }
1638
1639 /// Wrap `[s, e)` in `kind` via twig and return the byte span the *content*
1640 /// (not the delimiters) occupies afterwards. Markdown/Djot inline delimiters
1641 /// are symmetric (`**`…`**`, `_`…`_`, `` ` ``…`` ` ``), so the bytes twig
1642 /// added split evenly around the content — half the growth on each side.
1643 fn wrap_span(&mut self, s: usize, e: usize, kind: InlineKind) -> (usize, usize) {
1644 match self.editor.toggle_inline(s, e, kind) {
1645 Ok(change) => {
1646 self.last_edit_kind = None;
1647 self.refresh();
1648 self.dirty = self.source != self.clean_source;
1649 let added = (change.new.end - change.new.start).saturating_sub(e - s);
1650 let half = added / 2;
1651 (change.new.start + half, change.new.end - half)
1652 }
1653 // Unsupported here (e.g. mark on Markdown): leave the text unwrapped
1654 // rather than lose the keystroke.
1655 Err(e2) => {
1656 self.status = Some(format!("{kind:?}: {e2}"));
1657 (s, e)
1658 }
1659 }
1660 }
1661
1662 /// The safe offset to splice a block-level break at, given a caret that may
1663 /// sit exactly between an inline mark's content and its own closing
1664 /// delimiter (`content_span.end == off < span.end` for some enclosing mark
1665 /// — the WYSIWYG caret's natural resting place at the end of `**bold**`
1666 /// with nothing following it on the line: the closing `**` renders no
1667 /// glyph of its own, so the caret's "end of line" offset lands right
1668 /// before it). Splicing a paragraph/list/quote break at `off` itself would
1669 /// sever the delimiter from its content, stranding it alone on the new
1670 /// line. Walks out to the *outermost* such mark's `span.end` instead, so
1671 /// nested marks closing at the same point (`**_x_**`) all clear together.
1672 /// A no-op everywhere else — mid-run, or past real trailing content, no
1673 /// mark's `content_span` ends exactly at `off`.
1674 fn skip_trailing_close_delims(&mut self, off: usize) -> usize {
1675 let off = off.min(self.source.len());
1676 self.editor
1677 .ancestors_at(off)
1678 .unwrap_or_default()
1679 .into_iter()
1680 .filter(|m| inline_kind(&m.kind).is_some())
1681 .filter(|m| off < m.span.end && m.content_span.as_ref().is_some_and(|c| c.end == off))
1682 .map(|m| m.span.end)
1683 .max()
1684 .unwrap_or(off)
1685 }
1686
1687 /// The offset a *delete* aimed at the character before `off` should stop at,
1688 /// when `off` is the start of a run's text and the bytes behind it are that
1689 /// run's opening delimiter. The rich view draws no glyph for a `**`, so the
1690 /// byte behind the caret at the start of a bold word is not a character the
1691 /// writer can see, let alone one they aimed Backspace at: taking it leaves
1692 /// `a *bold** c` — the styling gone and a literal asterisk in its place. The
1693 /// delete steps over the whole delimiter to the visible character in front of
1694 /// it instead. Walks out to the *outermost* mark opening there, so
1695 /// `**_x_**` clears every delimiter at once, and is a no-op anywhere else.
1696 fn skip_leading_open_delims(&mut self, off: usize) -> usize {
1697 let off = off.min(self.source.len());
1698 self.editor
1699 .ancestors_at(off)
1700 .unwrap_or_default()
1701 .into_iter()
1702 .filter(|m| inline_kind(&m.kind).is_some())
1703 .filter(|m| {
1704 m.span.start < off && m.content_span.as_ref().is_some_and(|c| c.start == off)
1705 })
1706 .map(|m| m.span.start)
1707 .min()
1708 .unwrap_or(off)
1709 }
1710
1711 /// `off` moved *inside* the run whose closing delimiters end there — the
1712 /// other offset the rich view draws in the same place, since a `**` renders
1713 /// no glyph of its own. `**bold**` has a caret home on each side of its
1714 /// closing delimiter, one column apart on screen and eight bytes and a whole
1715 /// run apart in the file, and a plain ← lands on the outer one whenever a
1716 /// space follows the phrase. The inner one is what the writer is pointing at
1717 /// there: the end of their bold word. Walks in through every mark closing at
1718 /// that point, innermost last, so `***both***` lands inside both. A no-op
1719 /// anywhere else — mid-run, or in prose, no mark's span ends at `off`.
1720 fn step_inside_close_delims(&mut self, off: usize) -> usize {
1721 let mut off = off.min(self.source.len());
1722 loop {
1723 let inner = self
1724 .editor
1725 .ancestors_at(prev_boundary(&self.source, off))
1726 .unwrap_or_default()
1727 .into_iter()
1728 .filter(|m| inline_kind(&m.kind).is_some() && m.span.end == off)
1729 .filter_map(|m| m.content_span.clone().map(|c| c.end))
1730 .filter(|&end| end < off)
1731 .max();
1732 match inner {
1733 Some(end) => off = end,
1734 None => return off,
1735 }
1736 }
1737 }
1738
1739 /// The mirror at the opening edge: `off` moved inside the run whose
1740 /// delimiters *start* there, onto the first character of its text. See
1741 /// [`step_inside_close_delims`](Self::step_inside_close_delims).
1742 fn step_inside_open_delims(&mut self, off: usize) -> usize {
1743 let mut off = off.min(self.source.len());
1744 loop {
1745 let inner = self
1746 .editor
1747 .ancestors_at(off)
1748 .unwrap_or_default()
1749 .into_iter()
1750 .filter(|m| inline_kind(&m.kind).is_some() && m.span.start == off)
1751 .filter_map(|m| m.content_span.clone().map(|c| c.start))
1752 .filter(|&start| start > off)
1753 .min();
1754 match inner {
1755 Some(start) => off = start,
1756 None => return off,
1757 }
1758 }
1759 }
1760
1761 /// The inline mark kinds whose span covers `off`, each with that span — the
1762 /// span-carrying sibling of [`marks_at`](Self::marks_at), which reports node
1763 /// ids instead. Used to shed a mark by stepping past the end of its run.
1764 fn mark_spans_at(&mut self, off: usize) -> Vec<(InlineKind, std::ops::Range<usize>)> {
1765 let off = off.min(self.source.len());
1766 self.editor
1767 .ancestors_at(off)
1768 .unwrap_or_default()
1769 .into_iter()
1770 .filter(|m| off < m.span.end)
1771 .filter_map(|m| inline_kind(&m.kind).map(|k| (k, m.span.clone())))
1772 .collect()
1773 }
1774
1775 /// Insert clipboard `text` at the caret, replacing the selection if there is
1776 /// one — always its own undo step, whatever its length.
1777 ///
1778 /// Provenance is the whole point, and only the caller has it. `insert` reads
1779 /// a lone character as a keystroke and folds it into the run around it,
1780 /// which is right for typing and wrong for a one-character paste: that paste
1781 /// would vanish mid-run on an undo it was never part of, and the characters
1782 /// the user actually typed would go with it. Length can't tell the two
1783 /// apart — `⌘V` of `x` and typing `x` are the same string — so the door the
1784 /// caller comes through is what says which happened.
1785 pub fn paste(&mut self, text: &str) {
1786 // Pasting against a block picture dissolves it exactly as typing does,
1787 // and for the same reason — see `open_paragraph_at_block_media`.
1788 self.open_paragraph_at_block_media(text);
1789 let (s, e) = self.selection().unwrap_or((self.caret, self.caret));
1790 self.splice(s, e, text, EditKind::Other);
1791 }
1792
1793 /// Replace `[start, end)` with `text` as one step of an IME composition —
1794 /// the same splice as [`edit`](Self::edit), but marked so the run of steps
1795 /// folds into a single undo.
1796 ///
1797 /// A composition is *one* act of writing. Typing `かんじ` and picking 感じ is a
1798 /// dozen calls here, each replacing the last one's provisional bytes, and an
1799 /// undo step per call means undoing a word means pressing ⌘Z until the reading
1800 /// unspools backwards through kana — the intermediate states were never text
1801 /// the user wrote. Only the frontend knows a call is provisional (the bytes
1802 /// look like any other edit), so the door the caller comes through is what
1803 /// says so, exactly as it is for [`paste`](Self::paste) versus
1804 /// [`insert`](Self::insert).
1805 ///
1806 /// Pair with [`end_composition`](Self::end_composition), or the *next*
1807 /// composition folds into this one.
1808 pub fn edit_composing(&mut self, start: usize, end: usize, text: &str) {
1809 self.splice(start, end, text, EditKind::Compose);
1810 }
1811
1812 /// Close the open composition run, so the next one is its own undo step.
1813 /// Call when the IME commits or withdraws a composition.
1814 ///
1815 /// Only clears a *composition* run: a frontend that reports an end it never
1816 /// began (some IMEs unmark unprompted) would otherwise split the run of
1817 /// typing around it into two undo steps for no reason the user can see.
1818 pub fn end_composition(&mut self) {
1819 if self.last_edit_kind == Some(EditKind::Compose) {
1820 self.last_edit_kind = None;
1821 }
1822 }
1823
1824 // ── the clipboard's rich flavor ──────────────────────────────────────────
1825
1826 /// The selection rendered as HTML, for the clipboard's `text/html` flavor —
1827 /// what lets a paste into Docs/Mail/Slack keep its formatting. `None` when
1828 /// nothing is selected, or when the selection doesn't render (the caller
1829 /// still has [`selected_text`](Self::selected_text), which is what to publish
1830 /// as `text/plain` either way).
1831 ///
1832 /// **The fragment is a source substring, and that is the honest limit here.**
1833 /// It's parsed standalone, so a selection whose meaning depends on its
1834 /// surroundings converts as what it literally says rather than what it looks
1835 /// like on screen: half a list item is a paragraph, a row torn out of a table
1836 /// is the text of a row, the `**` of a bold run selected without its closing
1837 /// `**` is two asterisks. Every one of those still *renders* — there's no
1838 /// error to report — it just renders as the fragment and not as the document.
1839 /// Widening the range to whole blocks would publish text the user didn't
1840 /// select, which is a worse lie than a fragment being a fragment; the plain
1841 /// flavor has the same substring, so the two flavors at least agree.
1842 pub fn selection_html(&mut self) -> Option<String> {
1843 let (start, end) = self.selection()?;
1844 let inline = self.selection_is_inline(start, end);
1845 let html = html::render_fragment(&self.source[start..end], self.format)?;
1846 Some(match inline {
1847 true => html::strip_sole_paragraph(html),
1848 false => html,
1849 })
1850 }
1851
1852 /// Paste the clipboard's `text/html` flavor, converting it to this document's
1853 /// format first. Its own undo step, like any [`paste`](Self::paste).
1854 ///
1855 /// Returns whether it landed. `false` means the HTML didn't convert to
1856 /// anything worth pasting — the caller should fall back to the plain flavor
1857 /// rather than treat it as an error. The `html` module has the full list of
1858 /// what that covers: a table twig won't build, markup it doesn't recognise,
1859 /// an empty result.
1860 pub fn paste_html(&mut self, html: &str) -> bool {
1861 match html::parse_fragment(html, self.format) {
1862 Some(source) => {
1863 self.paste(&source);
1864 true
1865 }
1866 None => false,
1867 }
1868 }
1869
1870 /// Does the selection live *inside* a single top-level block?
1871 ///
1872 /// The question [`selection_html`](Self::selection_html) needs and the
1873 /// fragment can't answer: `**bold**` renders as `<p><strong>bold</strong></p>`
1874 /// whether the user selected one word of a sentence or a whole paragraph, and
1875 /// only the document knows which. Selecting a word and pasting into Docs
1876 /// should extend the line you paste into; selecting the paragraph should make
1877 /// a paragraph. So a selection strictly within one block is inline (its `<p>`
1878 /// is an artifact of standalone parsing), and one that covers a whole block —
1879 /// or spans two — keeps its structure.
1880 ///
1881 /// Reads the block from twig rather than guessing from the bytes:
1882 /// `ancestors_at` is `[doc, block, …inline]`, so index 1 is the top-level
1883 /// block containing an offset, and two ends inside the same one cannot have
1884 /// crossed a block boundary.
1885 fn selection_is_inline(&mut self, start: usize, end: usize) -> bool {
1886 // The last *character*, not `end - 1`: the selection's end is exclusive
1887 // and may sit mid-codepoint's-worth of bytes past the last char.
1888 let Some((off, _)) = self.source[start..end].char_indices().next_back() else {
1889 return false;
1890 };
1891 let (Some(head), Some(tail)) =
1892 (self.top_block_span(start), self.top_block_span(start + off))
1893 else {
1894 return false;
1895 };
1896 head == tail && !(start <= head.start && end >= head.end)
1897 }
1898
1899 /// The byte span of the top-level block containing `offset`, or `None` at an
1900 /// offset that belongs to no block (the blank line between two of them).
1901 fn top_block_span(&mut self, offset: usize) -> Option<std::ops::Range<usize>> {
1902 self.editor
1903 .ancestors_at(offset)
1904 .ok()?
1905 .get(1)
1906 .map(|m| m.span.clone())
1907 }
1908
1909 // ── indentation ──────────────────────────────────────────────────────────
1910
1911 /// One indent level.
1912 ///
1913 /// Two spaces, not the four both frontends type for Tab today, because in a
1914 /// markdown document four columns isn't a width — it's a *meaning*. Four
1915 /// spaces at the head of a line is markdown's indented-code-block marker, so
1916 /// one Tab on a paragraph would reparse it into code and style it as such;
1917 /// two cannot, and the line stays the prose it was. Two is also exactly
1918 /// where a `- ` bullet's content starts, so an indented line lands under its
1919 /// parent item's text instead of beside it — the column a list-aware indent
1920 /// has to hit anyway, which keeps this width from being relitigated later.
1921 const INDENT: &'static str = " ";
1922
1923 /// Indent the selected lines — or the caret's line, with no selection — by
1924 /// one level (Tab).
1925 pub fn indent(&mut self) {
1926 self.reindent(true);
1927 // Nesting changes an ordered list's numbering (the nested item restarts,
1928 // its old siblings resume) — keep the source markers in step.
1929 self.renumber_here();
1930 // Nesting an empty `-` item under a text line reparses that text as a
1931 // setext heading; swap the dash for a `*` before it can (a no-op unless
1932 // the collapse actually happened).
1933 self.avoid_setext_collapse();
1934 }
1935
1936 /// Take one indent level back off the selected lines, or the caret's line
1937 /// (Shift+Tab). A line with no indentation is left exactly as it is.
1938 ///
1939 /// A line with *less* than a full level gives back what it has rather than
1940 /// refusing: outdent's job is to walk a line left, and real documents — hand
1941 /// written, or reflowed by some other editor — are full of indentation that
1942 /// was never a clean multiple of anything. Refusing there would strand the
1943 /// line at a depth Shift+Tab couldn't undo.
1944 pub fn outdent(&mut self) {
1945 self.reindent(false);
1946 self.renumber_here();
1947 }
1948
1949 /// The body of [`indent`](Self::indent) / [`outdent`](Self::outdent).
1950 ///
1951 /// One splice across the whole line range, never one per line: a Tab is one
1952 /// thing the user did, so it has to be one undo step and one reparse. Per
1953 /// line, twig would reparse the document once per line and leave a stack of
1954 /// steps that Shift+⌘Z walks back one line at a time.
1955 fn reindent(&mut self, add: bool) {
1956 let (sel_start, sel_end) = self.selection().unwrap_or((self.caret, self.caret));
1957 let start = source_line_range(&self.source, sel_start).start;
1958 let end = source_line_range(&self.source, sel_end).end;
1959 let region = self.source[start..end].to_string();
1960 let lines: Vec<&str> = region.split('\n').collect();
1961 // A blank line has no text to move, and padding it would leave nothing
1962 // but trailing whitespace — but Tab on a blank line *is* a request for
1963 // indentation to type into, so the skip only applies where the op has
1964 // other lines to do real work on.
1965 let skip_blank = add && lines.len() > 1;
1966
1967 let mut out = String::with_capacity(region.len() + lines.len() * Self::INDENT.len());
1968 let mut deltas: Vec<isize> = Vec::with_capacity(lines.len());
1969 let mut line_off = start;
1970 for (i, full) in lines.iter().enumerate() {
1971 if i > 0 {
1972 out.push('\n');
1973 }
1974 // A list item moves by having its whole leading prefix *replaced*,
1975 // never by having spaces pushed in front of the line. twig spells
1976 // both prefixes, so the quote markers, the parent's indent and an
1977 // ordered marker's extra column all come out right without leaf
1978 // measuring any of them — and a line that only looks like an item
1979 // (a Djot continuation) reports no marker and is left to the plain
1980 // path, where a Tab is just a Tab.
1981 let marker = self.list_marker_on_line(line_off);
1982 let own = marker
1983 .as_ref()
1984 .map(|m| m.marker_start - m.line_start)
1985 .unwrap_or(0);
1986 let delta = if add {
1987 if skip_blank && full.trim().is_empty() {
1988 out.push_str(full);
1989 0
1990 } else if marker.is_some() && self.first_item_of_list(line_off) {
1991 // The first item of a list has no preceding sibling to nest
1992 // under, so a Tab here can't spell a sub-list — twig would
1993 // reparse the shoved-over marker as the same list, only
1994 // indented, which Shift+Tab then can't cleanly undo. Leave the
1995 // item where it is, the way every list editor refuses to
1996 // over-indent a list's first line.
1997 out.push_str(full);
1998 0
1999 } else if marker.is_some() {
2000 // Nesting means standing where a *continuation* of this line
2001 // would stand: past the parent's marker, inside its content
2002 // column. That is `continuation_prefix`, less a checkbox.
2003 let new = self.nesting_prefix_at(line_off);
2004 let delta = new.len() as isize - own as isize;
2005 out.push_str(&new);
2006 out.push_str(&full[own..]);
2007 delta
2008 } else {
2009 out.push_str(Self::INDENT);
2010 out.push_str(full);
2011 Self::INDENT.len() as isize
2012 }
2013 } else if marker.is_some() {
2014 // Unnesting is the mirror: stand where the parent item's own
2015 // line starts, which drops exactly the level it contributed.
2016 let new = self.outdent_prefix_at(line_off);
2017 let delta = new.len() as isize - own as isize;
2018 out.push_str(&new);
2019 out.push_str(&full[own..]);
2020 delta
2021 } else {
2022 // A plain line gives back the ordinary step.
2023 let strip = outdent_width(full, Self::INDENT.len());
2024 out.push_str(&full[strip..]);
2025 -(strip as isize)
2026 };
2027 deltas.push(delta);
2028 line_off += full.len() + 1;
2029 }
2030 // Nothing to give back. Returning before the splice keeps an outdent at
2031 // column zero from spending an undo step on a document it never changed.
2032 if deltas.iter().all(|d| *d == 0) {
2033 return;
2034 }
2035
2036 // Every line's text keeps its offset *within the line*, so the caret is
2037 // remapped by its column, not by its byte offset — which the prefixes on
2038 // the lines above it have already invalidated.
2039 let remap = |off: usize| -> usize {
2040 let (mut old_ls, mut new_ls) = (start, start);
2041 for (line, delta) in lines.iter().zip(&deltas) {
2042 let old_le = old_ls + line.len();
2043 let new_len = (line.len() as isize + delta) as usize;
2044 if off <= old_le {
2045 let col = (off - old_ls) as isize;
2046 return new_ls + ((col + delta).max(0) as usize).min(new_len);
2047 }
2048 old_ls = old_le + 1;
2049 new_ls += new_len + 1;
2050 }
2051 start + out.len()
2052 };
2053 let placed = match self.selection() {
2054 // Keep the rewritten region selected, the way a container toggle
2055 // keeps its own: it leaves a second Tab aimed at the same lines
2056 // rather than at whatever the shifted offsets now happen to cover.
2057 Some(_) => (start + out.len(), Some(start)),
2058 None => (remap(self.caret), None),
2059 };
2060
2061 // A rolled-back splice leaves the old source in place, where every offset
2062 // computed above addresses text that was never written.
2063 if !self.splice(start, end, &out, EditKind::Other) {
2064 return;
2065 }
2066 // `splice` re-anchors to the end of the `Change`, which for a whole-region
2067 // rewrite is the last line's end — nowhere the caret was. Place it, then
2068 // re-record the caret so this is the state redo restores, not the one
2069 // `splice` left behind from the `Change`.
2070 self.caret = placed.0.min(self.source.len());
2071 self.anchor = placed.1;
2072 self.clamp_caret();
2073 self.record_caret();
2074 }
2075
2076 /// The Enter key.
2077 ///
2078 /// In source view it's a literal newline. In WYSIWYG it's **AST-aware**: a
2079 /// bare `\n` is only a markdown soft break (same paragraph), so the block the
2080 /// caret is in decides what actually gets written.
2081 ///
2082 /// - paragraph → twig's [`Editor::split_block`], which parts the
2083 /// block at the caret and reopens its container
2084 /// - list item → likewise: the next item, its indent, quote
2085 /// prefix and `[ ]` box all reproduced by twig —
2086 /// except an *empty* item, which exits the list
2087 /// - block quote → likewise: a new paragraph inside the quote
2088 /// - heading → a new *paragraph*, not another heading
2089 /// - code block → a literal newline (stay in the block)
2090 /// - blank line → a literal newline (one Backspace undoes it)
2091 /// - [`LineFlow::Preserve`] → a single soft break, which renders as a
2092 /// visible line
2093 ///
2094 /// Where `split_block` is used it replaces markup leaf used to spell by hand,
2095 /// and it is better at it: it drops the whitespace the caret was sitting in
2096 /// front of instead of stranding it at the head of the second half, and it
2097 /// knows continuations leaf's marker scan never covered — a checklist item
2098 /// continues as an *unchecked* checklist item rather than a plain bullet.
2099 ///
2100 /// The exceptions above are exceptions because `split_block` is either wrong
2101 /// there or refuses: parting a fence yields two fences with the code split
2102 /// between them, parting a heading yields a second heading where every editor
2103 /// gives a paragraph, and a blank line, an empty item, a setext heading and a
2104 /// table all report an error rather than a split.
2105 pub fn newline(&mut self) {
2106 if self.view == View::Source {
2107 self.insert_raw("\n");
2108 return;
2109 }
2110 // Enter over a selection replaces it with a paragraph break.
2111 if let Some((s, e)) = self.selection() {
2112 self.splice(s, e, "\n\n", EditKind::Other);
2113 return;
2114 }
2115 // A caret resting exactly between an inline mark's content and its own
2116 // closing delimiter (`**bold**` with nothing after it on the line —
2117 // the WYSIWYG caret's natural end-of-line position) must not splice a
2118 // block break there: every path below eventually does via
2119 // `insert_raw`/`self.caret`, and splicing before the hidden closing
2120 // delimiter would strand it alone on the new line.
2121 self.caret = self.skip_trailing_close_delims(self.caret);
2122 // The block the caret is in. `block_offset_for_caret` nudges off a line
2123 // end (where the caret sits at the doc level); on a bare line (e.g. an
2124 // empty list item) fall back to the caret so the enclosing list/quote is
2125 // still visible in the ancestors.
2126 let off = self.block_offset_for_caret().unwrap_or(self.caret);
2127 let kinds: Vec<Kind> = self
2128 .editor
2129 .ancestors_at(off)
2130 .map(|c| c.into_iter().map(|m| m.kind).collect())
2131 .unwrap_or_default();
2132 let has = |k: Kind| kinds.contains(&k);
2133
2134 if has(Kind::CodeBlock) {
2135 self.insert_raw("\n");
2136 return;
2137 }
2138 // An *empty* list item exits the list — the standard double-Enter — which
2139 // `split_block` reports as an error rather than a split (there is no
2140 // content to part), so it stays leaf's. `list_marker_on_line` is itself
2141 // the AST gate — it answers from the tree, so a `- ` that reads as a
2142 // marker byte-for-byte but opens no item (a setext underline, a Djot
2143 // continuation line) never reaches here.
2144 if let Some(marker) = self.list_marker_on_line(self.caret)
2145 && self.item_is_empty(&marker)
2146 {
2147 self.exit_list(&marker);
2148 return;
2149 }
2150 // On an *empty* paragraph line, a lone Enter should add a single blank line,
2151 // not another full paragraph break — so it moves down one line and one
2152 // Backspace undoes it, not two. (`split_block` errors here too.)
2153 let line_start = self.source[..self.caret].rfind('\n').map_or(0, |i| i + 1);
2154 let line_end = self.source[self.caret..]
2155 .find('\n')
2156 .map_or(self.source.len(), |i| self.caret + i);
2157 if self.source[line_start..line_end].trim().is_empty() {
2158 self.insert_raw("\n");
2159 return;
2160 }
2161 // In `Preserve` flow a soft break is a *visible* line the author means to
2162 // make, so Enter writes a single `\n` and typing continues the same
2163 // paragraph on the next line — the behaviour of an ordinary text editor.
2164 // A second Enter then lands on the blank line above and takes the
2165 // empty-line branch, so double-Enter still promotes to a full paragraph
2166 // break; and Backspace, which deletes a lone `\n` over a soft break,
2167 // undoes a single Enter symmetrically. In `Fold` flow a lone `\n` would
2168 // render as an invisible space, so Enter keeps making the paragraph break
2169 // that actually shows.
2170 //
2171 // Only in running prose. A list or a quote has a continuation of its own
2172 // to write, and a `\n` there is not a soft line but a lost container.
2173 let in_container = has(Kind::ListItem) || has(Kind::TaskListItem) || has(Kind::BlockQuote);
2174 if self.line_flow == LineFlow::Preserve && !in_container {
2175 self.insert_raw("\n");
2176 return;
2177 }
2178 // A heading gets a *paragraph*, never a second heading: Enter at the end
2179 // of a title is how every editor is asked for the body under it, and
2180 // `split_block` would repeat the `#` instead. Whitespace at the split
2181 // point goes with the break rather than opening the new paragraph, which
2182 // is what `split_block` does everywhere else.
2183 if has(Kind::Heading) {
2184 let mut end = self.caret;
2185 while self.source.as_bytes().get(end) == Some(&b' ') {
2186 end += 1;
2187 }
2188 self.splice(self.caret, end, "\n\n", EditKind::Other);
2189 return;
2190 }
2191 self.split_block_here();
2192 }
2193
2194 /// Part the block at the caret with twig's [`Editor::split_block`], leaving
2195 /// the caret in the second half.
2196 ///
2197 /// twig reopens whatever the first half was inside of — the bullet with its
2198 /// indent, the quote's `>`, a checklist item's `[ ]` — which is the whole
2199 /// reason this replaced the markup leaf used to spell from the line's bytes.
2200 /// It renumbers nothing, though: a new item mid-list is written with its
2201 /// neighbour's number, so [`renumber_here`](Self::renumber_here) still runs
2202 /// behind it, folded into the same undo step.
2203 ///
2204 /// Falls back to a plain paragraph break if twig declines, so an unhandled
2205 /// shape still moves the caret down rather than swallowing the keystroke.
2206 fn split_block_here(&mut self) {
2207 match self.editor.split_block(self.caret) {
2208 Ok(change) => {
2209 self.last_edit_kind = None;
2210 self.refresh();
2211 self.anchor = None;
2212 self.caret = change.new.end;
2213 self.dirty = self.source != self.clean_source;
2214 self.status = None;
2215 self.clamp_caret();
2216 self.record_caret();
2217 // Aimed at the new block's *start*: the caret twig leaves is one
2218 // past the marker it wrote, where there is no list in reach.
2219 self.renumber_at(change.new.start);
2220 }
2221 Err(_) => self.insert_raw("\n\n"),
2222 }
2223 }
2224
2225 /// Whether the item on the marker's line carries no content — the shape
2226 /// double-Enter reads as "I'm done with this list."
2227 fn item_is_empty(&self, line: &ListMarker) -> bool {
2228 let content_start = line.content_start().min(self.source.len());
2229 let line_end = self.source[self.caret..]
2230 .find('\n')
2231 .map(|i| self.caret + i)
2232 .unwrap_or(self.source.len());
2233 self.source[content_start..line_end.max(content_start)]
2234 .trim()
2235 .is_empty()
2236 }
2237
2238 /// Leave the list: replace the empty item's marker with a blank line, so the
2239 /// caret lands in a fresh paragraph below it.
2240 ///
2241 /// Inside a quote the blank line has to stay quoted (a bare one would end the
2242 /// quote), and the caret's new line keeps the `> ` it was already behind —
2243 /// leaving the list without also leaving the quote.
2244 fn exit_list(&mut self, line: &ListMarker) {
2245 let prefix = self.quote_prefix_at(line.marker_start);
2246 let blank = prefix.trim_end();
2247 self.splice(
2248 line.line_start,
2249 self.caret,
2250 &format!("{blank}\n{prefix}"),
2251 EditKind::Other,
2252 );
2253 }
2254
2255 /// What a line continuing the containers at `off` has to open with — the
2256 /// quote markers reproduced, each enclosing item's marker as its width in
2257 /// spaces. Also the column a nested item's marker stands in, which is what
2258 /// makes it Tab's answer.
2259 fn continuation_prefix_at(&mut self, off: usize) -> String {
2260 self.editor
2261 .document()
2262 .and_then(|mut d| d.continuation_prefix(off))
2263 .map(|p| p.text)
2264 .unwrap_or_default()
2265 }
2266
2267 /// The column a *nested list* may open at inside the item at `off` — which
2268 /// is not always where the item's own text continues.
2269 ///
2270 /// twig counts a task item's `[ ] ` box as part of its marker, correctly:
2271 /// it is markup a rich view hides, and the item's own wrapped text does
2272 /// stand past it. But a nested list may only open at the *list* marker's
2273 /// column, and four columns further in is an indented continuation of the
2274 /// paragraph instead — `- [ ] a` + ` - [ ] b` is one item, not two.
2275 /// So the box's own width goes back.
2276 ///
2277 /// The one place leaf still reads a checkbox's spelling. It goes when twig
2278 /// reports the list marker's column apart from the box; `checked` is what
2279 /// says a box is there at all, so only its width is being measured here.
2280 fn nesting_prefix_at(&mut self, off: usize) -> String {
2281 let cont = self.continuation_prefix_at(off);
2282 let Some(item) = self.innermost_list_item(off) else {
2283 return cont;
2284 };
2285 if item.checked.is_none() {
2286 return cont;
2287 }
2288 let box_width = item
2289 .marker_span
2290 .and_then(|m| self.source.get(m))
2291 .and_then(|marker| marker.rfind('[').map(|i| marker.len() - i))
2292 .unwrap_or(0);
2293 // The trailing columns are the ones the item's own marker contributed,
2294 // so trimming from the end leaves any quote prefix standing.
2295 cont[..cont.len().saturating_sub(box_width)].to_string()
2296 }
2297
2298 /// Where the line of the item *containing* the item at `off` begins — the
2299 /// prefix Shift+Tab moves back to, which gives up exactly the level the
2300 /// parent contributed. The quote prefix alone for a top-level item, which
2301 /// has no level left to give.
2302 fn outdent_prefix_at(&mut self, off: usize) -> String {
2303 let items: Vec<usize> = self
2304 .editor
2305 .document()
2306 .and_then(|mut d| d.ancestors_at_caret(off))
2307 .map(|c| {
2308 c.into_iter()
2309 .filter(|m| m.kind == Kind::ListItem || m.kind == Kind::TaskListItem)
2310 .map(|m| m.span.start)
2311 .collect()
2312 })
2313 .unwrap_or_default();
2314 // The second-innermost item is the parent; its own line's indent is the
2315 // target. `list_marker_on_line` gives that line's prefix directly.
2316 let parent = items.len().checked_sub(2).map(|i| items[i]);
2317 match parent.and_then(|p| self.list_marker_on_line(p)) {
2318 Some(m) => self.source[m.line_start..m.marker_start].to_string(),
2319 None => self.quote_prefix_at(off),
2320 }
2321 }
2322
2323 /// The block-quote prefix in force at `off` — `""` outside a quote, `"> "`
2324 /// inside one, `"> > "` inside two.
2325 ///
2326 /// Assembled from each enclosing quote's own [`FlatNode::marker_span`], so
2327 /// the `>` and the space after it are twig's spelling rather than leaf's.
2328 /// The whole line prefix can't answer this: it also carries the indent of
2329 /// whatever the quote holds, which a blank separator line must *not* repeat.
2330 fn quote_prefix_at(&mut self, off: usize) -> String {
2331 let Ok(chain) = self
2332 .editor
2333 .document()
2334 .and_then(|mut d| d.ancestors_at_caret(off))
2335 else {
2336 return String::new();
2337 };
2338 let quotes: Vec<usize> = chain
2339 .iter()
2340 .filter(|m| m.kind == Kind::BlockQuote)
2341 .map(|m| m.node_id as usize)
2342 .collect();
2343 let Ok(nodes) = self.editor.nodes() else {
2344 return String::new();
2345 };
2346 quotes
2347 .iter()
2348 .filter_map(|id| nodes.get(*id)?.marker_span.clone())
2349 .filter_map(|s| self.source.get(s))
2350 .collect()
2351 }
2352
2353 /// Whether the item at `off` sits inside another one — the test Backspace
2354 /// uses to choose between outdenting and dropping the marker.
2355 ///
2356 /// Counted from the AST rather than from the line's leading whitespace,
2357 /// which is indentation in Markdown and, in Djot, may be nothing at all.
2358 fn item_is_nested(&mut self, off: usize) -> bool {
2359 self.editor
2360 .document()
2361 .and_then(|mut d| d.ancestors_at_caret(off))
2362 .map(|c| {
2363 c.into_iter()
2364 .filter(|m| m.kind == Kind::ListItem || m.kind == Kind::TaskListItem)
2365 .count()
2366 > 1
2367 })
2368 .unwrap_or(false)
2369 }
2370
2371 /// The innermost list item containing `probe`, under twig's **caret**
2372 /// containment rule — a block's end is inside it.
2373 ///
2374 /// Half-open containment can't answer this. An empty item's span is exactly
2375 /// its marker, so the caret sitting after `- ` is one past the end and the
2376 /// item it is plainly in tests as out of reach; that is the shape
2377 /// double-Enter has to recognise to leave the list.
2378 fn innermost_list_item(&mut self, probe: usize) -> Option<FlatNode> {
2379 let chain = self
2380 .editor
2381 .document()
2382 .and_then(|mut d| d.ancestors_at_caret(probe))
2383 .ok()?;
2384 let id = chain
2385 .iter()
2386 .rev()
2387 .find(|m| m.kind == Kind::ListItem || m.kind == Kind::TaskListItem)?
2388 .node_id as usize;
2389 self.editor.nodes().ok()?.get(id).cloned()
2390 }
2391
2392 /// The list marker opening `off`'s line, per twig — `None` when that line
2393 /// opens no list item.
2394 ///
2395 /// [`Document::line_prefix`] is the whole hidden run from the line start:
2396 /// `> 1. ` is a quote's marker, an indent, and an item's marker together,
2397 /// and it is `None` on a *continuation* line, which opens nothing. That last
2398 /// case is the one leaf could never get right by reading bytes. `- a\n - b`
2399 /// is two items in Markdown and one in Djot, where a marker cannot interrupt
2400 /// a paragraph and ` - b` is literal text — identical bytes, and only the
2401 /// parser knows which document it is looking at.
2402 ///
2403 /// The item's own marker is separated out via its
2404 /// [`FlatNode::marker_span`], so `marker_start` splits the prefix into what
2405 /// the containers around it contribute and what the item does.
2406 fn list_marker_on_line(&mut self, off: usize) -> Option<ListMarker> {
2407 let off = off.min(self.source.len());
2408 let prefix = self.editor.document().ok()?.line_prefix(off).ok()??;
2409 // The prefix belongs to a list only when an item's marker closes it —
2410 // a heading's `# ` or a bare quote's `> ` is a prefix too.
2411 let item = self.innermost_list_item(prefix.end.min(self.source.len()))?;
2412 let marker = item.marker_span.clone()?;
2413 if marker.end != prefix.end {
2414 return None;
2415 }
2416 Some(ListMarker {
2417 line_start: prefix.start,
2418 marker_start: marker.start,
2419 text: self.source.get(prefix)?.to_string(),
2420 })
2421 }
2422
2423 /// Whether the list item on `line_start`'s line is the **first item** of its
2424 /// list — the one Tab must not nest, because nesting needs a preceding
2425 /// sibling to become the new parent and a first item has none. `false` for a
2426 /// line that isn't a list item, and for an item with a sibling above it (the
2427 /// one Tab *can* nest). Gated on the AST, not the marker bytes: `- ` reads
2428 /// the same in a setext underline that opens no list at all.
2429 fn first_item_of_list(&mut self, line_start: usize) -> bool {
2430 let Some(marker) = self.list_marker_on_line(line_start) else {
2431 return false;
2432 };
2433 // Probe just inside the marker, where the item's own node is in reach —
2434 // the marker offset itself can resolve to the enclosing list, not the
2435 // `list_item`, whose span starts at the marker.
2436 let probe = marker.content_start().min(self.source.len());
2437 let Some(item) = self.innermost_list_item(probe) else {
2438 return false;
2439 };
2440 let Ok(nodes) = self.editor.nodes() else {
2441 return false;
2442 };
2443 match item.parent {
2444 // First when the parent list opens with this very item.
2445 Some(pid) => nodes
2446 .get(pid.0 as usize)
2447 .is_some_and(|p| p.first_child == Some(item.id)),
2448 // A parentless item is trivially the first (and only) one.
2449 None => true,
2450 }
2451 }
2452
2453 pub fn backspace(&mut self) {
2454 if let Some((s, e)) = self.selection() {
2455 self.splice(s, e, "", EditKind::Other);
2456 return;
2457 }
2458 // WYSIWYG: Backspace at the very start of a list item's content is a
2459 // structural key, not a character delete — it walks the "un-indent, then
2460 // un-list" ladder every list editor gives that keystroke (outdent a
2461 // nested item, strip a top-level one's marker to a paragraph). In source
2462 // view the `- ` is visible text the user is deleting a byte of, so it
2463 // keeps its literal meaning there, like Enter does.
2464 if self.view != View::Source && self.backspace_list_start() {
2465 return;
2466 }
2467 // WYSIWYG: and the same at the start of a heading's content — the `# `
2468 // there is markup the rich view hides, not text the user typed.
2469 if self.view != View::Source && self.backspace_heading_start() {
2470 return;
2471 }
2472 // WYSIWYG: at a block picture's stops, a byte-at-a-time delete would take
2473 // the markup apart under a caret that cannot see it — see
2474 // `delete_around_block_media`.
2475 if self.view != View::Source && self.delete_around_block_media(false) {
2476 return;
2477 }
2478 // WYSIWYG: Backspace on a *blank line* deletes back to the previous caret
2479 // stop, not a single newline. On a line with no text of its own, the byte
2480 // before the caret is a `\n` that spells part of a block boundary — the gap
2481 // between two blocks, drawn but never a caret home. Removing just it strands
2482 // the caret in that gap and leaves an odd blank line the eye reads as one
2483 // separator but the caret can't land on: the "extra newline" left behind
2484 // after leaving a list (Enter, Enter) or a paragraph and pressing Backspace.
2485 // Deleting to the previous stop instead collapses the whole break at once,
2486 // landing the caret at the end of the block above. Two blank lines in a row
2487 // are one stop apart, so this still removes exactly one — the lone-Enter /
2488 // lone-Backspace symmetry the empty-line case is built on is untouched.
2489 if self.view != View::Source
2490 && self.caret > self.caret_floor()
2491 && self.caret_on_blank_line()
2492 && let Some(stop) = self.vmap.stop_before(self.caret)
2493 {
2494 let stop = stop.max(self.caret_floor());
2495 if stop < self.caret {
2496 self.splice(stop, self.caret, "", EditKind::Delete);
2497 return;
2498 }
2499 }
2500 if self.caret > self.caret_floor() {
2501 // An in-cell `<br>` draws as one newline glyph, so Backspace over it
2502 // takes the whole tag — a single-byte step would leave a broken `<br`
2503 // showing in the cell. Rich view only (source view edits the literal).
2504 if self.view != View::Source
2505 && let Some((start, end)) = self.cell_break_at(BreakEdge::Backward)
2506 {
2507 let start = start.max(self.caret_floor());
2508 if start < end {
2509 self.splice(start, end, "", EditKind::Delete);
2510 return;
2511 }
2512 }
2513 // Aim the delete at the character the writer can *see* behind the
2514 // caret, never at a delimiter the rich view drew nothing for. Two
2515 // steps, and either can apply: from the far side of a run's closing
2516 // `**` step back into the run (the caret is drawn at the end of its
2517 // word), and at the start of a run's text step out past its opening
2518 // `**` to the character in front of it, leaving the run standing.
2519 // Without them a plain Backspace unspells the phrase it is editing
2520 // and leaves a literal asterisk on screen.
2521 let end = if self.view == View::Source {
2522 self.caret
2523 } else {
2524 let inside = self.step_inside_close_delims(self.caret);
2525 self.skip_leading_open_delims(inside)
2526 .max(self.caret_floor())
2527 };
2528 // Never delete back across the floor — that would eat hidden
2529 // frontmatter the WYSIWYG caret can't even see.
2530 let mut prev = prev_boundary(&self.source, end).max(self.caret_floor());
2531 // Take a hidden escape backslash with the char it escapes: the rich
2532 // view draws `\*` as a single `*`, so Backspace over it must delete
2533 // both bytes, never strand the `\` as a lone visible backslash (the
2534 // mirror of the Hidden-mode typing that wrote the escape). Source view
2535 // shows the `\`, so there it is an ordinary character.
2536 if self.view != View::Source
2537 && prev > self.caret_floor()
2538 && self.is_hidden_escape(prev - 1)
2539 {
2540 prev -= 1;
2541 }
2542 if prev < end {
2543 self.splice(prev, end, "", EditKind::Delete);
2544 }
2545 }
2546 }
2547
2548 /// Whether the caret's own source line holds nothing but whitespace — an
2549 /// empty paragraph, or the blank line a block boundary is spelled with. The
2550 /// test for [`backspace`](Self::backspace)'s stop-wise delete: such a line has
2551 /// no text of its own, so the newline before the caret belongs to the gap
2552 /// between blocks rather than to any word the caret is editing.
2553 fn caret_on_blank_line(&self) -> bool {
2554 let line_start = self.source[..self.caret].rfind('\n').map_or(0, |i| i + 1);
2555 let line_end = self.source[self.caret..]
2556 .find('\n')
2557 .map_or(self.source.len(), |i| self.caret + i);
2558 self.source[line_start..line_end].trim().is_empty()
2559 }
2560
2561 /// The source span of an in-cell hard break (`<br>`) touching the caret on the
2562 /// `edge` side — the byte range to delete whole. A table row is one source
2563 /// line, so its break is spelled `<br>` yet drawn as a single newline glyph
2564 /// (see `wysiwyg.rs`); a delete over it must take every byte, or a one-byte
2565 /// step strands a broken `<br` in the cell. `Backward` matches a break ending
2566 /// at the caret (Backspace), `Forward` one starting at it (Delete). `None`
2567 /// when no such break is adjacent. Only the in-cell break is spelled `<br>`
2568 /// (an ordinary hard break is ` \n`), so the leading `<` alone tells them
2569 /// apart — no ancestor walk needed. Rich view only; source view shows the
2570 /// literal tag and deletes it a byte at a time.
2571 fn cell_break_at(&mut self, edge: BreakEdge) -> Option<(usize, usize)> {
2572 let caret = self.caret;
2573 let nodes = self.nodes();
2574 let src = self.source.as_bytes();
2575 nodes
2576 .iter()
2577 .find(|n| {
2578 n.kind == Kind::HardBreak
2579 && n.span.start < n.span.end
2580 && src.get(n.span.start) == Some(&b'<')
2581 && match edge {
2582 BreakEdge::Backward => n.span.end == caret,
2583 BreakEdge::Forward => n.span.start == caret,
2584 }
2585 })
2586 .map(|n| (n.span.start, n.span.end))
2587 }
2588
2589 /// Whether the source byte at `off` is a backslash twig consumed as an escape
2590 /// (hidden in the rich view), as against a literal backslash (drawn). A
2591 /// backslash escapes exactly an ASCII-punctuation character (the CommonMark /
2592 /// Djot rule twig follows), so `\` + punctuation is the whole test — no AST
2593 /// round-trip needed.
2594 fn is_hidden_escape(&self, off: usize) -> bool {
2595 let b = self.source.as_bytes();
2596 b.get(off) == Some(&b'\\') && b.get(off + 1).is_some_and(u8::is_ascii_punctuation)
2597 }
2598
2599 /// Backspace's list behaviour: when the caret sits exactly at the start of a
2600 /// list item's content (right after its marker), outdent the item if it's
2601 /// nested, else strip the marker so it becomes a paragraph. Returns whether
2602 /// it acted — `false` leaves Backspace its ordinary character delete.
2603 fn backspace_list_start(&mut self) -> bool {
2604 let Some(marker) = self.list_marker_on_line(self.caret) else {
2605 return false;
2606 };
2607 // Only right after the marker. That the line opens a real item is
2608 // already settled: `list_marker_on_line` answers from the tree.
2609 if self.caret != marker.content_start() {
2610 return false;
2611 }
2612 if self.item_is_nested(marker.marker_start) {
2613 // Nested: give back one level, keeping the marker and carrying the
2614 // caret with it.
2615 self.outdent();
2616 } else {
2617 // Top level: drop the marker, leaving a paragraph, then renumber the
2618 // siblings the removed item was counted among. Only the marker goes —
2619 // a quote prefix in front of it still has a quote to hold up.
2620 self.splice(marker.marker_start, self.caret, "", EditKind::Other);
2621 self.renumber_here();
2622 }
2623 true
2624 }
2625
2626 /// Backspace's heading behaviour: with the caret exactly at the start of an
2627 /// ATX heading's content — right after the `#` marker the rich view hides —
2628 /// strip the marker so the line becomes a paragraph. The peer of
2629 /// [`backspace_list_start`](Self::backspace_list_start)'s ladder, and the same
2630 /// reasoning: hidden block markup is structure, so the keystroke over it is
2631 /// structural.
2632 ///
2633 /// Without this the ordinary delete takes the space out of `# Title` and
2634 /// leaves `#Title`, which is no longer a heading at all — the hash the view
2635 /// had been hiding surfaces as literal text the user has to delete a second
2636 /// time, having never typed it. A closing sequence (`# Title #`, hidden at the
2637 /// other end) goes with the marker for the same reason.
2638 ///
2639 /// Returns whether it acted; `false` leaves Backspace its character delete.
2640 fn backspace_heading_start(&mut self) -> bool {
2641 let caret = self.caret;
2642 // The heading whose content opens exactly at the caret. A bare `#` has no
2643 // content span at all — its content starts (and ends) where the line does.
2644 let Some((span, content_end, marker)) = self.nodes().iter().find_map(|n| {
2645 let (start, end) = match &n.content_span {
2646 Some(c) => (c.start, c.end),
2647 None => (n.span.end, n.span.end),
2648 };
2649 (n.kind == Kind::Heading && start == caret)
2650 .then(|| (n.span.clone(), end, n.marker_span.clone()))
2651 }) else {
2652 return false;
2653 };
2654 // twig reports the marker's own extent, so there is nothing to walk back
2655 // over and no `#` in this file. A setext heading has no marker — its
2656 // content opens the line — so it falls through to the ordinary delete,
2657 // as does anything else sitting at a content start.
2658 // `m.end == caret` is what excludes a setext heading, whose marker is the
2659 // underline *after* the content rather than a prefix before it.
2660 let Some(marker) = marker.filter(|m| m.end == caret) else {
2661 return false;
2662 };
2663 let start = marker.start;
2664 // A closing `#` sequence is hidden too, so it can't be left behind. Only
2665 // when the tail really is one: trailing spaces alone are nothing to strip.
2666 let tail = &self.source[content_end..span.end];
2667 if tail.contains('#') && tail.chars().all(|c| c == '#' || c.is_whitespace()) {
2668 let kept = self.source[caret..content_end].to_string();
2669 self.splice(start, span.end, &kept, EditKind::Other);
2670 // The splice leaves the caret past the text it re-wrote; the caret
2671 // belongs where the content now starts, which is where it already was.
2672 self.caret = start;
2673 self.record_caret();
2674 } else {
2675 self.splice(start, caret, "", EditKind::Other);
2676 }
2677 true
2678 }
2679
2680 pub fn delete_forward(&mut self) {
2681 if let Some((s, e)) = self.selection() {
2682 self.splice(s, e, "", EditKind::Other);
2683 } else if self.caret < self.source.len() {
2684 // The mirror of Backspace's: forward-delete in front of a picture
2685 // would eat the `!` off its markup and leave a link where a photo was.
2686 if self.view != View::Source && self.delete_around_block_media(true) {
2687 return;
2688 }
2689 // Delete forward over an in-cell `<br>` takes the whole tag, the mirror
2690 // of Backspace's swallow (see `cell_break_at`) — else a byte-step
2691 // strands a broken `<br` in the cell.
2692 if self.view != View::Source
2693 && let Some((start, end)) = self.cell_break_at(BreakEdge::Forward)
2694 {
2695 self.splice(start, end, "", EditKind::Delete);
2696 return;
2697 }
2698 // The mirror of Backspace's two steps: from in front of a run's
2699 // opening `**` step into it, onto the first letter of its text, and
2700 // at the end of a run's text step out past its closing `**` to the
2701 // character beyond. Either way Delete takes the character it looks
2702 // like it is pointing at, and never a delimiter drawn as nothing.
2703 // The caret then settles back inside the run it was standing in —
2704 // see `settle_inside_close_delims`.
2705 let from = if self.view == View::Source {
2706 self.caret
2707 } else {
2708 let inside = self.step_inside_open_delims(self.caret);
2709 self.skip_trailing_close_delims(inside)
2710 };
2711 let next = next_boundary(&self.source, from);
2712 if from < next {
2713 self.splice(from, next, "", EditKind::Delete);
2714 }
2715 }
2716 }
2717
2718 /// Delete from the caret back to the start of the previous word (⌥⌫ /
2719 /// Ctrl+⌫). Deletes the selection instead when one is active.
2720 pub fn delete_word_back(&mut self) {
2721 if let Some((s, e)) = self.selection() {
2722 self.splice(s, e, "", EditKind::Other);
2723 } else {
2724 // A word back from just past a picture is a word *of its markup*, and
2725 // a word back from in front of one runs through the paragraph break
2726 // into the prose above — dissolving the picture either way. See
2727 // `delete_around_block_media`.
2728 if self.view != View::Source && self.delete_around_block_media(false) {
2729 return;
2730 }
2731 let start = self.word_left_from(self.caret).max(self.caret_floor());
2732 if start < self.caret {
2733 let (s, e) = self.widen_over_emptied_inlines(start, self.caret);
2734 self.splice(s, e, "", EditKind::Delete);
2735 }
2736 }
2737 }
2738
2739 /// Delete from the caret forward to the end of the next word (⌥⌦ /
2740 /// Ctrl+Del). Deletes the selection instead when one is active.
2741 pub fn delete_word_forward(&mut self) {
2742 if let Some((s, e)) = self.selection() {
2743 self.splice(s, e, "", EditKind::Other);
2744 } else {
2745 // The mirror: a word forward from in front of a picture is its markup.
2746 if self.view != View::Source && self.delete_around_block_media(true) {
2747 return;
2748 }
2749 let end = self.word_right_from(self.caret);
2750 if end > self.caret {
2751 let (s, e) = self.widen_over_emptied_inlines(self.caret, end);
2752 self.splice(s, e, "", EditKind::Delete);
2753 }
2754 }
2755 }
2756
2757 /// Delete from the caret back to the start of its line (⌘⌫). Deletes the
2758 /// selection instead when one is active, as every other delete here does.
2759 ///
2760 /// The line is the view's own — the one Home and End work on, so in WYSIWYG
2761 /// a soft-wrapped row is a line. It is not Home's *target*, though: Home
2762 /// stops at the first character and this takes the indentation with it, the
2763 /// way Cocoa's `deleteToBeginningOfLine:` does. Stopping at the text would
2764 /// leave an indent behind that nothing can then ask to delete, where a caret
2765 /// left at column 0 is one press of Home away from either.
2766 pub fn delete_to_line_start(&mut self) {
2767 if let Some((s, e)) = self.selection() {
2768 self.splice(s, e, "", EditKind::Other);
2769 return;
2770 }
2771 // Never back across the floor: hidden frontmatter isn't on this line, or
2772 // on any line the WYSIWYG caret can see.
2773 let (start, _) = self.line_span();
2774 let start = start.max(self.caret_floor());
2775 if start < self.caret {
2776 let (s, e) = self.widen_over_emptied_inlines(start, self.caret);
2777 self.splice(s, e, "", EditKind::Delete);
2778 }
2779 }
2780
2781 /// Kill from the caret to the end of its line (^K). Deletes the selection
2782 /// instead when one is active.
2783 ///
2784 /// At the end of the line it does nothing, rather than pulling the line
2785 /// below up into this one. Joining has no meaning to give it in both views
2786 /// at once: a WYSIWYG line ends at a soft wrap as often as at a newline, and
2787 /// there is nothing there to delete, while the newline a *source* line ends
2788 /// with is only half of the blank line that separates two paragraphs —
2789 /// deleting one leaves a soft break, which is not the join it looks like.
2790 /// The views agreeing is worth more than emacs' second press, and Delete is
2791 /// already the key that joins.
2792 pub fn delete_to_line_end(&mut self) {
2793 if let Some((s, e)) = self.selection() {
2794 self.splice(s, e, "", EditKind::Other);
2795 return;
2796 }
2797 let (_, end) = self.line_span();
2798 if end > self.caret {
2799 let (s, e) = self.widen_over_emptied_inlines(self.caret, end);
2800 self.splice(s, e, "", EditKind::Delete);
2801 }
2802 }
2803
2804 /// Grow a WYSIWYG word-delete to swallow any inline node it empties.
2805 ///
2806 /// A glyph-space range covers what the user can see, which for `**bold**` is
2807 /// the word and never the delimiters around it — so deleting the word on its
2808 /// own leaves `a **** c`, markup wrapped around nothing. They asked for the
2809 /// word, and the styling was the word's; the two go together. Only the
2810 /// node's delimiters are taken, and those are hidden here anyway, so nothing
2811 /// visible outside the range is lost.
2812 ///
2813 /// Repeated to a fixed point: emptying `***bold***` empties the emph inside
2814 /// the strong, and only then is the strong empty too.
2815 fn widen_over_emptied_inlines(&mut self, start: usize, end: usize) -> (usize, usize) {
2816 if self.view == View::Source {
2817 return (start, end);
2818 }
2819 let nodes = self.nodes();
2820 let (mut s, mut e) = (start, end);
2821 loop {
2822 let mut grew = false;
2823 for n in nodes.iter().filter(|n| wysiwyg::is_inline(n)) {
2824 let Some(text) = inline_content_span(n, &self.source) else {
2825 continue;
2826 };
2827 // Some of its text survives, so the node still has a job.
2828 if text.start < s || text.end > e {
2829 continue;
2830 }
2831 if n.span.start < s || n.span.end > e {
2832 s = s.min(n.span.start);
2833 e = e.max(n.span.end);
2834 grew = true;
2835 }
2836 }
2837 if !grew {
2838 return (s, e);
2839 }
2840 }
2841 }
2842
2843 /// One splice of document text, keeping the **mark-edge rule**: an inline
2844 /// mark's content never begins or ends with whitespace. In Markdown and Djot
2845 /// a delimiter standing against a space is not a delimiter at all — `**bold **`
2846 /// is four literal asterisks around a word, and a rich view drawing the
2847 /// document faithfully has no choice but to show them. That is correct
2848 /// rendering of what the file says, and nobody typing a space after a bold
2849 /// word meant to say it.
2850 ///
2851 /// So the space goes *outside* the run instead — `**bold** ` — which is the
2852 /// same document to a reader and a live one to a parser. The caret follows it
2853 /// out and keeps the marks armed (see [`rearm`](Self::rearm)), so the next
2854 /// character rejoins the run (see [`rejoin_run`](Self::rejoin_run)) and the
2855 /// writer sees one unbroken bold phrase, never a flash of raw syntax.
2856 ///
2857 /// Every ordinary edit — typing, deleting, pasting, an IME step — comes
2858 /// through here, so the rule holds however the whitespace arrives at the
2859 /// edge. The repair is decided *after* the plain edit, by asking whether the
2860 /// mark actually died: a code span's backticks aren't whitespace-sensitive
2861 /// (`` `code ` `` is still code), and nothing is re-spelled when nothing broke.
2862 fn splice(&mut self, start: usize, end: usize, text: &str, kind: EditKind) -> bool {
2863 let fix = self.mark_edge_fix(start, end, text);
2864 if !self.splice_exact(start, end, text, kind) {
2865 return false;
2866 }
2867 if let Some(fix) = fix {
2868 self.repair_mark_edges(fix);
2869 }
2870 if text.is_empty() && end > start {
2871 self.settle_inside_close_delims();
2872 }
2873 true
2874 }
2875
2876 /// After a delete, take a caret left standing past a run's closing delimiters
2877 /// back inside the run.
2878 ///
2879 /// A delete leaves the caret where the deleted bytes began, and when those
2880 /// bytes were the last thing after a marked phrase — the space the mark-edge
2881 /// rule pushed out of `**bold** `, say — that spot is the far side of the
2882 /// closing `**`. The rich view has nothing to draw there: the delimiters are
2883 /// hidden, so the caret shows at the end of the word either way, and the two
2884 /// offsets are one place on screen with two different meanings. Typing at the
2885 /// outer one lands past the run, so the writer who backspaced a space out of
2886 /// their bold phrase watches the next character come out plain, and the
2887 /// toolbar button go dark, with the caret never appearing to move.
2888 ///
2889 /// The end of the run's text is the caret's home there — a delete that took
2890 /// away everything after a phrase leaves the caret at the end of that phrase,
2891 /// which is inside it — so it settles onto that
2892 /// ([`step_inside_close_delims`](Self::step_inside_close_delims) does the
2893 /// walk, through every mark closing at the point): the word stays bold, the
2894 /// button stays lit, and the next character carries on the phrase.
2895 ///
2896 /// Rich view only, and only where a mark really closes at the caret — mid-run
2897 /// or in plain prose no span ends there and the caret stays put. The opening
2898 /// edge is left alone on purpose: a caret in front of a run inherits from the
2899 /// text on its left, which is the plain text outside.
2900 fn settle_inside_close_delims(&mut self) {
2901 if self.view != View::Wysiwyg {
2902 return;
2903 }
2904 let at = self.step_inside_close_delims(self.caret);
2905 if at != self.caret {
2906 self.caret = at;
2907 self.clear_pending();
2908 self.record_caret();
2909 }
2910 }
2911
2912 /// The splice exactly as asked, with no mark-edge repair — for the callers
2913 /// that are *writing* the delimiters themselves ([`insert_with_marks`](Self::insert_with_marks)
2914 /// and [`rejoin_run`](Self::rejoin_run)) and place their own offsets around
2915 /// the bytes they inserted.
2916 ///
2917 /// One `edit_range` through twig, then re-anchor the caret from the returned
2918 /// `Change` and refresh the cached source. A reparse-breaking edit (rare for
2919 /// Markdown/Djot) leaves the document untouched and reports.
2920 ///
2921 /// Returns whether the edit landed — for a caller that has offsets of its
2922 /// own to place afterwards, which a rolled-back splice would leave pointing
2923 /// into text that never came to exist.
2924 fn splice_exact(&mut self, start: usize, end: usize, text: &str, kind: EditKind) -> bool {
2925 // The read-only gate, for every edit at once — see the field.
2926 if self.read_only {
2927 return false;
2928 }
2929 // twig records an undo step for every edit; when this one continues a
2930 // run of the same kind (typing, deleting), tell twig to fold it into the
2931 // step before it so the whole run undoes at once.
2932 let coalesce = kind != EditKind::Other && self.last_edit_kind == Some(kind);
2933 // Hand twig the pre-edit caret before the splice, so the undo step it
2934 // retires carries where the caret was standing.
2935 self.record_caret();
2936 match self.editor.edit_range(start, end, text) {
2937 Ok(change) => {
2938 if coalesce {
2939 let _ = self.editor.coalesce_last_undo();
2940 }
2941 self.last_edit_kind = Some(kind);
2942 self.refresh();
2943 self.caret = change.new.end;
2944 self.anchor = None;
2945 self.goal_col = None;
2946 self.clear_pending();
2947 self.dirty = self.source != self.clean_source;
2948 self.status = None;
2949 // And the post-edit caret, so a later redo restores it.
2950 self.record_caret();
2951 true
2952 }
2953 // The edit was rolled back, so twig's history did not move and
2954 // neither may ours: pushing here would leave a step with no edit
2955 // under it and shift every later undo onto the wrong caret.
2956 Err(e) => {
2957 self.status = Some(format!("edit: {e}"));
2958 false
2959 }
2960 }
2961 }
2962
2963 /// The re-spelling that would keep the mark-edge rule for the edit
2964 /// `[start, end)` → `text`, or `None` when the edit leaves no whitespace
2965 /// against a delimiter and the plain splice is already right. Computed
2966 /// *before* the edit, while the run's spans and delimiters can still be read
2967 /// off the document; applied afterwards, and only if the mark really died —
2968 /// see [`repair_mark_edges`](Self::repair_mark_edges).
2969 ///
2970 /// Rich view only. Source view is for typing raw markup, where a space put
2971 /// against a `**` is exactly the character it looks like.
2972 fn mark_edge_fix(&mut self, start: usize, end: usize, text: &str) -> Option<MarkEdgeFix> {
2973 if self.view != View::Wysiwyg || start > end || end > self.source.len() {
2974 return None;
2975 }
2976 // Every inline mark standing over the edit, outermost first, with the
2977 // content span that says where its delimiters are.
2978 let chain: Vec<(InlineKind, std::ops::Range<usize>, std::ops::Range<usize>)> = self
2979 .editor
2980 .ancestors_at(start)
2981 .unwrap_or_default()
2982 .into_iter()
2983 .filter_map(|m| {
2984 let kind = inline_kind(&m.kind)?;
2985 let content = m.content_span.clone()?;
2986 Some((kind, m.span.clone(), content))
2987 })
2988 .collect();
2989 // The innermost run whose *content* holds the whole edit: the one whose
2990 // text is being changed, rather than one the edit merely sits under.
2991 let (kind, span, content) = chain
2992 .iter()
2993 .rev()
2994 .find(|(_, _, c)| c.start <= start && end <= c.end)?
2995 .clone();
2996 // What that content becomes. Whitespace at either end of it is what
2997 // would put out the mark.
2998 let body = format!(
2999 "{}{text}{}",
3000 &self.source[content.start..start],
3001 &self.source[end..content.end]
3002 );
3003 let (lead, trail) = if body.trim().is_empty() {
3004 // Nothing but whitespace left: there is no content to mark at all,
3005 // and the delimiters go with it rather than closing on a space.
3006 (body.len(), 0)
3007 } else {
3008 (
3009 body.len() - body.trim_start().len(),
3010 body.len() - body.trim_end().len(),
3011 )
3012 };
3013 // Nothing against a delimiter, and something still between them: the
3014 // plain edit stands. An emptied run is broken just as surely (`**b**`
3015 // with the `b` deleted is the literal `****`) and is re-spelt as the
3016 // nothing it now says.
3017 if lead == 0 && trail == 0 && !body.is_empty() {
3018 return None;
3019 }
3020 // Marks that open or close exactly where this one does — `***both***` is
3021 // two runs sharing an edge — spell their delimiters as one run of bytes,
3022 // so the whitespace has to clear all of them together.
3023 let (mut open_at, mut close_at) = (span.start, span.end);
3024 for _ in 0..chain.len() {
3025 match chain.iter().find(|(_, _, c)| c.start == open_at) {
3026 Some((_, s, _)) => open_at = s.start,
3027 None => break,
3028 }
3029 }
3030 for _ in 0..chain.len() {
3031 match chain.iter().find(|(_, _, c)| c.end == close_at) {
3032 Some((_, s, _)) => close_at = s.end,
3033 None => break,
3034 }
3035 }
3036 let open = &self.source[open_at..content.start];
3037 let close = &self.source[content.end..close_at];
3038 let core = &body[lead..body.len() - trail];
3039 let respelt = if core.is_empty() {
3040 body.clone()
3041 } else {
3042 format!(
3043 "{}{open}{core}{close}{}",
3044 &body[..lead],
3045 &body[body.len() - trail..]
3046 )
3047 };
3048 // The caret sits just past the inserted text within the new content —
3049 // which, when that lands in the whitespace, is now outside the delimiters.
3050 let pos = (start - content.start) + text.len();
3051 let caret = if core.is_empty() || pos <= lead {
3052 open_at + pos
3053 } else if pos >= lead + core.len() {
3054 open_at + lead + open.len() + core.len() + close.len() + (pos - lead - core.len())
3055 } else {
3056 open_at + lead + open.len() + (pos - lead)
3057 };
3058 Some(MarkEdgeFix {
3059 kind,
3060 probe: content.start,
3061 start: open_at,
3062 end: close_at + text.len() - (end - start),
3063 text: respelt,
3064 caret,
3065 // The marks in force here, resolved against any armed sticky delta —
3066 // what the writer is typing in, and so what has to still be true on
3067 // the far side of the delimiter the caret just stepped over.
3068 want: chain
3069 .iter()
3070 .filter(|(_, s, _)| start < s.end)
3071 .map(|(k, _, _)| *k)
3072 .collect::<InlineMarks>()
3073 .xor(self.pending_here()),
3074 })
3075 }
3076
3077 /// Apply a [`MarkEdgeFix`] — but only if the edit it was computed for really
3078 /// did break the mark. Whether whitespace at a delimiter is fatal is the
3079 /// format's business, not leaf's: `**bold **` is no longer strong, while
3080 /// `` `code ` `` is still perfectly good verbatim, and Djot's braced spellings
3081 /// don't care either. Asking the parser afterwards settles it for every kind
3082 /// and format at once, and costs a re-spelling only where one is due.
3083 ///
3084 /// The repair rides along with the edit that caused it — one undo step puts
3085 /// back what the writer typed, not a delimiter shuffle they never saw.
3086 fn repair_mark_edges(&mut self, fix: MarkEdgeFix) {
3087 if fix.end > self.source.len() {
3088 return;
3089 }
3090 if self.marks_at(fix.probe).iter().any(|(k, _)| *k == fix.kind) {
3091 return; // still a mark: these delimiters don't mind the whitespace
3092 }
3093 let resumed = self.last_edit_kind;
3094 if !self.splice_exact(fix.start, fix.end, &fix.text, EditKind::Other) {
3095 return;
3096 }
3097 let _ = self.editor.coalesce_last_undo();
3098 // The keystroke owns the undo step, so the run of typing it belongs to
3099 // keeps coalescing over the repair rather than breaking in two here.
3100 self.last_edit_kind = resumed;
3101 self.caret = fix.caret.min(self.source.len());
3102 self.anchor = None;
3103 self.goal_col = None;
3104 self.rearm(fix.want);
3105 self.clamp_caret();
3106 self.record_caret();
3107 }
3108
3109 /// Arm whatever sticky delta reproduces `want` at the caret — the marks the
3110 /// writer is typing in, carried across an edit that moved the caret out of
3111 /// the run holding them. Arms nothing when the caret already stands in
3112 /// exactly those marks, but still remembers the spot, so a further ⌘b starts
3113 /// a clean delta here (see [`toggle`](Self::toggle)).
3114 fn rearm(&mut self, want: InlineMarks) {
3115 let here: InlineMarks = self
3116 .marks_at(self.caret)
3117 .into_iter()
3118 .map(|(k, _)| k)
3119 .collect();
3120 self.pending_marks = want.xor(here);
3121 self.pending_at = Some(self.caret);
3122 }
3123
3124 /// Insert `text` at `at` as a *literal* run via twig's `insert_literal`,
3125 /// which backslash-escapes any character that would otherwise open markup in
3126 /// this format and position (`*` → `\*`, a line-start `#` → `\#`). The mirror
3127 /// of [`splice`](Self::splice) for the Hidden reveal mode's typing path, with
3128 /// the same caret re-anchor, coalescing, and rollback contract. `at` must be
3129 /// a collapsed point — a selection is deleted by the caller first, since
3130 /// `insert_literal` inserts rather than replaces.
3131 fn insert_literal_at(
3132 &mut self,
3133 at: usize,
3134 text: &str,
3135 kind: EditKind,
3136 force_coalesce: bool,
3137 ) -> bool {
3138 // `force_coalesce` folds this into the immediately preceding edit (the
3139 // selection-delete of an overwrite) so the pair is one undo step; else it
3140 // coalesces only when it continues a run of the same-kind typing.
3141 let coalesce =
3142 force_coalesce || (kind != EditKind::Other && self.last_edit_kind == Some(kind));
3143 // The mark-edge rule holds for typed text however it is spelled — see
3144 // `splice`. Only an insert twig passed through unchanged can use it,
3145 // since a fix is measured in the bytes that actually land, and an escape
3146 // adds bytes this couldn't have counted.
3147 let fix = self.mark_edge_fix(at, at, text);
3148 self.record_caret();
3149 match self.editor.insert_literal(at, text) {
3150 Ok(change) => {
3151 if coalesce {
3152 let _ = self.editor.coalesce_last_undo();
3153 }
3154 self.last_edit_kind = Some(kind);
3155 self.refresh();
3156 self.caret = change.new.end;
3157 self.anchor = None;
3158 self.goal_col = None;
3159 self.clear_pending();
3160 self.dirty = self.source != self.clean_source;
3161 self.status = None;
3162 self.record_caret();
3163 if let Some(fix) = fix.filter(|_| change.new.end - change.new.start == text.len()) {
3164 self.repair_mark_edges(fix);
3165 }
3166 true
3167 }
3168 Err(e) => {
3169 self.status = Some(format!("edit: {e}"));
3170 false
3171 }
3172 }
3173 }
3174
3175 /// After a structural list edit (a new item, a nest/unnest), renumber the
3176 /// ordered list the caret sits in so its source markers run `1, 2, 3, …`
3177 /// again — a raw splice leaves them stale (`1. 2. 2. 3.`). twig does the
3178 /// renumber as its own edit; fold it into the edit that triggered it so the
3179 /// two undo as one, and only when it actually changed the source (a no-op or
3180 /// a caret outside any ordered list must not coalesce the real edit into the
3181 /// step before it).
3182 fn renumber_here(&mut self) {
3183 self.renumber_at(self.caret);
3184 }
3185
3186 /// [`renumber_here`](Self::renumber_here) aimed somewhere other than the
3187 /// caret — for an edit that leaves the caret one past the item it just wrote,
3188 /// where twig resolves no list to renumber.
3189 fn renumber_at(&mut self, off: usize) {
3190 let before = self.source.clone();
3191 if self.editor.renumber_ordered_lists(off).is_err() {
3192 return; // not inside an ordered list — nothing to renumber
3193 }
3194 self.refresh();
3195 if self.source != before {
3196 let _ = self.editor.coalesce_last_undo();
3197 self.dirty = self.source != self.clean_source;
3198 self.clamp_caret();
3199 self.record_caret();
3200 }
3201 }
3202
3203 /// Repair the one trap a list edit can spring on itself. An *empty* `-`
3204 /// sub-item written directly beneath a text line reparses that text as a
3205 /// setext heading — `- hello\n - ` is `<h2>hello</h2>`, because a lone `-`
3206 /// is also a setext-H2 underline (twig is right; pandoc agrees). `*` and `+`
3207 /// bullets can't underline anything, so swap the dash for a `*`: the item
3208 /// stays an empty nested bullet, the parent stays prose, and the source
3209 /// round-trips instead of hiding a heading the user never asked for. Folded
3210 /// into the triggering edit's undo step, the way renumbering is.
3211 ///
3212 /// Gated on the collapse having actually happened (the swapped dash was
3213 /// swallowed into a `heading`), so a real setext heading the author wrote —
3214 /// or a `- x` with content, which can't underline anything — is never
3215 /// touched. This has to live in the *edit*, not the renderer: leaving the
3216 /// hazardous bytes on disk and only painting over them would ship a file
3217 /// every other CommonMark tool reads as a heading.
3218 ///
3219 /// This one keeps its own byte scan, and has to: the hazard is precisely
3220 /// that the dash stopped being a list marker, so [`list_marker_on_line`] —
3221 /// which asks twig which lines open an item — reports nothing here. There is
3222 /// no node to ask about. It is also the last Markdown spelling leaf writes on
3223 /// purpose rather than for want of an answer; once twig spells continuations
3224 /// itself, avoiding the trap becomes twig's, and this goes.
3225 ///
3226 /// [`list_marker_on_line`]: Self::list_marker_on_line
3227 fn avoid_setext_collapse(&mut self) {
3228 let caret = self.caret.min(self.source.len());
3229 let line_start = self.source[..caret].rfind('\n').map_or(0, |i| i + 1);
3230 let bytes = self.source.as_bytes();
3231 let mut dash = line_start;
3232 while matches!(bytes.get(dash), Some(b' ' | b'\t')) {
3233 dash += 1;
3234 }
3235 // A dash bullet is the only marker that doubles as a setext underline.
3236 if bytes.get(dash) != Some(&b'-') {
3237 return;
3238 }
3239 // Only an *empty* item is a bare underline; `- x` carries content and
3240 // can't fold the line above into a heading.
3241 let line_end = self.source[dash..]
3242 .find('\n')
3243 .map_or(self.source.len(), |i| dash + i);
3244 if !self.source[dash + 1..line_end].trim().is_empty() {
3245 return;
3246 }
3247 // The tell: that dash was swallowed into a `heading`. A properly nested
3248 // empty item sits under a `list_item`, with no heading in reach. Probe
3249 // the dash byte itself (well inside the heading), not the caret, whose
3250 // end-of-line offset can fall on the half-open span boundary.
3251 let collapsed = self
3252 .editor
3253 .ancestors_at(dash)
3254 .map(|c| c.into_iter().any(|m| m.kind == Kind::Heading))
3255 .unwrap_or(false);
3256 if !collapsed {
3257 return;
3258 }
3259 let caret = self.caret;
3260 if self.splice(dash, dash + 1, "*", EditKind::Other) {
3261 // Same width, so the caret keeps its column; fold into the edit that
3262 // triggered this so Tab stays one undo step.
3263 let _ = self.editor.coalesce_last_undo();
3264 self.caret = caret.min(self.source.len());
3265 self.clamp_caret();
3266 self.record_caret();
3267 }
3268 }
3269
3270 fn snapshot(&self) -> CaretState {
3271 CaretState {
3272 caret: self.caret,
3273 anchor: self.anchor,
3274 }
3275 }
3276
3277 /// Hand twig the current caret and selection as the blob for the live
3278 /// document state. Called before an edit — so the step twig retires records
3279 /// where the caret was, and undo can restore it — and again once the op has
3280 /// placed the caret, so redo restores where the edit left it.
3281 ///
3282 /// This is the whole of leaf's undo-caret bookkeeping now. twig carries the
3283 /// caret through its own history, so coalescing falls out for free (folding
3284 /// two twig steps into one drops the intermediate blob, keeping the run's
3285 /// first) and the parallel stacks that had to march in lockstep — and could
3286 /// silently drift out of it — are gone.
3287 fn record_caret(&mut self) {
3288 let _ = self.editor.set_caret_blob(&self.snapshot().to_blob());
3289 }
3290
3291 /// Toggle an inline mark over the selection (Bold / Italic / Code / …). Keeps
3292 /// the toggled region selected so a second press cleanly reverses it.
3293 pub fn toggle(&mut self, kind: InlineKind) {
3294 // Ahead of the no-selection branch below: arming a mark for text not yet
3295 // typed is a promise `insert` cannot keep in a format with no delimiters
3296 // to spell it with. Per *kind*, not per format — Markdown spells three
3297 // of the eight marks, djot all eight, HTML seven.
3298 if self.refuse_unsupported(&format!("{kind:?}"), Gesture::ToggleInline(kind)) {
3299 return;
3300 }
3301 let Some((s, e)) = self.selection() else {
3302 // No selection: arm the mark for the next text typed here, the way a
3303 // word processor does. `⌘b`, type, `⌘b` again toggles bold on and off
3304 // in the flow of typing without ever selecting anything — the delta
3305 // is realised onto the freshly typed text by `insert`. A fresh caret
3306 // position starts the delta over from the marks actually in force.
3307 if self.pending_at != Some(self.caret) {
3308 self.pending_marks = InlineMarks::empty();
3309 self.pending_at = Some(self.caret);
3310 }
3311 self.pending_marks.flip(kind);
3312 self.status = None;
3313 return;
3314 };
3315 // Whitespace at the edge of a selection is not part of what was chosen —
3316 // a double-click takes the space after the word with it — and a mark
3317 // cannot close against one anyway: `**word **` is four literal asterisks
3318 // (the mark-edge rule, see `splice`). Mark the words, leave the spaces.
3319 let picked = &self.source[s..e];
3320 let (s, e) = (
3321 s + (picked.len() - picked.trim_start().len()),
3322 e - (picked.len() - picked.trim_end().len()),
3323 );
3324 if s >= e {
3325 self.status = Some(format!("{kind:?}: nothing selected to mark"));
3326 return;
3327 }
3328 // Styling a selection is a one-shot act, not a sticky mode.
3329 self.clear_pending();
3330 self.record_caret();
3331 match self.editor.toggle_inline(s, e, kind) {
3332 Ok(change) => {
3333 self.last_edit_kind = None; // structural edit is its own undo step
3334 self.refresh();
3335 self.anchor = Some(change.new.start);
3336 self.caret = change.new.end;
3337 self.dirty = self.source != self.clean_source;
3338 self.status = None;
3339 self.record_caret();
3340 }
3341 Err(e) => self.status = Some(format!("{kind:?}: {e}")),
3342 }
3343 }
3344
3345 /// Convert the block at the caret to a heading level or paragraph.
3346 pub fn set_block(&mut self, kind: BlockKind) {
3347 if self.refuse_unsupported(&format!("{kind:?}"), Gesture::SetBlock) {
3348 return;
3349 }
3350 self.record_caret();
3351 // A blank line has no node to convert, and twig opens a block there
3352 // rather than declining — so the caret's own offset is the right thing
3353 // to hand it when `block_offset_for_caret` finds nothing.
3354 let offset = self.block_offset_for_caret().unwrap_or(self.caret);
3355 match self.editor.set_block(offset, kind) {
3356 Ok(change) => {
3357 self.last_edit_kind = None;
3358 self.refresh();
3359 // Opening a block on a blank line writes a marker the caret
3360 // belongs *after*; converting an existing one moves nothing.
3361 self.caret = self.caret.max(change.new.end);
3362 self.clamp_caret();
3363 self.anchor = None;
3364 self.dirty = self.source != self.clean_source;
3365 self.status = None;
3366 self.record_caret();
3367 }
3368 Err(e) => self.status = Some(format!("{kind:?}: {e}")),
3369 }
3370 }
3371
3372 /// Whether `off` is inside a text block (paragraph, heading, code block…).
3373 fn has_block_at(&mut self, off: usize) -> bool {
3374 self.editor.ancestors_at(off).ok().is_some_and(|chain| {
3375 chain
3376 .iter()
3377 .any(|m| !wysiwyg::is_inline_kind(&m.kind) && !is_block_container(&m.kind))
3378 })
3379 }
3380
3381 /// The offset to hand twig's `set_block`: the caret when it is already inside
3382 /// a block, otherwise nudged onto the previous character (a caret at a line
3383 /// end sits at the doc level, outside the block). `None` when the caret is on
3384 /// a blank line — a new paragraph with no block node to convert.
3385 fn block_offset_for_caret(&mut self) -> Option<usize> {
3386 let caret = self.caret.min(self.source.len());
3387 if self.has_block_at(caret) {
3388 return Some(caret);
3389 }
3390 // Nudge to the previous character — but never across a newline: that would
3391 // target the previous block, and a blank line genuinely has no block.
3392 if let Some((i, ch)) = self.source[..caret].char_indices().next_back()
3393 && ch != '\n'
3394 && self.has_block_at(i)
3395 {
3396 return Some(i);
3397 }
3398 None
3399 }
3400
3401 /// The heading level of the text block at the caret, or `None` when that
3402 /// block is not a heading.
3403 pub fn current_heading_level(&mut self) -> Option<u32> {
3404 let caret = self.caret;
3405 self.nodes()
3406 .into_iter()
3407 .filter(|n| n.kind == Kind::Heading)
3408 .find(|n| n.span.start <= caret && caret <= n.span.end)
3409 .and_then(|n| n.level)
3410 }
3411
3412 /// The inline marks in force at the caret (or over the selection) — what a
3413 /// toolbar draws lit, and the block-level [`Doc::current_heading_level`]'s
3414 /// inline counterpart. Cheap enough to call every frame: one twig
3415 /// `ancestors_at` query per caret (two with a selection), each walking root
3416 /// → deepest node at one offset. It never snapshots the tree the way
3417 /// `current_heading_level` does, and the returned set is a `Copy` bitset, so
3418 /// the only allocation is twig's own small ancestor `Vec`.
3419 ///
3420 /// **A selection reports a mark only when the mark covers *all* of it.**
3421 /// That's what every real toolbar means by an active button — Bold lit over
3422 /// a half-bold selection would claim a press turns bold *off*, when
3423 /// [`Doc::toggle`] hands the range to twig and gets the whole thing bolded.
3424 /// Whole-coverage is asked as "is the same mark node standing over both the
3425 /// first and the last character?": inline nodes are contiguous, so one node
3426 /// covering both ends covers every byte between them. Two touching runs
3427 /// (`**a****b**`) are two nodes, and correctly light nothing.
3428 ///
3429 /// At a bare caret a mark is active when the caret stands inside the mark's
3430 /// span — `span.start <= caret < span.end`, delimiters included, which is
3431 /// what makes the boundaries behave. In `a **bold** b` the offsets from the
3432 /// opening `*` (2) through the last byte of the closing `**` (9) are all
3433 /// bold, so the WYSIWYG caret both before `b` and after `d` (the delimiters
3434 /// are hidden, and those offsets are 4 and 8) reports bold — matching where
3435 /// typing would actually land inside the marked run. The offset one past the
3436 /// mark (10) is the text after it and reports nothing, at the end of the
3437 /// buffer exactly as in the middle.
3438 pub fn active_inline_marks(&mut self) -> InlineMarks {
3439 let Some((start, end)) = self.selection() else {
3440 // The marks actually in force at the caret, flipped by any armed
3441 // sticky delta — so `⌘b` at a bare caret lights the Bold button
3442 // immediately, before a single character is typed.
3443 let base: InlineMarks = self
3444 .marks_at(self.caret)
3445 .into_iter()
3446 .map(|(k, _)| k)
3447 .collect();
3448 return base.xor(self.pending_here());
3449 };
3450 // The selection's *last character*, not its exclusive end: `end` is the
3451 // offset one past the selection, which for a selection ending exactly at
3452 // a mark's close is already outside it (`[4,10)` of `a **bold** b` is
3453 // entirely bold, but offset 10 is the space after).
3454 let last = prev_boundary(&self.source, end);
3455 let head = self.marks_at(start);
3456 let tail = self.marks_at(last);
3457 head.into_iter()
3458 .filter(|m| tail.contains(m))
3459 .map(|(k, _)| k)
3460 .collect()
3461 }
3462
3463 /// The inline marks whose span covers `off`, each with the id of the node
3464 /// carrying it — the id is what lets a selection tell one mark node from
3465 /// another of the same kind.
3466 fn marks_at(&mut self, off: usize) -> Vec<(InlineKind, u32)> {
3467 let off = off.min(self.source.len());
3468 self.editor
3469 .ancestors_at(off)
3470 .unwrap_or_default()
3471 .into_iter()
3472 // `span.end` is the offset one *past* the mark, so it isn't in it.
3473 // twig already resolves a boundary to whatever starts there — in
3474 // `**bold** x` offset 8 is the following text, not the strong — but
3475 // when nothing follows, the tie has nobody to break for and the
3476 // chain still ends at the mark. That would make the answer at the
3477 // last offset of the document depend on whether the file happens to
3478 // end in a newline; the rule is `span.start <= off < span.end`, and
3479 // it's the same rule at the end of a buffer as in the middle.
3480 .filter(|m| off < m.span.end)
3481 .filter_map(|m| inline_kind(&m.kind).map(|k| (k, m.node_id)))
3482 .collect()
3483 }
3484
3485 /// Toggle a heading at the caret: if the block is already this heading level,
3486 /// revert it to a paragraph; otherwise convert it to this heading level.
3487 /// This gives the heading commands the same toggle feel as bold/italic/code —
3488 /// re-applying a heading a line already has turns it back into body text.
3489 pub fn toggle_heading(&mut self, level: u32) {
3490 if self.current_heading_level() == Some(level) {
3491 self.set_block(BlockKind::Paragraph);
3492 } else {
3493 self.set_block(BlockKind::Heading(level));
3494 }
3495 }
3496
3497 /// Toggle a block quote around the selection, or around the block at the
3498 /// caret — the toolbar's Quote button.
3499 pub fn toggle_blockquote(&mut self) {
3500 self.toggle_container(BlockContainerKind::BlockQuote);
3501 }
3502
3503 /// Toggle a numbered (`ordered`) or bulleted list over the selection, or
3504 /// over the block at the caret — one op with the kind as a flag, the way
3505 /// `toggle_heading` takes its level, so a frontend needs no twig type to
3506 /// name the two buttons.
3507 ///
3508 /// Pressing the *other* list's button while in a list converts in place
3509 /// rather than nesting, so the pair reads as one three-state control
3510 /// (bulleted / numbered / neither) rather than two independent wrappers.
3511 pub fn toggle_list(&mut self, ordered: bool) {
3512 self.toggle_container(if ordered {
3513 BlockContainerKind::OrderedList
3514 } else {
3515 BlockContainerKind::BulletList
3516 });
3517 }
3518
3519 // ── Task list items ──────────────────────────────────────────────────────
3520 // The checkbox in `- [x] done`. twig owns all three gestures: the box is
3521 // inline content of the item's first paragraph rather than part of its
3522 // marker, so adding or removing one must leave the item's continuation
3523 // indentation alone, and an item inside a quote is found past the quote
3524 // markers. leaf names the gesture and the offset; the spelling is twig's.
3525
3526 /// Whether the list item at the caret carries a checkbox, and which way it
3527 /// faces — `Some(true)` ticked, `Some(false)` empty, `None` for a plain list
3528 /// item or no item at all. What a toolbar reads to light its checkbox button.
3529 pub fn task_checked_at_caret(&mut self) -> Option<bool> {
3530 self.task_checked_at(self.caret)
3531 }
3532
3533 /// [`task_checked_at_caret`](Self::task_checked_at_caret) for an arbitrary
3534 /// offset — what a frontend asks before deciding a click landed on a box.
3535 pub fn task_checked_at(&mut self, offset: usize) -> Option<bool> {
3536 self.innermost_list_item(offset.min(self.source.len()))?
3537 .checked
3538 }
3539
3540 /// Tick or untick the task item at the caret (the checkbox's keyboard half).
3541 /// A no-op with a reported reason when the caret is in no task item — minting
3542 /// a box here is [`toggle_task_item`](Self::toggle_task_item)'s job.
3543 pub fn toggle_task_checked(&mut self) {
3544 self.toggle_task_at(self.caret);
3545 }
3546
3547 /// Tick or untick the task item covering `offset` — what a *click* on a
3548 /// rendered checkbox is. Separate from the caret form because a click carries
3549 /// its own offset and must not first move the caret there: ticking a box
3550 /// three paragraphs away should not take the cursor with it.
3551 pub fn toggle_task_at(&mut self, offset: usize) {
3552 if self.refuse_unsupported("task", Gesture::ToggleTaskChecked) {
3553 return;
3554 }
3555 let offset = offset.min(self.source.len());
3556 self.record_caret();
3557 match self.editor.toggle_task_checked(offset) {
3558 Ok(_) => self.after_task_edit(),
3559 Err(e) => self.status = Some(format!("task: {e}")),
3560 }
3561 }
3562
3563 /// Give the list item at the caret a checkbox, or take its checkbox away —
3564 /// the gesture that converts between a plain bullet and a task. A new box
3565 /// arrives unticked.
3566 pub fn toggle_task_item(&mut self) {
3567 if self.refuse_unsupported("task", Gesture::ToggleTaskItem) {
3568 return;
3569 }
3570 let caret = self.caret.min(self.source.len());
3571 self.record_caret();
3572 match self.editor.toggle_task_item(caret) {
3573 Ok(_) => self.after_task_edit(),
3574 Err(e) => self.status = Some(format!("task: {e}")),
3575 }
3576 }
3577
3578 /// Settle after a task gesture. The caret rides its old byte offset and is
3579 /// clamped back in: a box is three or four bytes on the item's first line, so
3580 /// text after it shifts by that much at most, and `clamp_caret` lands it on a
3581 /// real stop either way.
3582 fn after_task_edit(&mut self) {
3583 self.last_edit_kind = None;
3584 self.refresh();
3585 self.anchor = None;
3586 self.dirty = self.source != self.clean_source;
3587 self.status = None;
3588 self.clamp_caret();
3589 self.record_caret();
3590 }
3591
3592 // ── Tables ───────────────────────────────────────────────────────────────
3593 // A table is a grid, and twig edits it as one — add/remove/move a row or
3594 // column, set a column's alignment — re-spelling the whole table in a single
3595 // splice. Every gesture is anchored at the caret's cell. leaf just names the
3596 // gesture and re-reads the result; the whole table's numbering, borders, and
3597 // delimiter are twig's to keep straight.
3598
3599 /// Whether the caret is inside a table — what a frontend asks to enable or
3600 /// disable its table controls.
3601 ///
3602 /// An HTML `<table>` still answers `true`: the caret really is in a table,
3603 /// and the reason the grid controls stay dark there is
3604 /// [`Capabilities::table`], which is a fact about the document's format
3605 /// rather than about the caret. A frontend needs both.
3606 pub fn caret_in_table(&mut self) -> bool {
3607 let caret = self.caret.min(self.source.len());
3608 self.editor
3609 .ancestors_at(caret)
3610 .map(|c| c.into_iter().any(|m| m.kind == Kind::Table))
3611 .unwrap_or(false)
3612 }
3613
3614 /// One grid op, guarded and settled — the shared body of the seven below.
3615 ///
3616 /// The guard is why this exists rather than seven copies of the same three
3617 /// lines, and it is the one guard leaf cannot delegate to twig. The table
3618 /// editor is the gesture family that consults no `Syntax` table (it spells a
3619 /// grid, not a delimiter) and therefore the one twig's `Format::supports`
3620 /// deliberately has no variant for: handed an HTML `<table>` it rebuilds the
3621 /// grid as a *pipe table* and reports success, swapping the element out for
3622 /// `| a | b |` and taking the rest of the document's markup with it. Nothing
3623 /// downstream could tell that from a successful edit — the splice is real,
3624 /// the reparse succeeds, `dirty` is honest — which is what makes it worth
3625 /// stopping at the door rather than detecting after the fact. See
3626 /// [`spells_pipe_tables`].
3627 fn table_op(
3628 &mut self,
3629 what: &str,
3630 op: impl FnOnce(&mut Editor, usize) -> Result<(), twig::Error>,
3631 ) {
3632 if self.refuse_unless(what, spells_pipe_tables(self.format)) {
3633 return;
3634 }
3635 self.record_caret();
3636 let at = self.caret;
3637 let r = op(&mut self.editor, at);
3638 self.apply_table(r, what);
3639 }
3640
3641 /// Insert an empty row below (`below`) or above the caret's row.
3642 pub fn table_insert_row(&mut self, below: bool) {
3643 self.table_op("table row", |e, at| e.table_insert_row(at, below));
3644 }
3645
3646 /// Delete the caret's row (not the header, not the last body row).
3647 pub fn table_delete_row(&mut self) {
3648 self.table_op("table row", |e, at| e.table_delete_row(at));
3649 }
3650
3651 /// Insert an empty column right (`right`) or left of the caret's column.
3652 pub fn table_insert_column(&mut self, right: bool) {
3653 self.table_op("table column", |e, at| e.table_insert_column(at, right));
3654 }
3655
3656 /// Delete the caret's column (unless it is the only one).
3657 pub fn table_delete_column(&mut self) {
3658 self.table_op("table column", |e, at| e.table_delete_column(at));
3659 }
3660
3661 /// Set the caret's column to `alignment`.
3662 pub fn table_set_alignment(&mut self, alignment: Alignment) {
3663 self.table_op("table alignment", |e, at| {
3664 e.table_set_alignment(at, alignment)
3665 });
3666 }
3667
3668 /// Move the caret's row one place down (`down`) or up, within the body rows.
3669 pub fn table_move_row(&mut self, down: bool) {
3670 self.table_op("table row", |e, at| e.table_move_row(at, down));
3671 }
3672
3673 /// Move the caret's column one place right (`right`) or left.
3674 pub fn table_move_column(&mut self, right: bool) {
3675 self.table_op("table column", |e, at| e.table_move_column(at, right));
3676 }
3677
3678 /// Settle the caret and document flags after a table op (or report its
3679 /// error). twig re-spells the whole table, so the caret rides its old byte
3680 /// offset and is clamped back into the rebuilt bytes — near enough to where
3681 /// it was, since the op preserves the cells' content and order around it.
3682 fn apply_table(&mut self, result: Result<(), twig::Error>, what: &str) {
3683 match result {
3684 Ok(()) => {
3685 self.last_edit_kind = None;
3686 self.refresh();
3687 self.anchor = None;
3688 self.clamp_caret();
3689 self.dirty = self.source != self.clean_source;
3690 self.status = None;
3691 self.record_caret();
3692 }
3693 Err(e) => self.status = Some(format!("{what}: {e}")),
3694 }
3695 }
3696
3697 /// One `toggle_block_container` over the block-level target.
3698 ///
3699 /// leaf says *where*; twig decides everything else — which blocks the range
3700 /// covers, whether that means wrapping, unwrapping, nesting or converting,
3701 /// and how this document's format spells the prefix. The rule that a
3702 /// container only comes off when the range covers every block it holds is
3703 /// what the re-anchoring below is built around.
3704 fn toggle_container(&mut self, kind: BlockContainerKind) {
3705 if self.refuse_unsupported(&format!("{kind:?}"), Gesture::ToggleBlockContainer(kind)) {
3706 return;
3707 }
3708 let selected = self.selection();
3709 // A blank line holds no block, and twig opens an *empty* container on one
3710 // — since 3.2.0; it used to decline the range with `NotFound`, which is
3711 // why this used to lend it a scratch paragraph to wrap. Worth knowing
3712 // here because the line-for-line caret mapping below cannot describe it:
3713 // opening one under a paragraph writes the blank line the format needs
3714 // above the marker too, so the rewritten region has a line the old one
3715 // didn't, and "the same line, the same distance from its end" lands on
3716 // that new blank instead of in the container.
3717 let opened_empty = selected.is_none() && self.block_offset_for_caret().is_none();
3718 // Without a selection the target is the caret's own block, resolved the
3719 // way `set_block` resolves it — a caret at a line end sits at the doc
3720 // level and has to be nudged back onto the block it looks like it's in.
3721 // An empty range is enough: twig widens to the whole lines it touches.
3722 let (start, end) = match selected {
3723 Some(range) => range,
3724 None => {
3725 let off = self.block_offset_for_caret().unwrap_or(self.caret);
3726 (off, off)
3727 }
3728 };
3729 self.record_caret();
3730 match self.editor.toggle_block_container(start, end, kind) {
3731 Ok(change) => {
3732 // Read the caret's place out of the *pre-edit* source, before
3733 // `refresh` swaps that source out from under it.
3734 let place = (selected.is_none() && !opened_empty)
3735 .then(|| self.caret_line_tail(&change.old));
3736 self.last_edit_kind = None; // structural edit is its own undo step
3737 self.refresh();
3738 match place {
3739 // Both land the caret at the far end of what twig wrote, and
3740 // differ only in what they leave selected.
3741 //
3742 // From a selection: select what the container now holds, the
3743 // way `toggle` keeps its marked region selected — and for a
3744 // stronger reason than symmetry: a container comes *off* only
3745 // a range covering every block it holds, so a selection left
3746 // on its old bytes (now short by a prefix per line) would nest
3747 // on the second press instead of reversing the first.
3748 //
3749 // From a blank line: nothing to select, and the end of the
3750 // region is exactly past the bare `> ` / `- ` twig wrote —
3751 // the caret standing inside the container that was asked for.
3752 None => {
3753 self.anchor = (!opened_empty).then_some(change.new.start);
3754 self.caret = change.new.end;
3755 }
3756 Some(place) => {
3757 self.anchor = None;
3758 self.caret = self.line_tail_offset(&change.new, place);
3759 }
3760 }
3761 self.dirty = self.source != self.clean_source;
3762 self.status = None;
3763 self.clamp_caret();
3764 self.record_caret();
3765 }
3766 Err(e) => self.status = Some(format!("{kind:?}: {e}")),
3767 }
3768 }
3769
3770 /// The caret's place inside the region a container toggle is rewriting, in
3771 /// the only terms the rewrite preserves: which of the region's lines it sits
3772 /// on, and how many bytes of that line lie ahead of it.
3773 ///
3774 /// A container's markup goes in at column 0 and never touches what follows
3775 /// on the line, so that pair survives the edit exactly where a byte offset
3776 /// does not — a caret left on its old offset slides back by one prefix per
3777 /// line above it, which on a hard-wrapped paragraph parks it *inside* the
3778 /// `> ` it just asked for.
3779 fn caret_line_tail(&self, old: &std::ops::Range<usize>) -> (usize, usize) {
3780 let caret = self.caret.clamp(old.start, old.end);
3781 let line = self.source[old.start..caret].matches('\n').count();
3782 let end = self.source[caret..old.end]
3783 .find('\n')
3784 .map_or(old.end, |i| caret + i);
3785 (line, end - caret)
3786 }
3787
3788 /// [`caret_line_tail`](Self::caret_line_tail) undone against the rewritten
3789 /// region: the offset `tail` bytes back from the end of the region's `line`.
3790 ///
3791 /// Both walks are clamped rather than trusted, because the one op that does
3792 /// *not* keep a region's lines one-to-one is stripping a list — twig blows
3793 /// the items back apart with blank lines between them — and a caret landing
3794 /// on the nearest line of the right item beats one landing out of the region
3795 /// entirely.
3796 fn line_tail_offset(
3797 &self,
3798 new: &std::ops::Range<usize>,
3799 (line, tail): (usize, usize),
3800 ) -> usize {
3801 let region = &self.source[new.start.min(self.source.len())..new.end.min(self.source.len())];
3802 let mut start = 0;
3803 for _ in 0..line {
3804 match region[start..].find('\n') {
3805 Some(i) => start += i + 1,
3806 None => break,
3807 }
3808 }
3809 let end = region[start..]
3810 .find('\n')
3811 .map_or(region.len(), |i| start + i);
3812 new.start + end.saturating_sub(tail).max(start)
3813 }
3814
3815 /// Link the selection to `destination` — the toolbar's Link button. With no
3816 /// selection it acts at the caret, which re-points a link the caret is
3817 /// already standing in (twig replaces an existing link's destination and
3818 /// keeps its text) and otherwise spells a link that has no text of its own:
3819 /// an autolink (`<https://x.dev>`) where the destination is one, and
3820 /// `[destination](destination)` where it isn't.
3821 ///
3822 /// `destination` reaches twig raw. Escaping it is format knowledge and the
3823 /// two formats genuinely disagree — Markdown ends a destination at the first
3824 /// space and moves it into `<…>`, djot reads that `<…>` as part of the URL
3825 /// itself — so the side holding the document is the side that gets to spell
3826 /// it. A destination twig can't carry at all (one with a newline) comes back
3827 /// as an error rather than a quietly rewritten URL.
3828 pub fn insert_link(&mut self, destination: &str) {
3829 if self.refuse_unsupported("link", Gesture::InsertLink) {
3830 return;
3831 }
3832 let (start, end) = self.selection().unwrap_or((self.caret, self.caret));
3833 self.record_caret();
3834 match self.editor.insert_link(start, end, destination) {
3835 Ok(change) => {
3836 self.last_edit_kind = None;
3837 self.refresh();
3838 match self.link_text_span(change.new.start) {
3839 // A link with text of its own: select it, so typing replaces
3840 // a `[dest](dest)`'s stand-in label and a second press
3841 // re-points what the first one linked.
3842 Some(text) => {
3843 self.anchor = (text.start != text.end).then_some(text.start);
3844 self.caret = text.end;
3845 }
3846 // An autolink is finished the moment it's written — its text
3847 // *is* the URL. Leaving it selected would aim the next press
3848 // at the one shape twig still wraps instead of re-points.
3849 None => {
3850 self.anchor = None;
3851 self.caret = change.new.end;
3852 }
3853 }
3854 self.dirty = self.source != self.clean_source;
3855 self.status = None;
3856 self.clamp_caret();
3857 self.record_caret();
3858 }
3859 Err(e) => self.status = Some(format!("link: {e}")),
3860 }
3861 }
3862
3863 /// Insert a block-level image at the caret: ``. Any
3864 /// selection becomes the alt text (so "select a caption, insert image" labels
3865 /// it); with no selection, `alt` is used — empty for none. The caret lands
3866 /// just past the inserted image.
3867 ///
3868 /// Both halves go through twig (`insert_literal` for the alt text,
3869 /// `insert_image` for the image), so neither is spelled here. That used to be a
3870 /// `format!`, and it was wrong the first time an app inserted a real filename:
3871 /// Markdown ends a destination at the first space, so `` is
3872 /// not an image at all — and the fix is per-format, since moving into the
3873 /// `<…>` form is exactly wrong for Djot, where `<…>` becomes the URL itself.
3874 pub fn insert_image(&mut self, destination: &str, alt: &str) {
3875 if self.refuse_unsupported("image", Gesture::InsertImage) {
3876 return;
3877 }
3878 let (start, end) = self.selection().unwrap_or((self.caret, self.caret));
3879 self.record_caret();
3880 // With no selection and an explicit `alt`, the alt text has to exist in the
3881 // document before it can be the image's — and it is raw caller input, so
3882 // it goes in through `insert_literal`, which escapes it for the format
3883 // rather than letting a `]` in someone's caption close the image early.
3884 let (start, end) = if start == end && !alt.is_empty() {
3885 match self.editor.insert_literal(start, alt) {
3886 Ok(change) => (change.new.start, change.new.end),
3887 Err(e) => {
3888 self.status = Some(format!("image: {e}"));
3889 return;
3890 }
3891 }
3892 } else {
3893 (start, end)
3894 };
3895 match self.editor.insert_image(start, end, destination) {
3896 Ok(change) => {
3897 self.last_edit_kind = None;
3898 self.refresh();
3899 // Just past the image, nothing selected — where a caret belongs
3900 // after inserting one.
3901 self.anchor = None;
3902 self.caret = change.new.end;
3903 self.dirty = self.source != self.clean_source;
3904 self.status = None;
3905 self.clamp_caret();
3906 self.record_caret();
3907 }
3908 Err(e) => self.status = Some(format!("image: {e}")),
3909 }
3910 }
3911
3912 /// Insert a block-level image, video, or audio at the caret. The image case
3913 /// is [`insert_image`](Self::insert_image); video and audio are spelled as
3914 /// HTML elements, which is the only spelling Markdown and Djot have for them:
3915 ///
3916 /// ```text
3917 /// <video src="clip.mp4" controls>alt</video>
3918 /// <audio src="take.mp3" controls>alt</audio>
3919 /// ```
3920 ///
3921 /// HTML rather than a `::video{…}` directive deliberately. A directive means
3922 /// something only to an app that knows the vocabulary, so the document would
3923 /// read as literal punctuation everywhere else; `<video>` is what every other
3924 /// renderer already understands, and what leaf's own reader picks back up
3925 /// through `html_elements` promotion (see [`parse_extensions`]).
3926 ///
3927 /// The one-line spelling needs twig ≥ 2.5.1, which widened CommonMark's
3928 /// HTML-block tag list to cover `<video>`/`<audio>`/`<picture>` under
3929 /// `html_elements`. Before that only the multi-line form parsed as a block at
3930 /// all, and this wrote three lines to work around it.
3931 ///
3932 /// `controls` is always written: a player with no transport is a still frame
3933 /// the reader can't do anything with. Any selection becomes the element's
3934 /// fallback text, exactly as it becomes an image's alt.
3935 ///
3936 /// The same verbatim-insertion caveat as [`insert_image`](Self::insert_image)
3937 /// applies, and bites harder here: a `"` in `destination` closes the
3938 /// attribute. A frontend taking these from a file picker is fine; one taking
3939 /// them from free text should keep them tame.
3940 ///
3941 /// [`MediaInfo`]: crate::MediaInfo
3942 pub fn insert_media(&mut self, kind: MediaKind, destination: &str, alt: &str) {
3943 if kind == MediaKind::Image {
3944 return self.insert_image(destination, alt);
3945 }
3946 // Gated on the *image* gesture, not on one of its own — there isn't one,
3947 // since the bytes below are spelled here rather than by twig, and an HTML
3948 // document would in fact parse them. The button is one control with three
3949 // kinds behind it, and two of them working in a format where the third
3950 // cannot is a worse surface than three that agree — especially as
3951 // `insert_image` is the kind anyone reaches for first.
3952 if self.refuse_unsupported("media", Gesture::InsertImage) {
3953 return;
3954 }
3955 let (start, end) = self.selection().unwrap_or((self.caret, self.caret));
3956 let alt_text = self
3957 .selected_text()
3958 .map(str::to_string)
3959 .unwrap_or_else(|| alt.to_string());
3960 let tag = match kind {
3961 MediaKind::Audio => "audio",
3962 _ => "video",
3963 };
3964 let markup = format!("<{tag} src=\"{destination}\" controls>{alt_text}</{tag}>");
3965 self.edit(start, end, &markup);
3966 }
3967
3968 /// Insert a thematic break at the caret — the toolbar's Horizontal Rule
3969 /// button. Spelling and placement are both twig's; leaf used to write `---`
3970 /// itself, which was the Markdown spelling in a djot document too.
3971 ///
3972 /// A rule is a block, so `insert_thematic_break` alone has nowhere to put one
3973 /// mid-paragraph and lands it after the caret's whole block. To get a rule
3974 /// *at* the caret — the paragraph parted in two around it, which is what a
3975 /// rule button is understood to do — the paragraph is first divided with
3976 /// `split_block` and the rule then aimed at the **first** half. Aiming it at
3977 /// the offset `split_block` returns puts the rule after the *second* half
3978 /// instead, which is a rule in the right document and the wrong place.
3979 ///
3980 /// Only a plain paragraph is split. Everywhere else the rule simply lands
3981 /// after the block, which is both twig's own answer and the better one:
3982 /// splitting a fenced code block would leave two fences with a rule between
3983 /// them, and splitting a list item would mint an item nobody asked for on the
3984 /// way to a rule that lands after the list regardless. A table and a setext
3985 /// heading refuse the split outright, so they take the same path by
3986 /// themselves.
3987 pub fn insert_thematic_break(&mut self) {
3988 if self.refuse_unsupported("thematic break", Gesture::InsertThematicBreak) {
3989 return;
3990 }
3991 self.caret = self.skip_trailing_close_delims(self.caret);
3992 // A selection is replaced by the rule, so collapse it first and let the
3993 // split-and-rule below run from the caret it leaves behind.
3994 if let Some((s, e)) = self.selection() {
3995 self.splice(s, e, "", EditKind::Other);
3996 }
3997 self.anchor = None;
3998 self.record_caret();
3999 let at = self.caret;
4000 if self.caret_in_bare_paragraph() {
4001 // A failure here is not fatal: the rule still lands after the block,
4002 // which is exactly what this call was trying to improve on.
4003 let _ = self.editor.split_block(at);
4004 }
4005 match self.editor.insert_thematic_break(at) {
4006 Ok(change) => {
4007 self.last_edit_kind = None;
4008 self.refresh();
4009 self.anchor = None;
4010 self.caret = change.new.end;
4011 self.dirty = self.source != self.clean_source;
4012 self.status = None;
4013 self.clamp_caret();
4014 self.record_caret();
4015 }
4016 Err(e) => self.status = Some(format!("thematic break: {e}")),
4017 }
4018 }
4019
4020 /// Whether the caret sits in a paragraph and nothing else — no list item, no
4021 /// quote, no fence, no table. The one shape where parting the block around
4022 /// the caret is unambiguously what a rule button means; see
4023 /// [`insert_thematic_break`](Self::insert_thematic_break) for why every other
4024 /// container is left to take the rule after itself.
4025 fn caret_in_bare_paragraph(&mut self) -> bool {
4026 let caret = self.caret.min(self.source.len());
4027 let Ok(chain) = self.editor.ancestors_at(caret) else {
4028 return false;
4029 };
4030 let mut in_para = false;
4031 for m in chain {
4032 match m.kind {
4033 Kind::Para => in_para = true,
4034 Kind::ListItem
4035 | Kind::TaskListItem
4036 | Kind::BlockQuote
4037 | Kind::CodeBlock
4038 | Kind::Table => return false,
4039 _ => {}
4040 }
4041 }
4042 in_para
4043 }
4044
4045 /// The destination of the link under the caret — what a Link prompt shows so
4046 /// ⌘K on an existing link edits its URL instead of asking for it again.
4047 /// `None` when the caret stands in no link.
4048 ///
4049 /// An autolink carries no separate destination: its text *is* the URL, so
4050 /// that's what comes back for one.
4051 pub fn link_destination_at_caret(&mut self) -> Option<String> {
4052 self.link_destination_at(self.caret)
4053 }
4054
4055 /// The destination of the link at `off`.
4056 /// [`link_destination_at_caret`](Self::link_destination_at_caret) for a place
4057 /// the caret isn't.
4058 ///
4059 /// The offset form exists for the same reason
4060 /// [`footnote_at`](Self::footnote_at)'s does: a frontend drawing a *piece* of
4061 /// the document somewhere else — a footnote's text in a popover, say — has
4062 /// rows and runs but no caret in them, and still needs to know which of those
4063 /// runs a reader can follow.
4064 pub fn link_destination_at(&mut self, off: usize) -> Option<String> {
4065 self.nodes()
4066 .into_iter()
4067 .filter(|n| matches!(n.kind.as_str(), "link" | "url" | "email"))
4068 .filter(|n| n.span.start <= off && off < n.span.end)
4069 .max_by_key(|n| n.span.start)
4070 .and_then(|n| n.destination.or(n.text))
4071 }
4072
4073 /// Where the locator `id` lands in this document — the `#v2` half of a
4074 /// `chapter.dj#v2`, resolved to the block it names. `None` when nothing here
4075 /// answers to it.
4076 ///
4077 /// The other end of a link, and the reason this exists: without it a
4078 /// destination has only file granularity, so following a citation into a
4079 /// chapter drops the reader at the top of it to hunt for the verse. Which is
4080 /// also why it is a *document* query rather than a caret one — the document
4081 /// being asked is usually not the one the reader is in.
4082 ///
4083 /// Three readings, tried in order, because the same `#some-heading` is
4084 /// written three ways across the formats leaf opens:
4085 ///
4086 /// 1. **A declared id**, exactly as written: djot's `{#v1}` on a block, and
4087 /// the auto-ids djot mints for its headings. The only exact answer, so it
4088 /// goes first — a document that says `{#v1}` has settled the question.
4089 /// 2. **A declared id, slugged.** djot spells a heading's auto-id
4090 /// `Some-Heading-Here`; nearly every tool that *writes* a link to one
4091 /// spells it `#some-heading-here`. Comparing slugs is what lets a link
4092 /// authored anywhere land on a djot heading.
4093 /// 3. **A heading's text, slugged.** Markdown has no ids at all — twig mints
4094 /// none and `{#custom}` is literal text in a Markdown heading — so for
4095 /// the format most vaults are written in, the heading's own words are the
4096 /// only thing a fragment can name. This is the rule every Markdown
4097 /// renderer already follows, which is what makes `#a-heading` mean in
4098 /// diaryx what it means on the web.
4099 ///
4100 /// Ties go to the earliest match, then to the widest: a duplicated id is the
4101 /// document's mistake and the first one is the answer every anchor
4102 /// implementation gives, while preferring the wider span picks the section
4103 /// over the heading that opens it — more for a peek to show, same place to
4104 /// land.
4105 pub fn locate(&mut self, id: &str) -> Option<Landing> {
4106 let id = id.trim();
4107 if id.is_empty() {
4108 return None;
4109 }
4110 let nodes = self.nodes();
4111
4112 // Earliest wins, then widest. `Reverse` on the end because `min_by_key`
4113 // is picking, among nodes that start together, the one that ends last.
4114 let pick = |matches: &mut dyn Iterator<Item = &FlatNode>| {
4115 matches
4116 .min_by_key(|n| (n.span.start, std::cmp::Reverse(n.span.end)))
4117 .map(|n| Landing {
4118 start: n.span.start,
4119 end: n.span.end,
4120 })
4121 };
4122
4123 if let Some(landing) = pick(&mut nodes.iter().filter(|n| declared_id(n) == Some(id))) {
4124 return Some(landing);
4125 }
4126 let want = slug(id);
4127 if want.is_empty() {
4128 return None;
4129 }
4130 if let Some(landing) = pick(
4131 &mut nodes
4132 .iter()
4133 .filter(|n| declared_id(n).map(slug).as_deref() == Some(&*want)),
4134 ) {
4135 return Some(landing);
4136 }
4137
4138 // A heading by its words. Its span is one line, so the end comes from
4139 // where the *section* it opens gives out — the next heading that is not
4140 // under it, or the end of the document. A Markdown heading has no
4141 // section node to ask (twig only builds those for djot), and a peek that
4142 // showed the heading alone would answer "what does that say" with the
4143 // title of the thing it says.
4144 let heading = nodes
4145 .iter()
4146 .filter(|n| n.kind == Kind::Heading)
4147 .filter(|n| {
4148 n.content_span
4149 .clone()
4150 .and_then(|s| self.source.get(s))
4151 .is_some_and(|text| slug(text) == want)
4152 })
4153 .min_by_key(|n| n.span.start)?;
4154 let level = heading.level.unwrap_or(u32::MAX);
4155 let end = nodes
4156 .iter()
4157 .filter(|n| n.kind == Kind::Heading)
4158 .filter(|n| n.span.start > heading.span.start)
4159 .filter(|n| n.level.unwrap_or(u32::MAX) <= level)
4160 .map(|n| n.span.start)
4161 .min()
4162 .unwrap_or(self.source.len());
4163 Some(Landing {
4164 start: heading.span.start,
4165 end,
4166 })
4167 }
4168
4169 /// Write a footnote at the caret — the toolbar's Footnote button, and the
4170 /// one gesture in the footnote story that *authors* rather than follows.
4171 ///
4172 /// Both halves go in as one twig edit: the `[^1]` where the caret is, and
4173 /// the `[^1]:` definition at the end of the document. Half a footnote is not
4174 /// a footnote — a bare reference with nothing defining it renders as literal
4175 /// brackets — so a single button that wrote only the reference would leave
4176 /// the author to hand-spell the other half in a document that had just
4177 /// stopped showing them what the first half meant. One edit also means one
4178 /// undo takes both back.
4179 ///
4180 /// The definition's body is left empty and **the caret lands in it**, which
4181 /// is the whole point of pressing the button: nobody wants a reference to a
4182 /// note they have not written yet. Getting back to where they were writing
4183 /// is [`footnote_definition_at_caret`](Self::footnote_definition_at_caret) —
4184 /// the same return leg a reader following a reference already uses, so the
4185 /// author is left standing on the near end of a round trip that works.
4186 ///
4187 /// A selection collapses to its *end* rather than being replaced: a
4188 /// reference annotates the words before it, so "select the claim, add a
4189 /// footnote" should mark that claim, not consume it.
4190 pub fn insert_footnote(&mut self) {
4191 if self.refuse_unsupported("footnote", Gesture::InsertFootnote) {
4192 return;
4193 }
4194 let at = self.selection().map_or(self.caret, |(_, end)| end);
4195 self.anchor = None;
4196 self.caret = at;
4197 self.record_caret();
4198 let label = self.next_footnote_label();
4199 match self.editor.insert_footnote(at, &label) {
4200 Ok(change) => {
4201 self.last_edit_kind = None;
4202 self.refresh();
4203 self.anchor = None;
4204 // `change.new` runs from the reference to the end of the
4205 // document, so its start is the `[^1]` just written and
4206 // `footnote_at` resolves it to the note the same way a reader's
4207 // tap does — and to the note's *body*, which is already a caret
4208 // stop even when it is empty (the `[^1]:` marker draws as `[1] `
4209 // and has none), so this needs no snap on top. The fallback is
4210 // the reference's own offset: a format that spelled the pair some
4211 // way leaf can't read back should still leave the caret on the
4212 // edit rather than at the far end of a document it just grew.
4213 self.caret = self
4214 .footnote_at(change.new.start)
4215 .and_then(|note| note.offset)
4216 .unwrap_or(change.new.start);
4217 self.dirty = self.source != self.clean_source;
4218 self.status = None;
4219 self.clamp_caret();
4220 self.record_caret();
4221 }
4222 Err(e) => self.status = Some(format!("footnote: {e}")),
4223 }
4224 }
4225
4226 /// The label to give a footnote the author has not named: the lowest counting
4227 /// number no footnote in the document is already wearing.
4228 ///
4229 /// twig takes the label rather than minting one, because it holds no opinion
4230 /// about what a document's footnotes should be called — and it is right not
4231 /// to. Numbering them is what every author of a numbered note expects, and
4232 /// re-using a taken number would silently point the new reference at somebody
4233 /// else's note (twig reuses an existing definition rather than appending a
4234 /// second one, which is the right rule for citing a note twice on purpose and
4235 /// exactly the wrong accident to have by default).
4236 ///
4237 /// *References* are counted alongside definitions, not just definitions: a
4238 /// document carrying a dangling `[^2]` has a 2 that means something to
4239 /// whoever wrote it, and minting a definition for it here would answer a
4240 /// question nobody asked. Non-numeric labels (`[^why]`) are left out of the
4241 /// count entirely — they take no number, so they block none.
4242 fn next_footnote_label(&mut self) -> String {
4243 let mut taken: Vec<u32> = wysiwyg::footnote_definitions(&mut self.editor)
4244 .into_iter()
4245 .filter_map(|note| wysiwyg::footnote_label(&self.source, note.span.start))
4246 .filter_map(|label| label.parse().ok())
4247 .collect();
4248 taken.extend(
4249 self.nodes()
4250 .into_iter()
4251 .filter(|n| n.kind == Kind::FootnoteReference)
4252 .filter_map(|n| wysiwyg::footnote_reference_label(&self.source, n.span))
4253 .filter_map(|label| label.parse::<u32>().ok()),
4254 );
4255 (1..).find(|n| !taken.contains(n)).unwrap_or(1).to_string()
4256 }
4257
4258 /// The footnote reference under the caret, resolved to the note it names.
4259 /// [`footnote_at`](Self::footnote_at) at the caret's offset.
4260 pub fn footnote_at_caret(&mut self) -> Option<FootnoteRef> {
4261 self.footnote_at(self.caret)
4262 }
4263
4264 /// The footnote reference at `off`, resolved to the note it names — what a
4265 /// frontend shows when a reader activates a `[^1]`.
4266 ///
4267 /// A reference is not a link node, so
4268 /// [`link_destination_at_caret`](Self::link_destination_at_caret) does not
4269 /// (and should not) answer for one: a link names a destination to leave for,
4270 /// a reference names a note that is already in this document. Following one
4271 /// is a move within the page, which is why this hands back an `offset`
4272 /// rather than something to open.
4273 ///
4274 /// Offset-based rather than caret-only because the gesture that wants this
4275 /// most is the one that must not move the caret: a pointer hovering a `[1]`
4276 /// asks what note it names without disturbing where the reader was typing.
4277 /// The caret is just the offset a click already placed —
4278 /// [`footnote_at_caret`](Self::footnote_at_caret) passes it.
4279 ///
4280 /// `None` when `off` stands in no reference. A reference whose note the
4281 /// document never defines is *not* `None` — it answers with the label it
4282 /// looked for and no text, which is what lets a frontend say so instead of
4283 /// silently doing nothing.
4284 pub fn footnote_at(&mut self, off: usize) -> Option<FootnoteRef> {
4285 // Innermost-wins by latest start, the rule its link sibling uses.
4286 let span = self
4287 .nodes()
4288 .into_iter()
4289 .filter(|n| n.kind == Kind::FootnoteReference)
4290 .filter(|n| n.span.start <= off && off < n.span.end)
4291 .max_by_key(|n| n.span.start)?
4292 .span;
4293 let label = wysiwyg::footnote_reference_label(&self.source, span)?.to_string();
4294
4295 // The note itself. Definitions are roots beside `doc` rather than
4296 // children of it, so they're asked for directly — see
4297 // `wysiwyg::footnote_definitions`.
4298 let note = wysiwyg::footnote_definitions(&mut self.editor)
4299 .into_iter()
4300 .find(|m| wysiwyg::footnote_label(&self.source, m.span.start) == Some(&label));
4301 let Some(note) = note else {
4302 return Some(FootnoteRef {
4303 label,
4304 text: None,
4305 offset: None,
4306 end: None,
4307 });
4308 };
4309 let body = wysiwyg::footnote_body_span(&self.source, note.span.clone());
4310 Some(FootnoteRef {
4311 label,
4312 text: body
4313 .clone()
4314 .and_then(|b| self.source.get(b))
4315 .map(str::to_string),
4316 // The body's start, not the definition's — see `FootnoteRef::offset`.
4317 offset: body.clone().map(|b| b.start),
4318 end: body.map(|b| b.end),
4319 })
4320 }
4321
4322 /// The footnote *definition* the caret stands in, and where the reference
4323 /// that names it is. [`footnote_definition_at`](Self::footnote_definition_at)
4324 /// at the caret's offset.
4325 pub fn footnote_definition_at_caret(&mut self) -> Option<FootnoteDef> {
4326 self.footnote_definition_at(self.caret)
4327 }
4328
4329 /// The footnote definition spanning `off`, and where the reference that
4330 /// names it is — the return leg of [`footnote_at`](Self::footnote_at).
4331 ///
4332 /// The mirror image, deliberately: the same gesture that takes a reader from
4333 /// `[1]` down to the note takes them from the note back up to `[1]`, so
4334 /// following a footnote is a round trip rather than a fall. It needs no
4335 /// memory of how the reader arrived — the document says where the reference
4336 /// is — which is what makes it work for a reader who scrolled to the notes
4337 /// themselves, and what keeps it right after an edit moves either end.
4338 ///
4339 /// `None` when `off` stands in no definition. A definition nothing cites is
4340 /// *not* `None`, for [`FootnoteRef`]'s reason in reverse: it answers with
4341 /// its label and no offset, so a frontend can say "nothing refers to this"
4342 /// rather than offer a jump that goes nowhere.
4343 pub fn footnote_definition_at(&mut self, off: usize) -> Option<FootnoteDef> {
4344 // Definitions are roots beside `doc`, so `nodes()` — which walks the
4345 // document body — never reports one. They're asked for directly, the way
4346 // `footnote_at` asks for the note it resolves to.
4347 //
4348 // Closed at the end, unlike the half-open test its neighbours use. A
4349 // definition's span stops at its last content byte — the newline ending
4350 // the line is outside it — so `span.end` is the caret stop at the end of
4351 // the note's own row, not the first byte of anything after. Excluding it
4352 // meant the one caret an author is guaranteed to have, the one left
4353 // sitting at the end of the note they just typed, was in no definition at
4354 // all: writing a note and then asking to go back to its reference
4355 // answered nothing. Two definitions in a row still can't both match —
4356 // there is a blank line between them — and `max_by_key` decides anyway.
4357 let note = wysiwyg::footnote_definitions(&mut self.editor)
4358 .into_iter()
4359 .filter(|m| m.span.start <= off && off <= m.span.end)
4360 .max_by_key(|m| m.span.start)?;
4361 let label = wysiwyg::footnote_label(&self.source, note.span.start)?.to_string();
4362
4363 // The earliest reference carrying this label. `min` rather than a `find`,
4364 // because `nodes()` reports a flattened walk whose order is twig's
4365 // business, not document order. Bound first: the walk needs `&mut self`
4366 // and reading the labels back out needs `&self.source`.
4367 let nodes = self.nodes();
4368 let offset = nodes
4369 .into_iter()
4370 .filter(|n| n.kind == Kind::FootnoteReference)
4371 .filter(|n| {
4372 wysiwyg::footnote_reference_label(&self.source, n.span.clone()) == Some(&*label)
4373 })
4374 // Past the `[^`, onto the label — see `FootnoteDef::offset`.
4375 .map(|n| n.span.start + 2)
4376 .min();
4377 Some(FootnoteDef { label, offset })
4378 }
4379
4380 /// The destination of the image under the caret — what an image prompt shows
4381 /// so editing an existing image starts from its current URL instead of blank,
4382 /// the image analogue of [`link_destination_at_caret`](Self::link_destination_at_caret).
4383 /// `None` when the caret stands in no image. A caret resting just after a
4384 /// block image (its trailing stop) is still "in" it — the half-open span test
4385 /// excludes that offset, which is the intended precision: past the image is
4386 /// past it.
4387 pub fn image_destination_at_caret(&mut self) -> Option<String> {
4388 let off = self.caret;
4389 self.nodes()
4390 .into_iter()
4391 .filter(|n| n.kind == Kind::Image)
4392 .filter(|n| n.span.start <= off && off < n.span.end)
4393 .max_by_key(|n| n.span.start)
4394 .and_then(|n| n.destination)
4395 }
4396
4397 /// The language of the fenced code block the caret stands in — what a
4398 /// language prompt shows so editing it starts from the current value rather
4399 /// than blank. `None` when the caret is in no code block, or in one whose
4400 /// fence carries no language (or an indented block, which has no fence).
4401 pub fn code_language_at_caret(&mut self) -> Option<String> {
4402 let start = self.code_block_start_at_caret()?;
4403 wysiwyg::code_language(&self.source, start)
4404 }
4405
4406 /// Whether the caret stands in a fenced code block — the one a language
4407 /// prompt could edit. A frontend gates its "set language" affordance on this
4408 /// (an indented block, which can't carry a language, reports `false`).
4409 pub fn caret_in_fenced_code(&mut self) -> bool {
4410 self.code_block_start_at_caret()
4411 .is_some_and(|start| wysiwyg::code_info_span(&self.source, start).is_some())
4412 }
4413
4414 /// Set (or clear, with `""`) the language of the fenced code block the caret
4415 /// is in — the prompt's confirm. A no-op when the caret is in no fenced
4416 /// block, and a reported error for a language the format's fence cannot
4417 /// carry.
4418 ///
4419 /// twig rewrites the info string, so the fence's own width — measured
4420 /// against a body neither side touches — is kept, and a language holding a
4421 /// space, a line end or the fence character is refused rather than written
4422 /// out to reparse as something else. Leaf used to splice over the info span
4423 /// itself and `trim()` the input, which handled the one bad case it had
4424 /// thought of.
4425 pub fn set_code_language(&mut self, lang: &str) {
4426 if self.refuse_unsupported("code language", Gesture::SetCodeLanguage) {
4427 return;
4428 }
4429 if self.code_block_start_at_caret().is_none() {
4430 return;
4431 }
4432 let lang = lang.trim();
4433 // `None` clears the info string; `Some("")` asks for an empty one. Both
4434 // write a bare fence, and the prompt's empty value means "clear".
4435 let want = (!lang.is_empty()).then_some(lang);
4436 self.record_caret();
4437 match self.editor.set_code_language(self.caret, want) {
4438 Ok(_) => {
4439 self.last_edit_kind = None;
4440 self.refresh();
4441 self.anchor = None;
4442 self.dirty = self.source != self.clean_source;
4443 self.status = None;
4444 self.clamp_caret();
4445 self.record_caret();
4446 }
4447 Err(e) => self.status = Some(format!("code language: {e}")),
4448 }
4449 }
4450
4451 /// The `span.start` of the code block covering the caret — the anchor
4452 /// [`wysiwyg::code_info_span`] reads the fence from. `None` when the caret is
4453 /// in none.
4454 fn code_block_start_at_caret(&mut self) -> Option<usize> {
4455 let off = self.caret;
4456 self.nodes()
4457 .into_iter()
4458 .filter(|n| n.kind == Kind::CodeBlock && n.span.start <= off && off <= n.span.end)
4459 .max_by_key(|n| n.span.start)
4460 .map(|n| n.span.start)
4461 }
4462
4463 /// The source range of the text inside the link covering `off` — what sits
4464 /// between its `[` and `]`. `None` when twig reports no link there.
4465 fn link_text_span(&mut self, off: usize) -> Option<std::ops::Range<usize>> {
4466 self.nodes()
4467 .into_iter()
4468 // Two links can touch (`[a](x)[b](y)`), and then one's `span.end` is
4469 // the other's `span.start`; the link that starts latest at or before
4470 // `off` is the one `off` is actually in.
4471 .filter(|n| n.kind == Kind::Link && n.span.start <= off && off < n.span.end)
4472 .max_by_key(|n| n.span.start)
4473 .and_then(|n| n.content_span)
4474 }
4475
4476 // ── undo / redo ───────────────────────────────────────────────────────────
4477 // twig owns the history of *bytes* (it owns the buffer) and now carries the
4478 // caret through it too: `record_caret` stashes each state's caret in twig's
4479 // opaque per-step blob, and undo/redo hand it back with the source they
4480 // restore. So leaf keeps no history of its own — no parallel stacks to march
4481 // in lockstep and silently drift out of it.
4482
4483 /// Undo the last edit step (⌘Z / ^Z), putting the caret and selection back
4484 /// where they were when that step began.
4485 pub fn undo(&mut self) {
4486 if self.read_only {
4487 return;
4488 }
4489 match self.editor.undo() {
4490 Ok(Some(change)) => self.after_history(change),
4491 Ok(None) => self.status = Some("nothing to undo".into()),
4492 Err(e) => self.status = Some(format!("undo: {e}")),
4493 }
4494 }
4495
4496 /// Redo the last undone edit step (⇧⌘Z / ^Y), putting the caret and
4497 /// selection back where that step originally left them.
4498 pub fn redo(&mut self) {
4499 if self.read_only {
4500 return;
4501 }
4502 match self.editor.redo() {
4503 Ok(Some(change)) => self.after_history(change),
4504 Ok(None) => self.status = Some("nothing to redo".into()),
4505 Err(e) => self.status = Some(format!("redo: {e}")),
4506 }
4507 }
4508
4509 /// Refresh the cached source and put the caret back where the step being
4510 /// undone/redone had it, clearing any active run.
4511 ///
4512 /// The caret comes from twig's blob for the restored state (what
4513 /// `record_caret` stored). `change` is only the fallback for a state with no
4514 /// blob — a caret at the end of the restored text, which is where this always
4515 /// landed before the blobs were kept. It is the edit site, not where the user
4516 /// was standing, so it's a floor and not the behaviour: undoing should hand
4517 /// back the document *and* the place you were working, which for an edit made
4518 /// anywhere but under the caret are two different places.
4519 fn after_history(&mut self, change: Change) {
4520 self.refresh();
4521 match self
4522 .editor
4523 .caret_blob()
4524 .ok()
4525 .and_then(|b| CaretState::from_blob(&b))
4526 {
4527 Some(state) => {
4528 self.caret = state.caret.min(self.source.len());
4529 self.anchor = state.anchor.map(|a| a.min(self.source.len()));
4530 }
4531 None => {
4532 self.caret = change.new.end.min(self.source.len());
4533 self.anchor = None;
4534 }
4535 }
4536 self.goal_col = None;
4537 self.last_edit_kind = None;
4538 self.dirty = self.source != self.clean_source;
4539 self.status = None;
4540 self.clamp_caret();
4541 }
4542
4543 // ── the file ──────────────────────────────────────────────────────────────
4544
4545 #[cfg(feature = "fs")]
4546 pub fn save(&mut self) {
4547 if self.is_untitled() {
4548 // No path to write and no name to invent: ⌘S on an untitled document
4549 // is a Save As, and only a frontend has a picker to ask with. Say so
4550 // rather than failing at the filesystem with an empty path.
4551 self.status = Some("untitled — save as…".into());
4552 return;
4553 }
4554 let path = self.path.clone();
4555 if self.write(&path) {
4556 self.mark_saved();
4557 }
4558 }
4559
4560 /// Save As: write the document to `path` and *move* it there — `self.path`
4561 /// becomes `path`, and every later [`Doc::save`] writes the new file. That's
4562 /// what Save As means; a copy would leave the user editing a document whose
4563 /// name is no longer where their keystrokes go.
4564 ///
4565 /// The move only happens if the bytes actually landed. A failed write leaves
4566 /// the path, `dirty`, and the disk watermark exactly as they were, with the
4567 /// same `save failed: …` status a failed [`Doc::save`] sets — the document
4568 /// must never come away believing it was saved.
4569 ///
4570 /// An existing `path` is overwritten, and the caller is the one that knows
4571 /// whether to ask first: a Save As picker has already run that prompt, and a
4572 /// second confirmation from down here would be the same question twice.
4573 ///
4574 /// `format` does **not** follow the new extension. The buffer is parsed as
4575 /// the format it was opened with, and re-reading it as another one is a
4576 /// conversion — a different, lossy operation that would throw away the undo
4577 /// history — not a rename. So `notes.md` saved as `notes.dj` holds Markdown
4578 /// in a `.dj` file, and `format_name()` keeps honestly saying `markdown`
4579 /// until it's reopened.
4580 #[cfg(feature = "fs")]
4581 pub fn save_as(&mut self, path: PathBuf) {
4582 if !self.write(&path) {
4583 return;
4584 }
4585 self.path = path;
4586 self.mark_saved();
4587 }
4588
4589 /// Put `source` on disk at `path`, reporting whether it got there. The one
4590 /// place leaf writes a document, so a save and a Save As can't disagree
4591 /// about what a failure looks like.
4592 #[cfg(feature = "fs")]
4593 fn write(&mut self, path: &Path) -> bool {
4594 match std::fs::write(path, self.source.as_bytes()) {
4595 Ok(()) => true,
4596 Err(e) => {
4597 self.status = Some(format!("save failed: {e}"));
4598 false
4599 }
4600 }
4601 }
4602
4603 /// Re-base the document's saved watermark to the current bytes: clears
4604 /// `dirty`, records `source` as the new clean state (so undoing back to here
4605 /// clears the flag again), and re-stamps the on-disk hash.
4606 ///
4607 /// [`Doc::save`]/[`Doc::save_as`] call this after a write lands. It is also
4608 /// the hook a **filesystem-free host** calls itself once it has persisted
4609 /// [`Doc::source`] its own way (a browser download, `localStorage`, a backend
4610 /// `PUT`) — which is why it is public and touches no filesystem: the bytes
4611 /// are already where that host wants them, and this just tells the model they
4612 /// are safe.
4613 pub fn mark_saved(&mut self) {
4614 self.clean_source = self.source.clone();
4615 self.dirty = false;
4616 // The bytes on disk are now ours, so this is the new watermark: without
4617 // re-stamping it, every save would report its own work as an external
4618 // change forever after.
4619 self.disk_hash = Some(hash_bytes(self.source.as_bytes()));
4620 self.status = Some(format!("saved {}", self.file_name()));
4621 }
4622
4623 /// What the file looks like now against the bytes leaf last read or wrote.
4624 ///
4625 /// Reads the file and hashes it (see `disk_hash` for why it isn't an mtime),
4626 /// so this is a filesystem round-trip, not a per-frame question — ask it
4627 /// when a window regains focus, on a timer, or before a save.
4628 ///
4629 /// This *only* reports the file. Whether the document also has unsaved edits
4630 /// is `dirty`, and the interesting case is the conjunction: `dirty` plus
4631 /// [`DiskState::Changed`] means a save overwrites someone's work and a
4632 /// [`Doc::reload`] discards the user's. leaf-core deliberately won't choose —
4633 /// it has no way to ask — so it hands a frontend both halves and lets it put
4634 /// the question to the person who can answer it.
4635 #[cfg(feature = "fs")]
4636 pub fn disk_state(&self) -> DiskState {
4637 let Some(want) = self.disk_hash else {
4638 return DiskState::Untitled;
4639 };
4640 match std::fs::read(&self.path) {
4641 Ok(bytes) if hash_bytes(&bytes) == want => DiskState::Unchanged,
4642 Ok(_) => DiskState::Changed,
4643 Err(e) if e.kind() == std::io::ErrorKind::NotFound => DiskState::Missing,
4644 Err(_) => DiskState::Unreadable,
4645 }
4646 }
4647
4648 /// Re-read the file and replace the document with what's there — the other
4649 /// answer to a [`DiskState::Changed`].
4650 ///
4651 /// **Discards unsaved changes and the undo history, unconditionally.** It
4652 /// doesn't check `dirty` first: a frontend that wants to protect unsaved
4653 /// work asks (`dirty` + [`Doc::disk_state`]) *before* calling this, and one
4654 /// reloading a clean document shouldn't have to argue with a guard. The
4655 /// history goes because twig's undo stack belongs to the buffer, and these
4656 /// are different bytes — replaying a step recorded against the old ones onto
4657 /// them would corrupt the document, and nothing here can honestly rebase it.
4658 ///
4659 /// The caret keeps its byte offset, clamped to the new length; the selection
4660 /// is dropped. Anything cleverer would be a lie: leaf doesn't know how the
4661 /// file changed, so it can't know where the caret "still" is. Clamping keeps
4662 /// it where the user left it in the common case (a change further down the
4663 /// file, or none in the text they're sitting in), and never puts it
4664 /// somewhere invalid. A selection has two such offsets and no such excuse —
4665 /// silently reinterpreting one over changed bytes would arm the *next*
4666 /// keystroke to delete something the user never selected.
4667 ///
4668 /// Nothing is touched unless the whole reload succeeds; a failure leaves the
4669 /// document alone with a status.
4670 #[cfg(feature = "fs")]
4671 pub fn reload(&mut self) {
4672 if self.is_untitled() {
4673 self.status = Some("no file to reload".into());
4674 return;
4675 }
4676 let bytes = match std::fs::read(&self.path) {
4677 Ok(b) => b,
4678 Err(e) => {
4679 self.status = Some(format!("reload failed: {e}"));
4680 return;
4681 }
4682 };
4683 let Ok(source) = String::from_utf8(bytes) else {
4684 self.status = Some("reload failed: file is not UTF-8".into());
4685 return;
4686 };
4687 // Reparse rather than splice the difference in: leaf doesn't know what
4688 // changed, and `format` is the format this document is, not what the
4689 // (unchanged) name now says — see `save_as`.
4690 let editor = match new_editor(source.as_bytes(), self.format) {
4691 Ok(ed) => ed,
4692 Err(e) => {
4693 self.status = Some(format!("reload failed: {e}"));
4694 return;
4695 }
4696 };
4697 self.editor = editor;
4698 self.disk_hash = Some(hash_bytes(source.as_bytes()));
4699 self.clean_source = source.clone();
4700 self.source = source;
4701 // Reload replaces the text without going through `refresh`, so it has to
4702 // move the revision itself or every frontend would keep painting the old
4703 // file from cache.
4704 self.revision += 1;
4705 self.caret = self.caret.min(self.source.len());
4706 self.anchor = None;
4707 self.goal_col = None;
4708 self.last_edit_kind = None;
4709 self.dirty = false;
4710 self.status = Some(format!("reloaded {}", self.file_name()));
4711 self.clamp_caret();
4712 }
4713
4714 /// Re-read the source from twig after it has changed the document. The one
4715 /// funnel every edit, undo, and redo comes through — so it's where the
4716 /// revision moves, and anything cached against the text dies here.
4717 fn refresh(&mut self) {
4718 if let Ok(s) = self.editor.source_str() {
4719 self.source = s;
4720 }
4721 self.revision += 1;
4722 self.clamp_caret();
4723 }
4724
4725 // ── caret movement ─────────────────────────────────────────────────────────
4726 // `extend` grows the selection (Shift+motion): it pins the anchor on the
4727 // first extended step and moves only the caret; an un-extended motion drops
4728 // the selection.
4729
4730 /// Place the caret at byte `offset` (clamped to a char boundary), extending
4731 /// the selection when `extend` is set. The public form of `move_to`, for a
4732 /// frontend that hit-tests pixels straight to a source offset.
4733 pub fn place_caret(&mut self, offset: usize, extend: bool) {
4734 self.goal_col = None;
4735 let before = self.caret;
4736 // A pixel hit-test can land between the visible caret stops — in the
4737 // blank gap a paragraph break is drawn with, or inside a hidden delimiter.
4738 // Snap to the nearest real stop so the caret can't come to rest where it
4739 // would draw in one place and type in another. The `(row, col)` click
4740 // path (`click`) already snaps this way through `offset_of_pos`; the
4741 // source view reaches every byte, so it snaps to nothing.
4742 let target = match self.view {
4743 View::Wysiwyg => self.vmap.snap_to_stop(offset.min(self.source.len())),
4744 // The source view reaches every byte, so there is no stop to snap
4745 // to — but "every byte" still means every *character* boundary. A
4746 // caret resting inside a multi-byte character draws nowhere real
4747 // and panics the next time anything slices there.
4748 View::Source => {
4749 let mut o = offset.min(self.source.len());
4750 while o > 0 && !self.source.is_char_boundary(o) {
4751 o -= 1;
4752 }
4753 o
4754 }
4755 };
4756 self.move_to(target, extend);
4757 self.clamp_caret();
4758 self.debug_assert_on_a_stop(before);
4759 }
4760
4761 /// Select the whole document (⌘A / Ctrl+A) — everything reachable in the
4762 /// active view, so in WYSIWYG it starts below hidden frontmatter (copy won't
4763 /// grab the metadata) while the source view still selects the literal whole.
4764 pub fn select_all(&mut self) {
4765 self.anchor = Some(self.caret_floor());
4766 self.caret = self.source.len();
4767 self.goal_col = None;
4768 self.last_edit_kind = None;
4769 self.status = None;
4770 }
4771
4772 /// Select the word (or whitespace / punctuation run) at `offset` — the
4773 /// double-click gesture. Anchors on the run's start with the caret at its
4774 /// end so a following Shift-motion extends from the far edge.
4775 pub fn select_word_at(&mut self, offset: usize) {
4776 let (s, e) = word_range_at(&self.source, offset.min(self.source.len()));
4777 self.anchor = Some(s);
4778 self.caret = e;
4779 self.goal_col = None;
4780 self.last_edit_kind = None;
4781 self.status = None;
4782 self.clamp_caret();
4783 }
4784
4785 /// Select the whole enclosing text block (paragraph, heading, list item's
4786 /// text…) at `offset` — the triple-click gesture. Reads the range straight
4787 /// from the AST (twig's `content_span`), so it selects the entire *logical*
4788 /// paragraph even when that paragraph soft-wraps across several visual rows —
4789 /// where a visual-row-based select breaks down, because one source offset at
4790 /// a wrap boundary belongs to two rows at once.
4791 pub fn select_block_at(&mut self, offset: usize) {
4792 let off = offset.min(self.source.len());
4793 let range = self
4794 .editor
4795 .ancestors_at(off)
4796 .ok()
4797 .and_then(|chain| {
4798 // Ancestors run root → deepest; the deepest node that is neither
4799 // an inline span nor a multi-block container is the text block
4800 // the caret sits in (a paragraph, a heading, a code block…).
4801 chain
4802 .into_iter()
4803 .rev()
4804 .find(|m| !wysiwyg::is_inline_kind(&m.kind) && !is_block_container(&m.kind))
4805 .map(|m| m.content_span.unwrap_or(m.span))
4806 })
4807 .unwrap_or_else(|| source_line_range(&self.source, off));
4808 self.anchor = Some(range.start.min(self.source.len()));
4809 self.caret = range.end.min(self.source.len());
4810 self.goal_col = None;
4811 self.last_edit_kind = None;
4812 self.status = None;
4813 self.clamp_caret();
4814 }
4815
4816 /// The lowest source offset the caret may occupy in the active view. In
4817 /// WYSIWYG, leading frontmatter is hidden and unreachable, so the floor is
4818 /// the first rendered offset; the source view reaches everything, so it's 0.
4819 fn caret_floor(&self) -> usize {
4820 match self.view {
4821 View::Wysiwyg => self.vmap.content_start.min(self.source.len()),
4822 View::Source => 0,
4823 }
4824 }
4825
4826 /// Land in a table cell with its whole content selected — the anchor at the
4827 /// cell's start, the caret at its end — so a Tab/Return hop into a cell reads
4828 /// like tabbing into a form field: the text comes up selected, so typing
4829 /// replaces it and an arrow collapses to an edge. An empty cell (`start ==
4830 /// end`) collapses to a plain caret home (an empty selection is no selection).
4831 fn select_cell(&mut self, start: usize, end: usize) {
4832 let floor = self.caret_floor();
4833 self.anchor = Some(start.min(self.source.len()).max(floor));
4834 self.caret = end.min(self.source.len()).max(floor);
4835 self.goal_col = None;
4836 self.status = None;
4837 self.last_edit_kind = None;
4838 self.clear_pending();
4839 }
4840
4841 fn move_to(&mut self, offset: usize, extend: bool) {
4842 if extend {
4843 if self.anchor.is_none() {
4844 self.anchor = Some(self.caret);
4845 }
4846 } else {
4847 self.anchor = None;
4848 }
4849 self.caret = offset.min(self.source.len()).max(self.caret_floor());
4850 self.status = None;
4851 // A caret move ends the current typing/deletion run, so the next edit
4852 // starts a fresh undo group rather than coalescing across the gap.
4853 self.last_edit_kind = None;
4854 // Moving away disarms any sticky mark — "start bold" applies only where
4855 // it was asked for, not wherever the caret next lands.
4856 self.clear_pending();
4857 }
4858
4859 // In the source view, motion walks source bytes / source lines. In the
4860 // WYSIWYG view it walks the rendered glyph grid (the visual map), which is
4861 // what steps the caret cleanly over hidden delimiters.
4862
4863 pub fn move_left(&mut self, extend: bool) {
4864 self.goal_col = None;
4865 if !extend && let Some((s, _e)) = self.selection() {
4866 self.move_to(s, false);
4867 return;
4868 }
4869 let target = match self.view {
4870 View::Source => {
4871 if self.caret > 0 {
4872 prev_boundary(&self.source, self.caret)
4873 } else {
4874 0
4875 }
4876 }
4877 // Walks caret *stops*, not columns: decoration (a table border, a
4878 // cell's padding) is stepped over in one press, and a hidden
4879 // delimiter never holds the caret up.
4880 View::Wysiwyg => self.vmap.stop_before(self.caret).unwrap_or(self.caret),
4881 };
4882 let before = self.caret;
4883 self.move_to(target, extend);
4884 self.debug_assert_on_a_stop(before);
4885 }
4886
4887 pub fn move_right(&mut self, extend: bool) {
4888 self.goal_col = None;
4889 if !extend && let Some((_s, e)) = self.selection() {
4890 self.move_to(e, false);
4891 return;
4892 }
4893 let target = match self.view {
4894 View::Source => {
4895 if self.caret < self.source.len() {
4896 next_boundary(&self.source, self.caret)
4897 } else {
4898 self.caret
4899 }
4900 }
4901 View::Wysiwyg => self.vmap.stop_after(self.caret).unwrap_or(self.caret),
4902 };
4903 let before = self.caret;
4904 self.move_to(target, extend);
4905 self.debug_assert_on_a_stop(before);
4906 }
4907
4908 /// Move to the start of the previous word (⌥← / Ctrl+←).
4909 pub fn move_word_left(&mut self, extend: bool) {
4910 self.goal_col = None;
4911 let before = self.caret;
4912 let target = self.word_left_from(self.caret);
4913 self.move_to(target, extend);
4914 self.debug_assert_on_a_stop(before);
4915 }
4916
4917 /// Move to the end of the next word (⌥→ / Ctrl+→).
4918 pub fn move_word_right(&mut self, extend: bool) {
4919 self.goal_col = None;
4920 let before = self.caret;
4921 let target = self.word_right_from(self.caret);
4922 self.move_to(target, extend);
4923 self.debug_assert_on_a_stop(before);
4924 }
4925
4926 // Word boundaries are found in the space the *view* is in. The source view
4927 // walks the source, because there the source is what's rendered. WYSIWYG
4928 // walks the rendered text instead: `**` is invisible to the user, so it has
4929 // to be invisible to word motion too — a caret parked inside one draws in
4930 // the column after `bold` and types two bytes earlier, and a word-delete
4931 // that stops there shreds the markup into `a ** c`.
4932
4933 /// The word boundary to the left of `off` in the active view's space.
4934 fn word_left_from(&self, off: usize) -> usize {
4935 match self.view {
4936 View::Source => prev_word(&self.source, off),
4937 View::Wysiwyg => self.glyph_word_left(off),
4938 }
4939 }
4940
4941 /// The word boundary to the right of `off` in the active view's space.
4942 fn word_right_from(&self, off: usize) -> usize {
4943 match self.view {
4944 View::Source => next_word(&self.source, off),
4945 View::Wysiwyg => self.glyph_word_right(off),
4946 }
4947 }
4948
4949 /// The character class of the glyph drawn at stop `off`.
4950 ///
4951 /// Read from the source, because a stop points at the source byte its glyph
4952 /// came from — the source *is* where the rendered character is written. What
4953 /// makes the walk glyph space rather than source space is that it only ever
4954 /// visits stops, and the hidden bytes between them have none.
4955 fn class_at(&self, off: usize) -> Class {
4956 self.source
4957 .get(off..)
4958 .and_then(|s| s.chars().next())
4959 .map_or(Class::Space, classify)
4960 }
4961
4962 /// [`next_word`] in glyph space: skip any leading separators, then consume
4963 /// the following word run, with the stop table standing in for the source's
4964 /// characters.
4965 fn glyph_word_right(&self, from: usize) -> usize {
4966 let Some(mut off) = self.vmap.stop_at_or_after(from) else {
4967 return from;
4968 };
4969 let mut in_word = false;
4970 loop {
4971 match self.class_at(off) {
4972 Class::Word => in_word = true,
4973 _ if in_word => return off,
4974 _ => {}
4975 }
4976 match self.vmap.stop_after(off) {
4977 Some(next) => off = next,
4978 None => return off,
4979 }
4980 }
4981 }
4982
4983 /// [`prev_word`] in glyph space: skip separators walking left, then consume
4984 /// the preceding word run.
4985 fn glyph_word_left(&self, from: usize) -> usize {
4986 let Some(mut off) = self.vmap.stop_at_or_before(from) else {
4987 return from;
4988 };
4989 let mut in_word = false;
4990 while let Some(prev) = self.vmap.stop_before(off) {
4991 match self.class_at(prev) {
4992 Class::Word => in_word = true,
4993 _ if in_word => return off,
4994 _ => {}
4995 }
4996 off = prev;
4997 }
4998 off
4999 }
5000
5001 /// After a motion that walks the visual map, the caret must be *on* the map.
5002 /// A stop is the only offset where the caret draws and edits in the same
5003 /// place, and it's the invariant both a caret parked inside an emoji and one
5004 /// parked inside a `**` were quietly breaking.
5005 ///
5006 /// Only when the caret actually moved: a walk with nowhere to go leaves it
5007 /// where it was, which is wherever the floor or a frontend put it rather
5008 /// than somewhere this motion chose.
5009 fn debug_assert_on_a_stop(&self, before: usize) {
5010 debug_assert!(
5011 self.view != View::Wysiwyg
5012 || self.vmap.num_rows() == 0
5013 || self.caret == before
5014 || self.vmap.is_stop(self.caret),
5015 "motion left the caret at {}, which is not a caret stop: it would draw in \
5016 one place and type in another",
5017 self.caret
5018 );
5019 }
5020
5021 // Up and Down run off the ends of the document rather than stopping dead at
5022 // them: Up from the first row lands at the document's start, Down from the
5023 // last at its end. That's Cocoa's rule (`moveUp:`/`moveDown:` past the edge
5024 // are `moveToBeginningOfDocument:`/`moveToEndOfDocument:`), and holding ↓
5025 // reaching the end of the text is what a reader means by it.
5026 //
5027 // The views used to disagree here by accident rather than by decision: the
5028 // source view fell into the edge behaviour through `row_col_to_offset`
5029 // clamping an out-of-range row to the end of the string, while WYSIWYG had
5030 // no row below to walk to and did nothing at all. They share the rule now,
5031 // each in its own space — the source view reaches every byte, WYSIWYG only
5032 // the offsets it draws.
5033
5034 pub fn move_up(&mut self, extend: bool) {
5035 let (row, col) = self.caret_pos();
5036 let goal = self.goal_col.unwrap_or(col);
5037 let target = match self.view {
5038 View::Source => match row.checked_sub(1) {
5039 Some(r) => row_col_to_offset(&self.source, r, goal),
5040 None => self.reachable_start(),
5041 },
5042 // A table's border rules are drawn but hold no caret, so Up steps
5043 // over them to the row that does.
5044 View::Wysiwyg => match self.vmap.navigable_above(row) {
5045 Some(r) => self.row_target(r, goal),
5046 None => self.reachable_start(),
5047 },
5048 };
5049 self.step_vertical(target, goal, extend);
5050 }
5051
5052 pub fn move_down(&mut self, extend: bool) {
5053 let (row, col) = self.caret_pos();
5054 let goal = self.goal_col.unwrap_or(col);
5055 let target = match self.view {
5056 View::Source => match self.source_row_below(row) {
5057 Some(r) => row_col_to_offset(&self.source, r, goal),
5058 None => self.reachable_end(),
5059 },
5060 View::Wysiwyg => match self.vmap.navigable_below(row) {
5061 Some(r) => self.row_target(r, goal),
5062 None => self.reachable_end(),
5063 },
5064 };
5065 self.step_vertical(target, goal, extend);
5066 }
5067
5068 /// Land a vertical motion at `target`, latching the `goal` column it aimed
5069 /// with so the rest of the run keeps aiming there.
5070 ///
5071 /// A motion with nowhere to go changes *nothing*, the goal column included:
5072 /// the latch used to run before the early return at the top of the document,
5073 /// so an Up that did nothing still armed a column, and the next Down aimed
5074 /// at one the caret had never been in.
5075 fn step_vertical(&mut self, target: usize, goal: usize, extend: bool) {
5076 let before = self.caret;
5077 if target == before {
5078 return;
5079 }
5080 self.goal_col = Some(goal);
5081 self.move_to(target, extend);
5082 self.debug_assert_on_a_stop(before);
5083 }
5084
5085 /// The source line below `row`, or `None` when `row` is the last one. Lines
5086 /// are counted by newline, so a trailing one leaves a real, empty last line
5087 /// for the caret to sit on — the document ends below it, not on it.
5088 fn source_row_below(&self, row: usize) -> Option<usize> {
5089 let last = self.source.bytes().filter(|&b| b == b'\n').count();
5090 (row < last).then_some(row + 1)
5091 }
5092
5093 /// Where a vertical motion aiming at the `goal` column lands on visual row
5094 /// `r`: the column clamped to the row, mapped to its offset, then held
5095 /// inside the row's own [bounds](Self::row_bounds) — a wrapped row's last
5096 /// column belongs to the row below, and a gutter's column 0 points at the
5097 /// block rather than at this row.
5098 fn row_target(&self, r: usize, goal: usize) -> usize {
5099 let (start, end) = self.row_bounds(r);
5100 self.vmap
5101 .offset_of_pos(r, goal.min(self.vmap.row_width(r)))
5102 .clamp(start, end)
5103 }
5104
5105 /// The first and last offsets the caret can reach in the active view.
5106 ///
5107 /// Not the same span in both: the source view shows every byte, so it can
5108 /// reach every byte. WYSIWYG reaches only what it draws — hidden frontmatter
5109 /// sits below the first stop, and a document's trailing newline is drawn
5110 /// nowhere and so sits past the last.
5111 fn reachable_start(&self) -> usize {
5112 match self.view {
5113 View::Source => 0,
5114 View::Wysiwyg => self.vmap.stop_at_or_after(0).unwrap_or(self.caret),
5115 }
5116 }
5117
5118 fn reachable_end(&self) -> usize {
5119 match self.view {
5120 View::Source => self.source.len(),
5121 View::Wysiwyg => self
5122 .vmap
5123 .stop_at_or_before(self.source.len())
5124 .unwrap_or(self.caret),
5125 }
5126 }
5127
5128 /// The `[start, end]` offsets visual row `r` *draws* — everything on it,
5129 /// including the space a soft wrap ate off its end, which is drawn on this
5130 /// row however much the offset past it belongs to the next one.
5131 fn row_span(&self, r: usize) -> (usize, usize) {
5132 let start = self
5133 .vmap
5134 .row_start(r)
5135 .unwrap_or_else(|| self.vmap.offset_of_pos(r, 0));
5136 let end = self.vmap.offset_of_pos(r, self.vmap.row_width(r));
5137 (start.min(end), end)
5138 }
5139
5140 /// [`row_span`](Self::row_span) narrowed to where the caret can stand: a
5141 /// soft wrap's shared offset opens the row below (see `pos_of_offset`), so
5142 /// this row's last position is the one before it — the offset before the
5143 /// space the wrap ate, where the caret draws just past the row's last word
5144 /// and types there too.
5145 ///
5146 /// Aiming at the shared offset instead is what stalled End: it is the row's
5147 /// last *column*, so End pressed on the row reached it and then read back as
5148 /// the row below's start, where a second press ran on to that row's end and
5149 /// the next to the one after — End walking down the paragraph a row a press.
5150 fn row_bounds(&self, r: usize) -> (usize, usize) {
5151 let (start, end) = self.row_span(r);
5152 let wraps = self
5153 .vmap
5154 .navigable_below(r)
5155 .and_then(|b| self.vmap.row_start(b))
5156 .is_some_and(|off| off == end);
5157 match wraps {
5158 true => (start, self.vmap.stop_before(end).unwrap_or(end).max(start)),
5159 false => (start, end),
5160 }
5161 }
5162
5163 /// The `[start, end]` of the line Home and End aim at: the visual row in
5164 /// WYSIWYG, the logical line in the source view. Both ends are caret stops.
5165 ///
5166 /// A soft-wrapped row is a line here, because it is one to the eye and the
5167 /// eye is what these keys are aimed by — a reader pressing End means the end
5168 /// of the line they can see. (`select_block_at` wants the opposite and reads
5169 /// the AST for it: a triple-click grabs the whole paragraph, however many
5170 /// rows it folds into.)
5171 fn line_bounds(&self) -> (usize, usize) {
5172 let (row, _) = self.caret_pos();
5173 match self.view {
5174 View::Source => {
5175 let start = line_start(&self.source, row);
5176 (start, line_end_from(&self.source, start))
5177 }
5178 View::Wysiwyg => self.row_bounds(row),
5179 }
5180 }
5181
5182 /// The same line as [`line_bounds`](Self::line_bounds), as far as it is
5183 /// *drawn* — what a kill takes.
5184 ///
5185 /// The two part only at a soft wrap, over the space the wrap ate: the caret
5186 /// can't stand after it (that offset opens the row below, and End stopping
5187 /// there would walk), but it is on this row, and a kill that spared it would
5188 /// leave a double space behind where the row's text had been. Deleting it
5189 /// joins nothing — a wrap is drawn, not written.
5190 fn line_span(&self) -> (usize, usize) {
5191 let (row, _) = self.caret_pos();
5192 match self.view {
5193 View::Source => self.line_bounds(),
5194 View::Wysiwyg => self.row_span(row),
5195 }
5196 }
5197
5198 /// The first offset in `[start, end]` holding something other than
5199 /// whitespace, or `end` when the line holds nothing else — where Home aims.
5200 ///
5201 /// Walks the space the view is in, as word motion does: WYSIWYG steps stops,
5202 /// so a hidden delimiter is never taken for the line's first character (nor
5203 /// landed on), and the source view steps the source it is showing.
5204 fn first_non_space(&self, start: usize, end: usize) -> usize {
5205 let mut off = start;
5206 while off < end {
5207 if self.class_at(off) != Class::Space {
5208 return off;
5209 }
5210 off = match self.view {
5211 View::Source => next_boundary(&self.source, off),
5212 View::Wysiwyg => match self.vmap.stop_after(off) {
5213 Some(next) => next,
5214 None => return end,
5215 },
5216 };
5217 }
5218 end
5219 }
5220
5221 /// Home: to the first character on the line, or to column 0 when the caret
5222 /// is already on it — the two-press toggle every editor spells this way.
5223 /// The indentation is somewhere the caret has to be able to reach and almost
5224 /// never where a reader is headed, so it costs the second press.
5225 pub fn move_home(&mut self, extend: bool) {
5226 self.goal_col = None;
5227 let (start, end) = self.line_bounds();
5228 let text = self.first_non_space(start, end);
5229 let target = if self.caret == text { start } else { text };
5230 let before = self.caret;
5231 self.move_to(target, extend);
5232 self.debug_assert_on_a_stop(before);
5233 }
5234
5235 /// End: to the end of the line.
5236 pub fn move_end(&mut self, extend: bool) {
5237 self.goal_col = None;
5238 let (_, end) = self.line_bounds();
5239 let before = self.caret;
5240 self.move_to(end, extend);
5241 self.debug_assert_on_a_stop(before);
5242 }
5243
5244 /// Hop to the next (Tab) or previous (Shift+Tab) table cell, landing with the
5245 /// cell's whole content selected (see [`Self::select_cell`]). Returns `false`
5246 /// when the caret isn't in a table, or is already in the last/first cell — the
5247 /// frontend then does whatever Tab normally does (indent), so Tab keeps its
5248 /// meaning everywhere else.
5249 pub fn cell_hop(&mut self, forward: bool) -> bool {
5250 let Some((grid, r, c)) = self.table_grid_at(self.caret) else {
5251 return false;
5252 };
5253 // Flatten to document (row-major) order and step one cell either way.
5254 let i: usize = grid[..r].iter().map(Vec::len).sum::<usize>() + c;
5255 let flat: Vec<(usize, usize)> = grid.into_iter().flatten().collect();
5256 let next = if forward {
5257 i.checked_add(1)
5258 } else {
5259 i.checked_sub(1)
5260 };
5261 let Some(&(start, end)) = next.and_then(|j| flat.get(j)) else {
5262 return false; // at the table's edge; leave Tab to the frontend
5263 };
5264 self.select_cell(start, end);
5265 true
5266 }
5267
5268 /// Move the caret to the cell directly above (`down == false`) or below in
5269 /// the same column, landing with the cell's whole content selected (see
5270 /// [`Self::select_cell`]). Returns `false` at the grid's top/bottom edge (or
5271 /// when the caret isn't in a table), so the frontend can fall through — the
5272 /// vertical counterpart of [`cell_hop`].
5273 ///
5274 /// A ragged row that is short a column clamps to its last cell, so Down never
5275 /// falls out of the table over a gap the row above happened to have.
5276 pub fn cell_move_vertical(&mut self, down: bool) -> bool {
5277 let Some((grid, r, c)) = self.table_grid_at(self.caret) else {
5278 return false;
5279 };
5280 let target = match down {
5281 true => r + 1,
5282 false if r == 0 => return false,
5283 false => r - 1,
5284 };
5285 let Some(row) = grid.get(target) else {
5286 return false;
5287 };
5288 let Some(&(start, end)) = row.get(c).or_else(|| row.last()) else {
5289 return false;
5290 };
5291 self.select_cell(start, end);
5292 true
5293 }
5294
5295 /// The table containing `off` as a row-major grid of `(start, end)` cell
5296 /// caret homes, plus the `(row, col)` the caret sits in — `None` when `off`
5297 /// isn't in a table. Read straight off the visual map's laid-out grid, so
5298 /// every cell (an empty one included, whose derived home twig gives no
5299 /// `content_span` for) is present and in the order Tab walks them.
5300 // Grid, row, column — three returns that only ever travel together, and a
5301 // named type for the pair of them would be read at one call site.
5302 #[allow(clippy::type_complexity)]
5303 fn table_grid_at(&self, off: usize) -> Option<(Vec<Vec<(usize, usize)>>, usize, usize)> {
5304 for t in &self.vmap.tables {
5305 let mut pos = None;
5306 let grid: Vec<Vec<(usize, usize)>> = t
5307 .grid
5308 .iter()
5309 .enumerate()
5310 .map(|(r, row)| {
5311 row.cells
5312 .iter()
5313 .enumerate()
5314 .map(|(c, cell)| {
5315 if pos.is_none() && off >= cell.start && off <= cell.end {
5316 pos = Some((r, c));
5317 }
5318 (cell.start, cell.end)
5319 })
5320 .collect()
5321 })
5322 .collect();
5323 if let Some((r, c)) = pos {
5324 return Some((grid, r, c));
5325 }
5326 }
5327 None
5328 }
5329
5330 // ── table key policy ──────────────────────────────────────────────────────
5331 // The three keys a table gives its own meaning — Tab, Return, Shift+Return —
5332 // as one policy every frontend shares, rather than each re-deriving it. Each
5333 // reports whether it acted *as a table key*; a `false` hands the key back to
5334 // the frontend's ordinary handling (indent, newline) so it keeps its meaning
5335 // everywhere else.
5336
5337 /// Tab / Shift+Tab inside a table. Tab steps to the next cell, appending a
5338 /// fresh row and entering it when it runs off the last one; Shift+Tab steps
5339 /// back and simply stays put at the very first cell. `false` when the caret
5340 /// isn't in a table.
5341 pub fn cell_tab(&mut self, forward: bool) -> bool {
5342 if !self.caret_in_table() {
5343 return false;
5344 }
5345 if self.cell_hop(forward) {
5346 return true;
5347 }
5348 // Off the last cell: grow the table by a row and step into its first
5349 // cell. (Shift+Tab at the first cell has nowhere to go and just holds.)
5350 if forward {
5351 self.append_row_and_enter(0);
5352 }
5353 true
5354 }
5355
5356 /// Return inside a table: drop to the cell below in the same column,
5357 /// appending a new row when the caret is already in the last one. `false`
5358 /// when the caret isn't in a table, so the frontend inserts a newline.
5359 pub fn cell_return(&mut self) -> bool {
5360 if !self.caret_in_table() {
5361 return false;
5362 }
5363 if self.cell_move_vertical(true) {
5364 return true;
5365 }
5366 // Already on the last row: grow one below and drop into the same column.
5367 let col = self.table_grid_at(self.caret).map_or(0, |(_, _, c)| c);
5368 self.append_row_and_enter(col);
5369 true
5370 }
5371
5372 /// Append a row below the caret's (last) row and land in `col` of it. The
5373 /// caret is in the last row, so twig's "insert below" makes the fresh row the
5374 /// table's new last — but twig re-spells the whole table, moving every byte,
5375 /// so the destination is read back from the rebuilt grid by the table's
5376 /// position (stable across a row insert), not from the pre-edit caret.
5377 fn append_row_and_enter(&mut self, col: usize) {
5378 let table = self.caret_table_index();
5379 self.table_insert_row(true);
5380 self.rebuild_map();
5381 let Some((start, end)) = table
5382 .and_then(|ti| self.vmap.tables.get(ti))
5383 .and_then(|t| t.grid.last())
5384 .and_then(|row| row.cells.get(col.min(row.cells.len().saturating_sub(1))))
5385 .map(|cell| (cell.start, cell.end))
5386 else {
5387 return;
5388 };
5389 self.select_cell(start, end);
5390 }
5391
5392 /// The index, among the document's tables, of the one the caret sits in —
5393 /// `None` when it's in none. Used to re-find a table after an edit re-spells
5394 /// it (a row insert leaves the table order unchanged).
5395 fn caret_table_index(&self) -> Option<usize> {
5396 let off = self.caret;
5397 self.vmap.tables.iter().position(|t| {
5398 t.grid
5399 .iter()
5400 .any(|row| row.cells.iter().any(|c| off >= c.start && off <= c.end))
5401 })
5402 }
5403
5404 /// Shift+Return inside a table: insert a hard line break *within* the current
5405 /// cell, via twig's `insert_line_break`. `false` when the caret isn't in a
5406 /// table, so the frontend inserts an ordinary line break.
5407 ///
5408 /// A table row is a single source line, so the newline-spelled hard break
5409 /// can't live in a cell. twig spells the in-cell break the format's way
5410 /// (`<br>` for Markdown) and reparses it as a *semantic* `hard_break`, so the
5411 /// break round-trips as structure the renderer reads back as a line — not the
5412 /// opaque raw HTML the old raw-splice left behind.
5413 ///
5414 /// Djot has no idiomatic in-cell break, so twig refuses it
5415 /// (`UnsupportedFormat`) rather than emit a `<br>` that any other djot reader
5416 /// would render as the literal text `<br>`. The gesture is still *consumed*
5417 /// there — returning `false` would let the frontend insert a real newline,
5418 /// which splits the one-line row — it just leaves the cell unchanged and says
5419 /// so on the status line. A rollback (`EditConflict`) is swallowed the same.
5420 ///
5421 /// Which formats refuse is [`Capabilities::cell_line_break`], and the two
5422 /// have to be read together: djot is not the only `false`, and naming it in
5423 /// the message was already a guess that HTML — which spells the break as its
5424 /// own `<br>` — would have made wrong.
5425 pub fn cell_line_break(&mut self) -> bool {
5426 if !self.caret_in_table() {
5427 return false;
5428 }
5429 self.record_caret();
5430 match self.editor.insert_line_break(self.caret) {
5431 Ok(change) => {
5432 self.last_edit_kind = None;
5433 self.refresh();
5434 self.caret = change.new.end;
5435 self.anchor = None;
5436 self.goal_col = None;
5437 self.clamp_caret();
5438 self.dirty = self.source != self.clean_source;
5439 self.status = None;
5440 self.record_caret();
5441 }
5442 Err(twig::Error::UnsupportedFormat) => {
5443 self.status = Some(format!(
5444 "in-cell line breaks aren't supported in {}",
5445 self.format_name()
5446 ));
5447 }
5448 Err(_) => {}
5449 }
5450 true
5451 }
5452
5453 /// Rebuild the visual map at the width the last build used. A structural edit
5454 /// bumps the revision and swaps the source in, but leaves the *map* stale;
5455 /// when a single gesture edits and then moves over the result (Tab appending
5456 /// a row, then stepping into it), the move needs the map to already show the
5457 /// edit rather than waiting for the frontend's next frame.
5458 fn rebuild_map(&mut self) {
5459 let wrap = self.vmap_key.as_ref().and_then(|(_, w, _)| *w);
5460 self.build_map(wrap);
5461 }
5462
5463 /// Move the caret to the very start of the document (⌘↑ on macOS,
5464 /// Ctrl+Home on Windows/Linux).
5465 pub fn move_doc_start(&mut self, extend: bool) {
5466 self.goal_col = None;
5467 self.move_to(0, extend);
5468 }
5469
5470 /// Move the caret to the very end of the document (⌘↓ on macOS,
5471 /// Ctrl+End on Windows/Linux).
5472 pub fn move_doc_end(&mut self, extend: bool) {
5473 self.goal_col = None;
5474 let end = self.source.len();
5475 self.move_to(end, extend);
5476 }
5477
5478 /// Point the caret at the body cell `(row, col)` the mouse landed on —
5479 /// `col` being a cell of the terminal grid, which is what a display column
5480 /// is. A click on the far cell of a wide character lands at that
5481 /// character's start; the mapping's own doc-comments carry the rule.
5482 pub fn click(&mut self, row: usize, col: usize, extend: bool) {
5483 self.goal_col = None;
5484 let target = match self.view {
5485 View::Source => row_col_to_offset(&self.source, row, col),
5486 View::Wysiwyg => self.vmap.offset_of_pos(row, col),
5487 };
5488 let before = self.caret;
5489 self.move_to(target, extend);
5490 self.debug_assert_on_a_stop(before);
5491 }
5492
5493 /// Settle `scroll` for a frame about to be drawn: follow the caret onto the
5494 /// screen if it has moved since the last frame, and never scroll past the
5495 /// last of `rows`.
5496 ///
5497 /// Only if it has *moved* — that's the whole point. Revealing the caret on
5498 /// every frame ties the viewport to it, and a scroll wheel that fights the
5499 /// caret for the viewport loses: the view snaps back the instant it tries to
5500 /// pass the caret's row, so the document can't be scrolled beyond what's
5501 /// already on screen. A caret move is the frontend's cue to follow; a scroll
5502 /// with the caret sitting still is the reader's cue to leave it alone.
5503 pub fn follow_caret(&mut self, caret_row: usize, height: usize, rows: usize) {
5504 if self.drawn_caret != Some(self.caret) {
5505 if caret_row < self.scroll {
5506 self.scroll = caret_row;
5507 } else if height > 0 && caret_row >= self.scroll + height {
5508 self.scroll = caret_row + 1 - height;
5509 }
5510 self.drawn_caret = Some(self.caret);
5511 }
5512 self.scroll = self.scroll.min(rows.saturating_sub(1));
5513 }
5514
5515 /// The caret's screen position `(row, col)` in the active view's grid, with
5516 /// `col` a display column: the cell to draw the caret in, which on a line of
5517 /// `你好` or emoji is not the count of characters before it.
5518 pub fn caret_pos(&self) -> (usize, usize) {
5519 match self.view {
5520 View::Source => offset_to_row_col(&self.source, self.caret),
5521 View::Wysiwyg => self.vmap.pos_of_offset(self.caret),
5522 }
5523 }
5524
5525 fn clamp_caret(&mut self) {
5526 if self.caret > self.source.len() {
5527 self.caret = self.source.len();
5528 }
5529 // In WYSIWYG the caret can't sit inside hidden frontmatter; lift it (and
5530 // any selection anchor) to the first rendered offset.
5531 let floor = self.caret_floor();
5532 if self.caret < floor {
5533 self.caret = floor;
5534 }
5535 if let Some(a) = self.anchor
5536 && a < floor
5537 {
5538 self.anchor = Some(floor);
5539 }
5540 while self.caret > 0 && !self.source.is_char_boundary(self.caret) {
5541 self.caret -= 1;
5542 }
5543 }
5544}
5545
5546// ── byte-offset ⇄ (row, col) helpers ─────────────────────────────────────────
5547
5548// Left/right motion and backspace/delete step by *grapheme cluster*, not
5549// codepoint, so an emoji (a ZWJ sequence) or a base letter plus its combining
5550// marks moves and deletes as the single character a user sees. Grapheme
5551// boundaries are a superset of char boundaries, so the caret stays valid for twig.
5552
5553/// How an insert of `text` groups for undo: a single typed character folds into
5554/// the run of typing around it, while a newline or a multi-character insert is a
5555/// step of its own.
5556fn typed_edit_kind(text: &str) -> EditKind {
5557 if text.chars().take(2).count() == 1 && text != "\n" {
5558 EditKind::Insert
5559 } else {
5560 EditKind::Other
5561 }
5562}
5563
5564fn prev_boundary(s: &str, i: usize) -> usize {
5565 let mut cursor = GraphemeCursor::new(i, s.len(), true);
5566 cursor.prev_boundary(s, 0).ok().flatten().unwrap_or(0)
5567}
5568
5569fn next_boundary(s: &str, i: usize) -> usize {
5570 let mut cursor = GraphemeCursor::new(i, s.len(), true);
5571 cursor.next_boundary(s, 0).ok().flatten().unwrap_or(s.len())
5572}
5573
5574// ── word boundaries ──────────────────────────────────────────────────────────
5575// The shared primitive behind word-wise motion, word deletion, and
5576// double-click-to-select-a-word. A "word" is a maximal run of one character
5577// class; whitespace and punctuation are their own classes, so motion skips
5578// cleanly between them the way native text fields do.
5579
5580#[derive(PartialEq, Eq, Clone, Copy)]
5581enum Class {
5582 Word,
5583 Space,
5584 Other,
5585}
5586
5587/// The source range of an inline node's own visible text — the part of it a
5588/// WYSIWYG caret can reach, as against the delimiters that only spell it.
5589/// `None` for a node with no interior to empty (a `str`, a break).
5590///
5591/// twig reports no `content_span` for `verbatim`/`inline_math`, whose text sits
5592/// one delimiter in from the span — the same place the renderer maps it to. A
5593/// longer fence (`` ``a`` ``) breaks that assumption, so the guess is checked
5594/// against the source rather than trusted: a range guessed wrong here is text
5595/// deleted wrong.
5596fn inline_content_span(n: &FlatNode, source: &str) -> Option<std::ops::Range<usize>> {
5597 if let Some(span) = n.content_span.clone() {
5598 return Some(span);
5599 }
5600 match n.kind.as_str() {
5601 "verbatim" | "inline_math" => {
5602 let text = n.text.as_ref()?;
5603 let start = n.span.start + 1;
5604 let range = start..start + text.len();
5605 (source.get(range.clone()) == Some(text.as_str())).then_some(range)
5606 }
5607 _ => None,
5608 }
5609}
5610
5611/// The `id` a node declares, or `None` for one that declares none — the
5612/// attribute djot writes for a `{#v1}` and mints for a heading.
5613///
5614/// A bare attribute (`{#v1 hidden}`'s `hidden`) has no value, and a bare `id`
5615/// names nothing, so it reads as absent rather than as the empty string.
5616fn declared_id(n: &FlatNode) -> Option<&str> {
5617 n.attrs.iter().find(|(k, _)| k == "id")?.1.as_deref()
5618}
5619
5620/// A heading's words reduced to the form a link fragment spells them in:
5621/// lowercase, runs of anything else collapsed to a single `-`, with none left
5622/// dangling at either end. `## Some Heading Here` → `some-heading-here`.
5623///
5624/// The rule every Markdown renderer follows, and applied to djot's own auto-ids
5625/// too so that `#some-heading-here` and `#Some-Heading-Here` are one question.
5626/// Unicode-aware (`is_alphanumeric`, not an ASCII test), because a heading in
5627/// any other language is still a heading someone will link to. Underscores
5628/// survive for the same reason they do on the web: they are word characters
5629/// wherever identifiers are written.
5630fn slug(text: &str) -> String {
5631 let mut out = String::new();
5632 let mut pending = false;
5633 for c in text.chars() {
5634 if c.is_alphanumeric() || c == '_' {
5635 if pending && !out.is_empty() {
5636 out.push('-');
5637 }
5638 pending = false;
5639 out.extend(c.to_lowercase());
5640 } else {
5641 pending = true;
5642 }
5643 }
5644 out
5645}
5646
5647fn is_block_container(kind: &Kind) -> bool {
5648 matches!(
5649 kind,
5650 Kind::Doc
5651 | Kind::Section
5652 | Kind::BlockQuote
5653 | Kind::BulletList
5654 | Kind::OrderedList
5655 | Kind::TaskList
5656 | Kind::ListItem
5657 | Kind::TaskListItem
5658 // Every `container` — a directive in any of its three forms, or a
5659 // promoted HTML element. A *text* directive is really inline, so
5660 // claiming it here is a small overreach, and the deliberate one this
5661 // function's kind-only peer `is_inline_kind` documents: the pair is
5662 // consulted together, and answering "block container" for something
5663 // inline is what keeps an ancestor walk from stopping short of the
5664 // paragraph that actually holds it.
5665 | Kind::Container
5666 )
5667}
5668
5669/// The `[start, end)` byte range of the source line containing `off` (newline
5670/// excluded) — the fallback when `off` sits outside any AST block (e.g. a blank
5671/// line between paragraphs).
5672fn source_line_range(s: &str, off: usize) -> std::ops::Range<usize> {
5673 let off = off.min(s.len());
5674 let start = s[..off].rfind('\n').map(|p| p + 1).unwrap_or(0);
5675 let end = s[off..].find('\n').map(|p| off + p).unwrap_or(s.len());
5676 start..end
5677}
5678
5679/// How many leading bytes an outdent takes off `line`: a whole indent level
5680/// where the line has one, and whatever it has where it has less.
5681///
5682/// A leading tab counts as a level on its own. It's indentation some other
5683/// editor wrote, and one tab is one level everywhere it came from — measuring it
5684/// in spaces it doesn't contain would leave it untouchable.
5685fn outdent_width(line: &str, unit: usize) -> usize {
5686 if line.starts_with('\t') {
5687 return 1;
5688 }
5689 line.bytes().take(unit).take_while(|b| *b == b' ').count()
5690}
5691
5692/// A list marker found at the head of a line, together with everything before it
5693/// that a sibling line has to repeat.
5694///
5695/// The three offsets differ only inside a block quote, where `> - b` opens with
5696/// a `> ` quote marker the line's own text doesn't own. Outside one they collapse:
5697/// `line_start == marker_start`, and `text` is the plain `" - "`.
5698#[derive(Clone, Debug)]
5699struct ListMarker {
5700 /// The line's first byte.
5701 line_start: usize,
5702 /// Where the marker proper begins, past any quote prefix. The offset to hand
5703 /// the AST: a quoted item's span opens at its bullet, not at the `>`.
5704 marker_start: usize,
5705 /// `line_start` through the marker's trailing space — quote prefix, indent
5706 /// and bullet together, which is what the next item's line opens with.
5707 text: String,
5708}
5709
5710impl ListMarker {
5711 /// Where the item's content starts — one past the marker's trailing space.
5712 fn content_start(&self) -> usize {
5713 self.line_start + self.text.len()
5714 }
5715}
5716
5717fn classify(c: char) -> Class {
5718 if c == '_' || c.is_alphanumeric() {
5719 Class::Word
5720 } else if c.is_whitespace() {
5721 Class::Space
5722 } else {
5723 Class::Other
5724 }
5725}
5726
5727/// The offset at the end of the next word to the right of `i` (⌥→ / Ctrl+→):
5728/// skip any leading separators, then consume the following word run.
5729fn next_word(s: &str, i: usize) -> usize {
5730 let mut off = i;
5731 let mut in_word = false;
5732 for c in s[i..].chars() {
5733 if classify(c) == Class::Word {
5734 in_word = true;
5735 } else if in_word {
5736 break;
5737 }
5738 off += c.len_utf8();
5739 }
5740 off
5741}
5742
5743/// The offset at the start of the word to the left of `i` (⌥← / Ctrl+←):
5744/// skip separators walking left, then consume the preceding word run.
5745fn prev_word(s: &str, i: usize) -> usize {
5746 let mut off = i;
5747 let mut in_word = false;
5748 for c in s[..i].chars().rev() {
5749 if classify(c) == Class::Word {
5750 in_word = true;
5751 } else if in_word {
5752 break;
5753 }
5754 off -= c.len_utf8();
5755 }
5756 off
5757}
5758
5759/// The `[start, end)` run of same-class characters surrounding `off` — the
5760/// word (or whitespace/punctuation run) a double-click selects. At end-of-text
5761/// the run ending there is used.
5762fn word_range_at(s: &str, off: usize) -> (usize, usize) {
5763 if s.is_empty() {
5764 return (0, 0);
5765 }
5766 let off = off.min(s.len());
5767 let reference = if off < s.len() {
5768 s[off..].chars().next()
5769 } else {
5770 s[..off].chars().next_back()
5771 };
5772 let Some(rc) = reference else {
5773 return (off, off);
5774 };
5775 let class = classify(rc);
5776
5777 let mut start = off;
5778 for c in s[..start].chars().rev() {
5779 if classify(c) == class {
5780 start -= c.len_utf8();
5781 } else {
5782 break;
5783 }
5784 }
5785 let mut end = off;
5786 for c in s[end..].chars() {
5787 if classify(c) == class {
5788 end += c.len_utf8();
5789 } else {
5790 break;
5791 }
5792 }
5793 (start, end)
5794}
5795
5796/// `(row, col)` of byte offset `off`, `col` counted in *display columns* from
5797/// the line's start — terminal cells, not characters, so the column names the
5798/// cell the caret is drawn in even on a line of `你好` or emoji.
5799fn offset_to_row_col(s: &str, off: usize) -> (usize, usize) {
5800 let off = off.min(s.len());
5801 let mut row = 0;
5802 let mut line_start = 0;
5803 for (i, &b) in s.as_bytes().iter().enumerate() {
5804 if i >= off {
5805 break;
5806 }
5807 if b == b'\n' {
5808 row += 1;
5809 line_start = i + 1;
5810 }
5811 }
5812 (row, wysiwyg::text_width(&s[line_start..off]))
5813}
5814
5815/// The byte offset at display column `col` of `row` (clamped to that line's
5816/// end) — the inverse of [`offset_to_row_col`], which it has to agree with.
5817///
5818/// A column landing *inside* a character — the second cell of `你`, or any cell
5819/// but the first of an emoji — resolves to that character's start, which is the
5820/// column the caret would have been drawn at to begin with. So both cells of a
5821/// wide character mean the character, and every offset survives the round trip
5822/// out to a column and back. The walk steps by grapheme cluster for the same
5823/// reason the caret does: a cluster is the character, and the cells belong to it
5824/// rather than to the codepoints spelling it.
5825fn row_col_to_offset(s: &str, row: usize, col: usize) -> usize {
5826 let start = line_start(s, row);
5827 let end = line_end_from(s, start);
5828 let mut off = start;
5829 let mut at = 0; // the display column `off` sits at
5830 while off < end {
5831 let next = next_boundary(s, off).min(end);
5832 let cells = wysiwyg::text_width(&s[off..next]);
5833 if at + cells > col {
5834 break; // `col` is one of this cluster's own cells
5835 }
5836 at += cells;
5837 off = next;
5838 }
5839 off
5840}
5841
5842fn line_start(s: &str, row: usize) -> usize {
5843 if row == 0 {
5844 return 0;
5845 }
5846 let mut r = 0;
5847 for (i, &b) in s.as_bytes().iter().enumerate() {
5848 if b == b'\n' {
5849 r += 1;
5850 if r == row {
5851 return i + 1;
5852 }
5853 }
5854 }
5855 s.len()
5856}
5857
5858fn line_end_from(s: &str, start: usize) -> usize {
5859 s[start..].find('\n').map(|p| start + p).unwrap_or(s.len())
5860}
5861
5862/// twig's node-kind name for an inline mark, back to the [`InlineKind`] a
5863/// frontend names when it calls [`Doc::toggle`] — the inverse of the mapping
5864/// twig applies writing the mark out, so the toolbar can light the same button
5865/// that made the node.
5866///
5867/// `None` for every other kind, including the inline nodes that aren't marks at
5868/// all (`str`, `link`, `image`, the math and break kinds): they're things a
5869/// caret stands in, not formatting a button toggles.
5870fn inline_kind(kind: &Kind) -> Option<InlineKind> {
5871 Some(match kind {
5872 Kind::Strong => InlineKind::Strong,
5873 Kind::Emph => InlineKind::Emph,
5874 Kind::Verbatim => InlineKind::Verbatim,
5875 Kind::Mark => InlineKind::Mark,
5876 Kind::Superscript => InlineKind::Superscript,
5877 Kind::Subscript => InlineKind::Subscript,
5878 Kind::Insert => InlineKind::Insert,
5879 Kind::Delete => InlineKind::Delete,
5880 _ => return None,
5881 })
5882}
5883
5884/// A watermark for a file's contents (see `Doc::disk_hash`).
5885///
5886/// `DefaultHasher` is not stable across Rust releases, which doesn't matter: a
5887/// watermark is compared only against one taken by the same process moments
5888/// earlier, and never outlives it. 64 bits leaves a collision — an external edit
5889/// that hashes to exactly what leaf wrote — at odds no filesystem race gets near.
5890fn hash_bytes(bytes: &[u8]) -> u64 {
5891 use std::hash::{Hash, Hasher};
5892 let mut h = std::collections::hash_map::DefaultHasher::new();
5893 bytes.hash(&mut h);
5894 h.finish()
5895}
5896
5897#[cfg(feature = "fs")]
5898fn detect_format(path: &Path) -> Result<Format> {
5899 let ext = path
5900 .extension()
5901 .and_then(|e| e.to_str())
5902 .unwrap_or("")
5903 .to_ascii_lowercase();
5904 Ok(match ext.as_str() {
5905 "dj" | "djot" => Format::Djot,
5906 "md" | "markdown" => Format::Markdown,
5907 "xml" => Format::Xml,
5908 "html" | "htm" => Format::Html,
5909 other => return Err(anyhow!("unknown document extension: .{other}")),
5910 })
5911}
5912
5913#[cfg(test)]
5914mod tests {
5915 use super::*;
5916
5917 /// A document open in `view`. WYSIWYG motion reads the visual map, which the
5918 /// renderer stamps each frame, so the map is built here too — a WYSIWYG doc
5919 /// without one is a view no user is ever in.
5920 fn doc_in(view: View, name: &str, body: &str) -> Doc {
5921 // The fixture name doubles as the temp file's, so two tests picking the
5922 // same one raced under the parallel runner and read each other's body —
5923 // a green suite proving the wrong thing. The counter makes that
5924 // unreachable rather than asking every future caller to notice.
5925 static SEQ: std::sync::atomic::AtomicUsize = std::sync::atomic::AtomicUsize::new(0);
5926 let seq = SEQ.fetch_add(1, std::sync::atomic::Ordering::Relaxed);
5927 let mut p = std::env::temp_dir();
5928 p.push(format!("leaf_test_{name}_{seq}.md"));
5929 std::fs::write(&p, body).unwrap();
5930 let mut d = Doc::open(p).unwrap();
5931 d.view = view;
5932 if view == View::Wysiwyg {
5933 d.build_visual(80);
5934 }
5935 d
5936 }
5937
5938 // Source-view document for the source-behaviour tests. `Doc::open` now
5939 // defaults to WYSIWYG (leaf's default view), so pin the source view here;
5940 // `wysiwyg_doc` builds the rich-text variant on top of this.
5941 fn doc_with(name: &str, body: &str) -> Doc {
5942 doc_in(View::Source, name, body)
5943 }
5944
5945 /// Every visual row's drawn text — what the reader actually sees, which is
5946 /// the only thing the reveal preference is supposed to change.
5947 fn drawn_rows(d: &Doc) -> Vec<String> {
5948 d.vmap
5949 .rows
5950 .iter()
5951 .map(|r| r.glyphs.iter().map(|g| g.ch).collect())
5952 .collect()
5953 }
5954
5955 /// Put the caret at the first byte of `needle` and rebuild, so the row under
5956 /// it becomes the revealed line.
5957 fn caret_at(d: &mut Doc, needle: &str) {
5958 d.caret = d.source.find(needle).expect("needle in source");
5959 d.build_visual(80);
5960 }
5961
5962 #[test]
5963 fn blockquote_after_a_list_is_not_bulleted() {
5964 // twig nests a following top-level block quote under the `bullet_list`
5965 // (a direct child, not a `list_item`). The map must render it de-nested —
5966 // `│ quote`, never `• │ quote` — with a blank separator, like any block
5967 // that follows a list. Regression for the "combined list + blockquote" bug.
5968 let mut d = doc_in(View::Wysiwyg, "bq_after_list", "- item\n\n> quote\n");
5969 d.build_visual(80);
5970 let rows: Vec<String> = d
5971 .vmap
5972 .rows
5973 .iter()
5974 .map(|r| r.glyphs.iter().map(|g| g.ch).collect())
5975 .collect();
5976 assert!(
5977 rows.iter().any(|r| r == "│ quote"),
5978 "block quote should render on its own gutter, got rows: {rows:?}"
5979 );
5980 assert!(
5981 !rows.iter().any(|r| r.contains('•') && r.contains('│')),
5982 "no row should carry both a bullet and a quote gutter, got rows: {rows:?}"
5983 );
5984 }
5985
5986 // ── the map is built at most once per (revision, wrap) ───────────────────
5987 //
5988 // A frontend repaints for reasons that have nothing to do with the text — a
5989 // blinking caret, a scroll — and rebuilding the map is O(document). These
5990 // pin *that the cache fires*, which a passing suite can't tell you: a cache
5991 // that never hits is invisible to every other test in this file.
5992 //
5993 // The probe is to wreck the built map and ask for it again. A rebuild
5994 // repairs it; a cache hit hands the wreckage straight back. Nothing else
5995 // can distinguish the two from outside.
5996
5997 #[test]
5998 fn a_rebuild_with_nothing_changed_reuses_the_map() {
5999 let mut d = doc_in(View::Wysiwyg, "cache_hit", "# Title\n\nbody\n");
6000 d.build_visual(80);
6001 assert!(!d.vmap.rows.is_empty());
6002 d.vmap.rows.clear(); // wreck it
6003 d.build_visual(80);
6004 assert!(
6005 d.vmap.rows.is_empty(),
6006 "the map was rebuilt though nothing changed — the cache never fired"
6007 );
6008 }
6009
6010 #[test]
6011 fn an_edit_rebuilds_the_map() {
6012 let mut d = doc_in(View::Wysiwyg, "cache_edit", "# Title\n\nbody\n");
6013 d.build_visual(80);
6014 let before = d.revision();
6015 d.vmap.rows.clear();
6016 d.insert("x");
6017 d.build_visual(80);
6018 assert!(d.revision() > before, "an edit must move the revision");
6019 assert!(
6020 !d.vmap.rows.is_empty(),
6021 "an edited document must not paint from a stale map"
6022 );
6023 }
6024
6025 #[test]
6026 fn a_width_change_rebuilds_the_map() {
6027 // The map is a function of the wrap width too, so a resize is a miss
6028 // even though the text is untouched.
6029 let mut d = doc_in(
6030 View::Wysiwyg,
6031 "cache_width",
6032 "one two three four five six\n",
6033 );
6034 d.build_visual(80);
6035 d.vmap.rows.clear();
6036 d.build_visual(12);
6037 assert!(!d.vmap.rows.is_empty(), "a resize must rebuild the map");
6038 // And the unwrapped map is its own key, not the same as any width.
6039 d.vmap.rows.clear();
6040 d.build_visual_unwrapped();
6041 assert!(!d.vmap.rows.is_empty(), "unwrapped is a different map");
6042 }
6043
6044 #[test]
6045 fn a_motion_does_not_rebuild_the_map() {
6046 // The whole point: moving the caret changes nothing the map is built
6047 // from. If a motion bumped the revision, every arrow key would cost a
6048 // full rebuild and the cache would be worthless.
6049 let mut d = doc_in(View::Wysiwyg, "cache_motion", "# Title\n\nbody text\n");
6050 d.build_visual(80);
6051 let rev = d.revision();
6052 d.move_right(false);
6053 d.move_right(true);
6054 d.move_down(false);
6055 assert_eq!(d.revision(), rev, "a motion must not move the revision");
6056 d.vmap.rows.clear();
6057 d.build_visual(80);
6058 assert!(
6059 d.vmap.rows.is_empty(),
6060 "a motion should not rebuild the map"
6061 );
6062 }
6063
6064 #[test]
6065 fn saving_does_not_rebuild_the_map() {
6066 // Saving changes `dirty`, not the text.
6067 let mut d = doc_in(View::Wysiwyg, "cache_save", "# Title\n\nbody\n");
6068 d.insert("x");
6069 d.build_visual(80);
6070 let rev = d.revision();
6071 d.save();
6072 assert_eq!(d.revision(), rev, "a save must not move the revision");
6073 assert!(!d.dirty, "the save should have cleaned the document");
6074 }
6075
6076 #[test]
6077 fn a_reload_rebuilds_the_map() {
6078 // Reload replaces the text without going through `refresh`, so it has to
6079 // move the revision itself — else the editor paints the old file.
6080 let mut d = doc_in(View::Wysiwyg, "cache_reload", "# Title\n\nbody\n");
6081 d.build_visual(80);
6082 let rev = d.revision();
6083 std::fs::write(&d.path, "# Other\n\nwholly new\n").unwrap();
6084 d.reload();
6085 assert!(d.revision() > rev, "a reload must move the revision");
6086 d.build_visual(80);
6087 let text: String = d
6088 .vmap
6089 .rows
6090 .iter()
6091 .flat_map(|r| r.glyphs.iter().map(|g| g.ch))
6092 .collect();
6093 assert!(
6094 text.contains("wholly new"),
6095 "the reloaded text should be on screen, got {text:?}"
6096 );
6097 }
6098
6099 // ── golden-case harness ──────────────────────────────────────────────────
6100 // The pattern the whole parity suite can reuse: write a fixture with the
6101 // caret marked by `|`, run one action, and compare the rendered result —
6102 // also caret-marked — against the expected string. One readable line per
6103 // behavior, and it exercises the exact `Doc` ops both frontends call.
6104
6105 /// Split a `|`-marked fixture into `(source, caret_offset)`.
6106 fn parse_caret(marked: &str) -> (String, usize) {
6107 let caret = marked.find('|').expect("fixture needs a `|` caret marker");
6108 (marked.replacen('|', "", 1), caret)
6109 }
6110
6111 /// Render a doc's source with `|` at the caret (and `[`…`]` around any
6112 /// selection) so a result reads like the fixtures.
6113 fn render_caret(d: &Doc) -> String {
6114 // (offset, rank, char); rank keeps coincident markers ordered `[ | ]`
6115 // so the caret always renders inside its own selection.
6116 let mut marks: Vec<(usize, u8, char)> = vec![(d.caret, 1, '|')];
6117 if let Some((s, e)) = d.selection() {
6118 marks.push((s, 0, '['));
6119 marks.push((e, 2, ']'));
6120 }
6121 // Insert right-to-left: descending offset, then descending rank.
6122 marks.sort_by(|a, b| b.0.cmp(&a.0).then(b.1.cmp(&a.1)));
6123 let mut out = d.source.clone();
6124 for (at, _, ch) in marks {
6125 out.insert(at, ch);
6126 }
6127 out
6128 }
6129
6130 /// Load a `|`-marked fixture, run `action`, return the caret-marked result.
6131 fn golden(name: &str, marked: &str, action: impl FnOnce(&mut Doc)) -> String {
6132 golden_in(View::Source, name, marked, action)
6133 }
6134
6135 /// [`golden`] in a chosen view — the editing ops are the view's to share, so
6136 /// the same fixture has to read the same way in both.
6137 fn golden_in(view: View, name: &str, marked: &str, action: impl FnOnce(&mut Doc)) -> String {
6138 let (src, caret) = parse_caret(marked);
6139 let mut d = doc_in(view, name, &src);
6140 d.caret = caret;
6141 action(&mut d);
6142 render_caret(&d)
6143 }
6144
6145 #[test]
6146 fn word_motion_walks_word_by_word() {
6147 let g = |m, f: fn(&mut Doc)| golden("word_motion", m, f);
6148 assert_eq!(
6149 g("hello wor|ld", |d| d.move_word_left(false)),
6150 "hello |world"
6151 );
6152 assert_eq!(
6153 g("hello| world", |d| d.move_word_left(false)),
6154 "|hello world"
6155 );
6156 assert_eq!(
6157 g("hel|lo world", |d| d.move_word_right(false)),
6158 "hello| world"
6159 );
6160 assert_eq!(
6161 g("hello| world", |d| d.move_word_right(false)),
6162 "hello world|"
6163 );
6164 // Punctuation is its own class, so motion stops at the boundary.
6165 assert_eq!(g("|foo.bar", |d| d.move_word_right(false)), "foo|.bar");
6166 }
6167
6168 #[test]
6169 fn word_motion_extends_the_selection_when_asked() {
6170 assert_eq!(
6171 golden("word_sel", "hello |world", |d| d.move_word_right(true)),
6172 "hello [world|]"
6173 );
6174 }
6175
6176 #[test]
6177 fn delete_word_removes_a_whole_word() {
6178 let g = |m, f: fn(&mut Doc)| golden("del_word", m, f);
6179 assert_eq!(g("hello world|", |d| d.delete_word_back()), "hello |");
6180 assert_eq!(g("hello |world", |d| d.delete_word_forward()), "hello |");
6181 assert_eq!(g("foo |bar baz", |d| d.delete_word_back()), "|bar baz");
6182 }
6183
6184 // ── Home / End ───────────────────────────────────────────────────────────
6185
6186 #[test]
6187 fn home_toggles_between_the_line_s_text_and_its_margin() {
6188 // Source: the indentation is what the toggle is for. WYSIWYG resolves an
6189 // indent to the markup it spells everywhere it means one, so the fixture
6190 // with whitespace left to walk is a code block, which is verbatim.
6191 let g = |m, f: fn(&mut Doc)| golden("smart_home", m, f);
6192 assert_eq!(g(" inden|ted", |d| d.move_home(false)), " |indented");
6193 assert_eq!(g(" |indented", |d| d.move_home(false)), "| indented");
6194 assert_eq!(g("| indented", |d| d.move_home(false)), " |indented");
6195 // A line with no indentation has one place to go, so the toggle is a
6196 // no-op rather than a trip to nowhere.
6197 assert_eq!(g("hel|lo", |d| d.move_home(false)), "|hello");
6198 assert_eq!(g("|hello", |d| d.move_home(false)), "|hello");
6199
6200 let mut d = wysiwyg_doc("smart_home_wys", "```\n indented\n```\n");
6201 let indent = d.source.find(" indented").unwrap();
6202 d.caret = indent + 6; // inside "indented"
6203 d.move_home(false);
6204 assert_eq!(
6205 d.caret,
6206 indent + 4,
6207 "wysiwyg: Home aims at the code line's text"
6208 );
6209 d.move_home(false);
6210 assert_eq!(
6211 d.caret, indent,
6212 "wysiwyg: the second press takes the indent"
6213 );
6214 d.move_home(false);
6215 assert_eq!(d.caret, indent + 4, "wysiwyg: the toggle swaps back");
6216 }
6217
6218 #[test]
6219 fn end_takes_the_line_the_view_is_showing() {
6220 // The line differs by view for the same document, and that is the point:
6221 // a bare newline inside a paragraph is a soft break, which WYSIWYG draws
6222 // as a space on one row and the source view as two lines.
6223 let mut d = doc_with("end_src", "one two\nthree\n");
6224 d.caret = 1;
6225 d.move_end(false);
6226 assert_eq!(d.caret, 7, "source: the end of the source line");
6227
6228 let mut d = wysiwyg_doc("end_wys", "one two\nthree\n");
6229 d.caret = 1;
6230 d.move_end(false);
6231 assert_eq!(
6232 d.caret, 13,
6233 "wysiwyg: the end of the row, soft break and all"
6234 );
6235 }
6236
6237 #[test]
6238 fn home_and_end_extend_the_selection_when_asked() {
6239 for (view, tag) in VIEWS {
6240 let mut d = doc_in(view, &format!("home_end_ext_{tag}"), "hello world");
6241 d.caret = 6;
6242 d.move_end(true);
6243 assert_eq!(d.selection(), Some((6, 11)), "{tag}: End extends");
6244 let mut d = doc_in(view, &format!("home_ext_{tag}"), "hello world");
6245 d.caret = 6;
6246 d.move_home(true);
6247 assert_eq!(d.selection(), Some((0, 6)), "{tag}: Home extends");
6248 }
6249 }
6250
6251 // ── kill to the line's start / end ───────────────────────────────────────
6252
6253 #[test]
6254 fn kill_to_the_line_start_and_end_in_both_views() {
6255 for (view, tag) in VIEWS {
6256 // The gap that reads as a paragraph break in each view: the source
6257 // view's lines are the renderer's rows only where the source says so.
6258 let gap = if view == View::Source { "\n" } else { "\n\n" };
6259 let mut d = doc_in(
6260 view,
6261 &format!("kill_end_{tag}"),
6262 &format!("one two{gap}three\n"),
6263 );
6264 d.caret = 3;
6265 d.delete_to_line_end();
6266 assert_eq!(
6267 d.source,
6268 format!("one{gap}three\n"),
6269 "{tag}: ^K to the line's end"
6270 );
6271 assert_eq!(d.caret, 3, "{tag}: the caret stays where it kills from");
6272
6273 let mut d = doc_in(
6274 view,
6275 &format!("kill_start_{tag}"),
6276 &format!("one two{gap}three\n"),
6277 );
6278 d.caret = 7; // the end of the first line
6279 d.delete_to_line_start();
6280 assert_eq!(
6281 d.source,
6282 format!("{gap}three\n"),
6283 "{tag}: ⌘⌫ to the line's start"
6284 );
6285 assert_eq!(d.caret, 0, "{tag}");
6286 }
6287 }
6288
6289 #[test]
6290 fn a_kill_at_the_line_s_edge_leaves_the_lines_joined() {
6291 // The decision: at the boundary both kills do nothing, rather than
6292 // eating the line break. "Line" is the view's own — in WYSIWYG it ends
6293 // at a soft wrap as often as at a newline, where there is nothing
6294 // written to delete — and a source newline is only half of the blank
6295 // line between two paragraphs, so taking it leaves a soft break rather
6296 // than the join it looks like. Backspace and Delete are the keys for it.
6297 for (view, tag) in VIEWS {
6298 let gap = if view == View::Source { "\n" } else { "\n\n" };
6299 let src = format!("one{gap}three\n");
6300 let mut d = doc_in(view, &format!("kill_edge_end_{tag}"), &src);
6301 d.caret = 3; // the end of "one"
6302 d.delete_to_line_end();
6303 assert_eq!(
6304 d.source, src,
6305 "{tag}: ^K at the line's end joined it to the next"
6306 );
6307
6308 let mut d = doc_in(view, &format!("kill_edge_start_{tag}"), &src);
6309 d.caret = 3 + gap.len(); // the start of "three"
6310 d.delete_to_line_start();
6311 assert_eq!(
6312 d.source, src,
6313 "{tag}: ⌘⌫ at the line's start joined it to the last"
6314 );
6315 }
6316 }
6317
6318 #[test]
6319 fn a_kill_takes_the_selection_when_there_is_one() {
6320 // What every other delete here does with one, so these two as well.
6321 for (view, tag) in VIEWS {
6322 for (name, kill) in [
6323 (
6324 "end",
6325 (|d: &mut Doc| d.delete_to_line_end()) as fn(&mut Doc),
6326 ),
6327 ("start", |d: &mut Doc| d.delete_to_line_start()),
6328 ] {
6329 let mut d = doc_in(view, &format!("kill_sel_{name}_{tag}"), "one two three\n");
6330 d.anchor = Some(4);
6331 d.caret = 7; // "two"
6332 kill(&mut d);
6333 assert_eq!(
6334 d.source, "one three\n",
6335 "{tag}: {name} ignored the selection"
6336 );
6337 assert_eq!(d.selection(), None, "{tag}: {name}");
6338 }
6339 }
6340 }
6341
6342 #[test]
6343 fn a_kill_takes_the_markup_it_empties_with_it() {
6344 // The same hazard a word-delete has: a WYSIWYG range covers what the
6345 // user can see, which for `**bold**` is the word and never the
6346 // delimiters, so a kill that stopped at the text would leave `a ****` —
6347 // markup wrapped around nothing.
6348 let mut d = wysiwyg_doc("kill_widen", "a **bold**\n");
6349 d.caret = d.source.find("bold").unwrap();
6350 d.delete_to_line_end();
6351 assert_eq!(d.source, "a \n");
6352 }
6353
6354 #[test]
6355 fn a_kill_is_undone_in_one_step() {
6356 for (view, tag) in VIEWS {
6357 let mut d = doc_in(view, &format!("kill_undo_{tag}"), "one two three\n");
6358 d.caret = 3;
6359 d.delete_to_line_end();
6360 assert_eq!(d.source, "one\n", "{tag}");
6361 d.undo();
6362 assert_eq!(d.source, "one two three\n", "{tag}: a kill takes one undo");
6363 }
6364 }
6365
6366 #[test]
6367 fn select_block_grabs_the_whole_paragraph_from_any_wrapped_row() {
6368 // Regression: triple-click used move_home/move_end over visual rows, so
6369 // it only worked on a paragraph's first row (a wrap-boundary offset maps
6370 // to the earlier row). select_block_at reads the AST, so every offset in
6371 // the paragraph selects the whole thing.
6372 let body = "one two three four five six seven eight\n";
6373 let mut d = doc_with("sel_block", body);
6374 d.view = View::Wysiwyg;
6375 d.build_visual(12); // force the paragraph to wrap into several rows
6376 assert!(d.vmap.num_rows() > 1, "test needs a wrapped paragraph");
6377 let para = (0, "one two three four five six seven eight".len());
6378 for off in [0usize, 8, 19, 28, 38] {
6379 d.caret = 0;
6380 d.anchor = None;
6381 d.select_block_at(off);
6382 assert_eq!(
6383 d.selection(),
6384 Some(para),
6385 "offset {off} should select the paragraph"
6386 );
6387 }
6388 }
6389
6390 #[test]
6391 fn select_block_uses_content_span_for_a_heading() {
6392 let mut d = doc_with("sel_head", "# Title\n\nbody\n");
6393 d.select_block_at(4); // inside "Title"
6394 // content_span excludes the "# " marker.
6395 assert_eq!(d.selected_text(), Some("Title"));
6396 d.select_block_at(10); // inside "body"
6397 assert_eq!(d.selected_text(), Some("body"));
6398 }
6399
6400 #[test]
6401 fn select_all_spans_the_document() {
6402 let mut d = doc_with("sel_all", "abc\n\ndef\n");
6403 d.select_all();
6404 assert_eq!(d.selection(), Some((0, d.source.len())));
6405 }
6406
6407 #[test]
6408 fn select_word_at_picks_the_surrounding_word() {
6409 let mut d = doc_with("sel_word", "hello world\n");
6410 d.select_word_at(8); // inside "world"
6411 assert_eq!(d.selection(), Some((6, 11)));
6412 // Double-clicking at end-of-word still grabs the word to its left.
6413 d.select_word_at(5); // the space between the words
6414 assert_eq!(d.selection(), Some((5, 6)));
6415 }
6416
6417 #[test]
6418 fn word_helpers_respect_utf8_boundaries() {
6419 // "café" is 5 bytes ('é' is two); motion must land on char boundaries.
6420 assert_eq!(
6421 golden("utf8", "|café ok", |d| d.move_word_right(false)),
6422 "café| ok"
6423 );
6424 assert_eq!(golden("utf8b", "café |ok", |d| d.delete_word_back()), "|ok");
6425 }
6426
6427 #[test]
6428 fn typing_inserts_at_the_caret_and_advances_it() {
6429 let mut d = doc_with("type", "hello\n");
6430 d.insert("Hi ");
6431 assert_eq!(d.source, "Hi hello\n");
6432 assert_eq!(d.caret, 3);
6433 assert!(d.dirty);
6434 }
6435
6436 #[test]
6437 fn backspace_deletes_the_char_before_the_caret() {
6438 let mut d = doc_with("bs", "hello\n");
6439 d.caret = 3; // after "hel"
6440 d.backspace();
6441 assert_eq!(d.source, "helo\n");
6442 assert_eq!(d.caret, 2);
6443 }
6444
6445 #[test]
6446 fn typing_replaces_the_selection() {
6447 let mut d = doc_with("replace", "a word b\n");
6448 d.anchor = Some(2);
6449 d.caret = 6; // "word" selected
6450 d.insert("X");
6451 assert_eq!(d.source, "a X b\n");
6452 assert_eq!(d.caret, 3);
6453 assert_eq!(d.anchor, None);
6454 }
6455
6456 #[test]
6457 fn toggle_bold_wraps_then_unwraps_the_selection() {
6458 let mut d = doc_with("bold", "a word b\n");
6459 d.anchor = Some(2);
6460 d.caret = 6;
6461 d.toggle(InlineKind::Strong);
6462 assert_eq!(d.source, "a **word** b\n");
6463 // The toggled region stays selected, so a second toggle reverses it.
6464 d.toggle(InlineKind::Strong);
6465 assert_eq!(d.source, "a word b\n");
6466 }
6467
6468 #[test]
6469 fn toggle_code_wraps_then_unwraps_the_selection() {
6470 let mut d = doc_with("code_rt", "a word b\n");
6471 d.anchor = Some(2);
6472 d.caret = 6;
6473 d.toggle(InlineKind::Verbatim);
6474 assert_eq!(d.source, "a `word` b\n");
6475 d.toggle(InlineKind::Verbatim);
6476 assert_eq!(d.source, "a word b\n");
6477 }
6478
6479 #[test]
6480 fn sticky_bold_with_no_selection_wraps_the_next_typed_text() {
6481 // ⌘b at a bare caret, then type: the text comes out bold with no
6482 // selection ever made — the word-processor "start bold here" gesture.
6483 let mut d = doc_with("sticky_wrap", "xy\n");
6484 d.caret = 1; // between x and y
6485 d.toggle(InlineKind::Strong);
6486 assert_eq!(d.source, "xy\n", "arming a mark must not edit the document");
6487 d.insert("A");
6488 assert_eq!(d.source, "x**A**y\n");
6489 }
6490
6491 #[test]
6492 fn sticky_bold_lights_the_toolbar_before_any_typing() {
6493 // The button must light the instant ⌘b is pressed, or the mode is
6494 // invisible until the first character lands.
6495 let mut d = doc_with("sticky_light", "xy\n");
6496 d.caret = 1;
6497 assert!(!d.active_inline_marks().contains(InlineKind::Strong));
6498 d.toggle(InlineKind::Strong);
6499 assert!(d.active_inline_marks().contains(InlineKind::Strong));
6500 }
6501
6502 #[test]
6503 fn sticky_bold_toggled_off_types_normally_again() {
6504 // ⌘b, type, ⌘b, type: the first run is bold, the second is not — all
6505 // in the flow of typing, the exact sequence the user described.
6506 let mut d = doc_with("sticky_off", "\n");
6507 d.caret = 0;
6508 d.toggle(InlineKind::Strong);
6509 d.insert("a");
6510 d.insert("b"); // continues inside the run, no re-arming
6511 assert_eq!(d.source, "**ab**\n");
6512 d.toggle(InlineKind::Strong); // ⌘b again — shed bold
6513 d.insert("c");
6514 assert_eq!(d.source, "**ab**c\n");
6515 }
6516
6517 #[test]
6518 fn continued_typing_after_a_sticky_run_stays_in_the_run() {
6519 // Once a mark is realised the caret sits inside the run, so plain typing
6520 // extends it rather than starting a second, adjacent bold span.
6521 let mut d = doc_with("sticky_cont", "\n");
6522 d.caret = 0;
6523 d.toggle(InlineKind::Emph);
6524 d.insert("h");
6525 d.insert("i");
6526 assert_eq!(d.source, "*hi*\n");
6527 }
6528
6529 #[test]
6530 fn moving_the_caret_disarms_a_sticky_mark() {
6531 // Arming a mark and then moving away must not style text elsewhere.
6532 let mut d = doc_with("sticky_disarm", "xy\n");
6533 d.caret = 0;
6534 d.toggle(InlineKind::Strong);
6535 d.move_right(false); // caret 0 → 1, disarms
6536 assert!(!d.active_inline_marks().contains(InlineKind::Strong));
6537 d.insert("A");
6538 assert_eq!(d.source, "xAy\n", "the mark must not follow the caret");
6539 }
6540
6541 #[test]
6542 fn stacked_sticky_marks_apply_together() {
6543 // ⌘b then ⌘i before typing: the text comes out both bold and italic.
6544 let mut d = doc_with("sticky_stack", "\n");
6545 d.caret = 0;
6546 d.toggle(InlineKind::Strong);
6547 d.toggle(InlineKind::Emph);
6548 d.insert("x");
6549 // Land the caret on the styled character and confirm both marks are live.
6550 d.anchor = Some(d.source.find('x').unwrap());
6551 d.caret = d.anchor.unwrap() + 1;
6552 let marks = d.active_inline_marks();
6553 assert!(marks.contains(InlineKind::Strong), "bold: {}", d.source);
6554 assert!(marks.contains(InlineKind::Emph), "italic: {}", d.source);
6555 }
6556
6557 // ── the mark-edge rule (see `Doc::splice`) ───────────────────────────────
6558
6559 #[test]
6560 fn a_space_typed_in_a_bold_run_never_leaves_the_delimiters_showing() {
6561 // The reported bug, keystroke for keystroke: ⌘b, "bold", space, "hey".
6562 // The space inside the run made `**bold **`, which is *not* bold — four
6563 // literal asterisks — so the rich view drew them, correctly and
6564 // uselessly, until the next character happened to close the run again.
6565 let mut d = wysiwyg_doc("edge_typing", "a \n");
6566 d.caret = 2;
6567 d.toggle(InlineKind::Strong);
6568 for c in "bold".chars() {
6569 d.insert(&c.to_string());
6570 }
6571 assert_eq!(d.source, "a **bold**\n");
6572 d.insert(" ");
6573 assert_eq!(
6574 d.source, "a **bold** \n",
6575 "the space belongs outside the run"
6576 );
6577 assert!(
6578 d.active_inline_marks().contains(InlineKind::Strong),
6579 "bold is still what's being typed, so the button stays lit"
6580 );
6581 // What the writer is looking at while all this happens: their words.
6582 d.build_visual(80);
6583 let drawn: String = d.vmap.rows[0].glyphs.iter().map(|g| g.ch).collect();
6584 assert_eq!(drawn, "a bold ", "no delimiter ever surfaces: {}", d.source);
6585 for c in "hey".chars() {
6586 d.insert(&c.to_string());
6587 }
6588 assert_eq!(
6589 d.source, "a **bold hey**\n",
6590 "one bold phrase, not two runs"
6591 );
6592 }
6593
6594 #[test]
6595 fn typing_past_a_space_can_still_leave_the_bold_behind() {
6596 // The other half: the marks stay armed across the space, so ⌘b turns
6597 // them off again there and the next word is plain — the run isn't
6598 // rejoined by a caret that was told not to.
6599 let mut d = wysiwyg_doc("edge_shed", "\n");
6600 d.caret = 0;
6601 d.toggle(InlineKind::Strong);
6602 for c in "bold ".chars() {
6603 d.insert(&c.to_string());
6604 }
6605 assert_eq!(d.source, "**bold** \n");
6606 d.toggle(InlineKind::Strong);
6607 assert!(!d.active_inline_marks().contains(InlineKind::Strong));
6608 d.insert("x");
6609 assert_eq!(d.source, "**bold** x\n");
6610 }
6611
6612 #[test]
6613 fn a_space_typed_first_of_all_still_leaves_the_mark_armed() {
6614 // ⌘b and then a space before any word: the space is not marked (nothing
6615 // is), and the word after it is.
6616 let mut d = wysiwyg_doc("edge_space_first", "a\n");
6617 d.caret = 1;
6618 d.toggle(InlineKind::Strong);
6619 d.insert(" ");
6620 assert_eq!(d.source, "a \n");
6621 assert!(d.active_inline_marks().contains(InlineKind::Strong));
6622 d.insert("b");
6623 assert_eq!(d.source, "a **b**\n");
6624 }
6625
6626 #[test]
6627 fn a_space_typed_at_either_edge_of_an_existing_mark_steps_outside_it() {
6628 let mut d = wysiwyg_doc("edge_tail", "x **bold**\n");
6629 d.caret = 8; // the caret's home at the end of the run's text
6630 d.insert(" ");
6631 assert_eq!(
6632 d.source, "x **bold** \n",
6633 "the space lands past the delimiters"
6634 );
6635 assert_eq!(d.caret, 11, "and the caret stands past it, outside the run");
6636
6637 let mut d = wysiwyg_doc("edge_head", "x **bold** y\n");
6638 d.caret = 4; // in front of the "b"
6639 d.insert(" ");
6640 assert_eq!(d.source, "x **bold** y\n");
6641 assert_eq!(d.caret, 3, "in front of the run, where the space was typed");
6642 }
6643
6644 #[test]
6645 fn a_delete_that_backs_a_space_onto_a_delimiter_moves_the_delimiter() {
6646 // Backspace over the last letter of a bold phrase.
6647 let mut d = wysiwyg_doc("edge_bksp", "a **bold h**\n");
6648 d.caret = 10; // past the "h"
6649 d.backspace();
6650 assert_eq!(d.source, "a **bold** \n");
6651 assert_eq!(d.caret, 11, "the caret keeps the place on screen it had");
6652 assert!(d.active_inline_marks().contains(InlineKind::Strong));
6653 d.insert("x");
6654 assert_eq!(d.source, "a **bold x**\n", "and typing rejoins the run");
6655 }
6656
6657 #[test]
6658 fn deleting_the_last_of_a_run_takes_its_delimiters_with_it() {
6659 // `**b**` with the `b` gone is `****`: two delimiters with nothing to
6660 // mark, which is only text. The marks live on in the caret instead.
6661 let mut d = wysiwyg_doc("edge_empty", "a **b** c\n");
6662 d.caret = 5;
6663 d.backspace();
6664 assert_eq!(d.source, "a c\n");
6665 assert!(d.active_inline_marks().contains(InlineKind::Strong));
6666 d.insert("x");
6667 assert_eq!(d.source, "a **x** c\n");
6668 }
6669
6670 #[test]
6671 fn typing_over_a_whole_bold_word_keeps_it_bold() {
6672 let mut d = wysiwyg_doc("edge_replace", "a **bold** c\n");
6673 d.anchor = Some(4);
6674 d.caret = 8; // the word, not its delimiters
6675 d.insert("x");
6676 assert_eq!(d.source, "a **x** c\n");
6677 }
6678
6679 #[test]
6680 fn a_code_span_keeps_the_space_it_is_given() {
6681 // Backticks are not whitespace-sensitive the way `**` is: `` `code ` ``
6682 // is still verbatim, so nothing is re-spelt. The repair asks the parser
6683 // rather than a table of kinds, and this is the answer it gets.
6684 let mut d = wysiwyg_doc("edge_code", "a `code` c\n");
6685 d.caret = 7;
6686 d.insert(" ");
6687 assert_eq!(d.source, "a `code ` c\n");
6688 }
6689
6690 #[test]
6691 fn a_delete_from_a_runs_outer_edge_reaches_into_the_run() {
6692 // A run's closing delimiter has a caret home on each side of it, one
6693 // column apart on screen — and a plain ← off the space after a bold word
6694 // lands on the outer one. The character drawn behind the caret there is
6695 // still the last letter of the phrase, so that is what Backspace takes;
6696 // the byte behind it is a `*` nobody can see.
6697 let mut d = wysiwyg_doc("edge_outer_close", "**bold** x\n");
6698 d.caret = 9;
6699 d.move_left(false);
6700 assert_eq!(d.caret, 8, "← rests past the delimiters, not inside them");
6701 d.backspace();
6702 assert_eq!(
6703 d.source, "**bol** x\n",
6704 "a letter of the phrase, not its `*`"
6705 );
6706 assert_eq!(d.caret, 5);
6707
6708 // And the mirror in front of the opening delimiter, where Delete's
6709 // character is the first letter of the run.
6710 let mut d = wysiwyg_doc("edge_outer_open", "x**bold**\n");
6711 d.caret = 1;
6712 d.delete_forward();
6713 assert_eq!(d.source, "x**old**\n");
6714 assert_eq!(d.caret, 3, "inside the run, in front of what is left of it");
6715 }
6716
6717 #[test]
6718 fn a_delete_at_a_run_edge_never_eats_a_delimiter() {
6719 // The byte beside the caret at either edge of a bold word is a `*` the
6720 // rich view draws nothing for. Taking it is not the character delete the
6721 // key was pressed for — it unspells the run and puts a literal asterisk
6722 // on screen (`a *bold** c`). The visible character is the one that goes.
6723 let mut d = wysiwyg_doc("edge_open_bksp", "a **bold** c\n");
6724 d.caret = 4; // in front of the "b"
6725 d.backspace();
6726 assert_eq!(d.source, "a**bold** c\n", "the space goes, the run stands");
6727
6728 let mut d = wysiwyg_doc("edge_close_del", "a **bold** c\n");
6729 d.caret = 8; // past the "d"
6730 d.delete_forward();
6731 assert_eq!(d.source, "a **bold**c\n");
6732 assert_eq!(d.caret, 8, "and the caret stays inside the run");
6733 d.insert("x");
6734 assert_eq!(d.source, "a **boldx**c\n");
6735
6736 // A code span's backticks are hidden the same way, so they are covered
6737 // by the same rule and not by a list of kinds.
6738 let mut d = wysiwyg_doc("edge_open_code", "a `code` c\n");
6739 d.caret = 3;
6740 d.backspace();
6741 assert_eq!(d.source, "a`code` c\n");
6742 }
6743
6744 #[test]
6745 fn the_source_view_deletes_the_delimiter_byte_it_is_shown() {
6746 // The asterisks are on the screen there and the caret can stand between
6747 // them, so a delete takes exactly the byte it is aimed at.
6748 let mut d = doc_with("edge_open_src", "a **bold** c\n");
6749 d.caret = 4;
6750 d.backspace();
6751 assert_eq!(d.source, "a *bold** c\n");
6752
6753 let mut d = doc_with("edge_close_src", "a **bold** c\n");
6754 d.caret = 8;
6755 d.delete_forward();
6756 assert_eq!(d.source, "a **bold* c\n");
6757 }
6758
6759 #[test]
6760 fn backspacing_the_space_out_of_a_bold_phrase_leaves_the_caret_in_it() {
6761 // The reported bug, keystroke for keystroke: ⌘b, "bold", space, Backspace.
6762 // The space had stepped outside the run (the mark-edge rule), taking the
6763 // caret with it, so the delete put it back down on the far side of the
6764 // closing `**` — one place on screen, and the wrong side of it. Typing
6765 // came out plain and the toolbar went dark, with nothing to see.
6766 let mut d = wysiwyg_doc("edge_bksp_space", "\n");
6767 d.caret = 0;
6768 d.toggle(InlineKind::Strong);
6769 for c in "bold".chars() {
6770 d.insert(&c.to_string());
6771 }
6772 d.insert(" ");
6773 assert_eq!(d.source, "**bold** \n");
6774 d.backspace();
6775 assert_eq!(
6776 d.source, "**bold**\n",
6777 "the space goes, the delimiters stay"
6778 );
6779 assert_eq!(d.caret, 6, "and the caret comes back inside the run");
6780 assert!(
6781 d.active_inline_marks().contains(InlineKind::Strong),
6782 "so the button is still lit"
6783 );
6784 d.insert("x");
6785 assert_eq!(
6786 d.source, "**boldx**\n",
6787 "and the next character is still bold"
6788 );
6789 }
6790
6791 #[test]
6792 fn a_second_backspace_there_deletes_a_letter_of_the_phrase() {
6793 // What the stranded caret did next: the byte behind it was the closing
6794 // `*`, so a second press took that instead of a letter — `**bold*`, the
6795 // styling gone and an asterisk on the screen where the word had been.
6796 let mut d = wysiwyg_doc("edge_bksp_twice", "\n");
6797 d.caret = 0;
6798 d.toggle(InlineKind::Strong);
6799 for c in "bold ".chars() {
6800 d.insert(&c.to_string());
6801 }
6802 assert_eq!(d.source, "**bold** \n");
6803 d.backspace();
6804 d.backspace();
6805 assert_eq!(d.source, "**bol**\n", "the delete lands inside the run");
6806 assert_eq!(d.caret, 5);
6807 }
6808
6809 #[test]
6810 fn a_delete_that_ends_at_a_nested_run_settles_inside_every_delimiter() {
6811 // `***both***` closes two runs with one stack of asterisks: the caret has
6812 // to walk in through all of them, or it lands between the emph and the
6813 // strong and types half-marked.
6814 let mut d = wysiwyg_doc("edge_bksp_nested", "***both*** \n");
6815 d.caret = 11;
6816 d.backspace();
6817 assert_eq!(d.source, "***both***\n");
6818 assert_eq!(d.caret, 7, "past the last letter, inside both runs");
6819 d.insert("x");
6820 assert_eq!(d.source, "***bothx***\n");
6821 }
6822
6823 #[test]
6824 fn a_delete_that_ends_mid_run_leaves_the_caret_where_it_fell() {
6825 // The settle only moves a caret a run actually closed over. Ordinary
6826 // deletes — inside a run, or in plain prose — are untouched.
6827 let mut d = wysiwyg_doc("edge_bksp_mid", "a **bold** c\n");
6828 d.caret = 8;
6829 d.backspace();
6830 assert_eq!(d.source, "a **bol** c\n");
6831 assert_eq!(d.caret, 7);
6832
6833 let mut d = wysiwyg_doc("edge_bksp_plain", "plain\n");
6834 d.caret = 5;
6835 d.backspace();
6836 assert_eq!(d.source, "plai\n");
6837 assert_eq!(d.caret, 4);
6838 }
6839
6840 #[test]
6841 fn the_source_view_leaves_a_delete_where_it_landed() {
6842 // The delimiters are on the screen there, so the offset past them is a
6843 // place the caret can be seen to be — nothing to settle.
6844 let mut d = doc_with("edge_bksp_src", "**bold** \n");
6845 d.caret = 9;
6846 d.backspace();
6847 assert_eq!(d.source, "**bold**\n");
6848 assert_eq!(d.caret, 8);
6849 }
6850
6851 #[test]
6852 fn the_mark_edge_rule_clears_every_delimiter_of_a_nested_run() {
6853 // `***both***` closes two runs with one stack of asterisks; a space that
6854 // clears only the inner one lands against the outer's and breaks that
6855 // instead.
6856 let mut d = wysiwyg_doc("edge_nested", "a ***both***\n");
6857 d.caret = 9;
6858 d.insert(" ");
6859 assert_eq!(d.source, "a ***both*** \n");
6860 assert_eq!(d.caret, 13);
6861 d.insert("x");
6862 assert_eq!(d.source, "a ***both x***\n");
6863 }
6864
6865 #[test]
6866 fn the_mark_edge_repair_undoes_with_the_keystroke_that_caused_it() {
6867 // The delimiter shuffle is not an edit the writer made, so it is not a
6868 // step they have to undo past.
6869 let mut d = wysiwyg_doc("edge_undo", "a **bold**\n");
6870 d.caret = 8;
6871 d.insert(" ");
6872 assert_eq!(d.source, "a **bold** \n");
6873 d.undo();
6874 assert_eq!(d.source, "a **bold**\n");
6875 }
6876
6877 #[test]
6878 fn the_source_view_types_the_space_where_it_was_asked_to() {
6879 // The rule is a rich-view courtesy. In the source view the delimiters are
6880 // on the screen and the user is editing the bytes they can see.
6881 let mut d = doc_with("edge_src", "a **bold** c\n");
6882 d.caret = 8;
6883 d.insert(" ");
6884 assert_eq!(d.source, "a **bold ** c\n");
6885 }
6886
6887 #[test]
6888 fn toggling_a_mark_over_a_selection_leaves_its_edge_whitespace_out() {
6889 // Double-clicking a word takes the space after it; bolding that must not
6890 // spell `**word **`, which is not bold at all.
6891 let mut d = wysiwyg_doc("edge_sel", "a word b\n");
6892 d.anchor = Some(2);
6893 d.caret = 7; // "word "
6894 d.toggle(InlineKind::Strong);
6895 assert_eq!(d.source, "a **word** b\n");
6896 // And a selection of nothing but whitespace has no word to mark.
6897 let mut d = wysiwyg_doc("edge_sel_ws", "a word b\n");
6898 d.anchor = Some(6);
6899 d.caret = 7;
6900 d.toggle(InlineKind::Strong);
6901 assert_eq!(d.source, "a word b\n");
6902 assert!(d.status.is_some());
6903 }
6904
6905 #[test]
6906 fn set_block_turns_a_paragraph_into_a_heading_at_the_caret() {
6907 let mut d = doc_with("head_set", "hello\n");
6908 d.caret = 2; // caret inside the paragraph, no selection
6909 d.set_block(BlockKind::Heading(1));
6910 assert_eq!(d.source, "# hello\n");
6911 }
6912
6913 #[test]
6914 fn set_block_heading_works_in_wysiwyg_view() {
6915 // The app defaults to WYSIWYG; the caret is a source offset either way.
6916 let mut d = wysiwyg_doc("head_wys", "hello\n");
6917 d.caret = 2;
6918 d.set_block(BlockKind::Heading(1));
6919 assert_eq!(d.source, "# hello\n");
6920 }
6921
6922 #[test]
6923 fn toggle_heading_applies_switches_and_reverts() {
6924 let mut d = doc_with("head_toggle", "hello\n");
6925 d.caret = 2;
6926 d.toggle_heading(1);
6927 assert_eq!(d.source, "# hello\n"); // paragraph → H1
6928 d.toggle_heading(2);
6929 assert_eq!(d.source, "## hello\n"); // H1 → H2 (different level switches)
6930 d.toggle_heading(2);
6931 assert_eq!(d.source, "hello\n"); // same level reverts to paragraph
6932 }
6933
6934 #[test]
6935 fn preserve_enter_at_a_line_end_lands_the_caret_on_the_new_blank_line() {
6936 // Regression: Enter at the end of a soft-break line (mid-paragraph) opened
6937 // the blank line but the caret rendered on the *next* line, because the
6938 // separator was a non-navigable decoration row. In Preserve flow that
6939 // blank line is a real caret home — the caret must resolve onto it, and
6940 // typing there makes the soft break that continues the paragraph.
6941 let src = "line one:\nsecond line\n";
6942 let mut d = wysiwyg_doc("pre_enter_lineend", src);
6943 d.set_line_flow(LineFlow::Preserve);
6944 d.build_visual_unwrapped(); // the GUI path (pixel-wrapped)
6945 d.caret = 9; // the visual end of row 0, at the soft-break '\n'
6946 d.newline();
6947 d.build_visual_unwrapped();
6948 assert_eq!(d.source, "line one:\n\nsecond line\n");
6949 assert_eq!(
6950 d.caret, 10,
6951 "caret sits on the new blank line, not the next line"
6952 );
6953 // The blank line is row 1, and the caret resolves onto it — not row 2.
6954 assert_eq!(
6955 d.vmap.pos_of_offset(10),
6956 (1, 0),
6957 "caret renders on the blank row"
6958 );
6959 assert!(
6960 !d.vmap.rows[1].decoration,
6961 "the blank line is navigable in Preserve"
6962 );
6963 // Typing there makes a soft break: one paragraph, three lines.
6964 d.insert("new clause,");
6965 assert_eq!(d.source, "line one:\nnew clause,\nsecond line\n");
6966 }
6967
6968 #[test]
6969 fn preserve_enter_makes_a_soft_break_not_a_paragraph() {
6970 // Mid-paragraph: Enter splits the line with a single `\n`, a soft break
6971 // that keeps it one paragraph — where Fold would open a second paragraph.
6972 let mut d = wysiwyg_doc("pre_enter_mid", "abcdef\n");
6973 d.set_line_flow(LineFlow::Preserve);
6974 d.caret = 3;
6975 d.newline();
6976 assert_eq!(d.source, "abc\ndef\n", "mid-line Enter is a soft break");
6977
6978 // End-of-paragraph: Enter then typing continues the same paragraph on a
6979 // new line (a soft break), not a fresh paragraph.
6980 let mut d = wysiwyg_doc("pre_enter_end", "abc\n");
6981 d.set_line_flow(LineFlow::Preserve);
6982 d.caret = 3;
6983 d.newline();
6984 d.insert("def");
6985 assert_eq!(
6986 d.source, "abc\ndef\n",
6987 "end-of-line Enter + typing is a soft break"
6988 );
6989 }
6990
6991 #[test]
6992 fn preserve_double_enter_still_makes_a_paragraph() {
6993 // Two Enters in a row promote to a real paragraph break: the second lands
6994 // on the blank line the first opened and takes the empty-line branch.
6995 let mut d = wysiwyg_doc("pre_enter_dbl", "abc\n");
6996 d.set_line_flow(LineFlow::Preserve);
6997 d.caret = 3;
6998 d.newline();
6999 d.newline();
7000 d.insert("def");
7001 assert_eq!(
7002 d.source, "abc\n\ndef\n",
7003 "double Enter is a paragraph break"
7004 );
7005 }
7006
7007 #[test]
7008 fn preserve_backspace_joins_across_a_soft_break() {
7009 // Backspace is the symmetric undo of a Preserve Enter: over the `\n` of a
7010 // soft break it deletes the single newline and joins the two lines.
7011 let mut d = wysiwyg_doc("pre_bs", "abc\ndef\n");
7012 d.set_line_flow(LineFlow::Preserve);
7013 d.build_visual(80);
7014 d.caret = 4; // start of "def", just past the soft break
7015 d.backspace();
7016 assert_eq!(
7017 d.source, "abcdef\n",
7018 "Backspace joins across the soft break"
7019 );
7020 assert_eq!(d.caret, 3, "caret lands where the lines meet");
7021 }
7022
7023 #[test]
7024 fn fold_enter_still_starts_a_new_paragraph() {
7025 // The default flow is unchanged: a lone `\n` would render as an invisible
7026 // space, so Enter keeps opening the paragraph break that actually shows.
7027 let mut d = wysiwyg_doc("fold_enter", "abcdef\n");
7028 d.caret = 3;
7029 d.newline();
7030 assert_eq!(
7031 d.source, "abc\n\ndef\n",
7032 "Fold mid-line Enter is a paragraph break"
7033 );
7034 }
7035
7036 #[test]
7037 fn wysiwyg_one_enter_starts_a_new_paragraph() {
7038 // Regression: one Enter left the caret between the two newlines, so typing
7039 // made a soft break (one paragraph) and you needed a second Enter.
7040 let mut d = wysiwyg_doc("wys_enter", "abc\n");
7041 d.caret = 3;
7042 d.newline();
7043 d.insert("def");
7044 assert_eq!(d.source, "abc\n\ndef\n"); // two paragraphs, not "abc\ndef\n"
7045 }
7046
7047 #[test]
7048 fn enter_at_the_end_of_a_bold_run_keeps_its_closing_delimiter_attached() {
7049 // Regression: Enter at the caret's natural End-of-line resting place
7050 // after a bold run with nothing following it (on screen: right after
7051 // "bold", before the hidden closing "**") spliced the paragraph break
7052 // at that very byte offset — which sits *before* the closing "**" in
7053 // the source, since the delimiter is hidden and emits no glyph of its
7054 // own for `push_row`'s "end of row" fallback to count. That severed the
7055 // mark: "**bold**\n" became "**bold\n\n**\n", stranding the closing
7056 // "**" alone on the new line instead of leaving "**bold**" intact with
7057 // a fresh empty paragraph after it.
7058 let mut d = wysiwyg_doc("bold_eol_enter", "**bold**\n");
7059 d.move_end(false); // the WYSIWYG End key, from caret 0
7060 assert_eq!(
7061 d.caret, 6,
7062 "caret rests right after \"bold\", before the hidden \"**\""
7063 );
7064 d.newline();
7065 assert!(
7066 d.source.starts_with("**bold**"),
7067 "the closing ** must stay attached to \"bold\": got {:?}",
7068 d.source
7069 );
7070 assert_eq!(
7071 d.source, "**bold**\n\n\n",
7072 "a fresh empty paragraph follows the still-intact bold run"
7073 );
7074 }
7075
7076 #[test]
7077 fn source_view_enter_is_a_single_newline() {
7078 let mut d = doc_with("src_enter", "abc\n");
7079 d.caret = 3;
7080 d.newline();
7081 assert_eq!(d.source, "abc\n\n");
7082 }
7083
7084 #[test]
7085 fn heading_applies_at_the_end_of_a_paragraph() {
7086 // The caret at a line end sits at the doc level; set_block must still find
7087 // the block on that line.
7088 let mut d = doc_with("head_end", "abc\n");
7089 d.caret = 3; // end of "abc"
7090 d.toggle_heading(1);
7091 assert_eq!(d.source, "# abc\n");
7092 }
7093
7094 #[test]
7095 fn heading_on_an_empty_new_paragraph_creates_one() {
7096 let mut d = wysiwyg_doc("head_empty", "abc\n");
7097 d.caret = 3;
7098 d.newline(); // caret now on a fresh, empty paragraph
7099 d.toggle_heading(1);
7100 d.insert("Title");
7101 assert!(d.source.contains("# Title"), "got {:?}", d.source);
7102 }
7103
7104 #[test]
7105 fn a_heading_typed_on_a_blank_line_keeps_the_caret_on_its_own_row() {
7106 // The reported bug, end to end: click a blank line with another one under
7107 // it, press H1, type. The text landed in the heading and the caret's
7108 // offset was right (the source view drew it there), but the rich view
7109 // drew it two rows lower, on the trailing blank line — the empty `# `
7110 // heading had left every row below it short by the marker's two bytes,
7111 // and the blank line ended up claiming the heading's own end offset.
7112 let mut d = wysiwyg_doc("head_blank", "one\n\ntwo\n\n\n\n");
7113 d.build_visual_unwrapped();
7114 d.caret = d.vmap.offset_of_pos(4, 0); // the first of the two blank lines
7115 d.toggle_heading(1);
7116 for c in "title".chars() {
7117 d.insert(&c.to_string());
7118 d.build_visual_unwrapped(); // as a frontend does, one frame per key
7119 }
7120 assert_eq!(d.source, "one\n\ntwo\n\n# title\n\n");
7121 assert_eq!(
7122 d.caret_pos(),
7123 (4, 5),
7124 "the caret draws at the end of the heading"
7125 );
7126 }
7127
7128 #[test]
7129 fn clicking_an_empty_heading_types_after_its_marker() {
7130 // The same anchor from the other side: the empty heading's row is its own
7131 // caret home, so a click on it must land past the hidden `# `. Landing in
7132 // front of the hashes made the first keystroke un-heading the line.
7133 let mut d = wysiwyg_doc("head_click", "# \n");
7134 d.build_visual_unwrapped();
7135 d.caret = d.vmap.offset_of_pos(0, 0);
7136 d.insert("x");
7137 assert_eq!(d.source, "# x\n");
7138 }
7139
7140 #[test]
7141 fn wysiwyg_enter_after_a_heading_makes_a_paragraph() {
7142 let mut d = wysiwyg_doc("head_enter", "# Title\n");
7143 d.caret = 7; // end of the heading
7144 d.newline();
7145 d.insert("body");
7146 assert_eq!(d.source, "# Title\n\nbody\n");
7147 }
7148
7149 #[test]
7150 fn wysiwyg_enter_continues_a_bullet_list() {
7151 let mut d = wysiwyg_doc("wys_bullet", "- item\n");
7152 d.caret = 6; // end of "item"
7153 d.newline();
7154 d.insert("two");
7155 assert_eq!(d.source, "- item\n- two\n");
7156 }
7157
7158 #[test]
7159 fn wysiwyg_enter_increments_an_ordered_list() {
7160 let mut d = wysiwyg_doc("wys_ol", "1. one\n");
7161 d.caret = 6; // end of "one"
7162 d.newline();
7163 d.insert("two");
7164 assert_eq!(d.source, "1. one\n2. two\n");
7165 }
7166
7167 #[test]
7168 fn wysiwyg_backspace_after_leaving_a_list_collapses_the_gap_cleanly() {
7169 // Regression for the "extra newline" left between a list and the paragraph
7170 // below it. Enter, Enter leaves the list on a fresh empty paragraph
7171 // (`- item\n\n\n\nnext`, a navigable blank between the two blocks); one
7172 // Backspace should then take the caret cleanly back to the end of the list
7173 // item, `- item\n\nnext`, not delete a single newline and strand it on the
7174 // odd `- item\n\n\nnext` — a blank line the eye reads as one separator but
7175 // no caret can land on. The map is rebuilt between keystrokes exactly as a
7176 // frontend does, since Backspace reads the stop table to place the delete.
7177 let mut d = wysiwyg_doc("wys_exit_bksp", "- item\n\nnext\n");
7178 d.caret = 6; // end of "item"
7179 d.newline();
7180 d.build_visual(80);
7181 d.newline(); // leave the list onto a fresh empty paragraph
7182 d.build_visual(80);
7183 assert_eq!(
7184 d.source, "- item\n\n\n\nnext\n",
7185 "double-Enter opens the empty paragraph"
7186 );
7187 d.backspace();
7188 assert_eq!(
7189 d.source, "- item\n\nnext\n",
7190 "one Backspace collapses the whole gap"
7191 );
7192 assert_eq!(
7193 d.caret, 6,
7194 "and lands the caret back at the end of the list item"
7195 );
7196 }
7197
7198 #[test]
7199 fn wysiwyg_backspace_on_stacked_blank_lines_still_removes_just_one() {
7200 // The stop-wise delete must not over-reach when there is no block boundary
7201 // to cross: two blank lines in a row are one caret stop apart, so pressing
7202 // Enter on an empty line and then Backspace removes exactly the one newline
7203 // it added — the lone-Enter / lone-Backspace symmetry, preserved.
7204 let mut d = wysiwyg_doc("wys_stack", "abc\n\n\n");
7205 d.caret = 5; // the empty paragraph the first Enter already opened
7206 d.build_visual(80);
7207 d.newline();
7208 d.build_visual(80);
7209 assert_eq!(
7210 d.source, "abc\n\n\n\n",
7211 "Enter on the blank line adds one newline"
7212 );
7213 d.backspace();
7214 assert_eq!(
7215 d.source, "abc\n\n\n",
7216 "Backspace takes back exactly that one newline"
7217 );
7218 }
7219
7220 #[test]
7221 fn wysiwyg_enter_on_an_empty_list_item_exits_the_list() {
7222 let mut d = wysiwyg_doc("wys_exit", "- a\n- \n");
7223 d.caret = 6; // end of the empty "- " item
7224 d.newline();
7225 d.insert("p");
7226 assert_eq!(d.source, "- a\n\np\n");
7227 }
7228
7229 #[test]
7230 fn wysiwyg_enter_does_not_mistake_a_setext_underline_for_a_list() {
7231 // `text\n- \n` is a setext heading — the `- ` is its underline, not a
7232 // list item, though it reads as a `- ` marker byte-for-byte. Enter must
7233 // not take the list-exit path (which would splice the `- ` away as if
7234 // leaving an empty item); the AST guard sends it to a normal break and
7235 // leaves the underline intact.
7236 let mut d = wysiwyg_doc("wys_setext", "text\n- \n");
7237 assert!(
7238 d.nodes().iter().any(|n| n.kind == Kind::Heading),
7239 "precondition: twig parses this as a heading, not a list",
7240 );
7241 d.caret = 7; // on the `- ` underline line
7242 d.newline();
7243 assert!(
7244 d.source.contains("- "),
7245 "the setext underline survives, not spliced away as a list item: {:?}",
7246 d.source,
7247 );
7248 }
7249
7250 #[test]
7251 fn wysiwyg_enter_in_a_code_block_is_a_literal_newline() {
7252 let mut d = wysiwyg_doc("wys_code", "```\nabc\n```\n");
7253 d.caret = 7; // end of "abc" inside the fence
7254 d.newline();
7255 d.insert("def");
7256 assert_eq!(d.source, "```\nabc\ndef\n```\n");
7257 }
7258
7259 #[test]
7260 fn wysiwyg_enter_continues_a_block_quote() {
7261 // Enter opens a new *paragraph* inside the quote, not a second line of
7262 // the same one. `> quote\n> more` is a soft break, which under
7263 // `LineFlow::Fold` renders as a space — the keystroke would look like it
7264 // did nothing. The quoted blank line is what makes the break visible, and
7265 // it's the same thing Enter does in running prose.
7266 let mut d = wysiwyg_doc("wys_quote", "> quote\n");
7267 d.caret = 7; // end of "quote"
7268 d.newline();
7269 d.insert("more");
7270 assert_eq!(d.source, "> quote\n>\n> more\n");
7271 // Still one quote, now holding two paragraphs — not a quote and a stray
7272 // line that fell out of it.
7273 let quotes = d
7274 .nodes()
7275 .iter()
7276 .filter(|n| n.kind == Kind::BlockQuote)
7277 .count();
7278 assert_eq!(quotes, 1);
7279 }
7280
7281 #[test]
7282 fn set_block_makes_a_heading_at_the_caret() {
7283 let mut d = doc_with("head", "Title\n\nbody\n");
7284 d.caret = 0;
7285 d.set_block(BlockKind::Heading(2));
7286 assert_eq!(d.source, "## Title\n\nbody\n");
7287 d.set_block(BlockKind::Paragraph);
7288 assert_eq!(d.source, "Title\n\nbody\n");
7289 }
7290
7291 // ── block containers (quote / list) ──────────────────────────────────────
7292
7293 #[test]
7294 fn toggle_blockquote_wraps_the_block_at_the_caret_and_reverses() {
7295 let g = |m, f: fn(&mut Doc)| golden("quote", m, f);
7296 assert_eq!(g("hel|lo\n", |d| d.toggle_blockquote()), "> hel|lo\n");
7297 assert_eq!(g("> hel|lo\n", |d| d.toggle_blockquote()), "hel|lo\n");
7298 // A caret at a line end sits at the doc level; the block is still found.
7299 assert_eq!(g("hello|\n", |d| d.toggle_blockquote()), "> hello|\n");
7300 }
7301
7302 #[test]
7303 fn toggle_blockquote_keeps_the_caret_in_a_hard_wrapped_paragraph() {
7304 // Every source line of the paragraph gets its own `> `, so a caret left
7305 // on its old byte offset falls one prefix per line above it too far
7306 // back — inside the markup it just asked for rather than in its word.
7307 assert_eq!(
7308 golden("quote_wrap", "aaa\nb|bb\nccc\n", |d| d.toggle_blockquote()),
7309 "> aaa\n> b|bb\n> ccc\n"
7310 );
7311 }
7312
7313 #[test]
7314 fn toggle_blockquote_works_in_wysiwyg_view() {
7315 let g = |n, m, f: fn(&mut Doc)| golden_in(View::Wysiwyg, n, m, f);
7316 assert_eq!(
7317 g("q_wys", "hel|lo\n", |d| d.toggle_blockquote()),
7318 "> hel|lo\n"
7319 );
7320 assert_eq!(
7321 g("q_wys2", "> hel|lo\n", |d| d.toggle_blockquote()),
7322 "hel|lo\n"
7323 );
7324 }
7325
7326 #[test]
7327 fn toggle_list_makes_a_list_and_converts_between_the_kinds() {
7328 let g = |m, f: fn(&mut Doc)| golden("list", m, f);
7329 assert_eq!(g("hel|lo\n", |d| d.toggle_list(false)), "- hel|lo\n");
7330 assert_eq!(g("hel|lo\n", |d| d.toggle_list(true)), "1. hel|lo\n");
7331 // The *other* kind converts in place instead of nesting, which is what
7332 // makes the two buttons one three-state control.
7333 assert_eq!(g("- hel|lo\n", |d| d.toggle_list(true)), "1. hel|lo\n");
7334 assert_eq!(g("1. hel|lo\n", |d| d.toggle_list(false)), "- hel|lo\n");
7335 // Its own kind, over the only item the list holds, takes it off.
7336 assert_eq!(g("- hel|lo\n", |d| d.toggle_list(false)), "hel|lo\n");
7337 }
7338
7339 #[test]
7340 fn toggle_list_works_in_wysiwyg_view() {
7341 let g = |n, m, f: fn(&mut Doc)| golden_in(View::Wysiwyg, n, m, f);
7342 assert_eq!(
7343 g("l_wys", "hel|lo\n", |d| d.toggle_list(true)),
7344 "1. hel|lo\n"
7345 );
7346 assert_eq!(
7347 g("l_wys2", "1. hel|lo\n", |d| d.toggle_list(false)),
7348 "- hel|lo\n"
7349 );
7350 assert_eq!(
7351 g("l_wys3", "- hel|lo\n", |d| d.toggle_list(false)),
7352 "hel|lo\n"
7353 );
7354 }
7355
7356 #[test]
7357 fn a_list_over_a_selection_numbers_each_block_and_stays_selected() {
7358 // The selection has to grow with the markup: twig takes a container off
7359 // only a range covering every block it holds, so the second press can
7360 // reverse the first only if the result is what's selected.
7361 let mut d = doc_with("list_sel", "abc\n\ndef\n");
7362 d.select_all();
7363 d.toggle_list(true);
7364 assert_eq!(d.source, "1. abc\n\n2. def\n");
7365 assert_eq!(d.selection(), Some((0, d.source.len())));
7366 d.toggle_list(true);
7367 assert_eq!(d.source, "abc\n\ndef\n");
7368 }
7369
7370 #[test]
7371 fn toggle_blockquote_nests_a_partly_covered_quote() {
7372 // twig's rule: covering only some of a container's blocks nests, because
7373 // taking the quote off would drag its uncovered siblings out with it.
7374 let mut d = doc_with("quote_nest", "> a\n>\n> b\n");
7375 d.caret = 2; // in the first quoted paragraph only
7376 d.toggle_blockquote();
7377 assert_eq!(d.source, "> > a\n>\n> b\n");
7378 }
7379
7380 #[test]
7381 fn a_container_toggle_opens_an_empty_one_on_a_blank_line() {
7382 // A blank line used to be no block for twig to wrap —
7383 // `toggle_block_container` answered `NotFound` — so Quote and the list
7384 // buttons did nothing on the very line the H1 button works on, and leaf
7385 // lent twig a scratch paragraph to wrap and took it back out again.
7386 // twig 3.2.0 opens an empty container there itself, so what is left here
7387 // is where the caret lands: inside the marker that was just written.
7388 let mut d = doc_with("quote_blank", "\nabc\n");
7389 d.caret = 0;
7390 d.toggle_blockquote();
7391 assert_eq!(d.source, "> \nabc\n");
7392 assert_eq!(
7393 d.caret, 2,
7394 "the caret belongs inside the quote it just opened"
7395 );
7396 assert!(d.status.is_none(), "{:?}", d.status);
7397 assert!(d.dirty);
7398
7399 // And the paragraph below is still its own block: an empty container one
7400 // soft break from `abc` would take that paragraph into the quote with it.
7401 let mut d = wysiwyg_doc("quote_blank_rows", "\nabc\n");
7402 d.caret = 0;
7403 d.toggle_blockquote();
7404 d.build_visual(80);
7405 assert_eq!(drawn_rows(&d), ["│ ", "", "abc"]);
7406
7407 // The same from the other side: a blank line directly under a paragraph
7408 // earns the blank line an empty block needs, rather than being read as a
7409 // soft break inside that paragraph.
7410 let mut d = doc_with("list_blank_below", "abc\n");
7411 d.caret = 4;
7412 d.toggle_list(false);
7413 assert_eq!(d.source, "abc\n\n- ");
7414 assert_eq!(d.caret, 7);
7415 }
7416
7417 #[test]
7418 fn enter_at_the_end_of_a_quote_stays_in_the_quote() {
7419 // The gesture the rendering fix is for. `newline` inside a quote already
7420 // wrote the right source — `> a\n` becomes `> a\n>\n> \n`, twig's own
7421 // spelling — but the two marker lines it adds belonged to no node until
7422 // twig 3.2.0, so the gutter stopped at `a` and the line the writer had
7423 // just made drew as plain prose under the quote.
7424 let mut d = wysiwyg_doc("quote_enter", "> a\n");
7425 d.caret = 3; // past `a`, at the end of the quoted line
7426 d.newline();
7427 assert_eq!(d.source, "> a\n>\n> \n");
7428 d.build_visual(80);
7429 assert_eq!(drawn_rows(&d), ["│ a", "│ ", "│ "]);
7430 // And the caret is on the new line, not stranded on the old one.
7431 assert_eq!(d.caret, 8);
7432 }
7433
7434 #[test]
7435 fn opening_a_container_on_a_blank_line_is_one_undo_step() {
7436 // It was three edits — scratch, wrap, unscratch — coalesced into one, and
7437 // now it is twig's single edit. Either way one ⌘z has to put the blank
7438 // line back rather than undoing into a half-built document.
7439 for open in [
7440 &(|d: &mut Doc| d.toggle_blockquote()) as &dyn Fn(&mut Doc),
7441 &|d: &mut Doc| d.toggle_list(false),
7442 &|d: &mut Doc| d.toggle_list(true),
7443 ] {
7444 let mut d = doc_with("container_blank_undo", "a\n\n\n\nb\n");
7445 d.caret = 3;
7446 open(&mut d);
7447 assert_ne!(d.source, "a\n\n\n\nb\n");
7448 d.undo();
7449 assert_eq!(d.source, "a\n\n\n\nb\n");
7450 }
7451 }
7452
7453 #[test]
7454 fn a_container_toggle_is_one_undo_step() {
7455 let mut d = doc_with("quote_undo", "hello\n");
7456 d.caret = 3;
7457 d.insert("X"); // a typing run the structural edit must not fold into
7458 d.toggle_blockquote();
7459 assert_eq!(d.source, "> helXlo\n");
7460 d.undo();
7461 assert_eq!(d.source, "helXlo\n");
7462 }
7463
7464 // ── links ────────────────────────────────────────────────────────────────
7465
7466 #[test]
7467 fn insert_link_wraps_the_selection_and_leaves_its_text_selected() {
7468 let mut d = doc_with("link_sel", "word here\n");
7469 d.anchor = Some(0);
7470 d.caret = 4;
7471 d.insert_link("http://x.dev");
7472 assert_eq!(d.source, "[word](http://x.dev) here\n");
7473 // The text, not the destination — so a second press re-points the link
7474 // the first one made rather than nesting one inside it.
7475 assert_eq!(d.selected_text(), Some("word"));
7476 d.insert_link("http://y.dev");
7477 assert_eq!(d.source, "[word](http://y.dev) here\n");
7478 assert_eq!(d.selected_text(), Some("word"));
7479 }
7480
7481 #[test]
7482 fn insert_image_at_the_caret_spells_the_markup_and_lands_past_it() {
7483 let mut d = doc_with("img_caret", "before after\n");
7484 d.caret = 7; // between "before " and "after"
7485 d.insert_image("cat.png", "a cat");
7486 assert_eq!(d.source, "before after\n");
7487 // The caret sits just past the inserted image, nothing selected.
7488 assert_eq!(d.selection(), None);
7489 assert_eq!(d.caret, 7 + "".len());
7490 }
7491
7492 /// The bug a real vault hit: a filename with spaces in it. Markdown ends a
7493 /// destination at the first space, so the `format!` this used to be wrote
7494 /// something that was not an image at all — and the reader saw the markup as
7495 /// text. twig owns the spelling now, and moves it into the angle form.
7496 #[test]
7497 fn insert_image_spells_a_destination_with_spaces_so_it_stays_an_image() {
7498 let mut d = doc_with("img_space", "x\n");
7499 d.caret = 0;
7500 d.insert_image("Jesus Commands the Apostles to Rest.jpg", "");
7501 assert_eq!(
7502 d.source,
7503 "x\n"
7504 );
7505 // And it reads back as an image pointing at the unescaped path — the angle
7506 // brackets are spelling, not part of the destination.
7507 d.caret = 2;
7508 assert_eq!(
7509 d.image_destination_at_caret(),
7510 Some("Jesus Commands the Apostles to Rest.jpg".to_string())
7511 );
7512 }
7513
7514 /// A `)` in a caption or a filename must not close the image early.
7515 #[test]
7516 fn insert_image_escapes_a_paren_in_either_half() {
7517 let mut d = doc_with("img_paren", "x\n");
7518 d.caret = 0;
7519 d.insert_image("a)b.png", "");
7520 assert_eq!(d.source, "b.png)x\n");
7521 d.caret = 2;
7522 assert_eq!(d.image_destination_at_caret(), Some("a)b.png".to_string()));
7523 }
7524
7525 #[test]
7526 fn insert_image_uses_the_selection_as_alt_text() {
7527 let mut d = doc_with("img_sel", "caption here\n");
7528 d.anchor = Some(0);
7529 d.caret = 7; // "caption"
7530 d.insert_image("p.png", "ignored fallback");
7531 assert_eq!(d.source, " here\n");
7532 }
7533
7534 #[test]
7535 fn insert_image_with_no_alt_leaves_empty_brackets() {
7536 let mut d = doc_with("img_noalt", "\n");
7537 d.caret = 0;
7538 d.insert_image("logo.svg", "");
7539 assert_eq!(d.source, "\n");
7540 }
7541
7542 #[test]
7543 fn insert_media_spells_a_video_as_html_and_reads_it_back_as_a_block() {
7544 // The round trip is the point: it's no use writing markup the reader
7545 // can't pick up again. This is the pair that only holds from twig 2.5.1
7546 // on — before it, the one-line form went in fine and came back as a
7547 // paragraph of raw tags, publishing no media at all.
7548 let mut d = doc_with("vid_rt", "\n");
7549 d.caret = 0;
7550 d.insert_media(MediaKind::Video, "clip.mp4", "a clip");
7551 assert_eq!(
7552 d.source,
7553 "<video src=\"clip.mp4\" controls>a clip</video>\n"
7554 );
7555
7556 d.build_visual(80);
7557 assert_eq!(d.vmap.media.len(), 1, "reads back as one block media");
7558 assert_eq!(d.vmap.media[0].kind, MediaKind::Video);
7559 assert_eq!(d.vmap.media[0].destination, "clip.mp4");
7560 assert_eq!(d.vmap.media[0].alt, "a clip");
7561 }
7562
7563 #[test]
7564 fn insert_media_spells_audio_with_its_own_tag() {
7565 let mut d = doc_with("aud_rt", "\n");
7566 d.caret = 0;
7567 d.insert_media(MediaKind::Audio, "take.mp3", "");
7568 assert_eq!(d.source, "<audio src=\"take.mp3\" controls></audio>\n");
7569 d.build_visual(80);
7570 assert_eq!(d.vmap.media[0].kind, MediaKind::Audio);
7571 }
7572
7573 #[test]
7574 fn insert_media_uses_the_selection_as_fallback_text() {
7575 // The same courtesy `insert_image` does with alt: select a caption,
7576 // insert, and the caption labels the thing rather than being replaced.
7577 let mut d = doc_with("vid_sel", "the talk here\n");
7578 d.anchor = Some(0);
7579 d.caret = 8; // "the talk"
7580 d.insert_media(MediaKind::Video, "talk.mp4", "ignored fallback");
7581 assert_eq!(
7582 d.source,
7583 "<video src=\"talk.mp4\" controls>the talk</video> here\n"
7584 );
7585 }
7586
7587 #[test]
7588 fn insert_media_with_an_image_kind_is_just_insert_image() {
7589 let mut d = doc_with("img_via_media", "\n");
7590 d.caret = 0;
7591 d.insert_media(MediaKind::Image, "logo.svg", "x");
7592 assert_eq!(d.source, "\n");
7593 }
7594
7595 // ── thematic breaks ─────────────────────────────────────────────────────
7596
7597 /// The node the source parses as at `caret` — what confirms an inserted
7598 /// `---` actually reads back as a rule, not stray text or a setext heading.
7599 ///
7600 /// The *narrowest* node covering the offset. Every ancestor covers it too,
7601 /// and since twig 2.8 that includes the `doc` root, which now carries a real
7602 /// span (it reported none before, so taking the first match used to land on
7603 /// the block by luck and now always answers `"doc"`).
7604 fn kind_at(d: &mut Doc, caret: usize) -> Option<Kind> {
7605 d.nodes()
7606 .into_iter()
7607 .filter(|n| n.span.start <= caret && caret < n.span.end)
7608 .min_by_key(|n| n.span.end - n.span.start)
7609 .map(|n| n.kind)
7610 }
7611
7612 #[test]
7613 fn a_task_box_toggles_at_the_caret_and_reads_back() {
7614 let mut d = doc_with("task_toggle", "- [ ] todo\n- [x] done\n");
7615 d.caret = 8; // inside "todo"
7616 assert_eq!(d.task_checked_at_caret(), Some(false));
7617 d.toggle_task_checked();
7618 assert_eq!(d.source, "- [x] todo\n- [x] done\n");
7619 assert_eq!(d.task_checked_at_caret(), Some(true));
7620 d.toggle_task_checked();
7621 assert_eq!(d.source, "- [ ] todo\n- [x] done\n");
7622 }
7623
7624 #[test]
7625 fn a_click_toggles_a_box_without_taking_the_caret_with_it() {
7626 // The whole reason `toggle_task_at` exists apart from the caret form:
7627 // ticking a box elsewhere must not move the cursor out of what's being
7628 // typed.
7629 let mut d = doc_with("task_click", "- [ ] first\n- [ ] second\n");
7630 d.caret = 8; // inside "first"
7631 let second = d.source.find("second").unwrap();
7632 d.toggle_task_at(second);
7633 assert_eq!(d.source, "- [ ] first\n- [x] second\n");
7634 assert_eq!(d.caret, 8, "the caret stayed in the first item");
7635 }
7636
7637 #[test]
7638 fn a_plain_item_gains_and_loses_a_box() {
7639 let mut d = doc_with("task_mint", "- plain\n");
7640 d.caret = 4;
7641 assert_eq!(d.task_checked_at_caret(), None);
7642 d.toggle_task_item();
7643 assert_eq!(d.source, "- [ ] plain\n");
7644 assert_eq!(
7645 d.task_checked_at_caret(),
7646 Some(false),
7647 "a new box arrives unticked"
7648 );
7649 d.toggle_task_item();
7650 assert_eq!(d.source, "- plain\n");
7651 }
7652
7653 #[test]
7654 fn ticking_a_box_that_isnt_there_reports_rather_than_minting_one() {
7655 // `set checked` must not silently convert a bullet into a task — that is
7656 // `toggle_task_item`'s job, and twig refuses it here.
7657 let mut d = doc_with("task_none", "- plain\n");
7658 d.caret = 4;
7659 d.toggle_task_checked();
7660 assert_eq!(d.source, "- plain\n", "nothing written");
7661 assert!(
7662 d.status.is_some(),
7663 "the refusal should reach the status line"
7664 );
7665 }
7666
7667 #[test]
7668 fn a_task_item_in_a_quote_is_found_past_the_quote_marker() {
7669 let mut d = doc_with("task_quote", "> - [ ] nested\n");
7670 d.caret = d.source.find("nested").unwrap();
7671 assert_eq!(d.task_checked_at_caret(), Some(false));
7672 d.toggle_task_checked();
7673 assert_eq!(d.source, "> - [x] nested\n");
7674 }
7675
7676 #[test]
7677 fn insert_thematic_break_parts_the_paragraph_around_the_caret() {
7678 // A rule is a block, so twig's `insert_thematic_break` alone lands it
7679 // after the whole paragraph. `split_block` parts the paragraph first and
7680 // the rule is aimed at the *first* half, which is what a rule button is
7681 // understood to do — and what leaf spelled by hand until twig grew both
7682 // halves of the gesture.
7683 let mut d = doc_with("hr_mid", "before after\n");
7684 d.caret = 7; // between "before " and "after"
7685 d.insert_thematic_break();
7686 assert_eq!(d.source, "before \n\n---\n\nafter\n");
7687 assert_eq!(d.selection(), None);
7688 assert_eq!(
7689 kind_at(&mut d, "before \n\n".len()),
7690 Some(Kind::ThematicBreak)
7691 );
7692 }
7693
7694 #[test]
7695 fn insert_thematic_break_spells_the_rule_the_format_s_own_way() {
7696 // The whole point of delegating: `---` is Markdown's, `* * *` is djot's,
7697 // and leaf wrote the first into both until twig started spelling it.
7698 let mut md = doc_with("hr_md", "para\n");
7699 md.caret = 2;
7700 md.insert_thematic_break();
7701 assert_eq!(md.source, "pa\n\n---\n\nra\n");
7702
7703 let mut dj = Doc::from_source("para\n".into(), Format::Djot).unwrap();
7704 dj.caret = 2;
7705 dj.insert_thematic_break();
7706 assert_eq!(dj.source, "pa\n\n* * *\n\nra\n");
7707 }
7708
7709 #[test]
7710 fn enter_in_a_nested_list_item_keeps_the_new_item_nested() {
7711 // The same bytes are two documents. In Markdown ` - b` is a nested item
7712 // and the next one belongs beside it, at its indent. In Djot a list
7713 // marker can't interrupt a paragraph, so those bytes are literal text in
7714 // item `a` and there is only one item — writing ` - ` under it would add
7715 // no item at all, just more text, and the new sibling has to go to
7716 // column zero. Both spellings come out of the *enclosing item's* line.
7717 let mut md = wysiwyg_doc("enter_nested_md", "- a\n - b\n");
7718 md.caret = "- a\n - b".len();
7719 md.newline();
7720 assert_eq!(md.source, "- a\n - b\n - \n");
7721 assert_eq!(list_items(&mut md), 3);
7722
7723 let mut dj = Doc::from_source("- a\n - b\n".into(), Format::Djot).unwrap();
7724 dj.view = View::Wysiwyg;
7725 dj.build_visual(80);
7726 dj.caret = "- a\n - b".len();
7727 dj.newline();
7728 assert_eq!(dj.source, "- a\n - b\n- \n");
7729 assert_eq!(list_items(&mut dj), 2);
7730
7731 // Where Djot's nesting is real — opened by a blank line — the indent is
7732 // reproduced there too, and the two formats agree again.
7733 let mut dj = Doc::from_source("- a\n\n - b\n".into(), Format::Djot).unwrap();
7734 dj.view = View::Wysiwyg;
7735 dj.build_visual(80);
7736 dj.caret = "- a\n\n - b".len();
7737 dj.newline();
7738 assert_eq!(dj.source, "- a\n\n - b\n - \n");
7739 assert_eq!(list_items(&mut dj), 3);
7740 }
7741
7742 #[test]
7743 fn tab_nests_an_item_at_the_column_its_own_marker_asks_for() {
7744 // Tab replaces the line's whole prefix with the one twig spells, so the
7745 // quote markers, the parent's indent and an ordered marker's extra
7746 // column are all its answer rather than leaf's arithmetic.
7747 for (name, body, caret, want) in [
7748 ("bullet", "- a\n- b\n", 6, "- a\n - b\n"),
7749 ("ordered", "1. a\n2. b\n", 8, "1. a\n 1. b\n"),
7750 ("quoted", "> - a\n> - b\n", 10, "> - a\n> - b\n"),
7751 // A checkbox is markup the item's own text wraps past, but a nested
7752 // list may only open at the *list* marker's column — four in from
7753 // there is a paragraph continuation, and `- [ ] a\n - [ ] b`
7754 // parses as one item, not two.
7755 ("task", "- [ ] a\n- [ ] b\n", 14, "- [ ] a\n - [ ] b\n"),
7756 (
7757 "quoted task",
7758 "> - [ ] a\n> - [ ] b\n",
7759 18,
7760 "> - [ ] a\n> - [ ] b\n",
7761 ),
7762 ] {
7763 let mut doc = wysiwyg_doc(name, body);
7764 doc.caret = caret;
7765 doc.indent();
7766 assert_eq!(doc.source, want, "{name}");
7767 // The nesting is real, not just indented text.
7768 assert_eq!(list_items(&mut doc), 2, "{name}");
7769 }
7770 }
7771
7772 #[test]
7773 fn backspace_only_outdents_where_the_format_says_there_is_an_item() {
7774 // The same bytes, the two formats disagreeing, and a gesture that used
7775 // to read the bytes. ` - b` is a nested item in Markdown, so Backspace
7776 // at its marker outdents. In Djot a marker can't interrupt a paragraph,
7777 // so those bytes are literal text inside item `a` — there is nothing to
7778 // outdent, and treating them as a marker turned one item into two, a
7779 // structural edit from a keystroke that should delete one character.
7780 //
7781 // twig's `line_prefix` is what tells them apart: it reports the marker
7782 // on the Markdown line and nothing on the Djot one, which is a
7783 // continuation. No byte scan can reach that answer.
7784 let src = "- a\n - b\n";
7785 let at = "- a\n - ".len();
7786
7787 let mut md = Doc::from_source(src.into(), Format::Markdown).unwrap();
7788 md.view = View::Wysiwyg;
7789 md.build_visual(80);
7790 md.caret = at;
7791 md.backspace();
7792 assert_eq!(md.source, "- a\n- b\n");
7793 assert_eq!(list_items(&mut md), 2);
7794
7795 let mut dj = Doc::from_source(src.into(), Format::Djot).unwrap();
7796 dj.view = View::Wysiwyg;
7797 dj.build_visual(80);
7798 dj.caret = at;
7799 dj.backspace();
7800 assert_eq!(dj.source, "- a\n -b\n"); // an ordinary character delete
7801 assert_eq!(list_items(&mut dj), 1); // and the structure is untouched
7802 }
7803
7804 #[test]
7805 fn enter_in_a_checklist_item_starts_another_unchecked_one() {
7806 // Leaf used to spell the next item from the marker bytes it scanned, and
7807 // its scanner stopped at the bullet — so Enter in a checklist wrote `- `
7808 // and dropped out of the checklist. twig reproduces the whole
7809 // continuation, and a fresh item is always unticked however the one above
7810 // it stands.
7811 for (name, body, want) in [
7812 ("unchecked", "- [ ] a\n", "- [ ] a\n- [ ] \n"),
7813 ("checked", "- [x] a\n", "- [x] a\n- [ ] \n"),
7814 ] {
7815 let mut doc = wysiwyg_doc(name, body);
7816 doc.caret = body.trim_end_matches('\n').len();
7817 doc.newline();
7818 assert_eq!(doc.source, want, "{name}");
7819 // Both items are checklist items — the new one is a box, not the
7820 // plain bullet the old marker scan left behind — and it is unticked
7821 // whichever way the one above it faces.
7822 let boxes: Vec<Option<bool>> = doc
7823 .nodes()
7824 .iter()
7825 .filter(|n| n.kind == Kind::TaskListItem)
7826 .map(|n| n.checked)
7827 .collect();
7828 assert_eq!(boxes.len(), 2, "{name}");
7829 assert_eq!(boxes[1], Some(false), "{name}");
7830 }
7831 }
7832
7833 #[test]
7834 fn a_split_takes_the_space_the_caret_was_in_front_of() {
7835 // Splicing a break at the caret strands the space the words were parted
7836 // at on the head of the second block, where it reads as an indent nobody
7837 // typed. twig's split consumes it.
7838 for (name, body, caret, want) in [
7839 ("para", "one two\n", 3, "one\n\ntwo\n"),
7840 ("item", "- one two\n", 5, "- one\n- two\n"),
7841 ("quote", "> one two\n", 5, "> one\n>\n> two\n"),
7842 // A heading takes leaf's own path, which has to match.
7843 ("heading", "# one two\n", 5, "# one\n\ntwo\n"),
7844 ] {
7845 let mut doc = wysiwyg_doc(name, body);
7846 doc.caret = caret;
7847 doc.newline();
7848 assert_eq!(doc.source, want, "{name}");
7849 }
7850 }
7851
7852 #[test]
7853 fn enter_at_the_end_of_a_heading_opens_a_paragraph() {
7854 // The one place leaf keeps its own break: `split_block` repeats the `#`,
7855 // and Enter after a title is how the body under it is asked for.
7856 let mut doc = wysiwyg_doc("head_enter", "# Title\n");
7857 doc.caret = "# Title".len();
7858 doc.newline();
7859 doc.insert("body");
7860 assert_eq!(doc.source, "# Title\n\nbody\n");
7861 assert_eq!(
7862 doc.nodes()
7863 .iter()
7864 .filter(|n| n.kind == Kind::Heading)
7865 .count(),
7866 1
7867 );
7868 }
7869
7870 #[test]
7871 fn enter_in_a_quoted_list_item_starts_the_next_quoted_item() {
7872 // A quoted item's marker doesn't open its line, so a scan that starts at
7873 // column zero finds a `>` where it wanted a bullet, calls the line "not a
7874 // list" and hands Enter to the plain-quote branch — which writes `> ` and
7875 // drops the list. The next item has to carry the whole prefix.
7876 for (name, body, want) in [
7877 ("flat", "> - a\n", "> - a\n> - \n"),
7878 ("sibling", "> - a\n> - b\n", "> - a\n> - b\n> - \n"),
7879 ("nested", "> - a\n> - b\n", "> - a\n> - b\n> - \n"),
7880 ("ordered", "> 1. a\n> 2. b\n", "> 1. a\n> 2. b\n> 3. \n"),
7881 ("twice quoted", "> > - a\n", "> > - a\n> > - \n"),
7882 ] {
7883 let mut doc = wysiwyg_doc(name, body);
7884 doc.caret = body.trim_end_matches('\n').len();
7885 doc.newline();
7886 assert_eq!(doc.source, want, "{name}");
7887 // The marker isn't just spelled right, it parses as an item.
7888 assert_eq!(list_items(&mut doc), body.lines().count() + 1, "{name}");
7889 }
7890 }
7891
7892 #[test]
7893 fn an_empty_quoted_item_leaves_the_list_and_stays_in_the_quote() {
7894 // Double-Enter exits the list. Unquoted that means a blank line, but a
7895 // *bare* blank line would end the quote too and drop the caret out of it,
7896 // so the separator keeps its `>` and the caret's line keeps its `> `.
7897 let mut doc = wysiwyg_doc("quoted_exit", "> - a\n> - \n");
7898 doc.caret = "> - a\n> - ".len();
7899 doc.newline();
7900 assert_eq!(doc.source, "> - a\n>\n> \n");
7901 assert_eq!(list_items(&mut doc), 1);
7902 // What "still in the quote" means for the next keystroke: the caret sits
7903 // behind the prefix, and what's typed there lands inside the quote as a
7904 // paragraph of its own — not as more of item `a`.
7905 doc.insert("x");
7906 assert_eq!(doc.source, "> - a\n>\n> x\n");
7907 assert!(
7908 doc.editor
7909 .ancestors_at(doc.caret - 1)
7910 .is_ok_and(|c| c.into_iter().any(|m| m.kind == Kind::BlockQuote))
7911 );
7912 }
7913
7914 #[test]
7915 fn backspace_at_a_quoted_marker_takes_the_marker_and_leaves_the_quote() {
7916 // The marker is hidden block markup, so Backspace over it is structural —
7917 // but only the marker is the list's. Splicing from the line start would
7918 // take the `>` with it and silently unquote the line.
7919 let mut doc = wysiwyg_doc("quoted_bksp", "> - a\n");
7920 doc.caret = "> - ".len();
7921 doc.backspace();
7922 assert_eq!(doc.source, "> a\n");
7923 assert_eq!(list_items(&mut doc), 0);
7924
7925 // A nested one outdents instead, moving the bullet within the quote
7926 // rather than moving the quote.
7927 let mut doc = wysiwyg_doc("quoted_outdent", "> - a\n> - b\n");
7928 doc.caret = "> - a\n> - ".len();
7929 doc.backspace();
7930 assert_eq!(doc.source, "> - a\n> - b\n");
7931 assert_eq!(list_items(&mut doc), 2);
7932 }
7933
7934 #[test]
7935 fn only_a_bare_paragraph_is_parted_around_the_caret() {
7936 // The split is deliberately narrow. Parting a fenced block would leave
7937 // two fences with a rule between them, and parting a list item would
7938 // mint an item nobody asked for on the way to a rule that lands after
7939 // the list either way — so both keep the whole block intact and take the
7940 // rule after it. A caret in a quote is likewise left alone.
7941 for (name, body, caret, want) in [
7942 (
7943 "code",
7944 "```\nfn x() {}\n```\n",
7945 8,
7946 "```\nfn x() {}\n```\n\n---\n",
7947 ),
7948 ("list", "- one two\n", 6, "- one two\n\n---\n"),
7949 ("quote", "> one two\n", 6, "> one two\n>\n> ---\n"),
7950 ] {
7951 let mut d = doc_with(&format!("hr_narrow_{name}"), body);
7952 d.caret = caret;
7953 d.insert_thematic_break();
7954 assert_eq!(d.source, want, "{name}: the block should stay whole");
7955 }
7956 }
7957
7958 #[test]
7959 fn insert_thematic_break_replaces_the_selection() {
7960 // Now that the rule lands *at* the caret again, replacing the selection
7961 // is coherent once more: the text goes, and the rule takes its place.
7962 // The space the deletion left leading the second half is consumed by the
7963 // split rather than opening the new paragraph with it.
7964 let mut d = doc_with("hr_sel", "one two three\n");
7965 d.anchor = Some(4);
7966 d.caret = 7; // "two"
7967 d.insert_thematic_break();
7968 assert_eq!(d.source, "one \n\n---\n\nthree\n");
7969 assert_eq!(d.selection(), None);
7970 }
7971
7972 #[test]
7973 fn insert_thematic_break_clears_a_code_block_and_a_table_rather_than_refusing() {
7974 // Both are blocks the rule lands *after*. Leaf used to refuse a fence,
7975 // because writing `---` into one is code, not a rule — twig now walks out
7976 // to the block that owns the caret's line, so there is nothing to refuse.
7977 let mut code = doc_with("hr_code", "```\nfn x() {}\n```\n");
7978 code.caret = 5; // inside the fenced code
7979 code.insert_thematic_break();
7980 assert_eq!(code.source, "```\nfn x() {}\n```\n\n---\n");
7981 assert_eq!(code.status, None, "no refusal to report any more");
7982
7983 let mut table = doc_with("hr_table", "| a | b |\n|---|---|\n| 1 | 2 |\n");
7984 table.caret = 3; // in the header row
7985 table.insert_thematic_break();
7986 assert_eq!(table.source, "| a | b |\n|---|---|\n| 1 | 2 |\n\n---\n");
7987 }
7988
7989 #[test]
7990 fn insert_thematic_break_in_a_list_item_ends_the_list() {
7991 // The un-indented rule cannot continue the list, so it closes the list
7992 // and lands at the top level rather than nested inside it.
7993 let mut d = doc_with("hr_list", "- one\n- two\n");
7994 d.caret = "- one\n- tw".len(); // mid "two"
7995 d.insert_thematic_break();
7996 d.build_visual(80);
7997 let rule_at = d.source.find("---").unwrap();
7998 assert_eq!(kind_at(&mut d, rule_at), Some(Kind::ThematicBreak));
7999 assert!(
8000 !d.nodes().iter().any(|n| n.kind == Kind::BulletList
8001 && n.span.start <= rule_at
8002 && rule_at < n.span.end),
8003 "the rule must not be nested inside the list"
8004 );
8005 }
8006
8007 #[test]
8008 fn insert_thematic_break_in_a_blockquote_stays_in_the_quote() {
8009 // Leaf used to end the quote. twig gives the rule the quote's own prefix,
8010 // which is the document the gesture was actually asked for.
8011 let mut d = doc_with("hr_quote", "> hello\n");
8012 d.caret = 4; // inside the quoted text
8013 d.insert_thematic_break();
8014 assert_eq!(d.source, "> hello\n>\n> ---\n");
8015 d.build_visual(80);
8016 let rule_at = d.source.find("---").unwrap();
8017 assert_eq!(kind_at(&mut d, rule_at), Some(Kind::ThematicBreak));
8018 assert!(
8019 d.nodes().iter().any(|n| n.kind == Kind::BlockQuote
8020 && n.span.start <= rule_at
8021 && rule_at < n.span.end),
8022 "the rule belongs to the quote it was asked for"
8023 );
8024 }
8025
8026 // ── typing against a block picture ────────────────────────────────────────
8027
8028 /// A rendered-view document with the caret parked on one of the picture's two
8029 /// stops, and the map already built — the state a frontend is in between
8030 /// drawing a frame and the next keystroke.
8031 fn doc_at_picture(name: &str, src: &str, side: MediaStop) -> Doc {
8032 let mut d = doc_in(View::Wysiwyg, name, src);
8033 d.build_visual_unwrapped();
8034 let start = src.find("".len(),
8038 };
8039 d
8040 }
8041
8042 /// The block media the map publishes, after rebuilding it — "is this still a
8043 /// picture, or has it become a line of text with an image in it?"
8044 fn media_count(d: &mut Doc) -> usize {
8045 d.build_visual_unwrapped();
8046 d.vmap.media.len()
8047 }
8048
8049 #[test]
8050 fn typing_past_a_block_picture_opens_a_paragraph_under_it() {
8051 // The accident this prevents: tap the blank page under a photo (which
8052 // lands on the picture's trailing stop), type, and `xy` is a
8053 // paragraph with an *inline* image — the photo stops being drawn.
8054 let mut d = doc_at_picture("pic_after", "hi\n\n\n", MediaStop::After);
8055 d.insert("xy");
8056 assert_eq!(d.source, "hi\n\n\n\nxy\n");
8057 assert_eq!(media_count(&mut d), 1, "still a picture");
8058 }
8059
8060 #[test]
8061 fn typing_in_front_of_a_block_picture_opens_a_paragraph_above_it() {
8062 let mut d = doc_at_picture("pic_before", "hi\n\n\n", MediaStop::Before);
8063 d.insert("xy");
8064 assert_eq!(d.source, "hi\n\nxy\n\n\n");
8065 assert_eq!(media_count(&mut d), 1);
8066 }
8067
8068 #[test]
8069 fn a_picture_that_opens_the_document_still_takes_a_paragraph_above_it() {
8070 let mut d = doc_at_picture("pic_first", "\n", MediaStop::Before);
8071 d.insert("x");
8072 assert_eq!(d.source, "x\n\n\n");
8073 assert_eq!(media_count(&mut d), 1);
8074 }
8075
8076 #[test]
8077 fn one_undo_puts_the_picture_back_the_way_it_was_found() {
8078 // The opened paragraph is part of the keystroke, not an edit the writer
8079 // made — so it undoes with the character, not a step later.
8080 let mut d = doc_at_picture("pic_undo", "hi\n\n\n", MediaStop::After);
8081 d.insert("x");
8082 assert_eq!(d.source, "hi\n\n\n\nx\n");
8083 d.undo();
8084 assert_eq!(d.source, "hi\n\n\n");
8085 }
8086
8087 #[test]
8088 fn pasting_against_a_block_picture_opens_a_paragraph_too() {
8089 // ⌘V dissolves the picture exactly as a keystroke does.
8090 let mut d = doc_at_picture("pic_paste", "hi\n\n\n", MediaStop::After);
8091 d.paste("pasted");
8092 assert_eq!(d.source, "hi\n\n\n\npasted\n");
8093 assert_eq!(media_count(&mut d), 1);
8094 }
8095
8096 #[test]
8097 fn typing_beside_an_inline_image_is_ordinary_editing() {
8098 // An inline image has no placeholder row and no stops of its own. Opening
8099 // a paragraph mid-sentence would be the bug, not the fix.
8100 let mut d = doc_in(View::Wysiwyg, "pic_inline", "see  here\n");
8101 d.build_visual_unwrapped();
8102 d.caret = "see ".len();
8103 d.insert("!");
8104 assert_eq!(d.source, "see ! here\n");
8105 }
8106
8107 #[test]
8108 fn source_view_types_raw_markup_against_an_image_untouched() {
8109 // Source view is for writing the markup itself; a break inserted behind
8110 // the writer's back there would be the editor arguing with them.
8111 let mut d = doc_in(View::Source, "pic_src", "\n");
8112 d.caret = "".len();
8113 d.insert("x");
8114 assert_eq!(d.source, "x\n");
8115 }
8116
8117 #[test]
8118 fn typing_over_a_selection_that_starts_at_a_picture_stop_replaces_it() {
8119 // A selection is replaced, not joined into, so there is nothing to
8120 // protect: the range takes the picture with it.
8121 let mut d = doc_at_picture("pic_sel", "hi\n\n\n", MediaStop::Before);
8122 d.anchor = Some(d.caret);
8123 d.caret = d.source.find("".len();
8124 d.insert("x");
8125 assert_eq!(d.source, "hi\n\nx\n");
8126 }
8127
8128 #[test]
8129 fn backspace_past_a_block_picture_deletes_the_picture_not_its_last_byte() {
8130 // What this actually cost: a real vault's photo, to one stray Backspace.
8131 // The caret past `` was deleting the closing paren — invisible
8132 // in the rendered view — and the photo became the text `\n", MediaStop::After);
8134 d.backspace();
8135 assert_eq!(d.source, "hi\n");
8136 assert_eq!(media_count(&mut d), 0, "the picture went, in one piece");
8137 d.undo();
8138 assert_eq!(
8139 d.source, "hi\n\n\n",
8140 "and comes back in one piece"
8141 );
8142 }
8143
8144 #[test]
8145 fn backspace_in_front_of_a_block_picture_steps_out_instead_of_merging_it() {
8146 // Deleting the break here would join the picture to the paragraph above,
8147 // where it is an *inline* image and stops being drawn. Step over the
8148 // boundary; the next press deletes in the paragraph the caret reached.
8149 let mut d = doc_at_picture("pic_bs_before", "hi\n\n\n", MediaStop::Before);
8150 d.backspace();
8151 assert_eq!(d.source, "hi\n\n\n", "nothing deleted");
8152 assert_eq!(d.caret, 2, "the caret stepped up to the end of `hi`");
8153 d.backspace();
8154 assert_eq!(d.source, "h\n\n\n", "and now it deletes there");
8155 assert_eq!(media_count(&mut d), 1, "the picture was never at risk");
8156 }
8157
8158 #[test]
8159 fn forward_delete_in_front_of_a_block_picture_deletes_the_picture() {
8160 // The mirror. A byte-step here eats the `!` and leaves a link.
8161 let mut d = doc_at_picture("pic_del", "hi\n\n\n\nbye\n", MediaStop::Before);
8162 d.delete_forward();
8163 assert_eq!(d.source, "hi\n\nbye\n");
8164 assert_eq!(media_count(&mut d), 0);
8165 }
8166
8167 #[test]
8168 fn forward_delete_past_a_block_picture_steps_over_the_boundary() {
8169 let mut d = doc_at_picture(
8170 "pic_del_after",
8171 "hi\n\n\n\nbye\n",
8172 MediaStop::After,
8173 );
8174 d.delete_forward();
8175 assert_eq!(d.source, "hi\n\n\n\nbye\n", "nothing deleted");
8176 assert_eq!(
8177 d.caret,
8178 d.source.find("bye").unwrap(),
8179 "the caret stepped down to `bye`"
8180 );
8181 }
8182
8183 #[test]
8184 fn a_picture_that_is_the_whole_document_still_deletes_cleanly() {
8185 let mut d = doc_at_picture("pic_only", "\n", MediaStop::After);
8186 d.backspace();
8187 assert_eq!(d.source, "\n");
8188 assert_eq!(media_count(&mut d), 0);
8189 }
8190
8191 #[test]
8192 fn a_word_delete_takes_the_picture_whole_or_steps_out_of_it() {
8193 // ⌥⌫ past a picture would otherwise eat a "word" of its markup.
8194 let mut d = doc_at_picture("pic_wordbs", "hi there\n\n\n", MediaStop::After);
8195 d.delete_word_back();
8196 assert_eq!(d.source, "hi there\n");
8197
8198 // And in front of one it runs *through* the paragraph break into the
8199 // prose above, which merges the picture inline — so it steps out first,
8200 // and the second press deletes the word it was aimed at.
8201 let mut d = doc_at_picture("pic_wordbs2", "hi there\n\n\n", MediaStop::Before);
8202 d.delete_word_back();
8203 assert_eq!(d.source, "hi there\n\n\n");
8204 d.delete_word_back();
8205 assert_eq!(
8206 d.source, "hi \n\n\n",
8207 "the word above went, the picture stayed"
8208 );
8209 assert_eq!(media_count(&mut d), 1);
8210 }
8211
8212 #[test]
8213 fn source_view_deletes_raw_markup_against_an_image_untouched() {
8214 let mut d = doc_in(View::Source, "pic_src_del", "\n");
8215 d.caret = "".len();
8216 d.backspace();
8217 assert_eq!(d.source, ";
8218 }
8219
8220 #[test]
8221 fn image_destination_at_caret_reads_the_image_under_the_caret() {
8222 let mut d = doc_with("img_read", "\n");
8223 d.caret = 3; // inside the image markup
8224 assert_eq!(d.image_destination_at_caret(), Some("cat.png".to_string()));
8225 // Past the image, the caret is in no image.
8226 d.caret = "".len();
8227 assert_eq!(d.image_destination_at_caret(), None);
8228 }
8229
8230 #[test]
8231 fn set_media_rows_reserves_blank_filler_rows_the_frontend_paints_over() {
8232 // The image is one placeholder row by default, and `set_media_rows` grows
8233 // it to the height the frontend measured: the label row plus blank
8234 // `decoration` fillers that hold the vertical space a raster is drawn into.
8235 let mut d = wysiwyg_doc("img_rows", "intro\n\n\n\nend\n");
8236 assert_eq!(d.vmap.media.len(), 1);
8237 let img_row = d.vmap.media[0].rows_span.start;
8238 assert_eq!(
8239 d.vmap.media[0].rows_span,
8240 img_row..img_row + 1,
8241 "default is one row"
8242 );
8243
8244 d.set_media_rows(HashMap::from([("cat.png".to_string(), 4)]));
8245 d.build_visual(80);
8246 assert_eq!(d.vmap.media.len(), 1, "still one image, now taller");
8247 let span = d.vmap.media[0].rows_span.clone();
8248 assert_eq!(span.end - span.start, 4, "reserves the four rows asked for");
8249 // The label row carries the mark and its glyphs; the three below are blank
8250 // decoration — drawn, but no caret and no text.
8251 assert!(
8252 d.vmap.rows[span.start].media.is_some(),
8253 "mark rides the first row"
8254 );
8255 for r in (span.start + 1)..span.end {
8256 assert!(d.vmap.rows[r].decoration, "filler row {r} is decoration");
8257 assert!(d.vmap.rows[r].glyphs.is_empty(), "filler row {r} is blank");
8258 assert!(
8259 d.vmap.rows[r].media.is_none(),
8260 "only the first row is marked"
8261 );
8262 }
8263 }
8264
8265 #[test]
8266 fn a_taller_image_adds_no_caret_stops_and_motion_steps_over_its_fillers() {
8267 // The extra rows are pure spacers: the caret's only homes stay the stop in
8268 // front of the image and the one just past it, so walking the document top
8269 // to bottom visits the same offsets whether the image is 1 row or 5.
8270 let body = "ab\n\n\n\ncd\n";
8271 let stops_at = |rows: usize| -> Vec<usize> {
8272 let mut d = wysiwyg_doc("img_stops", body);
8273 if rows > 1 {
8274 d.set_media_rows(HashMap::from([("p.png".to_string(), rows)]));
8275 d.build_visual(80);
8276 }
8277 d.caret = 0;
8278 let mut seen = vec![d.caret];
8279 loop {
8280 d.move_right(false);
8281 if *seen.last().unwrap() == d.caret {
8282 break;
8283 }
8284 seen.push(d.caret);
8285 }
8286 seen
8287 };
8288 assert_eq!(
8289 stops_at(1),
8290 stops_at(5),
8291 "reserving rows must not add stops"
8292 );
8293 }
8294
8295 #[test]
8296 fn insert_link_repoints_the_link_at_a_bare_caret() {
8297 let mut d = doc_with("link_repoint", "[word](http://x.dev)\n");
8298 d.caret = 3; // in the link's text, nothing selected
8299 d.insert_link("http://y.dev");
8300 assert_eq!(d.source, "[word](http://y.dev)\n");
8301 assert_eq!(d.selected_text(), Some("word"));
8302 }
8303
8304 #[test]
8305 fn insert_link_on_an_empty_range_autolinks_a_url() {
8306 // A link with no text of its own is an autolink, and twig spells it —
8307 // `<…>` is the canonical form and needs no text typed into it, so the
8308 // caret lands after it rather than selecting a finished link.
8309 let mut d = doc_with("link_empty", "\n");
8310 d.caret = 0;
8311 d.insert_link("http://x.dev");
8312 assert_eq!(d.source, "<http://x.dev>\n");
8313 assert_eq!(d.selection(), None);
8314 assert_eq!(d.caret, 14);
8315 }
8316
8317 #[test]
8318 fn insert_link_on_an_empty_range_falls_back_for_a_non_url() {
8319 // `<./notes.md>` is literal text in both formats and `<foo>` is raw HTML
8320 // in Markdown, so a destination that can't autolink doubles as the text
8321 // instead — which is then selected, ready to be typed over.
8322 let mut d = doc_with("link_rel", "\n");
8323 d.caret = 0;
8324 d.insert_link("./notes.md");
8325 assert_eq!(d.source, "[./notes.md](./notes.md)\n");
8326 assert_eq!(d.selection(), Some((1, 11)));
8327 d.insert("Notes");
8328 assert_eq!(d.source, "[Notes](./notes.md)\n");
8329 }
8330
8331 #[test]
8332 fn insert_link_repoints_the_autolink_the_caret_stands_in() {
8333 // The autolink's text is its URL, so re-pointing replaces the whole
8334 // node — the caret must not splice a second link inside the first.
8335 let mut d = doc_with("link_repoint_auto", "see <https://x.dev> ok\n");
8336 d.caret = 10;
8337 d.insert_link("https://y.dev");
8338 assert_eq!(d.source, "see <https://y.dev> ok\n");
8339 }
8340
8341 #[test]
8342 fn code_language_reads_and_edits_through_the_fence() {
8343 let mut d = doc_with("code_lang", "```rust\nlet x = 1;\n```\n");
8344 d.caret = 10; // inside the code body
8345 assert_eq!(d.code_language_at_caret().as_deref(), Some("rust"));
8346 assert!(d.caret_in_fenced_code());
8347
8348 d.set_code_language("python");
8349 assert!(
8350 d.source.starts_with("```python\n"),
8351 "source: {:?}",
8352 d.source
8353 );
8354 assert_eq!(d.code_language_at_caret().as_deref(), Some("python"));
8355
8356 // Clearing it leaves a bare fence and no label.
8357 d.set_code_language("");
8358 assert!(d.source.starts_with("```\n"), "source: {:?}", d.source);
8359 assert_eq!(d.code_language_at_caret(), None);
8360
8361 // A caret outside any code block edits nothing.
8362 let mut p = doc_with("code_lang_none", "just prose\n");
8363 assert!(!p.caret_in_fenced_code());
8364 p.set_code_language("rust");
8365 assert_eq!(p.source, "just prose\n");
8366 }
8367
8368 #[test]
8369 fn a_language_the_fence_cannot_carry_is_refused_not_written() {
8370 // Markdown's info string ends at whitespace, so `two words` would write
8371 // a fence that reads back with a different language than the one asked
8372 // for. twig refuses it; leaf reports that and leaves the source alone.
8373 // The old splice trimmed the ends and wrote whatever was left.
8374 let mut d = doc_with("code_lang_bad", "```rust\nx\n```\n");
8375 d.caret = 10;
8376 d.set_code_language("two words");
8377 assert_eq!(d.source, "```rust\nx\n```\n", "source should be untouched");
8378 assert!(d.status.is_some(), "the refusal should be reported");
8379 assert_eq!(d.code_language_at_caret().as_deref(), Some("rust"));
8380 }
8381
8382 #[test]
8383 fn link_destination_at_caret_reads_both_spellings() {
8384 let mut d = doc_with("link_dest", "see [t](https://x.dev) ok\n");
8385 d.caret = 5;
8386 assert_eq!(
8387 d.link_destination_at_caret().as_deref(),
8388 Some("https://x.dev")
8389 );
8390 d.caret = 0;
8391 assert_eq!(d.link_destination_at_caret(), None);
8392
8393 // An autolink has no `destination`; its text is the URL.
8394 let mut a = doc_with("link_dest_auto", "see <https://x.dev> ok\n");
8395 a.caret = 10;
8396 assert_eq!(
8397 a.link_destination_at_caret().as_deref(),
8398 Some("https://x.dev")
8399 );
8400 a.caret = 21;
8401 assert_eq!(a.link_destination_at_caret(), None);
8402 }
8403
8404 #[test]
8405 fn locate_finds_the_block_a_declared_id_names() {
8406 // The Book of Mormon shape: one document per chapter, one `{#v…}` per
8407 // verse. The locator has to land on the *verse*, which is the whole
8408 // reason a link carries one.
8409 let src = "{#v1}\nI, Nephi, having been born of goodly parents.\n\n\
8410 {#v2}\nYea, I make a record in the language of my father.\n";
8411 let mut d = Doc::from_source(src.to_string(), Format::Djot).unwrap();
8412 let v2 = d.locate("v2").expect("the document declares `{#v2}`");
8413 assert_eq!(
8414 d.source[v2.start..v2.end].trim_end(),
8415 "Yea, I make a record in the language of my father."
8416 );
8417 // The attribute line is not part of it: `start` is a place to put a
8418 // caret, and `{#v2}` is markup the caret has no business landing in.
8419 assert!(d.source[..v2.start].ends_with("{#v2}\n"));
8420 assert_eq!(d.locate("v99"), None);
8421 }
8422
8423 #[test]
8424 fn locate_reads_a_heading_by_its_words_when_the_format_mints_no_ids() {
8425 // Markdown has no ids at all — twig mints none, and `{#custom}` in a
8426 // Markdown heading is literal text. So `#the-second-part` can only be
8427 // the heading's own words, which is the rule every Markdown renderer
8428 // already follows and therefore the one a link was authored against.
8429 let src = "# Title\n\nintro\n\n## The Second Part\n\nbody\n\n## Third\n\nmore\n";
8430 let mut d = doc_with("locate_md", src);
8431 let hit = d.locate("the-second-part").expect("the heading's slug");
8432 assert!(d.source[hit.start..].starts_with("## The Second Part"));
8433 // Bounded by the next heading that isn't under it, so a peek shows the
8434 // section rather than only its title.
8435 assert_eq!(
8436 &d.source[hit.start..hit.end],
8437 "## The Second Part\n\nbody\n\n"
8438 );
8439
8440 // A subsection does not end its parent: `# Title` runs to `## Third`'s
8441 // sibling only because there is no other `#`, so it covers the lot.
8442 let title = d.locate("title").expect("the top heading");
8443 assert_eq!(title.end, d.source.len());
8444 }
8445
8446 #[test]
8447 fn locate_reads_a_djot_auto_id_however_the_link_spelled_it() {
8448 // djot mints `Some-Heading-Here`; a link to it is written
8449 // `#some-heading-here` by nearly everything that writes links. Both
8450 // spellings are one question.
8451 let src = "## Some Heading Here\n\nbody\n";
8452 let mut d = Doc::from_source(src.to_string(), Format::Djot).unwrap();
8453 let exact = d.locate("Some-Heading-Here").expect("djot's own spelling");
8454 let slugged = d.locate("some-heading-here").expect("the link's spelling");
8455 assert_eq!(exact, slugged);
8456 // The section, not the heading line — there is more to show than a title.
8457 assert_eq!(&d.source[exact.start..exact.end], src);
8458 }
8459
8460 #[test]
8461 fn locate_ignores_an_empty_locator_and_one_that_slugs_to_nothing() {
8462 let mut d = doc_with("locate_empty", "# Title\n\nbody\n");
8463 assert_eq!(d.locate(""), None);
8464 assert_eq!(d.locate(" "), None);
8465 // All punctuation: it names nothing, and must not be read as "match the
8466 // first heading whose slug is also empty".
8467 assert_eq!(d.locate("!!!"), None);
8468 }
8469
8470 #[test]
8471 fn locate_gives_a_duplicated_id_to_the_first_block_that_claims_it() {
8472 // The document's mistake, and the answer every other anchor
8473 // implementation gives — the alternative is for a link to mean whichever
8474 // of the two a walk happened to reach first.
8475 let src = "{#dup}\nfirst.\n\n{#dup}\nsecond.\n";
8476 let mut d = Doc::from_source(src.to_string(), Format::Djot).unwrap();
8477 let hit = d.locate("dup").expect("the first `{#dup}`");
8478 assert_eq!(d.source[hit.start..hit.end].trim_end(), "first.");
8479 }
8480
8481 #[test]
8482 fn insert_footnote_writes_both_halves_and_lands_the_caret_in_the_note() {
8483 // The button's whole job: a reference where the caret was, a definition
8484 // to give it meaning, and the caret waiting in the empty note so the
8485 // next keystroke is the note's first word.
8486 let mut d = doc_with("fn_insert", "A claim and more.\n");
8487 d.caret = 7; // just past "A claim"
8488 d.insert_footnote();
8489 assert!(
8490 d.source.starts_with("A claim[^1] and more."),
8491 "{:?}",
8492 d.source
8493 );
8494 assert!(
8495 d.source.contains("[^1]:"),
8496 "the definition too: {:?}",
8497 d.source
8498 );
8499 assert_eq!(d.status, None);
8500
8501 let reference = d.source.find("[^1]").unwrap();
8502 let note = d
8503 .footnote_at(reference + 2)
8504 .expect("the reference just written");
8505 assert_eq!(note.label, "1");
8506 assert_eq!(note.text.as_deref(), Some(""), "the note starts empty");
8507 assert_eq!(Some(d.caret), note.offset, "the caret waits in the note");
8508 // …and typing there is typing into the note, not near it.
8509 d.insert("the note");
8510 assert_eq!(
8511 d.footnote_at(reference + 2).and_then(|f| f.text),
8512 Some("the note".to_string())
8513 );
8514 }
8515
8516 #[test]
8517 fn insert_footnote_numbers_past_the_notes_already_written() {
8518 // A second press must not hand back a label somebody else is using: twig
8519 // reuses a defined label rather than appending a rival definition, so a
8520 // repeat of `1` would quietly point the new reference at the old note.
8521 let mut d = doc_with("fn_insert_number", "One[^1] two.\n\n[^1]: first\n");
8522 d.caret = 7; // past `[^1]`, before " two."
8523 d.insert_footnote();
8524 assert!(d.source.starts_with("One[^1][^2] two."), "{:?}", d.source);
8525 assert_eq!(d.source.matches("[^2]:").count(), 1);
8526 }
8527
8528 #[test]
8529 fn insert_footnote_counts_a_dangling_reference_and_ignores_a_named_one() {
8530 // `[^2]` with no definition is still a 2 that means something to whoever
8531 // wrote it — stepping over it would mint a note for their reference. A
8532 // word label takes no number, so it blocks none.
8533 let mut d = doc_with("fn_insert_dangling", "a[^2] b[^why] c\n\n[^why]: named\n");
8534 d.caret = d.source.find(" c").unwrap();
8535 d.insert_footnote();
8536 assert!(d.source.contains("[^1]:"), "1 is free: {:?}", d.source);
8537 assert!(
8538 d.source.starts_with("a[^2] b[^why][^1] c"),
8539 "{:?}",
8540 d.source
8541 );
8542 }
8543
8544 #[test]
8545 fn insert_footnote_marks_the_selection_rather_than_replacing_it() {
8546 // A reference annotates the words before it. Consuming the selection —
8547 // which is what an insert normally does — would delete the very claim
8548 // the author selected in order to footnote.
8549 let mut d = doc_with("fn_insert_sel", "A claim and more.\n");
8550 d.anchor = Some(2);
8551 d.caret = 7; // "claim" selected
8552 d.insert_footnote();
8553 assert!(
8554 d.source.starts_with("A claim[^1] and more."),
8555 "{:?}",
8556 d.source
8557 );
8558 }
8559
8560 #[test]
8561 fn a_note_just_written_still_knows_where_its_reference_is() {
8562 // The authoring loop in one test: press the button, type the note, ask to
8563 // go back. The caret ends at the note's last byte — which is the *end* of
8564 // the definition's span, the one offset the query used to exclude — so
8565 // this is where the round trip either works or doesn't.
8566 let mut d = doc_with("fn_insert_return", "A claim and more.\n");
8567 d.caret = 7;
8568 d.insert_footnote();
8569 d.insert("the note");
8570 assert_eq!(d.source, "A claim[^1] and more.\n\n[^1]: the note\n");
8571 let back = d
8572 .footnote_definition_at_caret()
8573 .expect("still in the note we just typed");
8574 assert_eq!(back.label, "1");
8575 // …and following it lands on the reference's label, where a reader's
8576 // return leg lands.
8577 assert_eq!(back.offset, Some(9));
8578 assert_eq!(&d.source[9..10], "1");
8579 }
8580
8581 #[test]
8582 fn insert_footnote_takes_one_undo_for_both_halves() {
8583 // twig writes the pair as a single edit; the point of that is here.
8584 let before = "A claim and more.\n";
8585 let mut d = doc_with("fn_insert_undo", before);
8586 d.caret = 7;
8587 d.insert_footnote();
8588 assert_ne!(d.source, before);
8589 d.undo();
8590 assert_eq!(d.source, before, "one undo takes back both halves");
8591 }
8592
8593 #[test]
8594 fn insert_footnote_refuses_a_format_that_cannot_spell_one() {
8595 // HTML is authorable — it spells the inline marks — and has no footnote.
8596 // The refusal says so rather than writing brackets that would render as
8597 // brackets.
8598 let src = "<p>A claim.</p>\n";
8599 let mut d = Doc::from_source(src.to_string(), Format::Html).unwrap();
8600 assert!(!Capabilities::of(Format::Html).footnote);
8601 d.caret = 5;
8602 d.insert_footnote();
8603 assert_eq!(d.source, src, "nothing written");
8604 assert!(d.status.is_some_and(|s| s.starts_with("footnote:")));
8605 }
8606
8607 #[test]
8608 fn insert_footnote_leaves_the_caret_on_a_real_stop_in_the_rich_view() {
8609 // The empty body is the one place this could go wrong: the definition
8610 // renders as a `[1] ` marker the caret cannot occupy, so a caret aimed a
8611 // byte early would draw up in the paragraph above the note it belongs to.
8612 let mut d = doc_in(View::Wysiwyg, "fn_insert_stop", "A claim and more.\n");
8613 d.place_caret(7, false);
8614 d.insert_footnote();
8615 d.build_visual(80); // the frame a frontend draws after the edit
8616 assert_eq!(
8617 d.vmap.snap_to_stop(d.caret),
8618 d.caret,
8619 "the caret sits on a stop"
8620 );
8621 let (row, _) = d.caret_pos();
8622 assert!(
8623 drawn_rows(&d)[row].contains("[1]"),
8624 "the caret is on the note's row, not above it: {:?}",
8625 drawn_rows(&d)
8626 );
8627 }
8628
8629 #[test]
8630 fn footnote_at_caret_resolves_a_reference_to_its_note() {
8631 // `[^1]` spans 7..11; its label byte is at 9. The definition follows a
8632 // blank line, as one has to.
8633 let mut d = doc_with("fn_at_caret", "A claim[^1] and more.\n\n[^1]: the note\n");
8634 d.caret = 9;
8635 let f = d
8636 .footnote_at_caret()
8637 .expect("the caret stands in a reference");
8638 assert_eq!(f.label, "1");
8639 assert_eq!(f.text.as_deref(), Some("the note"));
8640 // The offset points at the note's first word, not at the definition's
8641 // `[` — the marker is decoration with no caret stop on it.
8642 assert_eq!(f.offset, Some(29));
8643 assert_eq!(&d.source[29..37], "the note");
8644 // …and `end` closes the range, so a frontend can ask which rendered rows
8645 // the note occupies rather than re-deriving them from the text.
8646 assert_eq!(f.end, Some(37));
8647 assert_eq!(&d.source[f.offset.unwrap()..f.end.unwrap()], "the note");
8648 }
8649
8650 /// Two definitions in a row: each is its own note, and neither reaches into
8651 /// the other.
8652 ///
8653 /// A djot definition's span used to run past the blank line into the first
8654 /// byte of whatever followed, so this answered `"first note.\n\n["` — and the
8655 /// offsets named the *next* note's rows too, showing a reader two footnotes
8656 /// when they had asked about one. twig 3.1 ends the span after the block's
8657 /// own last line; the test outlives the workaround leaf carried for it.
8658 #[test]
8659 fn footnote_at_stops_a_note_at_the_definition_after_it() {
8660 let src = "Claim[^2a] and [^2b].\n\n[^2a]: first note.\n\n[^2b]: second note.\n";
8661 for format in [Format::Markdown, Format::Djot] {
8662 let mut d = Doc::from_source(src.to_string(), format).unwrap();
8663 d.caret = 7;
8664 let f = d.footnote_at_caret().expect("a reference");
8665 assert_eq!(f.text.as_deref(), Some("first note."), "in {format:?}");
8666 assert_eq!(
8667 &src[f.offset.unwrap()..f.end.unwrap()],
8668 "first note.",
8669 "in {format:?}"
8670 );
8671 }
8672 }
8673
8674 /// The other side of that boundary: a blank line *inside* a definition is
8675 /// interior to it, and the note keeps its second paragraph.
8676 ///
8677 /// This is what the old body scan cost. It stopped at the first line not
8678 /// indented under the note — a blank line is not — so a two-paragraph note
8679 /// came back as its first paragraph, and "go to note" framed half of it.
8680 /// Reading the span twig gives is both simpler and right.
8681 #[test]
8682 fn footnote_at_keeps_a_notes_second_paragraph() {
8683 let src = "Claim[^1].\n\n[^1]: first para.\n\n second para.\n\nAfter.\n";
8684 let mut d = Doc::from_source(src.to_string(), Format::Djot).unwrap();
8685 d.caret = 7;
8686 let f = d.footnote_at_caret().expect("a reference");
8687 assert_eq!(f.text.as_deref(), Some("first para.\n\n second para."));
8688 // And it stops there — `After.` is the next block, not more note.
8689 assert_eq!(
8690 &src[f.offset.unwrap()..f.end.unwrap()],
8691 f.text.as_deref().unwrap()
8692 );
8693 assert!(!f.text.as_deref().unwrap().contains("After"));
8694 }
8695
8696 #[test]
8697 fn footnote_at_bounds_a_note_whose_body_is_empty() {
8698 // `[^1]:` with nothing after it. The range is empty rather than
8699 // inverted, and still points inside the definition — which is what keeps
8700 // a frontend's row lookup from walking off into the block above.
8701 let src = "A claim[^1].\n\n[^1]:\n";
8702 let mut d = doc_with("fn_empty_body", src);
8703 d.caret = 9;
8704 let f = d.footnote_at_caret().expect("a reference");
8705 assert_eq!(f.text.as_deref(), Some(""));
8706 assert_eq!(f.offset, f.end, "an empty note is an empty range");
8707 assert!(f.offset.unwrap() >= src.find("[^1]:").unwrap());
8708 }
8709
8710 #[test]
8711 fn footnote_at_caret_ignores_a_caret_that_stands_in_no_reference() {
8712 let mut d = doc_with(
8713 "fn_at_caret_none",
8714 "A claim[^1] and more.\n\n[^1]: the note\n",
8715 );
8716 d.caret = 2; // in the prose
8717 assert_eq!(d.footnote_at_caret(), None);
8718 }
8719
8720 #[test]
8721 fn footnote_at_caret_is_not_a_link_query_and_vice_versa() {
8722 // The two are deliberately separate: a reference names a note in this
8723 // document, a link names somewhere to leave for, and answering one with
8724 // the other is what made a reference click do nothing at all.
8725 let mut d = doc_with("fn_vs_link", "a[^1] b [t](https://x.dev)\n\n[^1]: note\n");
8726 d.caret = 3; // the `1` of `[^1]`
8727 assert!(d.footnote_at_caret().is_some());
8728 assert_eq!(
8729 d.link_destination_at_caret(),
8730 None,
8731 "a reference is not a link"
8732 );
8733
8734 d.caret = 10; // inside the link's label
8735 assert_eq!(d.footnote_at_caret(), None, "a link is not a reference");
8736 assert_eq!(
8737 d.link_destination_at_caret().as_deref(),
8738 Some("https://x.dev")
8739 );
8740 }
8741
8742 #[test]
8743 fn footnote_at_caret_reports_an_undefined_reference_rather_than_nothing() {
8744 // A `[^99]` the document never defines is a real state — a note deleted
8745 // out from under its reference — and the label is what lets a frontend
8746 // say so. `None` here would be indistinguishable from "not on a
8747 // reference", which is the wrong thing to tell a reader.
8748 let mut d = doc_with("fn_undefined", "A claim[^99] and more.\n");
8749 d.caret = 9;
8750 let f = d
8751 .footnote_at_caret()
8752 .expect("the reference is still a reference");
8753 assert_eq!(f.label, "99");
8754 assert_eq!(f.text, None);
8755 assert_eq!(f.offset, None);
8756 }
8757
8758 #[test]
8759 fn footnote_at_caret_reads_a_word_label_and_a_multiline_note() {
8760 // Labels are not always numbers, and a note's body runs past its first
8761 // line — the indented continuation belongs to the note, so it comes back
8762 // with it (source bytes, verbatim, as documented).
8763 let src = "see[^note] here\n\n[^note]: first line\n second line\n";
8764 let mut d = doc_with("fn_word_label", src);
8765 d.caret = 6;
8766 let f = d
8767 .footnote_at_caret()
8768 .expect("the caret stands in a reference");
8769 assert_eq!(f.label, "note");
8770 assert_eq!(f.text.as_deref(), Some("first line\n second line"));
8771 }
8772
8773 #[test]
8774 fn footnote_at_answers_for_an_offset_the_caret_is_nowhere_near() {
8775 // The point of the offset form: a pointer hovering a reference asks what
8776 // note it names, and must not drag the caret along to ask.
8777 let mut d = doc_with("fn_at_off", "A claim[^1] and more.\n\n[^1]: the note\n");
8778 d.caret = 0;
8779 let f = d.footnote_at(9).expect("offset 9 stands in the reference");
8780 assert_eq!(f.label, "1");
8781 assert_eq!(f.text.as_deref(), Some("the note"));
8782 assert_eq!(d.caret, 0, "asking must not move the caret");
8783 assert_eq!(d.footnote_at(2), None, "offset 2 is prose");
8784 }
8785
8786 #[test]
8787 fn footnote_definition_at_caret_points_back_at_the_reference() {
8788 // The return leg. `[^1]` spans 7..11, so its label — the only byte of it
8789 // the caret can rest on — is at 9.
8790 let mut d = doc_with("fn_def", "A claim[^1] and more.\n\n[^1]: the note\n");
8791 d.caret = 30; // inside the note's body
8792 let f = d
8793 .footnote_definition_at_caret()
8794 .expect("the caret stands in a definition");
8795 assert_eq!(f.label, "1");
8796 assert_eq!(f.offset, Some(9));
8797 assert_eq!(&d.source[7..11], "[^1]");
8798 }
8799
8800 #[test]
8801 fn footnote_definition_at_covers_where_a_go_to_note_actually_lands() {
8802 // The two legs have to meet: wherever `footnote_at` sends the caret, the
8803 // definition query must answer for — otherwise arriving at a note leaves
8804 // the reader somewhere the way back isn't offered.
8805 let src = "A claim[^1] and more.\n\n[^1]: the note\n";
8806 let mut d = doc_with("fn_def_marker", src);
8807 let landed = d.footnote_at(9).unwrap().offset.unwrap();
8808 assert_eq!(
8809 d.footnote_definition_at(landed).and_then(|f| f.offset),
8810 Some(9),
8811 "the note a reference sends you to offers the way back"
8812 );
8813 }
8814
8815 #[test]
8816 fn footnote_definition_at_caret_ignores_prose_and_the_reference_itself() {
8817 // The two queries answer for disjoint places, which is what lets one
8818 // gesture mean "down to the note" in one and "back up" in the other
8819 // without either having to remember which way the reader is going.
8820 let mut d = doc_with("fn_def_none", "A claim[^1] and more.\n\n[^1]: the note\n");
8821 d.caret = 2; // prose
8822 assert_eq!(d.footnote_definition_at_caret(), None);
8823 d.caret = 9; // the reference
8824 assert_eq!(d.footnote_definition_at_caret(), None);
8825 assert!(
8826 d.footnote_at_caret().is_some(),
8827 "which is the reference's own query"
8828 );
8829 }
8830
8831 #[test]
8832 fn footnote_definition_at_caret_reports_an_orphan_note_rather_than_nothing() {
8833 // Nothing cites `[^2]`. Answering `None` would say "you are not in a
8834 // note", which is false and leaves a frontend unable to explain why the
8835 // way back is missing.
8836 let src = "A claim[^1].\n\n[^1]: cited\n\n[^2]: orphan\n";
8837 let mut d = doc_with("fn_def_orphan", src);
8838 d.caret = src.find("orphan").unwrap();
8839 let f = d
8840 .footnote_definition_at_caret()
8841 .expect("an orphan is still a definition");
8842 assert_eq!(f.label, "2");
8843 assert_eq!(f.offset, None);
8844 }
8845
8846 #[test]
8847 fn footnote_definition_at_caret_returns_to_the_first_of_repeated_references() {
8848 // One label, cited twice. The first is where the reader most likely came
8849 // from, and the only answer that doesn't depend on how they got here.
8850 let src = "One[^a] and two[^a].\n\n[^a]: the note\n";
8851 let mut d = doc_with("fn_def_repeat", src);
8852 d.caret = src.find("the note").unwrap();
8853 let f = d.footnote_definition_at_caret().expect("a definition");
8854 assert_eq!(
8855 f.offset,
8856 Some(5),
8857 "the first `[^a]`'s label, not the second's"
8858 );
8859 assert_eq!(&src[3..7], "[^a]");
8860 }
8861
8862 #[test]
8863 fn footnote_navigation_is_a_round_trip_through_placed_carets() {
8864 // Down and back up, each leg found from the document rather than from a
8865 // memory of the other — so it still works for a reader who scrolled to
8866 // the notes instead of jumping there.
8867 //
8868 // `place_caret` rather than assigning `caret`, because that is what a
8869 // frontend calls: it snaps to a real caret stop, and a jump that lands
8870 // on a byte the caret can't rest on would arrive somewhere the return
8871 // leg no longer answers for. `build_map` first, since snapping is a
8872 // no-op until the map exists — which is exactly how this went unnoticed
8873 // when the offsets pointed at the `[^` markers.
8874 let mut d = doc_with("fn_round", "A claim[^1] and more.\n\n[^1]: the note\n");
8875 d.build_map(None);
8876 d.place_caret(9, false);
8877 let down = d
8878 .footnote_at_caret()
8879 .expect("a reference")
8880 .offset
8881 .expect("a note");
8882 d.place_caret(down, false);
8883 let up = d
8884 .footnote_definition_at_caret()
8885 .expect("a definition")
8886 .offset
8887 .expect("a reference");
8888 d.place_caret(up, false);
8889 assert_eq!(d.caret, up, "the way back is a stop the caret can occupy");
8890 assert_eq!(
8891 d.footnote_at_caret().expect("back on the reference").label,
8892 "1"
8893 );
8894 }
8895
8896 #[test]
8897 fn insert_link_hands_the_destination_to_twig_raw() {
8898 // Escaping is twig's, and format-specific: Markdown ends a destination
8899 // at the first space and needs the `<…>` form, where djot would read
8900 // those angle brackets as part of the URL.
8901 let mut d = doc_with("link_space", "word\n");
8902 d.anchor = Some(0);
8903 d.caret = 4;
8904 d.insert_link("a b");
8905 assert_eq!(d.source, "[word](<a b>)\n");
8906 }
8907
8908 #[test]
8909 fn insert_link_reports_a_destination_no_format_can_carry() {
8910 let mut d = doc_with("link_bad", "word\n");
8911 d.anchor = Some(0);
8912 d.caret = 4;
8913 d.insert_link("a\nb");
8914 assert_eq!(d.source, "word\n"); // untouched, not quietly rewritten
8915 assert!(
8916 d.status.is_some(),
8917 "InvalidArgument should reach the status line"
8918 );
8919 assert!(!d.dirty);
8920 }
8921
8922 #[test]
8923 fn insert_link_works_in_wysiwyg_view() {
8924 let mut d = wysiwyg_doc("link_wys", "word here\n");
8925 d.anchor = Some(0);
8926 d.caret = 4;
8927 d.insert_link("http://x.dev");
8928 assert_eq!(d.source, "[word](http://x.dev) here\n");
8929 assert_eq!(d.selected_text(), Some("word"));
8930 // The map the caret has to keep riding is rebuilt each frame; motion
8931 // over the fresh one must still land on a real stop (the debug_assert).
8932 d.build_visual(80);
8933 d.move_right(false);
8934 d.move_left(false);
8935 }
8936
8937 #[test]
8938 fn click_maps_a_row_col_to_a_byte_offset() {
8939 let mut d = doc_with("click", "ab\ncd\n");
8940 d.click(1, 1, false); // row 1 ("cd"), col 1 -> the 'd'
8941 assert_eq!(d.caret, 4);
8942 }
8943
8944 // A pixel-hit-test placement (the GUI's `place_caret`) must land on a caret
8945 // stop just as the `(row, col)` click path does, so the caret can never come
8946 // to rest in the blank gap between two paragraphs — where it would draw in one
8947 // place and type in another.
8948 #[test]
8949 fn place_caret_snaps_out_of_the_blank_gap_between_paragraphs() {
8950 // "A\n\nB": offset 2 is the gap the paragraph break is drawn with, not a
8951 // caret stop (stops are 0,1,3,4).
8952 let mut d = wysiwyg_doc("place_gap", "A\n\nB");
8953 assert!(!d.vmap.is_stop(2), "offset 2 should be an unreachable gap");
8954 d.place_caret(2, false);
8955 assert!(d.vmap.is_stop(d.caret), "caret {} is not a stop", d.caret);
8956 assert_eq!(d.caret, 1, "should snap to the end of the paragraph above");
8957 }
8958
8959 #[test]
8960 fn place_caret_dragging_through_the_gap_keeps_selection_on_stops() {
8961 let mut d = wysiwyg_doc("place_gap_drag", "A\n\nB");
8962 d.place_caret(0, false); // anchor at the start of "A"
8963 d.place_caret(2, true); // drag into the gap
8964 assert!(d.vmap.is_stop(d.caret), "caret {} is not a stop", d.caret);
8965 let (s, e) = d.selection().expect("a selection");
8966 assert!(
8967 d.vmap.is_stop(s) && d.vmap.is_stop(e),
8968 "selection {s}..{e} off a stop"
8969 );
8970 }
8971
8972 #[test]
8973 fn place_caret_on_a_real_stop_is_left_untouched() {
8974 let mut d = wysiwyg_doc("place_stop", "A\n\nB");
8975 d.place_caret(3, false); // the start of "B" — a genuine stop
8976 assert_eq!(d.caret, 3);
8977 }
8978
8979 // An *empty paragraph* (two blank lines, an intentional blank line the user
8980 // opened) is a real caret stop, unlike the gap — a click into it must stay.
8981 #[test]
8982 fn place_caret_rests_in_an_empty_paragraph() {
8983 let mut d = wysiwyg_doc("place_empty_para", "A\n\n\n\nB");
8984 let empty = 3; // the navigable empty row's offset (stops: 0,1,3,5,6)
8985 assert!(d.vmap.is_stop(empty));
8986 d.place_caret(empty, false);
8987 assert_eq!(d.caret, empty);
8988 }
8989
8990 fn wysiwyg_doc(name: &str, body: &str) -> Doc {
8991 doc_in(View::Wysiwyg, name, body)
8992 }
8993
8994 /// How many list items the source actually parses into — the check that a
8995 /// marker Leaf wrote is a marker the format agrees is one.
8996 fn list_items(doc: &mut Doc) -> usize {
8997 doc.editor
8998 .nodes()
8999 .unwrap()
9000 .iter()
9001 .filter(|n| n.kind == Kind::ListItem || n.kind == Kind::TaskListItem)
9002 .count()
9003 }
9004
9005 /// A from-scratch, cache-free WYSIWYG map for `source` — the ground truth the
9006 /// incremental (`build_spliced` / `build_cached`) path must always match.
9007 fn reference_map(source: &str) -> crate::wysiwyg::VisualMap {
9008 reference_map_revealing(source, None)
9009 }
9010
9011 /// [`reference_map`] with a reveal line — the ground truth for the
9012 /// `MarkupMode::Full` builds, where the map is a function of the caret's
9013 /// line as well as the text.
9014 fn reference_map_revealing(
9015 source: &str,
9016 reveal: Option<Range<usize>>,
9017 ) -> crate::wysiwyg::VisualMap {
9018 // The same parse `Doc` uses. With twig's plain defaults instead, the two
9019 // sides disagree on what the *document* is before the renderer is even
9020 // reached — a bare `:word` is a text directive to one and prose to the
9021 // other — and the mismatch reads as a splice bug that isn't one.
9022 let mut ed =
9023 twig::Editor::new_ext(source.as_bytes(), Format::Markdown, parse_extensions()).unwrap();
9024 let nodes = ed.nodes().unwrap();
9025 crate::wysiwyg::build(
9026 &nodes,
9027 source,
9028 None,
9029 false,
9030 &std::collections::HashMap::new(),
9031 reveal,
9032 )
9033 }
9034
9035 fn maps_differ(a: &crate::wysiwyg::VisualMap, b: &crate::wysiwyg::VisualMap) -> bool {
9036 if a.rows.len() != b.rows.len() {
9037 return true;
9038 }
9039 for (ra, rb) in a.rows.iter().zip(&b.rows) {
9040 if ra.end_src != rb.end_src || ra.glyphs.len() != rb.glyphs.len() {
9041 return true;
9042 }
9043 for (ga, gb) in ra.glyphs.iter().zip(&rb.glyphs) {
9044 if ga.ch != gb.ch || ga.src != gb.src {
9045 return true;
9046 }
9047 }
9048 }
9049 false
9050 }
9051
9052 #[test]
9053 fn incremental_build_matches_a_fresh_build_across_edits() {
9054 // Every `Doc` edit rebuilds through `build_spliced` (the single-block
9055 // fast path, gated on twig's `dirty_range`) or falls back to
9056 // `build_cached`. After each edit the map must be byte-identical to a
9057 // from-scratch build — this is the correctness net under the splice.
9058 let docs = [
9059 "# Title\n\nThe quick brown fox jumps.\n\nAnother paragraph here.\n\n- a\n- b\n",
9060 "para one\n\n> quote **bold** text\n> continued line\n\ntail paragraph\n",
9061 "alpha\n\nbeta\n\ngamma\n\ndelta\n\nepsilon\n\nzeta\n",
9062 // A footnote definition is a root beside `doc`, merged back into the
9063 // top-level list by `wysiwyg::top_blocks`. The random edits below
9064 // make and unmake definitions as they go (a deleted `:` turns one
9065 // back into a paragraph, and vice versa), which is exactly the
9066 // structural churn the splice path has to notice and bail out of.
9067 "text[^1] here\n\n[^1]: the note\n\nmore text[^b]\n\n[^b]: second\n",
9068 ];
9069 // A deterministic mix: mostly single characters (which stay inside one
9070 // block → splice), plus edits that reshape structure (a paragraph break,
9071 // a heading marker, a code fence → fallback), so both paths are exercised.
9072 let inserts = ["x", "y", "\n\n", "#", "`", " ", "z"];
9073 for src in docs {
9074 let mut d = wysiwyg_doc("diff", src);
9075 d.build_visual_unwrapped();
9076 wysiwyg::assert_maps_eq(&d.vmap, &reference_map(&d.source), "initial");
9077
9078 for step in 0..60usize {
9079 let len = d.source.len();
9080 let raw = (step * 13 + 5) % (len + 1);
9081 let pos = (raw..=len).find(|&i| d.source.is_char_boundary(i)).unwrap();
9082 let pre = d.source.clone();
9083 let action;
9084 if step % 3 == 0 && pos < len {
9085 let end = (pos + 1..=len)
9086 .find(|&i| d.source.is_char_boundary(i))
9087 .unwrap();
9088 action = format!("delete [{pos},{end})");
9089 d.edit(pos, end, "");
9090 } else {
9091 let ins = inserts[step % inserts.len()];
9092 action = format!("insert {ins:?} @ {pos}");
9093 d.edit(pos, pos, ins);
9094 }
9095 d.build_visual_unwrapped();
9096 if maps_differ(&d.vmap, &reference_map(&d.source)) {
9097 panic!(
9098 "FIRST MISMATCH at step {step}: {action}\n pre = {pre:?}\n post = {:?}",
9099 d.source
9100 );
9101 }
9102 }
9103 }
9104 }
9105
9106 #[test]
9107 fn incremental_build_matches_a_fresh_build_under_full_reveal() {
9108 // The same correctness net as `incremental_build_matches_a_fresh_build_
9109 // across_edits`, under `MarkupMode::Full` — where the map depends on
9110 // the caret's *line* as well as the text, so the two caches have a new
9111 // way to be wrong. Both are exercised: the block cache can hand back
9112 // rows built for a line that is no longer the revealed one, and the
9113 // splice path can reuse a suffix that still has yesterday's line raw.
9114 //
9115 // Caret motion is interleaved with the edits deliberately, because a
9116 // caret that only ever moved with the edit would never cross a line
9117 // without also dirtying it — the case where a stale reveal survives.
9118 let docs = [
9119 "# Title\n\n*one* and **two**\n\n[lk](http://x) and `code`\n\n- a *b*\n",
9120 "para *em* one\n\n> quote **bold** text\n\ntail ~~del~~ paragraph\n",
9121 ];
9122 let inserts = ["x", "*", "\n\n", "#", "`", " ", "_"];
9123 for src in docs {
9124 let mut d = wysiwyg_doc("reveal_diff", src);
9125 d.set_markup_mode(MarkupMode::Full);
9126
9127 for step in 0..60usize {
9128 let len = d.source.len();
9129 let raw = (step * 13 + 5) % (len + 1);
9130 let pos = (raw..=len).find(|&i| d.source.is_char_boundary(i)).unwrap();
9131 let pre = d.source.clone();
9132 let action;
9133 if step % 3 == 0 && pos < len {
9134 let end = (pos + 1..=len)
9135 .find(|&i| d.source.is_char_boundary(i))
9136 .unwrap();
9137 action = format!("delete [{pos},{end})");
9138 d.edit(pos, end, "");
9139 } else {
9140 let ins = inserts[step % inserts.len()];
9141 action = format!("insert {ins:?} @ {pos}");
9142 d.edit(pos, pos, ins);
9143 }
9144 // Walk the caret somewhere else in the document, independently
9145 // of where the edit landed.
9146 let want = (step * 29 + 11) % (d.source.len() + 1);
9147 d.caret = (want..=d.source.len())
9148 .find(|&i| d.source.is_char_boundary(i))
9149 .unwrap();
9150 d.build_visual_unwrapped();
9151
9152 let want = reference_map_revealing(&d.source, d.reveal_line());
9153 if maps_differ(&d.vmap, &want) {
9154 panic!(
9155 "FIRST MISMATCH at step {step}: {action}, caret {}\n pre = {pre:?}\n post = {:?}",
9156 d.caret, d.source
9157 );
9158 }
9159 }
9160 }
9161 }
9162
9163 #[test]
9164 fn caret_motion_across_lines_rebuilds_only_under_full() {
9165 // The cache-key change has to earn its keep in both directions: `Full`
9166 // must rebuild when the caret changes line (or the reveal would never
9167 // move), and the hidden modes must *not* (or every arrow key would pay
9168 // for a feature they don't use). The existing `cache_motion` test pins
9169 // the second for the default mode; this pins the pair against a mode
9170 // change alone.
9171 let body = "*one* here\n\n*two* there\n";
9172
9173 let mut full = doc_in(View::Wysiwyg, "motion_full", body);
9174 full.set_markup_mode(MarkupMode::Full);
9175 caret_at(&mut full, "one");
9176 let before = full.revision();
9177 caret_at(&mut full, "two");
9178 assert_eq!(full.revision(), before, "motion is not an edit");
9179 assert!(
9180 drawn_rows(&full).iter().any(|r| r == "*two* there"),
9181 "the map followed the caret: {:?}",
9182 drawn_rows(&full)
9183 );
9184
9185 let mut hidden = doc_in(View::Wysiwyg, "motion_hidden", body);
9186 caret_at(&mut hidden, "one");
9187 let key = hidden.vmap_key.clone();
9188 caret_at(&mut hidden, "two");
9189 assert_eq!(
9190 hidden.vmap_key, key,
9191 "a hidden mode rebuilds nothing on motion"
9192 );
9193 }
9194
9195 #[test]
9196 fn wysiwyg_down_crosses_a_paragraph_boundary() {
9197 // Regression: the blank separator row used to share the previous
9198 // paragraph's end offset, so Down got pinned at the boundary (while Up
9199 // still crossed). Both directions must step through it symmetrically.
9200 //
9201 // It's now stepped *over* rather than onto: the blank line between two
9202 // paragraphs is the boundary being drawn, not a line of the document, so
9203 // one press of Down crosses it. The goal column survives the crossing —
9204 // col 3 at the end of "abc" is col 3 at the end of "def".
9205 let mut d = wysiwyg_doc("wys_down", "abc\n\ndef\n");
9206 d.caret = 3; // end of "abc" (row 0)
9207 d.move_down(false);
9208 assert_eq!(d.caret_pos().0, 2, "Down should reach the second paragraph");
9209 assert_eq!(d.caret, 8); // end of "def", col 3 kept
9210 d.move_up(false);
9211 assert_eq!(d.caret_pos().0, 0, "Up should come back symmetrically");
9212 assert_eq!(d.caret, 3);
9213 }
9214
9215 #[test]
9216 fn wysiwyg_up_and_down_are_inverse_across_paragraphs() {
9217 // The second Up and the second Down here run off the ends of the
9218 // document, which is no longer a place a press is swallowed: they carry
9219 // the caret to the start and the end of the text. The claim in the
9220 // middle — that a Down retraces the Up that crossed the paragraph gap —
9221 // is the one this test is for, and it is asserted where it is made.
9222 let mut d = wysiwyg_doc("wys_updown", "abc\n\ndef\n");
9223 d.caret = 5; // start of "def"
9224 let start = d.caret_pos();
9225 d.move_up(false);
9226 assert_eq!(d.caret_pos().0, 0, "Up reaches the first paragraph");
9227 d.move_up(false);
9228 assert_eq!(d.caret, 0, "a second Up runs on to the document's start");
9229 d.move_down(false);
9230 assert_eq!(d.caret_pos(), start, "Down retraces Up exactly");
9231 d.move_down(false);
9232 assert_eq!(d.caret, 8, "a second Down runs on to the document's end");
9233 }
9234
9235 #[test]
9236 fn wysiwyg_new_paragraph_shows_before_typing() {
9237 // Regression: two Enters at the end of a paragraph produced trailing
9238 // newlines with no AST node, so the caret appeared stuck on the old line
9239 // until a character was typed. It must ride down onto the new line now.
9240 let mut d = doc_with("wys_newpara", "abc\n");
9241 d.view = View::Wysiwyg;
9242 d.caret = 3;
9243 d.insert("\n");
9244 d.insert("\n"); // source is now "abc\n\n\n", caret at 5
9245 assert_eq!(d.source, "abc\n\n\n");
9246 d.build_visual(80);
9247 let (row, _) = d.caret_pos();
9248 assert!(
9249 row >= 2,
9250 "caret should have moved down to the new line, got row {row}"
9251 );
9252 assert!(
9253 d.vmap.num_rows() >= 3,
9254 "the blank lines should render as rows"
9255 );
9256 }
9257
9258 #[test]
9259 fn wysiwyg_enter_between_paragraphs_lands_on_an_empty_line() {
9260 // The reported bug: Enter at the end of a paragraph that has another
9261 // paragraph below put the caret at the *start of the next paragraph* —
9262 // the empty paragraph it opened had no row, so the caret snapped onto
9263 // "World". It must now sit on its own empty line, with a blank spacer
9264 // above it (the paragraph gap).
9265 let mut d = wysiwyg_doc("wys_gap_mid", "Hello\n\nWorld\n");
9266 d.caret = 5; // end of "Hello"
9267 d.newline();
9268 d.build_visual(80);
9269 let (row, col) = d.caret_pos();
9270 assert_eq!(col, 0, "caret should start an empty line, not sit in text");
9271 assert_eq!(
9272 d.vmap.row_width(row),
9273 0,
9274 "caret's row must be empty, not 'World'"
9275 );
9276 assert!(
9277 row >= 2,
9278 "a blank spacer row should sit above the caret, got row {row}"
9279 );
9280 // The row above the caret is a real (empty) gap, and "Hello" stays put.
9281 assert_eq!(
9282 d.vmap.row_width(row - 1),
9283 0,
9284 "the row above the caret is a gap"
9285 );
9286 let row0: String = d.vmap.rows[0].glyphs.iter().map(|g| g.ch).collect();
9287 assert_eq!(row0, "Hello", "the paragraph above the caret must not move");
9288 }
9289
9290 #[test]
9291 fn wysiwyg_enter_at_eof_shows_a_gap_before_typing() {
9292 // At the document end a single Enter must also show the paragraph gap —
9293 // a blank spacer row above the caret — so the layout already matches how
9294 // it will look once the new paragraph has text.
9295 let mut d = wysiwyg_doc("wys_gap_eof", "Hello");
9296 d.caret = 5; // end of "Hello", no trailing newline
9297 d.newline(); // source becomes "Hello\n\n"
9298 d.build_visual(80);
9299 let (row, col) = d.caret_pos();
9300 assert_eq!(col, 0);
9301 assert!(
9302 row >= 2,
9303 "caret should sit below a blank spacer, got row {row}"
9304 );
9305 assert_eq!(
9306 d.vmap.row_width(row - 1),
9307 0,
9308 "the row above the caret is a gap"
9309 );
9310 }
9311
9312 #[test]
9313 fn wysiwyg_typing_after_enter_does_not_shift_the_caret_row() {
9314 // The spacer is view-only: typing the new paragraph must not reflow the
9315 // caret onto a different row — the transient view already matched the
9316 // settled one.
9317 let mut d = wysiwyg_doc("wys_no_reflow", "Hello\n\nWorld\n");
9318 d.caret = 5;
9319 d.newline();
9320 d.build_visual(80);
9321 let before = d.caret_pos();
9322 d.insert("New");
9323 d.build_visual(80);
9324 let after = d.caret_pos();
9325 assert_eq!(
9326 after.0, before.0,
9327 "typing must not move the caret to another row ({before:?} -> {after:?})"
9328 );
9329 }
9330
9331 #[test]
9332 fn wysiwyg_hides_frontmatter_from_the_caret_and_copy() {
9333 let fm = "---\ntitle: hi\n---\n";
9334 let body = format!("{fm}# leaf\n\nbody\n");
9335 let mut d = wysiwyg_doc("wys_fm", &body);
9336 // Opening lifts the caret out of the now-hidden frontmatter.
9337 assert_eq!(
9338 d.caret,
9339 fm.len(),
9340 "caret should start at the first real block"
9341 );
9342 // Left at the content start can't step back into frontmatter.
9343 d.move_left(false);
9344 assert_eq!(d.caret, fm.len(), "left must not enter frontmatter");
9345 // Doc-start lands on the content floor, not offset 0.
9346 d.move_doc_start(false);
9347 assert_eq!(d.caret, fm.len());
9348 // Select-all + copy never include the frontmatter bytes.
9349 d.select_all();
9350 let sel = d.selected_text().unwrap().to_string();
9351 assert!(!sel.contains("title"), "copy leaked frontmatter: {sel:?}");
9352 assert!(
9353 sel.starts_with("# leaf"),
9354 "selection should begin at content: {sel:?}"
9355 );
9356 }
9357
9358 #[test]
9359 fn wysiwyg_backspace_at_content_start_leaves_frontmatter_intact() {
9360 // Backspace deletes `prev_boundary..caret` directly; at the first real
9361 // block that boundary is inside the hidden frontmatter, so it must be a
9362 // no-op rather than eating the closing `---`.
9363 let fm = "---\ntitle: hi\n---\n";
9364 let body = format!("{fm}leaf\n");
9365 let mut d = wysiwyg_doc("wys_fm_bs", &body);
9366 assert_eq!(d.caret, fm.len());
9367 d.backspace();
9368 assert_eq!(d.source, body, "backspace must not touch frontmatter");
9369 d.delete_word_back();
9370 assert_eq!(
9371 d.source, body,
9372 "word-delete must not touch frontmatter either"
9373 );
9374 }
9375
9376 #[test]
9377 fn wysiwyg_edits_inside_a_vis_directive_block_without_disturbing_its_fences() {
9378 // diaryx's `:::vis{.audience}` visibility block — any `:::name{.class}`
9379 // fenced div, really, since core parses these on for every document
9380 // now (`parse_extensions`). The container is a `directive` node, an
9381 // `is_block_container` kind like `block_quote`, so the caret works
9382 // inside its child paragraph exactly as it would inside a quote: typing
9383 // edits the paragraph, and the `:::vis{...}` / `:::` fences round-trip
9384 // untouched.
9385 let body = ":::vis{.public .family}\nhello\n:::\nafter\n";
9386 let mut d = wysiwyg_doc("wys_vis", body);
9387 d.caret = body.find("hello").unwrap() + "hello".len();
9388 d.insert("!");
9389 assert_eq!(
9390 d.source, ":::vis{.public .family}\nhello!\n:::\nafter\n",
9391 "typing inside the block edits its content in place"
9392 );
9393 assert!(
9394 d.source.contains(":::vis{.public .family}"),
9395 "opening fence survives"
9396 );
9397 assert!(d.source.contains(":::\nafter"), "closing fence survives");
9398 }
9399
9400 #[test]
9401 fn source_view_still_reaches_frontmatter() {
9402 // The metadata is only *hidden*, never lost: the source view edits and
9403 // selects it in full, and it's always preserved on save.
9404 let fm = "---\ntitle: hi\n---\n";
9405 let body = format!("{fm}# leaf\n");
9406 let mut d = doc_with("src_fm", &body);
9407 d.select_all();
9408 let sel = d.selected_text().unwrap();
9409 assert!(
9410 sel.contains("title"),
9411 "source view should select everything"
9412 );
9413 d.move_doc_start(false);
9414 assert_eq!(d.caret, 0, "source view can reach offset 0");
9415 }
9416
9417 const TABLE: &str = "| Name | Qty |\n|:-----|----:|\n| Pear | 3 |\n| Fig | 12 |\n";
9418
9419 #[test]
9420 fn wysiwyg_right_crosses_a_cell_border_without_stalling() {
9421 // The border and padding between two cells all share one source offset,
9422 // so a column-stepping caret would sit on `│` and then stall there
9423 // forever. Right must step: end of "Name" -> start of "Qty".
9424 let mut d = wysiwyg_doc("tbl_right", TABLE);
9425 d.caret = TABLE.find("Name").unwrap() + 4; // just after "Name"
9426 d.move_right(false);
9427 assert_eq!(
9428 d.caret,
9429 TABLE.find("Qty").unwrap(),
9430 "should land in the next cell"
9431 );
9432 let (r, c) = d.caret_pos();
9433 assert_eq!(d.vmap.rows[r].glyphs[c].ch, 'Q');
9434 }
9435
9436 #[test]
9437 fn wysiwyg_left_crosses_back_to_the_previous_cell() {
9438 let mut d = wysiwyg_doc("tbl_left", TABLE);
9439 d.caret = TABLE.find("Qty").unwrap();
9440 d.move_left(false);
9441 assert_eq!(
9442 d.caret,
9443 TABLE.find("Name").unwrap() + 4,
9444 "end of the previous cell"
9445 );
9446 }
9447
9448 #[test]
9449 fn wysiwyg_down_steps_over_a_table_rule() {
9450 // Between the header and the first body row sits a `├───┼───┤` rule.
9451 // It's drawn but holds no caret, so one Down must reach "Pear".
9452 let mut d = wysiwyg_doc("tbl_down", TABLE);
9453 d.caret = TABLE.find("Name").unwrap();
9454 d.move_down(false);
9455 assert_eq!(
9456 d.caret,
9457 TABLE.find("Pear").unwrap(),
9458 "one Down reaches the body row"
9459 );
9460 d.move_down(false);
9461 assert_eq!(d.caret, TABLE.find("Fig").unwrap());
9462 }
9463
9464 #[test]
9465 fn wysiwyg_tab_walks_the_cells_and_shift_tab_walks_back() {
9466 let mut d = wysiwyg_doc("tbl_tab", TABLE);
9467 d.caret = TABLE.find("Name").unwrap();
9468 // A hop lands with the destination cell's whole content selected, the
9469 // caret at its end — so typing replaces the cell like a form field.
9470 assert!(d.cell_hop(true));
9471 assert_eq!(
9472 d.selected_text(),
9473 Some("Qty"),
9474 "the target cell comes up selected"
9475 );
9476 assert_eq!(d.caret, TABLE.find("Qty").unwrap() + "Qty".len());
9477 assert!(d.cell_hop(true), "Tab wraps onto the next row's first cell");
9478 assert_eq!(d.selected_text(), Some("Pear"));
9479 assert!(d.cell_hop(false));
9480 assert_eq!(d.selected_text(), Some("Qty"));
9481 }
9482
9483 #[test]
9484 fn tab_outside_a_table_is_not_a_cell_hop() {
9485 // `cell_hop` reports false so the frontend can indent as usual.
9486 let mut d = wysiwyg_doc("tbl_none", "just a paragraph\n");
9487 d.caret = 4;
9488 assert!(!d.cell_hop(true));
9489 assert_eq!(d.caret, 4, "a refused hop leaves the caret alone");
9490 }
9491
9492 #[test]
9493 fn tab_at_the_last_cell_declines_rather_than_leaving_the_table() {
9494 let mut d = wysiwyg_doc("tbl_edge", TABLE);
9495 d.caret = TABLE.rfind("12").unwrap(); // the final cell
9496 assert!(!d.cell_hop(true), "no cell after the last one");
9497 d.caret = TABLE.find("Name").unwrap();
9498 assert!(!d.cell_hop(false), "no cell before the first one");
9499 }
9500
9501 #[test]
9502 fn wysiwyg_vertical_cell_motion_holds_the_column() {
9503 // Down/Up step to the cell above/below in the *same column*, not back to
9504 // the top-left the way a naive row/col motion over the picture would.
9505 let mut d = wysiwyg_doc("tbl_vert", TABLE);
9506 d.caret = TABLE.find("Qty").unwrap();
9507 // Each vertical hop selects the destination cell, holding the column.
9508 assert!(d.cell_move_vertical(true));
9509 assert_eq!(d.selected_text(), Some("3"), "Down holds column 1");
9510 assert!(d.cell_move_vertical(true));
9511 assert_eq!(d.selected_text(), Some("12"), "Down again, still column 1");
9512 assert!(!d.cell_move_vertical(true), "no row below the last");
9513 assert!(d.cell_move_vertical(false));
9514 assert_eq!(d.selected_text(), Some("3"), "Up holds column 1");
9515 assert!(d.cell_move_vertical(false));
9516 assert_eq!(d.selected_text(), Some("Qty"), "Up onto the header");
9517 assert!(!d.cell_move_vertical(false), "no row above the header");
9518 }
9519
9520 #[test]
9521 fn tab_off_the_last_cell_grows_a_row_and_enters_it() {
9522 let mut d = wysiwyg_doc("tbl_grow", TABLE);
9523 d.caret = TABLE.rfind("12").unwrap();
9524 let rows_before = d.source.matches('\n').count();
9525 assert!(d.cell_tab(true), "acts as a table key");
9526 assert_eq!(
9527 d.source.matches('\n').count(),
9528 rows_before + 1,
9529 "a fresh row was appended"
9530 );
9531 assert!(d.caret_in_table(), "the caret entered the new row");
9532 // The caret sits in the new row's first cell — past the old last cell.
9533 assert!(d.caret > TABLE.rfind("12").unwrap());
9534 }
9535
9536 #[test]
9537 fn return_in_a_table_drops_a_cell_and_grows_a_row_at_the_bottom() {
9538 let mut d = wysiwyg_doc("tbl_ret", TABLE);
9539 d.caret = TABLE.find("Name").unwrap();
9540 assert!(d.cell_return(), "acts as a table key");
9541 assert_eq!(
9542 d.selected_text(),
9543 Some("Pear"),
9544 "Return drops one cell, selecting it"
9545 );
9546 // From the last row, Return appends a row and enters it.
9547 d.caret = TABLE.rfind("Fig").unwrap();
9548 let rows_before = d.source.matches('\n').count();
9549 assert!(d.cell_return());
9550 assert_eq!(d.source.matches('\n').count(), rows_before + 1);
9551 assert!(d.caret_in_table());
9552 }
9553
9554 #[test]
9555 fn return_and_tab_outside_a_table_decline() {
9556 let mut d = wysiwyg_doc("tbl_decline", "just a paragraph\n");
9557 d.caret = 4;
9558 assert!(!d.cell_return(), "no table: the frontend inserts a newline");
9559 assert!(!d.cell_tab(true), "no table: the frontend indents");
9560 assert!(
9561 !d.cell_line_break(),
9562 "no table: the frontend breaks the line"
9563 );
9564 }
9565
9566 #[test]
9567 fn shift_return_inserts_an_in_cell_break_the_renderer_reads_as_a_line() {
9568 let mut d = wysiwyg_doc("tbl_break", TABLE);
9569 d.caret = TABLE.find("Pear").unwrap() + 4; // just after "Pear"
9570 assert!(d.cell_line_break(), "acts as a table key");
9571 assert!(
9572 d.source.contains("Pear<br>"),
9573 "spelled as an inline <br>: {}",
9574 d.source
9575 );
9576 assert!(d.caret_in_table(), "still in the cell, past the break");
9577 // The break renders as a real line: the "Pear" cell now draws two lines,
9578 // so the table's picture is one row taller than a single-line table.
9579 d.build_visual(80);
9580 let table = &d.vmap.tables[0];
9581 let cell = &table.grid[1].cells[0]; // first body row, first column
9582 assert!(
9583 cell.glyphs.iter().any(|g| g.ch == '\n'),
9584 "the cell carries the break as a newline glyph for the frontend to split"
9585 );
9586 }
9587
9588 #[test]
9589 fn shift_return_in_a_markdown_cell_leaves_a_semantic_hard_break_not_raw_html() {
9590 // twig promotes the in-cell `<br>` to a `hard_break`, so the break reads
9591 // back as structure — the whole point of routing through insert_line_break
9592 // instead of splicing raw `<br>` bytes.
9593 let mut d = wysiwyg_doc("tbl_break_semantic", TABLE);
9594 d.caret = TABLE.find("Pear").unwrap() + 4;
9595 assert!(d.cell_line_break());
9596 let kinds: Vec<Kind> = d
9597 .editor
9598 .nodes()
9599 .unwrap()
9600 .iter()
9601 .map(|n| n.kind.clone())
9602 .collect();
9603 assert!(kinds.contains(&Kind::HardBreak), "got {kinds:?}");
9604 assert!(
9605 !kinds.contains(&Kind::RawInline),
9606 "still raw HTML: {kinds:?}"
9607 );
9608 }
9609
9610 #[test]
9611 fn backspace_over_an_in_cell_break_deletes_the_whole_br_not_a_byte() {
9612 // The `<br>` draws as one newline glyph, so Backspace over it must take
9613 // all four bytes — a one-byte delete would strand a visible `<br` in the
9614 // cell (the reported bug).
9615 let mut d = wysiwyg_doc("tbl_break_bs", TABLE);
9616 d.caret = TABLE.find("Pear").unwrap() + 4;
9617 assert!(d.cell_line_break());
9618 assert!(d.source.contains("Pear<br>"), "precondition: {}", d.source);
9619 d.backspace(); // caret sits just past the break
9620 assert!(
9621 !d.source.contains("<br"),
9622 "no half-deleted <br left: {}",
9623 d.source
9624 );
9625 assert!(
9626 d.source.contains("| Pear |"),
9627 "the cell is back to one line: {}",
9628 d.source
9629 );
9630 }
9631
9632 #[test]
9633 fn delete_forward_over_an_in_cell_break_deletes_the_whole_br() {
9634 let mut d = wysiwyg_doc("tbl_break_del", TABLE);
9635 d.caret = TABLE.find("Pear").unwrap() + 4;
9636 assert!(d.cell_line_break());
9637 d.caret = TABLE.find("Pear").unwrap() + 4; // back onto the break's start
9638 d.delete_forward();
9639 assert!(
9640 !d.source.contains("<br"),
9641 "no half-deleted <br: {}",
9642 d.source
9643 );
9644 assert!(
9645 d.source.contains("| Pear |"),
9646 "cell back to one line: {}",
9647 d.source
9648 );
9649 }
9650
9651 #[test]
9652 fn shift_return_in_a_djot_cell_is_swallowed_and_leaves_the_row_intact() {
9653 // Djot has no idiomatic in-cell break, so twig refuses it. The gesture is
9654 // still consumed (a real newline would split the one-line row), but the
9655 // cell must be left exactly as it was — no non-idiomatic `<br>` spliced in.
9656 let src = "| Name | Qty |\n|:-----|----:|\n| Pear | 3 |\n";
9657 let mut d = Doc::from_source(src.to_string(), Format::Djot).unwrap();
9658 d.caret = src.find("Pear").unwrap() + 4;
9659 assert!(d.caret_in_table(), "caret should be inside the djot table");
9660 assert!(
9661 d.cell_line_break(),
9662 "the key is consumed, not passed to the frontend"
9663 );
9664 assert_eq!(d.source, src, "the djot cell is left untouched");
9665 assert!(
9666 !d.source.contains("<br>"),
9667 "no non-idiomatic <br> spliced into djot"
9668 );
9669 assert!(
9670 d.status.is_some(),
9671 "the refusal is surfaced on the status line"
9672 );
9673 }
9674
9675 #[test]
9676 fn typing_in_a_cell_edits_that_cell() {
9677 // Editing comes free once offsets map correctly: the caret is a source
9678 // offset, so a normal splice lands inside the pipe table.
9679 let mut d = wysiwyg_doc("tbl_type", TABLE);
9680 d.caret = TABLE.find("Pear").unwrap() + 4;
9681 d.insert("s");
9682 assert!(d.source.contains("| Pears | 3 |"), "got {:?}", d.source);
9683 }
9684
9685 #[test]
9686 fn motion_and_delete_treat_an_emoji_as_one_character() {
9687 // 👨👩👧 is a single grapheme built from three emoji joined by ZWJ — 18
9688 // bytes, several codepoints. Right-arrow must clear it in one step, and
9689 // backspace must remove the whole cluster, not a stray joiner.
9690 let family = "👨👩👧";
9691 let mut d = doc_with("emoji", &format!("a{family}b\n"));
9692 d.caret = 1; // just after 'a', before the emoji
9693 d.move_right(false);
9694 assert_eq!(
9695 d.caret,
9696 1 + family.len(),
9697 "one step clears the whole cluster"
9698 );
9699 assert_eq!(&d.source[d.caret..d.caret + 1], "b");
9700
9701 d.backspace(); // delete the emoji as a unit
9702 assert_eq!(d.source, "ab\n");
9703 assert_eq!(d.caret, 1);
9704 }
9705
9706 #[test]
9707 fn motion_handles_a_combining_accent_as_one_character() {
9708 // "e" + U+0301 (combining acute) renders as one é.
9709 let mut d = doc_with("combining", "e\u{0301}x\n");
9710 d.caret = 0;
9711 d.move_right(false);
9712 assert_eq!(
9713 d.caret,
9714 "e\u{0301}".len(),
9715 "steps past base + combining mark"
9716 );
9717 }
9718
9719 #[test]
9720 fn undo_then_redo_round_trips_an_edit() {
9721 let mut d = doc_with("undo", "hello\n");
9722 d.caret = 5;
9723 d.insert("!");
9724 assert_eq!(d.source, "hello!\n");
9725 d.undo();
9726 assert_eq!(d.source, "hello\n");
9727 assert_eq!(d.caret, 5, "undo restores the caret");
9728 d.redo();
9729 assert_eq!(d.source, "hello!\n");
9730 }
9731
9732 #[test]
9733 fn a_run_of_typing_undoes_as_one_step() {
9734 let mut d = doc_with("coalesce", "\n");
9735 d.caret = 0;
9736 d.insert("a");
9737 d.insert("b");
9738 d.insert("c");
9739 assert_eq!(d.source, "abc\n");
9740 d.undo(); // the whole typed run, not just "c"
9741 assert_eq!(d.source, "\n");
9742 d.undo(); // nothing left — the run was one step
9743 assert_eq!(d.source, "\n");
9744 assert_eq!(d.status.as_deref(), Some("nothing to undo"));
9745 }
9746
9747 // ── IME composition ──────────────────────────────────────────────────────
9748
9749 #[test]
9750 fn a_composition_run_undoes_as_one_step() {
9751 let mut d = doc_with("compose", "\n");
9752 d.caret = 0;
9753 // What an IME does: each step replaces the last one's provisional bytes.
9754 d.edit_composing(0, 0, "k");
9755 d.edit_composing(0, 1, "か");
9756 d.edit_composing(0, 3, "かん");
9757 d.edit_composing(0, 6, "感"); // the commit
9758 d.end_composition();
9759 assert_eq!(d.source, "感\n");
9760 d.undo(); // the whole composition, not its last keystroke
9761 assert_eq!(d.source, "\n");
9762 assert_eq!(d.status.as_deref(), None, "the run was a single step");
9763 }
9764
9765 #[test]
9766 fn two_compositions_are_two_undo_steps() {
9767 let mut d = doc_with("compose_two", "\n");
9768 d.caret = 0;
9769 d.edit_composing(0, 0, "か");
9770 d.edit_composing(0, 3, "蚊");
9771 d.end_composition();
9772 d.edit_composing(3, 3, "き");
9773 d.edit_composing(3, 6, "木");
9774 d.end_composition();
9775 assert_eq!(d.source, "蚊木\n");
9776 d.undo();
9777 assert_eq!(d.source, "蚊\n", "only the second composition");
9778 d.undo();
9779 assert_eq!(d.source, "\n");
9780 }
9781
9782 #[test]
9783 fn a_composition_does_not_fold_into_the_typing_around_it() {
9784 let mut d = doc_with("compose_typing", "\n");
9785 d.caret = 0;
9786 d.insert("a");
9787 d.insert("b");
9788 d.edit_composing(2, 2, "か");
9789 d.edit_composing(2, 5, "蚊");
9790 d.end_composition();
9791 d.insert("c");
9792 assert_eq!(d.source, "ab蚊c\n");
9793 d.undo();
9794 assert_eq!(d.source, "ab蚊\n");
9795 d.undo();
9796 assert_eq!(d.source, "ab\n");
9797 d.undo();
9798 assert_eq!(d.source, "\n");
9799 }
9800
9801 #[test]
9802 fn ending_a_composition_that_never_began_leaves_a_typing_run_alone() {
9803 let mut d = doc_with("compose_spurious", "\n");
9804 d.caret = 0;
9805 d.insert("a");
9806 d.end_composition(); // an IME unmarking unprompted
9807 d.insert("b");
9808 assert_eq!(d.source, "ab\n");
9809 d.undo();
9810 assert_eq!(d.source, "\n", "still one typed run");
9811 }
9812
9813 // ── the clipboard's rich flavor ──────────────────────────────────────────
9814
9815 #[test]
9816 fn an_inline_selection_publishes_html_without_a_paragraph_wrapper() {
9817 let mut d = doc_with("sel_inline", "a **bold** c\n");
9818 d.anchor = Some(2);
9819 d.caret = 10; // `**bold**`, inside the paragraph
9820 assert_eq!(d.selection_html().as_deref(), Some("<strong>bold</strong>"));
9821 }
9822
9823 #[test]
9824 fn a_whole_block_selection_keeps_its_paragraph() {
9825 let mut d = doc_with("sel_block", "a **bold** c\n");
9826 d.anchor = Some(0);
9827 d.caret = 12; // the entire paragraph
9828 assert_eq!(
9829 d.selection_html().as_deref(),
9830 Some("<p>a <strong>bold</strong> c</p>")
9831 );
9832 }
9833
9834 #[test]
9835 fn a_multi_block_selection_keeps_its_structure() {
9836 let mut d = doc_with("sel_multi", "para\n\n- one\n- two\n");
9837 d.select_all();
9838 let html = d.selection_html().expect("renders");
9839 assert!(html.contains("<p>para</p>"), "{html:?}");
9840 assert!(html.contains("<li>one</li>"), "{html:?}");
9841 }
9842
9843 #[test]
9844 fn a_word_inside_a_heading_publishes_as_text_not_a_heading() {
9845 // The fragment `Head` is a paragraph standalone; the *document* says it
9846 // sits inside one block, so the wrapper is an artifact either way.
9847 let mut d = doc_with("sel_heading", "# Head line\n");
9848 d.anchor = Some(2);
9849 d.caret = 6;
9850 assert_eq!(d.selection_html().as_deref(), Some("Head"));
9851 }
9852
9853 #[test]
9854 fn no_selection_publishes_no_html() {
9855 let mut d = doc_with("sel_none", "a b\n");
9856 d.caret = 1;
9857 assert_eq!(d.selection_html(), None);
9858 }
9859
9860 #[test]
9861 fn pasting_html_converts_it_and_is_one_undo_step() {
9862 let mut d = doc_with("paste_html", "x\n");
9863 d.caret = 1;
9864 assert!(d.paste_html("<p>a <strong>b</strong> c</p>"));
9865 assert_eq!(d.source, "xa **b** c\n");
9866 d.undo();
9867 assert_eq!(d.source, "x\n", "the whole paste, in one step");
9868 }
9869
9870 #[test]
9871 fn pasting_html_replaces_the_selection() {
9872 let mut d = doc_with("paste_html_sel", "keep drop\n");
9873 d.anchor = Some(5);
9874 d.caret = 9;
9875 assert!(d.paste_html("<em>new</em>"));
9876 assert_eq!(d.source, "keep *new*\n");
9877 }
9878
9879 #[test]
9880 fn html_that_would_paste_garbage_declines_so_the_caller_falls_back() {
9881 let mut d = doc_with("paste_html_bad", "x\n");
9882 d.caret = 1;
9883 // twig builds no table from HTML; raw `<table>` in prose is worse than
9884 // the plain flavor the caller still holds.
9885 assert!(!d.paste_html("<table><tr><td>a</td></tr></table>"));
9886 assert_eq!(d.source, "x\n", "declined edits nothing");
9887 }
9888
9889 #[test]
9890 fn copy_then_paste_round_trips_through_the_html_flavor() {
9891 let mut d = doc_with("clip_round", "a **b** and [l](https://x.dev)\n");
9892 d.select_all();
9893 let html = d.selection_html().expect("renders");
9894 let mut into = doc_with("clip_round_dst", "\n");
9895 into.caret = 0;
9896 assert!(into.paste_html(&html));
9897 assert_eq!(into.source, "a **b** and [l](https://x.dev)\n");
9898 }
9899
9900 #[test]
9901 fn moving_the_caret_starts_a_new_undo_group() {
9902 let mut d = doc_with("break", "\n");
9903 d.caret = 0;
9904 d.insert("a");
9905 d.insert("b"); // "ab\n", caret at 2
9906 d.move_left(false); // breaks the run
9907 d.insert("X"); // "aXb\n"
9908 assert_eq!(d.source, "aXb\n");
9909 d.undo();
9910 assert_eq!(
9911 d.source, "ab\n",
9912 "first undo removes only the post-move insert"
9913 );
9914 d.undo();
9915 assert_eq!(d.source, "\n", "second undo removes the earlier run");
9916 }
9917
9918 #[test]
9919 fn undo_reverses_a_format_toggle() {
9920 let mut d = doc_with("fmt_undo", "a word b\n");
9921 d.anchor = Some(2);
9922 d.caret = 6;
9923 d.toggle(InlineKind::Strong);
9924 assert_eq!(d.source, "a **word** b\n");
9925 d.undo();
9926 assert_eq!(d.source, "a word b\n");
9927 }
9928
9929 #[test]
9930 fn undo_back_to_the_saved_state_clears_dirty() {
9931 let mut d = doc_with("dirty_undo", "hello\n");
9932 assert!(!d.dirty);
9933 d.caret = 5;
9934 d.insert("!");
9935 assert!(d.dirty);
9936 d.undo();
9937 assert!(
9938 !d.dirty,
9939 "undoing to the saved source is not a modification"
9940 );
9941 }
9942
9943 #[test]
9944 fn a_new_edit_invalidates_redo() {
9945 let mut d = doc_with("redo_inv", "\n");
9946 d.caret = 0;
9947 d.insert("a");
9948 d.undo();
9949 d.insert("b"); // diverges — the redo of "a" is now gone
9950 d.redo();
9951 assert_eq!(d.source, "b\n");
9952 }
9953
9954 #[test]
9955 fn undo_on_empty_history_is_a_no_op() {
9956 let mut d = doc_with("undo_empty", "hi\n");
9957 d.undo();
9958 assert_eq!(d.source, "hi\n");
9959 assert_eq!(d.status.as_deref(), Some("nothing to undo"));
9960 }
9961
9962 #[test]
9963 fn a_one_character_paste_is_its_own_undo_step() {
9964 for view in [View::Source, View::Wysiwyg] {
9965 let mut d = doc_in(view, "paste_step", "ab\n");
9966 d.caret = 0;
9967 d.insert("x");
9968 d.insert("y"); // a run of typing
9969 d.paste("z"); // one character, but pasted — not part of that run
9970 assert_eq!(d.source, "xyzab\n");
9971 d.undo();
9972 assert_eq!(d.source, "xyab\n", "the paste undoes on its own");
9973 assert_eq!(d.caret, 2, "and hands back the caret it found");
9974 d.undo();
9975 assert_eq!(d.source, "ab\n", "the typed run is still one step under it");
9976 }
9977 }
9978
9979 #[test]
9980 fn the_same_character_typed_still_joins_the_run() {
9981 // The other half of the pair: `z` is a keystroke here and a paste above,
9982 // and the two undo differently. Nothing about the *string* says which —
9983 // which is why provenance has to come from the door the caller uses.
9984 for view in [View::Source, View::Wysiwyg] {
9985 let mut d = doc_in(view, "typed_run", "ab\n");
9986 d.caret = 0;
9987 d.insert("x");
9988 d.insert("y");
9989 d.insert("z");
9990 d.undo();
9991 assert_eq!(d.source, "ab\n", "one run, one step");
9992 }
9993 }
9994
9995 #[test]
9996 fn undo_restores_the_caret_to_where_it_was_not_to_the_edit_site() {
9997 for view in [View::Source, View::Wysiwyg] {
9998 let mut d = doc_in(view, "undo_caret", "hello world\n");
9999 d.caret = 11; // standing at the end of "world", away from the edit
10000 d.edit(0, 5, "goodbye");
10001 assert_eq!(d.source, "goodbye world\n");
10002 d.undo();
10003 assert_eq!(d.source, "hello world\n");
10004 // The undone edit ends at offset 5; the user was at 11.
10005 assert_eq!(d.caret, 11, "the caret comes back with the bytes");
10006 }
10007 }
10008
10009 #[test]
10010 fn undo_restores_the_selection_the_edit_replaced() {
10011 for view in [View::Source, View::Wysiwyg] {
10012 let mut d = doc_in(view, "undo_sel", "a word b\n");
10013 d.anchor = Some(2);
10014 d.caret = 6; // "word" selected
10015 d.insert("X");
10016 assert_eq!(d.source, "a X b\n");
10017 d.undo();
10018 assert_eq!(d.source, "a word b\n");
10019 assert_eq!(d.selection(), Some((2, 6)), "the selection comes back too");
10020 }
10021 }
10022
10023 #[test]
10024 fn redo_restores_the_caret_the_edit_left_behind() {
10025 for view in [View::Source, View::Wysiwyg] {
10026 let mut d = doc_in(view, "redo_caret", "hello world\n");
10027 d.caret = 11;
10028 d.edit(0, 5, "goodbye");
10029 assert_eq!(d.caret, 7, "the edit left the caret after its new text");
10030 d.undo();
10031 d.redo();
10032 assert_eq!(d.source, "goodbye world\n");
10033 assert_eq!(d.caret, 7, "redo puts it back where the edit had it");
10034 }
10035 }
10036
10037 #[test]
10038 fn undoing_a_typed_run_restores_the_caret_from_before_the_whole_run() {
10039 for view in [View::Source, View::Wysiwyg] {
10040 let mut d = doc_in(view, "run_caret", "hi\n");
10041 d.caret = 2;
10042 d.insert("a");
10043 d.insert("b");
10044 d.insert("c");
10045 assert_eq!(d.source, "hiabc\n");
10046 d.undo();
10047 assert_eq!(d.source, "hi\n");
10048 assert_eq!(d.caret, 2, "before the run, not before its last keystroke");
10049 d.redo();
10050 assert_eq!(d.caret, 5, "and redo restores the end of the whole run");
10051 }
10052 }
10053
10054 #[test]
10055 fn undo_restores_the_caret_across_a_format_toggle() {
10056 // A toggle reaches twig without going through `splice`, so it has to
10057 // record its own step — miss it and every stack depth below it is off by
10058 // one, and undo starts handing back another edit's caret.
10059 for view in [View::Source, View::Wysiwyg] {
10060 let mut d = doc_in(view, "fmt_caret", "a word b\n");
10061 d.caret = 8;
10062 d.anchor = Some(2);
10063 d.caret = 6;
10064 d.toggle(InlineKind::Strong);
10065 assert_eq!(d.source, "a **word** b\n");
10066 d.undo();
10067 assert_eq!(d.source, "a word b\n");
10068 assert_eq!(
10069 d.selection(),
10070 Some((2, 6)),
10071 "the toggled selection comes back"
10072 );
10073 }
10074 }
10075
10076 #[test]
10077 fn an_edit_after_an_undo_truncates_the_caret_history_with_twigs() {
10078 // The drift that would never announce itself: twig drops its redo stack
10079 // on any fresh edit, so a leaf redo entry that outlives it would restore
10080 // a caret from the timeline that edit abandoned.
10081 for view in [View::Source, View::Wysiwyg] {
10082 let mut d = doc_in(view, "redo_trunc", "hello world\n");
10083 d.caret = 11;
10084 d.edit(0, 5, "goodbye"); // step A, caret 11 → 7
10085 d.undo();
10086 assert_eq!(d.caret, 11);
10087 d.caret = 0;
10088 d.insert("X"); // diverges: A's redo is gone from twig
10089 assert_eq!(d.source, "Xhello world\n");
10090
10091 d.redo();
10092 assert_eq!(d.source, "Xhello world\n", "nothing to redo onto");
10093 assert_eq!(d.status.as_deref(), Some("nothing to redo"));
10094 d.undo();
10095 assert_eq!(d.source, "hello world\n");
10096 assert_eq!(
10097 d.caret, 0,
10098 "the surviving step's caret, not the dropped one"
10099 );
10100 }
10101 }
10102
10103 #[test]
10104 fn indent_and_outdent_move_the_caret_line_with_its_text() {
10105 for view in [View::Source, View::Wysiwyg] {
10106 let g = |m, f: fn(&mut Doc)| golden_in(view, "indent_line", m, f);
10107 assert_eq!(g("he|llo\n", |d| d.indent()), " he|llo\n");
10108 assert_eq!(g(" he|llo\n", |d| d.outdent()), "he|llo\n");
10109 // Indentation the caret is standing *in* collapses to the line start
10110 // rather than dragging the caret into the text.
10111 assert_eq!(g("| hello\n", |d| d.outdent()), "|hello\n");
10112 // A line with none to give back is left exactly as it was.
10113 assert_eq!(g("he|llo\n", |d| d.outdent()), "he|llo\n");
10114 // Less than a full level gives back what it has.
10115 assert_eq!(g(" he|llo\n", |d| d.outdent()), "he|llo\n");
10116 // A tab is one level however many spaces it isn't.
10117 assert_eq!(g("\the|llo\n", |d| d.outdent()), "he|llo\n");
10118 }
10119 }
10120
10121 #[test]
10122 fn one_indent_level_leaves_a_paragraph_a_paragraph() {
10123 // Why the level is two spaces and not the four both frontends type
10124 // today. Four is markdown's indented-code-block marker, so a Tab on a
10125 // paragraph would silently restyle it as code — a width that changes
10126 // what the document *means* isn't an indent. Pinned because the number
10127 // is the kind of thing a later list-aware pass would reach for.
10128 let mut d = doc_with("indent_kind", "hello\n");
10129 d.caret = 2;
10130 d.indent();
10131 assert_eq!(d.source, " hello\n");
10132 assert!(
10133 d.nodes().iter().any(|n| n.kind == Kind::Para),
10134 "still prose after a Tab"
10135 );
10136 assert!(!d.nodes().iter().any(|n| n.kind == Kind::CodeBlock));
10137
10138 // The four-space level this replaces, for contrast: same text, and twig
10139 // reparses the paragraph into a code block.
10140 let mut wide = doc_with("indent_kind_4", " hello\n");
10141 wide.build_visual(80);
10142 assert!(
10143 wide.nodes().iter().any(|n| n.kind == Kind::CodeBlock),
10144 "four spaces is a code block, not an indented paragraph"
10145 );
10146 }
10147
10148 #[test]
10149 fn indent_nests_a_list_item_under_its_parent() {
10150 // Tab indents a list item by its own marker width, landing its marker at
10151 // the parent's content column so twig reparses it as a nested list.
10152 for view in [View::Source, View::Wysiwyg] {
10153 let mut d = doc_in(view, "indent_nest", "- a\n- b\n");
10154 d.caret = 6; // on the second item
10155 d.indent();
10156 assert_eq!(d.source, "- a\n - b\n");
10157 let lists = d
10158 .nodes()
10159 .iter()
10160 .filter(|n| n.kind == Kind::BulletList)
10161 .count();
10162 assert_eq!(lists, 2, "the indented item is a nested list");
10163 }
10164 }
10165
10166 #[test]
10167 fn indent_nests_an_ordered_item_at_its_marker_width() {
10168 // An ordered marker `1. ` is three columns wide, so a two-space step
10169 // (which nests a bullet) leaves it flat. Regression: Tab must use the
10170 // marker width, three, so the item actually nests — and the source
10171 // renumbers so the sub-list restarts at 1 and the outer list resumes.
10172 for view in [View::Source, View::Wysiwyg] {
10173 let mut d = doc_in(view, "indent_ord", "1. a\n2. b\n3. c\n");
10174 d.caret = d.source.find('b').unwrap();
10175 d.indent();
10176 assert_eq!(d.source, "1. a\n 1. b\n2. c\n");
10177 let lists = d
10178 .nodes()
10179 .iter()
10180 .filter(|n| n.kind == Kind::OrderedList)
10181 .count();
10182 assert_eq!(lists, 2, "the indented item is a nested ordered list");
10183 }
10184 }
10185
10186 #[test]
10187 fn indent_leaves_a_lists_first_item_put() {
10188 // The first item of a list has no sibling above it to nest under, so Tab
10189 // is a no-op there — the marker stays at column zero rather than being
10190 // shoved into indentation twig can't read as a sub-list.
10191 for view in [View::Source, View::Wysiwyg] {
10192 let mut d = doc_in(view, "indent_first", "- a\n- b\n");
10193 d.caret = 1; // on the FIRST item
10194 d.indent();
10195 assert_eq!(d.source, "- a\n- b\n", "the first item doesn't nest");
10196 // The sibling below still nests, proving the guard is per-item.
10197 d.caret = d.source.find('b').unwrap();
10198 d.indent();
10199 assert_eq!(d.source, "- a\n - b\n");
10200 }
10201 }
10202
10203 #[test]
10204 fn hidden_mode_keeps_typed_markup_literal() {
10205 // The Diaryx default: typing `*hi*` gives the characters, not emphasis —
10206 // twig escapes what would open markup, so the source is `\*hi\*` and the
10207 // AST is a plain string. Formatting is the commands' job in this mode.
10208 let mut d = doc_in(View::Wysiwyg, "hidden_literal", "");
10209 d.insert("*hi*");
10210 assert_eq!(d.source, "\\*hi\\*");
10211 assert!(
10212 d.nodes()
10213 .iter()
10214 .all(|n| n.kind != Kind::Emph && n.kind != Kind::Strong)
10215 );
10216 }
10217
10218 #[test]
10219 fn hidden_mode_escapes_a_line_start_block_marker() {
10220 // A `#`/`-`/`>` at a line start would open a block, so Hidden mode keeps
10221 // it literal too — a Diaryx user's "# 1 idea" stays prose, not a heading.
10222 let mut d = doc_in(View::Wysiwyg, "hidden_block", "");
10223 d.insert("# hi");
10224 assert_eq!(d.source, "\\# hi");
10225 assert!(d.nodes().iter().all(|n| n.kind != Kind::Heading));
10226 }
10227
10228 #[test]
10229 fn authoring_modes_keep_typed_markup_live() {
10230 // Both authoring rungs of the ladder: typing `*hi*` really is emphasis
10231 // (no escape), the same as source view — escaping is `None`'s alone, and
10232 // it's the axis, not the reveal, that decides.
10233 for (view, mode) in [
10234 (View::Wysiwyg, MarkupMode::Shortcuts),
10235 (View::Wysiwyg, MarkupMode::Full),
10236 (View::Source, MarkupMode::None),
10237 ] {
10238 let mut d = doc_in(view, "live_markup", "");
10239 d.set_markup_mode(mode);
10240 d.insert("*hi*");
10241 assert_eq!(d.source, "*hi*", "{mode:?} in {view:?} types raw markup");
10242 }
10243 }
10244
10245 #[test]
10246 fn hidden_mode_overwrite_undoes_in_one_step() {
10247 // Typing over a selection escapes the replacement *and* stays a single
10248 // undo — the selection-delete and the literal insert fold together, so
10249 // one undo brings the whole selection back, like a plain overwrite.
10250 let mut d = doc_in(View::Wysiwyg, "hidden_overwrite", "a word b\n");
10251 d.anchor = Some(2);
10252 d.caret = 6; // "word"
10253 d.insert("*");
10254 assert_eq!(d.source, "a \\* b\n", "the replacement is escaped");
10255 d.undo();
10256 assert_eq!(d.source, "a word b\n");
10257 assert_eq!(d.selection(), Some((2, 6)), "one undo, selection restored");
10258 }
10259
10260 #[test]
10261 fn backspace_over_an_escaped_char_takes_the_hidden_backslash_too() {
10262 // Type `*` in Hidden mode → `\*` (drawn as one `*`); one Backspace clears
10263 // the whole visual character, never stranding the hidden `\`.
10264 let mut d = doc_in(View::Wysiwyg, "bsp_escape", "");
10265 d.insert("*");
10266 assert_eq!(d.source, "\\*");
10267 d.backspace();
10268 assert_eq!(d.source, "", "the escape backslash went with the *");
10269 // A *literal* backslash (source view, no escape) is an ordinary char.
10270 let mut s = doc_in(View::Source, "bsp_lit", "a\\b\n");
10271 s.caret = 3; // after `b`
10272 s.backspace();
10273 assert_eq!(s.source, "a\\\n", "only the b is deleted, the \\ stays");
10274 }
10275
10276 #[test]
10277 fn hidden_mode_leaves_structural_markup_alone() {
10278 // Enter continues a bullet list by writing a real `- ` marker (an
10279 // `insert_raw`, not the typing path), so Hidden mode's escaping never
10280 // touches it — the list keeps working.
10281 let mut d = doc_in(View::Wysiwyg, "hidden_struct", "- item\n");
10282 d.caret = 6;
10283 d.newline();
10284 d.insert("two");
10285 assert_eq!(d.source, "- item\n- two\n");
10286 }
10287
10288 #[test]
10289 fn markup_mode_defaults_to_none_and_round_trips() {
10290 // Diaryx's default is the clean `None` surface; a markup-fluent
10291 // frontend can climb the ladder, and the choice sticks.
10292 let mut d = doc_in(View::Wysiwyg, "markup_mode", "hi\n");
10293 assert_eq!(d.markup_mode(), MarkupMode::None, "None by default");
10294 for mode in [MarkupMode::Shortcuts, MarkupMode::Full, MarkupMode::None] {
10295 d.set_markup_mode(mode);
10296 assert_eq!(d.markup_mode(), mode);
10297 }
10298 }
10299
10300 #[test]
10301 fn full_mode_reveals_only_the_caret_line() {
10302 // The mode's whole claim: the caret's line shows its raw delimiters and
10303 // every other line stays resolved. Two paragraphs with identical markup
10304 // so the only difference between the rows is where the caret is.
10305 let mut d = doc_in(
10306 View::Wysiwyg,
10307 "reveal_caret_line",
10308 "*one* here\n\n*two* there\n",
10309 );
10310 d.set_markup_mode(MarkupMode::Full);
10311
10312 caret_at(&mut d, "one");
10313 let rows = drawn_rows(&d);
10314 assert!(
10315 rows.iter().any(|r| r == "*one* here"),
10316 "caret's line raw: {rows:?}"
10317 );
10318 assert!(
10319 rows.iter().any(|r| r == "two there"),
10320 "other line resolved: {rows:?}"
10321 );
10322
10323 // Move to the other paragraph: the reveal follows, and the line just
10324 // left goes back to being resolved.
10325 caret_at(&mut d, "two");
10326 let rows = drawn_rows(&d);
10327 assert!(
10328 rows.iter().any(|r| r == "*two* there"),
10329 "caret's line raw: {rows:?}"
10330 );
10331 assert!(
10332 rows.iter().any(|r| r == "one here"),
10333 "left line resolved: {rows:?}"
10334 );
10335 }
10336
10337 #[test]
10338 fn hidden_modes_never_reveal_wherever_the_caret_is() {
10339 // The two rungs below `Full` share a rendering: delimiters stay hidden
10340 // even under the caret. `Shortcuts` differing from `None` only in what
10341 // typing does is exactly the point of splitting the axes.
10342 for mode in [MarkupMode::None, MarkupMode::Shortcuts] {
10343 let mut d = doc_in(View::Wysiwyg, "reveal_hidden", "*one* here\n");
10344 d.set_markup_mode(mode);
10345 caret_at(&mut d, "one");
10346 let rows = drawn_rows(&d);
10347 assert!(
10348 rows.iter().any(|r| r == "one here"),
10349 "{mode:?} hides: {rows:?}"
10350 );
10351 assert!(
10352 !rows.iter().any(|r| r.contains('*')),
10353 "{mode:?} shows no `*`: {rows:?}"
10354 );
10355 }
10356 }
10357
10358 #[test]
10359 fn revealed_delimiters_are_the_authors_own_spelling() {
10360 // Delimiters are re-read from the source rather than synthesized per
10361 // kind, so a line comes back spelled the way it was written: `_em_` does
10362 // not turn into `*em*`, and a two-backtick fence keeps both backticks.
10363 let body = "_em_ and __st__ and ``lit ` tick`` and [lk](http://x) and ~~del~~\n";
10364 let mut d = doc_in(View::Wysiwyg, "reveal_spelling", body);
10365 d.set_markup_mode(MarkupMode::Full);
10366 caret_at(&mut d, "em");
10367 let rows = drawn_rows(&d);
10368 assert!(
10369 rows.iter().any(|r| r == body.trim_end()),
10370 "the revealed line is its own source: {rows:?}"
10371 );
10372 }
10373
10374 #[test]
10375 fn revealed_heading_shows_its_hashes() {
10376 // The `# ` marker is a block-level prefix, not an inline delimiter, so
10377 // it takes its own path — but it reveals on the same rule.
10378 let mut d = doc_in(View::Wysiwyg, "reveal_heading", "# Title\n\nbody\n");
10379 d.set_markup_mode(MarkupMode::Full);
10380
10381 caret_at(&mut d, "Title");
10382 assert!(
10383 drawn_rows(&d).iter().any(|r| r == "# Title"),
10384 "{:?}",
10385 drawn_rows(&d)
10386 );
10387
10388 caret_at(&mut d, "body");
10389 let rows = drawn_rows(&d);
10390 assert!(
10391 rows.iter().any(|r| r == "Title"),
10392 "hashes hidden again: {rows:?}"
10393 );
10394 }
10395
10396 #[test]
10397 fn revealed_delimiters_are_caret_stops() {
10398 // A delimiter that is drawn but can't be reached is worse than one
10399 // that's hidden: the mode exists so the markup can be *edited*. Every
10400 // revealed byte must be somewhere the caret can stand.
10401 let mut d = doc_in(View::Wysiwyg, "reveal_stops", "*em* x\n");
10402 d.set_markup_mode(MarkupMode::Full);
10403 caret_at(&mut d, "em");
10404 let opener = d.source.find('*').unwrap();
10405 assert!(d.vmap.is_stop(opener), "the opening `*` is a caret stop");
10406 assert!(
10407 d.vmap.is_stop(opener + 3),
10408 "the closing `*` is a caret stop"
10409 );
10410 }
10411
10412 #[test]
10413 fn setext_heading_reveals_nothing_across_its_newline() {
10414 // A setext heading's underline is on another line, so it is not the
10415 // caret line's to reveal — and emitting it would inject a `\n` glyph
10416 // that splits the row where the author wrote no break.
10417 let mut d = doc_in(View::Wysiwyg, "reveal_setext", "Title\n=====\n\nbody\n");
10418 d.set_markup_mode(MarkupMode::Full);
10419 caret_at(&mut d, "Title");
10420 let rows = drawn_rows(&d);
10421 assert!(
10422 rows.iter().any(|r| r == "Title"),
10423 "title renders alone: {rows:?}"
10424 );
10425 assert!(
10426 !rows.iter().any(|r| r.contains('=')),
10427 "no underline leaks in: {rows:?}"
10428 );
10429 }
10430
10431 #[test]
10432 fn markup_mode_axes_split_the_ladder() {
10433 // The two behaviours the ladder spells: `Shortcuts` is the middle rung
10434 // that authors markup but still hides it, and it's the only rung where
10435 // the two axes disagree.
10436 assert!(!MarkupMode::None.authors());
10437 assert!(!MarkupMode::None.reveals_caret_line());
10438 assert!(MarkupMode::Shortcuts.authors());
10439 assert!(!MarkupMode::Shortcuts.reveals_caret_line());
10440 assert!(MarkupMode::Full.authors());
10441 assert!(MarkupMode::Full.reveals_caret_line());
10442 }
10443
10444 #[test]
10445 fn indenting_an_empty_dash_item_under_text_dodges_the_setext_collapse() {
10446 // Tabbing an empty `- ` under a text line would spell `- hello\n - `,
10447 // which twig (correctly, per CommonMark — pandoc agrees) reparses as a
10448 // setext H2. leaf swaps the dash for a `*` so the item stays an empty
10449 // nested bullet and `hello` stays prose: the file round-trips instead of
10450 // hiding a heading the user never asked for.
10451 for view in [View::Source, View::Wysiwyg] {
10452 let mut d = doc_in(view, "setext_guard", "- hello\n- \n");
10453 d.caret = d.source.find("- \n").unwrap() + 2; // after the empty marker
10454 d.indent();
10455 assert_eq!(d.source, "- hello\n * \n");
10456 assert!(
10457 d.nodes().iter().all(|n| n.kind != Kind::Heading),
10458 "no heading"
10459 );
10460 // And it's genuinely a nested list, not a flat one.
10461 assert_eq!(
10462 d.nodes()
10463 .iter()
10464 .filter(|n| n.kind == Kind::BulletList)
10465 .count(),
10466 2
10467 );
10468 }
10469 }
10470
10471 #[test]
10472 fn indenting_a_dash_item_with_content_keeps_its_dash() {
10473 // With content, `- x` can't be a setext underline, so there's nothing to
10474 // dodge: the marker stays a dash and nests as an ordinary sub-bullet.
10475 let mut d = doc_in(View::Wysiwyg, "setext_ok", "- hello\n- x\n");
10476 d.caret = d.source.find('x').unwrap();
10477 d.indent();
10478 assert_eq!(d.source, "- hello\n - x\n");
10479 }
10480
10481 #[test]
10482 fn the_setext_swap_undoes_as_one_step_with_the_indent() {
10483 // The dash→`*` repair coalesces into the Tab, so a single undo restores
10484 // the whole pre-Tab state rather than stranding a half-collapsed doc.
10485 let mut d = doc_in(View::Wysiwyg, "setext_undo", "- hello\n- \n");
10486 d.caret = d.source.find("- \n").unwrap() + 2;
10487 d.indent();
10488 assert_eq!(d.source, "- hello\n * \n");
10489 d.undo();
10490 assert_eq!(d.source, "- hello\n- \n", "one undo, not two");
10491 }
10492
10493 #[test]
10494 fn indent_leaves_a_nested_lists_first_item_put_too() {
10495 // The guard is about siblings, not depth: the first item of an *inner*
10496 // list (already nested under `a`) still has nothing before it at its own
10497 // level, so Tab can't take it deeper.
10498 let mut d = doc_in(View::Wysiwyg, "indent_first_nested", "- a\n - b\n - c\n");
10499 d.caret = d.source.find('b').unwrap();
10500 d.indent();
10501 assert_eq!(d.source, "- a\n - b\n - c\n", "inner first item holds");
10502 // But `c` (a sibling of `b`) nests under `b`.
10503 d.caret = d.source.find('c').unwrap();
10504 d.indent();
10505 assert_eq!(d.source, "- a\n - b\n - c\n");
10506 }
10507
10508 #[test]
10509 fn backspace_at_a_nested_item_start_outdents_it() {
10510 // Backspace with the caret right after a nested item's marker gives back
10511 // one level of nesting, the mirror of Tab — and renumbers the flattened
10512 // ordered list back to a clean run.
10513 let mut d = doc_in(View::Wysiwyg, "bsp_outdent", "1. a\n 1. b\n2. c\n");
10514 d.caret = d.source.find('b').unwrap(); // start of the nested item's content
10515 d.backspace();
10516 assert_eq!(d.source, "1. a\n2. b\n3. c\n");
10517 }
10518
10519 #[test]
10520 fn backspace_at_a_top_level_item_start_strips_the_marker() {
10521 // At the outermost level there's no nesting left to give back, so the same
10522 // keystroke drops the bullet and leaves a plain paragraph.
10523 let mut d = doc_in(View::Wysiwyg, "bsp_strip", "- a\n- b\n");
10524 d.caret = d.source.find('b').unwrap(); // right after `- `
10525 d.backspace();
10526 assert_eq!(d.source, "- a\nb\n", "the marker is gone, the text stays");
10527 }
10528
10529 #[test]
10530 fn backspace_mid_item_still_deletes_a_character() {
10531 // The list behaviour is armed only at the item's content start; anywhere
10532 // else Backspace is the ordinary character delete.
10533 let mut d = doc_in(View::Wysiwyg, "bsp_mid", "- ab\n");
10534 d.caret = d.source.find('b').unwrap(); // between `a` and `b`
10535 d.backspace();
10536 assert_eq!(d.source, "- b\n");
10537 }
10538
10539 #[test]
10540 fn backspace_at_a_heading_start_strips_the_marker() {
10541 // The `# ` is markup the rich view hides, so Backspace over it takes the
10542 // whole marker and leaves a paragraph. Deleting a byte of it instead left
10543 // `#Title` — no longer a heading, with the hash now literal text the user
10544 // never typed and has to delete again.
10545 let mut d = doc_in(View::Wysiwyg, "bsp_head", "## Title\n");
10546 d.caret = d.source.find('T').unwrap(); // right after `## `
10547 d.backspace();
10548 assert_eq!(d.source, "Title\n");
10549 assert_eq!(
10550 d.caret, 0,
10551 "the caret stays with the text it was in front of"
10552 );
10553 }
10554
10555 #[test]
10556 fn backspace_at_a_heading_start_keeps_the_block_around_it() {
10557 // Only the heading's own marker goes — the quote (or list) it sits in is
10558 // untouched, exactly as un-heading it should be.
10559 let mut d = doc_in(View::Wysiwyg, "bsp_head_quote", "> # Title\n");
10560 d.caret = d.source.find('T').unwrap();
10561 d.backspace();
10562 assert_eq!(d.source, "> Title\n");
10563 }
10564
10565 #[test]
10566 fn backspace_at_a_heading_start_takes_its_closing_sequence_too() {
10567 // `# Title #`'s trailing hashes are hidden at the other end; leaving them
10568 // behind would surface the same stray hash the marker delete just avoided.
10569 let mut d = doc_in(View::Wysiwyg, "bsp_head_closed", "# Title #\n");
10570 d.caret = d.source.find('T').unwrap();
10571 d.backspace();
10572 assert_eq!(d.source, "Title\n");
10573 // And it's one edit: a single undo puts the whole heading back.
10574 d.undo();
10575 assert_eq!(d.source, "# Title #\n");
10576 }
10577
10578 #[test]
10579 fn backspace_mid_heading_still_deletes_a_character() {
10580 // The heading behaviour is armed only at the content's start; anywhere
10581 // else Backspace is the ordinary character delete.
10582 let mut d = doc_in(View::Wysiwyg, "bsp_head_mid", "# ab\n");
10583 d.caret = d.source.find('b').unwrap();
10584 d.backspace();
10585 assert_eq!(d.source, "# b\n");
10586 }
10587
10588 #[test]
10589 fn source_view_backspace_still_edits_the_heading_marker_literally() {
10590 // In source view the `# ` is text on the screen the user is deleting a
10591 // byte of, so it keeps its literal meaning — the same split the list
10592 // ladder and Enter draw between the two views.
10593 let mut d = doc_with("bsp_head_src", "# Title\n");
10594 d.caret = d.source.find('T').unwrap();
10595 d.backspace();
10596 assert_eq!(d.source, "#Title\n");
10597 }
10598
10599 #[test]
10600 fn outdent_unnests_an_ordered_item_in_one_press() {
10601 // Shift+Tab gives back exactly the marker width the indent added, so a
10602 // nested ordered item unnests in a single press, and the flattened list
10603 // renumbers back to a clean 1, 2, 3.
10604 let mut d = doc_with("outdent_ord", "1. a\n 2. b\n3. c\n");
10605 d.caret = d.source.find('b').unwrap();
10606 d.outdent();
10607 assert_eq!(d.source, "1. a\n2. b\n3. c\n");
10608 let lists = d
10609 .nodes()
10610 .iter()
10611 .filter(|n| n.kind == Kind::OrderedList)
10612 .count();
10613 assert_eq!(lists, 1, "back to one flat list");
10614 }
10615
10616 #[test]
10617 fn table_insert_row_adds_a_row_below_the_caret() {
10618 let mut d = doc_with("tbl_ins_row", "| a | b |\n| --- | --- |\n| 1 | 2 |\n");
10619 d.caret = d.source.find('1').unwrap(); // in the body row
10620 d.table_insert_row(true);
10621 assert_eq!(d.source, "| a | b |\n| --- | --- |\n| 1 | 2 |\n| | |\n");
10622 }
10623
10624 #[test]
10625 fn table_insert_and_delete_column_at_the_caret() {
10626 let mut d = doc_with("tbl_col", "| a | b |\n| --- | --- |\n| 1 | 2 |\n");
10627 d.caret = d.source.find('a').unwrap(); // column 0
10628 d.table_insert_column(true); // add a column to the right of `a`
10629 assert_eq!(
10630 d.source,
10631 "| a | | b |\n| --- | --- | --- |\n| 1 | | 2 |\n"
10632 );
10633 d.caret = d.source.find('b').unwrap(); // now the third column
10634 d.table_delete_column();
10635 assert_eq!(d.source, "| a | |\n| --- | --- |\n| 1 | |\n");
10636 }
10637
10638 // ── ragged formats ───────────────────────────────────────────────────────
10639 // No format spells every gesture. HTML writes the inline marks as a tag pair
10640 // and no heading, list, quote or link; Markdown spells three of the eight
10641 // marks; djot spells all eight and no in-cell break. leaf asks twig per
10642 // gesture (`Doc::supports`) and refuses at the door, rather than letting each
10643 // op discover the fact on its own — one of them didn't.
10644
10645 /// An HTML document in the rich view, ready for a gesture.
10646 fn html_doc(body: &str) -> Doc {
10647 let mut d = Doc::from_source(body.to_string(), Format::Html).unwrap();
10648 d.view = View::Wysiwyg;
10649 d.build_visual(80);
10650 d
10651 }
10652
10653 #[test]
10654 fn a_table_gesture_leaves_an_html_table_alone() {
10655 // The regression this guard exists for. twig's table editor consults no
10656 // `Syntax` table — it spells a grid, not a delimiter — so it rebuilt an
10657 // HTML `<table>` as a *pipe table* and reported success: the whole
10658 // element replaced by `| a | b |`, silently, on one press of a toolbar
10659 // button. Every grid op went the same way.
10660 let src = "<table><tr><td>a</td><td>b</td></tr><tr><td>c</td><td>d</td></tr></table>\n";
10661 // A table of named operations, which is what it looks like.
10662 #[allow(clippy::type_complexity)]
10663 let ops: [(&str, &dyn Fn(&mut Doc)); 7] = [
10664 ("insert row", &|d: &mut Doc| d.table_insert_row(true)),
10665 ("delete row", &|d: &mut Doc| d.table_delete_row()),
10666 ("insert column", &|d: &mut Doc| d.table_insert_column(true)),
10667 ("delete column", &|d: &mut Doc| d.table_delete_column()),
10668 ("align", &|d: &mut Doc| {
10669 d.table_set_alignment(Alignment::Right)
10670 }),
10671 ("move row", &|d: &mut Doc| d.table_move_row(true)),
10672 ("move column", &|d: &mut Doc| d.table_move_column(true)),
10673 ];
10674 for (name, op) in ops {
10675 let mut d = html_doc(src);
10676 d.caret = d.source.find('a').unwrap();
10677 assert!(d.caret_in_table(), "{name}: the caret really is in a table");
10678 op(&mut d);
10679 assert_eq!(d.source, src, "{name} rewrote an HTML table");
10680 assert!(
10681 !d.dirty,
10682 "{name} marked the document dirty without editing it"
10683 );
10684 assert!(d.status.is_some(), "{name} refused without saying why");
10685 }
10686 }
10687
10688 #[test]
10689 fn the_block_gestures_html_cannot_spell_are_refused_with_a_reason() {
10690 // A heading is a wrapping tag pair carrying its level in both ends, a
10691 // quote wraps a range rather than prefixing each line, a link's
10692 // destination lives in an attribute — different *shapes*, not a
10693 // different alphabet, so twig spells none of them and neither does leaf.
10694 let src = "<h1>Title</h1>\n<p>Hello world</p>\n<ul><li>one</li></ul>\n";
10695 // A table of named operations, which is what it looks like.
10696 #[allow(clippy::type_complexity)]
10697 let ops: [(&str, &dyn Fn(&mut Doc)); 9] = [
10698 ("heading", &|d: &mut Doc| d.toggle_heading(2)),
10699 ("paragraph", &|d: &mut Doc| {
10700 d.set_block(BlockKind::Paragraph)
10701 }),
10702 ("quote", &|d: &mut Doc| d.toggle_blockquote()),
10703 ("list", &|d: &mut Doc| d.toggle_list(false)),
10704 ("task item", &|d: &mut Doc| d.toggle_task_item()),
10705 ("task tick", &|d: &mut Doc| d.toggle_task_checked()),
10706 ("link", &|d: &mut Doc| d.insert_link("https://example.dev")),
10707 ("image", &|d: &mut Doc| d.insert_image("pic.png", "alt")),
10708 ("video", &|d: &mut Doc| {
10709 d.insert_media(MediaKind::Video, "clip.mp4", "")
10710 }),
10711 ];
10712 for (name, op) in ops {
10713 let mut d = html_doc(src);
10714 let at = d.source.find("Hello").unwrap();
10715 d.caret = at;
10716 d.anchor = Some(at + 5); // a selection, for the ops that want one
10717 op(&mut d);
10718 assert_eq!(d.source, src, "{name} edited an HTML document");
10719 assert!(
10720 !d.dirty,
10721 "{name} marked the document dirty without editing it"
10722 );
10723 let status = d.status.as_deref().unwrap_or("");
10724 assert!(
10725 status.contains("html"),
10726 "{name}: the refusal should name the format, got {status:?}"
10727 );
10728 }
10729 }
10730
10731 #[test]
10732 fn html_spells_the_inline_marks_and_the_rule() {
10733 // The other half, and why one per-document flag stopped being enough:
10734 // ⌘B in an HTML document writes `<strong>` — the tag the serializer
10735 // already emits and the parser reads straight back as the same mark —
10736 // and the rule button writes an `<hr>`. Refusing these on the old
10737 // "HTML is parse-only" reading would now be leaf's own limitation.
10738 let mut d = html_doc("<p>Hello world</p>\n");
10739 let at = d.source.find("world").unwrap();
10740 d.caret = at;
10741 d.anchor = Some(at + 5);
10742 d.toggle(InlineKind::Strong);
10743 assert_eq!(d.source, "<p>Hello <strong>world</strong></p>\n");
10744 assert!(d.dirty);
10745 assert_eq!(d.status, None, "a supported gesture reports nothing");
10746
10747 // And off again — the toggle reverses, which is the property that makes
10748 // authoring in HTML worth offering rather than a one-way trip.
10749 d.toggle(InlineKind::Strong);
10750 assert_eq!(d.source, "<p>Hello world</p>\n");
10751
10752 let mut d = html_doc("<p>Hello world</p>\n");
10753 d.caret = d.source.find("world").unwrap();
10754 d.insert_thematic_break();
10755 assert!(d.source.contains("<hr>"), "got {:?}", d.source);
10756 }
10757
10758 #[test]
10759 fn a_mark_the_format_cannot_spell_arms_nothing() {
10760 // `toggle` with a collapsed caret doesn't reach twig at all — it arms a
10761 // sticky mark for the next text typed. Guarding only the twig call
10762 // leaves that path live, promising a highlight Markdown will never spell
10763 // and then swallowing the error inside `insert`. Markdown carries the
10764 // case now that HTML spells `<mark>`: `==mark==` is djot's alone.
10765 let mut d = doc_with("mark", "Hello world\n");
10766 d.view = View::Wysiwyg;
10767 d.build_visual(80);
10768 d.caret = d.source.find("world").unwrap();
10769 d.toggle(InlineKind::Mark);
10770 assert!(d.pending_marks.is_empty(), "no mark should be armed");
10771 assert!(d.status.as_deref().unwrap_or("").contains("markdown"));
10772 d.insert("X");
10773 assert_eq!(d.source, "Hello Xworld\n");
10774 }
10775
10776 #[test]
10777 fn html_documents_still_take_typed_text() {
10778 // The guard covers *markup* gestures and must not touch plain editing:
10779 // twig's splicer is language-neutral, and typing into an HTML document
10780 // is the thing that does work today.
10781 let mut d = html_doc("<p>Hello world</p>\n");
10782 d.caret = d.source.find("world").unwrap();
10783 d.insert("big ");
10784 assert_eq!(d.source, "<p>Hello big world</p>\n");
10785 assert!(d.dirty);
10786 d.backspace();
10787 assert_eq!(d.source, "<p>Hello bigworld</p>\n");
10788 d.undo();
10789 d.undo();
10790 assert_eq!(d.source, "<p>Hello world</p>\n");
10791 }
10792
10793 #[test]
10794 fn authorable_is_the_coarse_question_and_capabilities_the_useful_one() {
10795 // `authorable` only separates "there is a door in" from "there is not",
10796 // and HTML is on the near side of that line — which is exactly why a
10797 // toolbar must not be built from it.
10798 let html = Doc::from_source("<p>x</p>\n".into(), Format::Html).unwrap();
10799 assert!(html.authorable());
10800 assert!(
10801 !Doc::from_source("<r>x</r>".into(), Format::Xml)
10802 .unwrap()
10803 .authorable()
10804 );
10805
10806 let caps = html.capabilities();
10807 assert!(caps.bold && caps.italic && caps.code && caps.mark);
10808 assert!(caps.thematic_break && caps.cell_line_break);
10809 assert!(!caps.heading && !caps.blockquote && !caps.bullet_list);
10810 assert!(!caps.task && !caps.link && !caps.image && !caps.code_language);
10811 // The one flag that isn't twig's answer: an HTML `<table>` is a grid
10812 // twig's table editor would happily re-emit as `| a | b |`.
10813 assert!(!caps.table);
10814
10815 // The two lightweight formats spell everything leaf offers — and still
10816 // differ from each other, which is the other half of why one boolean
10817 // can't serve.
10818 for fmt in [Format::Markdown, Format::Djot] {
10819 let caps = Capabilities::of(fmt);
10820 assert!(
10821 caps.heading && caps.blockquote && caps.ordered_list,
10822 "{fmt:?}"
10823 );
10824 assert!(
10825 caps.task && caps.link && caps.image && caps.table,
10826 "{fmt:?}"
10827 );
10828 }
10829 assert!(Capabilities::of(Format::Djot).mark);
10830 assert!(!Capabilities::of(Format::Markdown).mark);
10831 assert!(Capabilities::of(Format::Markdown).cell_line_break);
10832 assert!(!Capabilities::of(Format::Djot).cell_line_break);
10833
10834 // A parse-only format answers no to every one of them, so the coarse
10835 // predicate and the record agree there.
10836 let caps = Capabilities::of(Format::Xml);
10837 assert!(!caps.bold && !caps.heading && !caps.table && !caps.thematic_break);
10838 }
10839
10840 #[test]
10841 fn a_refused_gesture_says_so_where_twig_would_have_said_it() {
10842 // The guard exists to name the *document's* format rather than twig's
10843 // internals, so the message has to survive being one leaf writes itself.
10844 // Checked against the gesture twig also refuses, since that is the pair
10845 // most at risk of drifting apart.
10846 let mut d = html_doc("<p>Hello</p>\n");
10847 d.caret = d.source.find("Hello").unwrap();
10848 d.set_code_language("zig");
10849 assert_eq!(
10850 d.status.as_deref(),
10851 Some("code language: not supported in html")
10852 );
10853 assert!(!d.dirty);
10854 }
10855
10856 #[test]
10857 fn table_set_alignment_respells_the_delimiter() {
10858 let mut d = doc_with("tbl_align", "| a | b |\n| --- | --- |\n| 1 | 2 |\n");
10859 d.caret = d.source.find('b').unwrap();
10860 d.table_set_alignment(Alignment::Right);
10861 assert_eq!(d.source, "| a | b |\n| --- | ---: |\n| 1 | 2 |\n");
10862 }
10863
10864 #[test]
10865 fn each_empty_table_cell_has_its_own_editable_home() {
10866 // Regression: an empty cell has no twig content_span, so both cells of a
10867 // `| | |` row collapsed onto the row's start (before the first `│`).
10868 // Typing there inserted *before* the table (`hello| | |`); nav couldn't
10869 // tell the cells apart. Each empty cell must now have a distinct home
10870 // inside it.
10871 let mut d = wysiwyg_doc("tbl_empty", "| a | b |\n| --- | --- |\n| | |\n");
10872 let (c0, c1) = {
10873 let cells = &d.vmap.tables[0].grid[1].cells;
10874 (cells[0].start, cells[1].start)
10875 };
10876 assert!(
10877 c0 < c1,
10878 "the two empty cells have distinct homes: {c0} < {c1}"
10879 );
10880 d.caret = c0;
10881 d.insert("x");
10882 assert_eq!(
10883 d.source, "| a | b |\n| --- | --- |\n| x | |\n",
10884 "typed inside the cell"
10885 );
10886 }
10887
10888 #[test]
10889 fn arrows_step_into_each_empty_table_cell() {
10890 let mut d = wysiwyg_doc("tbl_empty_nav", "| a | b |\n| --- | --- |\n| | |\n");
10891 let (c0, c1) = {
10892 let cells = &d.vmap.tables[0].grid[1].cells;
10893 (cells[0].start, cells[1].start)
10894 };
10895 d.caret = d.source.find('b').unwrap(); // in the header's second cell
10896 let mut seen = std::collections::HashSet::new();
10897 for _ in 0..6 {
10898 d.move_right(false);
10899 seen.insert(d.caret);
10900 }
10901 assert!(
10902 seen.contains(&c0),
10903 "right arrow reaches the first empty cell"
10904 );
10905 assert!(
10906 seen.contains(&c1),
10907 "right arrow reaches the second empty cell"
10908 );
10909 }
10910
10911 #[test]
10912 fn table_op_off_a_table_is_a_no_op_with_a_status() {
10913 let mut d = doc_with("tbl_none", "just text\n");
10914 d.caret = 3;
10915 d.table_insert_row(true);
10916 assert_eq!(d.source, "just text\n", "nothing changed");
10917 assert!(d.status.is_some(), "a status explains why");
10918 assert!(!d.caret_in_table());
10919 }
10920
10921 #[test]
10922 fn enter_in_an_ordered_list_renumbers_the_following_items() {
10923 // Inserting an item mid-list left the source markers stale (`1. 2. 2. 3.`);
10924 // the renumber pass keeps them sequential, matching what the view draws.
10925 let mut d = wysiwyg_doc("enter_renumber", "1. a\n2. b\n3. c\n");
10926 d.caret = d.source.find('a').unwrap() + 1; // end of item a
10927 d.newline();
10928 d.insert("x");
10929 assert_eq!(d.source, "1. a\n2. x\n3. b\n4. c\n");
10930 }
10931
10932 #[test]
10933 fn outdent_with_nothing_to_give_back_records_no_undo_step() {
10934 for view in [View::Source, View::Wysiwyg] {
10935 let mut d = doc_in(view, "outdent_noop", "hello\n");
10936 d.caret = 2;
10937 d.outdent();
10938 assert_eq!(d.source, "hello\n");
10939 assert!(!d.dirty, "a no-op is not a modification");
10940 d.undo();
10941 assert_eq!(
10942 d.status.as_deref(),
10943 Some("nothing to undo"),
10944 "spends no undo step"
10945 );
10946 assert_eq!(d.source, "hello\n");
10947 }
10948 }
10949
10950 #[test]
10951 fn indent_shifts_every_selected_line_and_keeps_them_selected() {
10952 for view in [View::Source, View::Wysiwyg] {
10953 let mut d = doc_in(view, "indent_sel", "one\n\ntwo\n");
10954 d.anchor = Some(0);
10955 d.caret = 7; // through "two"
10956 d.indent();
10957 assert_eq!(
10958 d.source, " one\n\n two\n",
10959 "the blank line keeps no trailing pad"
10960 );
10961 // Selected, so a second Tab lands on the same lines rather than on
10962 // whatever the shifted offsets now cover.
10963 assert_eq!(d.selection(), Some((0, 12)));
10964 d.indent();
10965 assert_eq!(d.source, " one\n\n two\n");
10966 }
10967 }
10968
10969 #[test]
10970 fn outdent_takes_what_each_line_has_and_leaves_the_rest_alone() {
10971 for view in [View::Source, View::Wysiwyg] {
10972 let mut d = doc_in(view, "outdent_sel", " two\n one\nnone\n");
10973 d.anchor = Some(0);
10974 d.caret = 15;
10975 d.outdent();
10976 assert_eq!(d.source, "two\none\nnone\n");
10977 }
10978 }
10979
10980 #[test]
10981 fn a_tab_undoes_as_one_step_however_many_lines_it_moved() {
10982 for view in [View::Source, View::Wysiwyg] {
10983 let mut d = doc_in(view, "indent_undo", "one\n\ntwo\n");
10984 d.anchor = Some(0);
10985 d.caret = 7;
10986 d.indent();
10987 assert_eq!(d.source, " one\n\n two\n");
10988 d.undo();
10989 assert_eq!(d.source, "one\n\ntwo\n", "one step, not one per line");
10990 assert_eq!(
10991 d.selection(),
10992 Some((0, 7)),
10993 "with the selection it was aimed at"
10994 );
10995 d.redo();
10996 assert_eq!(d.source, " one\n\n two\n");
10997 assert_eq!(
10998 d.selection(),
10999 Some((0, 12)),
11000 "redo replays the caret the indent placed, not the one splice left"
11001 );
11002 }
11003 }
11004
11005 #[test]
11006 fn vertical_motion_keeps_the_column() {
11007 let mut d = doc_with("move", "abcd\nef\n");
11008 d.caret = 3; // "abc|d" on row 0, col 3
11009 d.move_down(false); // row 1 "ef" only has cols 0..2 -> clamps to end
11010 assert_eq!(d.caret, 7); // just after "ef"
11011 }
11012
11013 // ── goal column ──────────────────────────────────────────────────────────
11014
11015 #[test]
11016 fn vertical_motion_goal_column_survives_a_short_line() {
11017 // Regression: re-deriving the column from the clamped position on
11018 // every step permanently forgets it once a short line clamps it.
11019 // Down through "xy" (2 cols) and into "ghijkl" must return to col 4.
11020 let g = |m, f: fn(&mut Doc)| golden("goalcol", m, f);
11021 assert_eq!(
11022 g("abcd|ef\nxy\nghijkl\n", |d| {
11023 d.move_down(false); // clamps to end of "xy"
11024 d.move_down(false); // restores col 4 on the long line
11025 }),
11026 "abcdef\nxy\nghij|kl\n"
11027 );
11028 }
11029
11030 #[test]
11031 fn goal_column_state_is_set_by_vertical_motion_and_cleared_by_horizontal() {
11032 let mut d = doc_with("goalcol_state", "abcdef\nxy\nghijkl\n");
11033 assert_eq!(d.goal_col, None);
11034 d.caret = 4; // row 0, col 4
11035 d.move_down(false); // clamps into "xy"; goal stays the original col
11036 assert_eq!(d.goal_col, Some(4));
11037 assert_eq!(d.caret_pos(), (1, 2));
11038
11039 // A horizontal motion drops the goal column...
11040 d.move_left(false);
11041 assert_eq!(d.goal_col, None);
11042
11043 // ...so the next vertical motion picks up the *new* column (1), not
11044 // the stale one (4).
11045 d.move_down(false);
11046 assert_eq!(d.goal_col, Some(1));
11047 assert_eq!(d.caret_pos(), (2, 1));
11048 }
11049
11050 #[test]
11051 fn editing_clears_the_goal_column() {
11052 let mut d = doc_with("goalcol_edit", "abcdef\nxy\nghijkl\n");
11053 d.caret = 4;
11054 d.move_down(false);
11055 assert_eq!(d.goal_col, Some(4));
11056 d.insert("Z");
11057 assert_eq!(d.goal_col, None);
11058 }
11059
11060 #[test]
11061 fn vertical_motion_on_an_empty_document_is_a_no_op() {
11062 let mut d = doc_with("empty_vert", "");
11063 d.move_down(false);
11064 assert_eq!(d.caret, 0);
11065 d.move_up(false);
11066 assert_eq!(d.caret, 0);
11067 }
11068
11069 // ── the document's edges ─────────────────────────────────────────────────
11070
11071 #[test]
11072 fn vertical_motion_at_the_document_edges_runs_to_them_in_both_views() {
11073 // The reproduction, and the disagreement: Down on the last line ran to
11074 // the end of the document in the source view — by accident, an
11075 // out-of-range row clamping to the end of the string — and did nothing
11076 // whatever in the view leaf opens in. One rule now, in both.
11077 for (view, tag) in VIEWS {
11078 let mut d = doc_in(view, &format!("edge_{tag}"), "abc");
11079 d.caret = 1;
11080 d.move_down(false);
11081 assert_eq!(d.caret, 3, "{tag}: Down on the last line runs to the end");
11082 d.move_up(false);
11083 assert_eq!(d.caret, 0, "{tag}: Up on the first line runs to the start");
11084 }
11085 }
11086
11087 #[test]
11088 fn vertical_motion_at_the_edges_carries_the_column_across_the_lines_between() {
11089 // Down off the bottom is a motion like any other, so it latches a goal
11090 // column — and Up comes back to the column the caret left, not to the
11091 // one the document's end happened to be in.
11092 for (view, tag) in VIEWS {
11093 let gap = if view == View::Source { "\n" } else { "\n\n" };
11094 let src = format!("abcdef{gap}ghijkl");
11095 let mut d = doc_in(view, &format!("edge_goal_{tag}"), &src);
11096 d.caret = 2; // row 0, col 2
11097 d.move_down(false);
11098 assert_eq!(d.caret_pos().1, 2, "{tag}: Down keeps the column");
11099 d.move_down(false);
11100 assert_eq!(
11101 d.caret,
11102 src.len(),
11103 "{tag}: Down off the bottom reaches the end"
11104 );
11105 d.move_up(false);
11106 assert_eq!(
11107 d.caret_pos().1,
11108 2,
11109 "{tag}: Up returns to the column Down left"
11110 );
11111 }
11112 }
11113
11114 #[test]
11115 fn vertical_motion_with_nowhere_to_go_latches_no_goal_column() {
11116 // `goal_col.get_or_insert` ran *before* the early return at row 0, so an
11117 // Up that did nothing still armed a goal column, and the next Down aimed
11118 // at a column the caret had never been in.
11119 for (view, tag) in VIEWS {
11120 let mut d = doc_in(view, &format!("noop_goal_{tag}"), "abc\n\ndef");
11121 d.caret = 0;
11122 d.move_up(false);
11123 assert_eq!(d.caret, 0, "{tag}: already at the start");
11124 assert_eq!(d.goal_col, None, "{tag}: a no-op Up latched a goal column");
11125
11126 d.caret = d.source.len();
11127 d.move_down(false);
11128 assert_eq!(d.caret, d.source.len(), "{tag}: already at the end");
11129 assert_eq!(
11130 d.goal_col, None,
11131 "{tag}: a no-op Down latched a goal column"
11132 );
11133 }
11134 }
11135
11136 // ── soft wrap ────────────────────────────────────────────────────────────
11137 // Every other test here builds the map at 80 columns, where no fixture is
11138 // long enough to fold. A wrap is where one offset belongs to two rows at
11139 // once, and it broke everything that asks the caret what row it is on.
11140
11141 /// The wrapped fixture these cases share, folded at 12 columns into
11142 /// `one two ` / `three four ` / `five six ` / `seven eight`.
11143 fn wrapped_doc(name: &str) -> Doc {
11144 let mut d = wysiwyg_doc(name, "one two three four five six seven eight");
11145 d.build_visual(12);
11146 d
11147 }
11148
11149 #[test]
11150 fn home_and_end_work_from_a_wrapped_row() {
11151 // The reproduction: offset 19 is the `f` of "five", the first character
11152 // of the third row — and also the offset the second row ends at. It
11153 // resolved to the *second* row, so End aimed at a place the caret was
11154 // already in and did nothing, while Home walked backwards onto a row the
11155 // caret had left.
11156 let mut d = wrapped_doc("wrap_home_end");
11157 d.caret = 19;
11158 assert_eq!(
11159 d.caret_pos(),
11160 (2, 0),
11161 "the wrap boundary opens the third row"
11162 );
11163 d.move_end(false);
11164 assert_eq!(d.caret, 27, "End stalled at the wrap boundary");
11165 d.move_home(false);
11166 assert_eq!(d.caret, 19, "Home left the row the caret was on");
11167 }
11168
11169 #[test]
11170 fn end_of_a_wrapped_row_stays_put_when_pressed_again() {
11171 // The row's end is the last offset that is only ever its own: the offset
11172 // past it opens the row below, and aiming there would send a second
11173 // press on to *that* row's end, and a third to the next — End walking
11174 // down the paragraph rather than sitting where it landed.
11175 let mut d = wrapped_doc("wrap_end_twice");
11176 d.caret = 12; // inside "three", on the second row
11177 d.move_end(false);
11178 assert_eq!(
11179 d.caret, 18,
11180 "the end of `three four`, before the space the wrap ate"
11181 );
11182 assert_eq!(d.caret_pos(), (1, 10), "drawn on the row it is the end of");
11183 d.move_end(false);
11184 assert_eq!(d.caret, 18, "a second End moved the caret");
11185 d.move_home(false);
11186 assert_eq!(d.caret, 8, "Home takes the row's own start");
11187 }
11188
11189 #[test]
11190 fn vertical_motion_crosses_a_soft_wrap() {
11191 // Down aimed at the row below's column 0, an offset that resolved *up*
11192 // to the row above's end — so it landed on the offset it already had and
11193 // the caret could never leave a paragraph's first row.
11194 let mut d = wrapped_doc("wrap_down");
11195 d.caret = 0;
11196 for (want, row) in [(8, 1), (19, 2), (28, 3), (39, 3)] {
11197 d.move_down(false);
11198 assert_eq!(d.caret, want, "Down stalled");
11199 assert_eq!(d.caret_pos().0, row, "Down landed on the wrong row");
11200 }
11201 d.move_down(false);
11202 assert_eq!(d.caret, 39, "the last row's Down runs to the end and stops");
11203
11204 // ...and back up, one row per press. The goal column is the end of the
11205 // last row, past every other row's width, so each press clamps to the
11206 // row's own last offset rather than to the one that opens the next.
11207 let mut d = wrapped_doc("wrap_up");
11208 d.caret = 39;
11209 for (want, pos) in [(27, (2, 8)), (18, (1, 10)), (7, (0, 7)), (0, (0, 0))] {
11210 d.move_up(false);
11211 assert_eq!(d.caret, want, "Up stalled");
11212 assert_eq!(d.caret_pos(), pos, "Up landed on the wrong row");
11213 }
11214 }
11215
11216 #[test]
11217 fn a_kill_on_a_wrapped_row_stops_at_the_row() {
11218 // The kills take the same line Home and End do, so in WYSIWYG they take
11219 // the visual row — and a soft wrap has no newline in it to delete, so
11220 // nothing is joined by reaching the end of one.
11221 let mut d = wrapped_doc("wrap_kill");
11222 d.caret = 19; // the `f` of "five", opening the third row
11223 d.delete_to_line_end();
11224 // The space the wrap ate goes with the row it was drawn on: sparing it
11225 // would leave "four seven", two spaces where the row had been.
11226 assert_eq!(d.source, "one two three four seven eight");
11227
11228 // Backwards from the row's last caret position — which is *before* that
11229 // space, so this one survives, being on the far side of the caret.
11230 let mut d = wrapped_doc("wrap_kill_back");
11231 d.caret = 27;
11232 d.delete_to_line_start();
11233 assert_eq!(d.source, "one two three four seven eight");
11234 }
11235
11236 // ── document start / end ────────────────────────────────────────────────
11237
11238 #[test]
11239 fn move_doc_start_and_end_jump_to_the_edges() {
11240 let g = |m, f: fn(&mut Doc)| golden("doc_edges", m, f);
11241 assert_eq!(
11242 g("hello\nwor|ld\n", |d| d.move_doc_start(false)),
11243 "|hello\nworld\n"
11244 );
11245 assert_eq!(
11246 g("hel|lo\nworld\n", |d| d.move_doc_end(false)),
11247 "hello\nworld\n|"
11248 );
11249 // Already at the edge: a no-op.
11250 assert_eq!(g("|hello\n", |d| d.move_doc_start(false)), "|hello\n");
11251 assert_eq!(g("hello|\n", |d| d.move_doc_end(false)), "hello\n|");
11252 }
11253
11254 #[test]
11255 fn move_doc_start_and_end_extend_the_selection() {
11256 assert_eq!(
11257 golden("doc_edges_ext_end", "hello wor|ld\n", |d| d
11258 .move_doc_end(true)),
11259 "hello wor[ld\n|]"
11260 );
11261 assert_eq!(
11262 golden("doc_edges_ext_start", "hello wor|ld\n", |d| d
11263 .move_doc_start(true)),
11264 "[|hello wor]ld\n"
11265 );
11266 }
11267
11268 #[test]
11269 fn move_doc_start_and_end_on_an_empty_document_are_a_no_op() {
11270 let mut d = doc_with("empty_edges", "");
11271 d.move_doc_end(false);
11272 assert_eq!(d.caret, 0);
11273 d.move_doc_start(false);
11274 assert_eq!(d.caret, 0);
11275 }
11276
11277 // ── arrow collapses an active selection ─────────────────────────────────
11278
11279 #[test]
11280 fn arrow_collapses_selection_to_its_near_edge() {
11281 let mut d = doc_with("collapse", "hello world\n");
11282
11283 // Forward selection (anchor before caret): Right -> end, Left -> start.
11284 d.anchor = Some(2);
11285 d.caret = 7;
11286 d.move_right(false);
11287 assert_eq!((d.caret, d.anchor), (7, None));
11288
11289 d.anchor = Some(2);
11290 d.caret = 7;
11291 d.move_left(false);
11292 assert_eq!((d.caret, d.anchor), (2, None));
11293
11294 // Backward selection (anchor after caret): edges are the same
11295 // regardless of which end the caret started on.
11296 d.anchor = Some(7);
11297 d.caret = 2;
11298 d.move_right(false);
11299 assert_eq!((d.caret, d.anchor), (7, None));
11300
11301 d.anchor = Some(7);
11302 d.caret = 2;
11303 d.move_left(false);
11304 assert_eq!((d.caret, d.anchor), (2, None));
11305 }
11306
11307 #[test]
11308 fn arrow_with_extend_keeps_growing_the_selection() {
11309 let mut d = doc_with("collapse_extend", "hello world\n");
11310 d.anchor = Some(2);
11311 d.caret = 7;
11312 d.move_right(true); // extend: no collapse, caret steps one further
11313 assert_eq!((d.caret, d.anchor), (8, Some(2)));
11314 }
11315
11316 #[test]
11317 fn arrow_without_a_selection_moves_one_character_as_before() {
11318 let mut d = doc_with("no_collapse", "hello\n");
11319 d.caret = 2;
11320 d.move_right(false);
11321 assert_eq!(d.caret, 3);
11322 d.move_left(false);
11323 assert_eq!(d.caret, 2);
11324 }
11325
11326 /// Press Right until it stops, collecting the offsets walked through. Every
11327 /// caret bug in the WYSIWYG view shows up here as a walk that ends early:
11328 /// two stops sharing one source offset can't be moved between, so the caret
11329 /// stalls on the first of them and the walk never reaches the rest.
11330 fn walk_right(d: &mut Doc) -> Vec<usize> {
11331 let mut seen = vec![d.caret];
11332 for _ in 0..2000 {
11333 let before = d.caret;
11334 d.move_right(false);
11335 if d.caret == before {
11336 break;
11337 }
11338 seen.push(d.caret);
11339 }
11340 seen
11341 }
11342
11343 #[test]
11344 fn the_caret_crosses_a_soft_break() {
11345 // A newline inside a paragraph is a `soft_break`, which twig gives no
11346 // span of its own — the space it renders as used to borrow the offset of
11347 // the character before it, and a caret can't move without changing
11348 // offset. Right must walk clean off the end of the first line.
11349 let mut d = wysiwyg_doc("soft_break_walk", "one two\nthree four\n");
11350 d.caret = 0;
11351 let seen = walk_right(&mut d);
11352 assert_eq!(seen, (0..=18).collect::<Vec<_>>(), "walk stalled: {seen:?}");
11353 }
11354
11355 #[test]
11356 fn line_flow_preserve_resplits_the_map_and_defaults_to_fold() {
11357 // The paragraph holds one soft break. Folded (the default) it lays out as
11358 // a single reflowed row; Preserve re-lays it as a row per source line.
11359 // The setter must invalidate the cached map for the change to show, and
11360 // again on the way back — so a round trip returns to the folded layout.
11361 let mut d = wysiwyg_doc("line_flow", "one two\nthree four\n");
11362 assert_eq!(d.line_flow(), LineFlow::Fold, "fold is the default");
11363 d.build_visual(80);
11364 assert_eq!(d.vmap.num_rows(), 1, "fold: one flowing row");
11365
11366 d.set_line_flow(LineFlow::Preserve);
11367 d.build_visual(80);
11368 assert_eq!(d.vmap.num_rows(), 2, "preserve: a row per source line");
11369
11370 d.set_line_flow(LineFlow::Fold);
11371 d.build_visual(80);
11372 assert_eq!(d.vmap.num_rows(), 1, "fold again: back to one row");
11373 }
11374
11375 #[test]
11376 fn the_caret_still_crosses_a_preserved_soft_break() {
11377 // Preserve renders the soft break as a row boundary rather than a space,
11378 // but the caret must still reach every offset — the break's own offset is
11379 // the first row's end stop, so Right walks clean off the end of line one
11380 // onto line two, exactly as it does when the break is folded.
11381 let mut d = wysiwyg_doc("preserve_walk", "one two\nthree four\n");
11382 d.set_line_flow(LineFlow::Preserve);
11383 d.build_visual(80);
11384 d.caret = 0;
11385 let seen = walk_right(&mut d);
11386 assert_eq!(seen, (0..=18).collect::<Vec<_>>(), "walk stalled: {seen:?}");
11387 }
11388
11389 #[test]
11390 fn the_caret_walks_a_code_block() {
11391 // Every glyph of a code block used to map to the block's start, so the
11392 // whole block was a single offset and the caret couldn't move inside it.
11393 let src = "```rust\nlet x = 1;\nfn f() {}\n```\n";
11394 let mut d = wysiwyg_doc("code_walk", src);
11395 d.caret = 0;
11396 let seen = walk_right(&mut d);
11397 // The fences are markup: hidden, and no caret stop. The code between
11398 // them is reached a character at a time.
11399 let code = src.find("let").unwrap()..src.find("\n```").unwrap();
11400 for off in code.clone() {
11401 assert!(seen.contains(&off), "offset {off} unreachable: {seen:?}");
11402 }
11403 assert!(seen.contains(&code.end), "no stop after the last line");
11404 }
11405
11406 #[test]
11407 fn the_caret_walks_an_indented_code_block() {
11408 // An indented block's text has the four-space indent stripped, so it
11409 // isn't a verbatim slice and its lines have to be re-found. The caret
11410 // lands on the code, never in the indent.
11411 let src = " indented\n code\n";
11412 let mut d = wysiwyg_doc("indent_code_walk", src);
11413 d.caret = 0;
11414 let seen = walk_right(&mut d);
11415 assert!(seen.contains(&src.find("indented").unwrap()));
11416 assert!(seen.contains(&src.find("code").unwrap()));
11417 assert!(
11418 !seen.contains(&0) || seen[0] == 0,
11419 "the caret starts where it was put"
11420 );
11421 // Nothing in the stripped indent is a stop.
11422 for off in [1, 2, 3] {
11423 assert!(!seen.contains(&off), "landed in the indent at {off}");
11424 }
11425 }
11426
11427 #[test]
11428 fn the_caret_leaves_a_tight_heading() {
11429 // "# H" with text directly under it: the heading row's end and the
11430 // separator row's end are the same offset. Right used to find the
11431 // separator's copy, set the caret to where it already was, and stop.
11432 let mut d = wysiwyg_doc("tight_heading_walk", "# H\ntext\n");
11433 d.caret = 2; // the "H"
11434 let seen = walk_right(&mut d);
11435 assert!(
11436 seen.len() > 2,
11437 "Right stalled at the heading's end: {seen:?}"
11438 );
11439 assert!(
11440 seen.contains(&8),
11441 "never reached the end of \"text\": {seen:?}"
11442 );
11443 }
11444
11445 #[test]
11446 fn the_caret_skips_the_gap_between_two_paragraphs() {
11447 // The blank line between two paragraphs is the boundary itself. The
11448 // caret used to be able to sit on it, and typing there landed in the
11449 // previous paragraph — "A\n\nB" became "A\nx\nB", one paragraph with a
11450 // soft break, so the text visibly snapped back up.
11451 let mut d = wysiwyg_doc("gap_skip", "A\n\nB\n");
11452 d.caret = 1; // the end of "A"
11453 d.move_right(false);
11454 assert_eq!(d.caret, 3, "Right stopped in the gap");
11455 d.insert("x");
11456 assert_eq!(d.source, "A\n\nxB\n", "typing landed outside B");
11457 }
11458
11459 #[test]
11460 fn down_from_a_paragraph_lands_on_the_next_one() {
11461 let mut d = wysiwyg_doc("gap_down", "A\n\nB\n");
11462 d.caret = 0;
11463 d.move_down(false);
11464 assert_eq!(d.caret, 3, "Down stopped in the gap");
11465 }
11466
11467 #[test]
11468 fn clicking_the_gap_lands_on_real_text() {
11469 // A click can still *reach* the gap — it's drawn, so it's clickable.
11470 // It has to resolve to somewhere the caret can be.
11471 let mut d = wysiwyg_doc("gap_click", "A\n\nB\n");
11472 d.click(1, 0, false); // the gap row
11473 assert!(
11474 d.caret == 1 || d.caret == 3,
11475 "click left the caret in the gap at {}",
11476 d.caret
11477 );
11478 d.insert("x");
11479 // Either edge of the boundary is a fair place to land; inside it isn't.
11480 assert!(
11481 d.source == "Ax\n\nB\n" || d.source == "A\n\nxB\n",
11482 "click in the gap typed into the boundary: {:?}",
11483 d.source
11484 );
11485 }
11486
11487 #[test]
11488 fn enter_opens_an_empty_paragraph_the_caret_can_type_into() {
11489 // Enter inserts a paragraph break, which leaves a blank line spare on
11490 // either side of a new one. That middle line is a real empty paragraph:
11491 // the caret lands there, and typing makes a paragraph rather than
11492 // extending a neighbour.
11493 let mut d = wysiwyg_doc("gap_enter", "A\n\nB\n");
11494 d.caret = 1;
11495 d.newline();
11496 assert_eq!(d.source, "A\n\n\n\nB\n");
11497 d.build_visual(80);
11498 let (row, _) = d.caret_pos();
11499 assert!(
11500 d.vmap.row_is_navigable(row),
11501 "the caret landed on a gap row"
11502 );
11503 d.insert("x");
11504 assert_eq!(
11505 d.source, "A\n\nx\n\nB\n",
11506 "the new paragraph merged into a neighbour"
11507 );
11508 }
11509
11510 #[test]
11511 fn enter_at_the_end_of_the_document_opens_a_paragraph_too() {
11512 let mut d = wysiwyg_doc("gap_eof", "A\n");
11513 d.caret = 1;
11514 d.newline();
11515 d.build_visual(80);
11516 let (row, _) = d.caret_pos();
11517 assert!(
11518 d.vmap.row_is_navigable(row),
11519 "the caret landed on a gap row"
11520 );
11521 d.insert("x");
11522 assert!(
11523 d.source.starts_with("A\n\n") && d.source.contains('x'),
11524 "typing at the end merged into A: {:?}",
11525 d.source
11526 );
11527 }
11528
11529 #[test]
11530 fn triple_click_selects_a_paragraph_across_its_soft_breaks() {
11531 // A paragraph broken over two source lines is one paragraph. Selecting
11532 // it must not stop at the newline inside it — that newline is markup the
11533 // rich-text view exists to hide.
11534 let src = "one two\nthree four\n\nnext\n";
11535 let mut d = wysiwyg_doc("triple_para", src);
11536 d.select_block_at(2);
11537 assert_eq!(
11538 d.selected_text(),
11539 Some("one two\nthree four"),
11540 "stopped at the soft break"
11541 );
11542 }
11543
11544 #[test]
11545 fn the_wheel_can_scroll_away_from_a_caret_that_stays_put() {
11546 // The reader scrolls down past the caret's row. Nothing moved the
11547 // caret, so the view must stay where it was put — the old code revealed
11548 // the caret every frame, which dragged the view straight back and made
11549 // the document unscrollable past the caret.
11550 let mut d = wysiwyg_doc("scroll_free", "a\n\nb\n\nc\n\nd\n\ne\n");
11551 d.caret = 0;
11552 d.follow_caret(0, 3, 9); // first frame: the caret is at the top
11553 d.scroll = 4; // the wheel
11554 d.follow_caret(0, 3, 9);
11555 assert_eq!(
11556 d.scroll, 4,
11557 "the wheel was overruled by a caret that never moved"
11558 );
11559 }
11560
11561 #[test]
11562 fn moving_the_caret_brings_the_view_back_to_it() {
11563 let mut d = wysiwyg_doc("scroll_follow", "a\n\nb\n\nc\n\nd\n\ne\n");
11564 d.caret = 0;
11565 d.follow_caret(0, 3, 9);
11566 d.scroll = 6; // scrolled away
11567 d.move_right(false); // ...and now the caret moves
11568 let (row, _) = d.caret_pos();
11569 d.follow_caret(row, 3, 9);
11570 assert!(
11571 d.scroll <= row && row < d.scroll + 3,
11572 "caret row {row} off screen at scroll {}",
11573 d.scroll
11574 );
11575 }
11576
11577 #[test]
11578 fn scrolling_stops_at_the_last_row() {
11579 let mut d = wysiwyg_doc("scroll_clamp", "a\n\nb\n");
11580 d.caret = 0;
11581 d.follow_caret(0, 3, 3); // a first frame, so the caret isn't "new"
11582 d.scroll = 999; // the wheel, spun hard
11583 d.follow_caret(0, 3, 3);
11584 assert_eq!(d.scroll, 2, "scrolled into the void past the document");
11585 }
11586
11587 #[test]
11588 fn every_cell_of_a_wide_table_is_reachable() {
11589 // A table whose cells are far wider than the surface: the columns are
11590 // cut to fit and the text wraps inside them, so no cell hangs off the
11591 // right edge where the caret can never go.
11592 let src = "| Ingredient | Notes |\n|---|---|\n\
11593 | flour milled coarse | sift it twice before folding it in |\n";
11594 let mut d = wysiwyg_doc("wide_table_walk", src);
11595 d.build_visual(30);
11596 d.caret = 0;
11597 let seen = walk_right(&mut d);
11598 for word in ["Ingredient", "Notes", "coarse", "folding"] {
11599 let at = src.find(word).unwrap();
11600 assert!(seen.contains(&at), "{word:?} at {at} unreachable: {seen:?}");
11601 }
11602 }
11603
11604 // ── view parity ──────────────────────────────────────────────────────────
11605 // `doc_with` pins the source view, so everything above tests a view users
11606 // never start in — `Doc::open` opens in WYSIWYG. These run the motion and
11607 // deletion golden cases through *both*, plus the WYSIWYG cases the two
11608 // can't share: where the source carries markup the rendered text is a
11609 // different string, and the views agreeing would itself be the bug.
11610
11611 const VIEWS: [(View, &str); 2] = [(View::Source, "source"), (View::Wysiwyg, "wysiwyg")];
11612
11613 /// Run `action` in both views on one `|`-marked fixture and assert they
11614 /// agree. Plain prose only: with no markup to hide, WYSIWYG renders the
11615 /// source verbatim, so the two views are looking at the same text and any
11616 /// disagreement is one of them having lost the plot.
11617 fn both_views(name: &str, marked: &str, action: fn(&mut Doc)) -> String {
11618 let (src, caret) = parse_caret(marked);
11619 let run = |view: View, tag: &str| {
11620 let mut d = doc_in(view, &format!("{name}_{tag}"), &src);
11621 d.caret = caret;
11622 action(&mut d);
11623 render_caret(&d)
11624 };
11625 let source = run(VIEWS[0].0, VIEWS[0].1);
11626 let wysiwyg = run(VIEWS[1].0, VIEWS[1].1);
11627 assert_eq!(source, wysiwyg, "the views disagree on {marked:?}");
11628 source
11629 }
11630
11631 #[test]
11632 fn word_motion_agrees_across_the_views_on_plain_prose() {
11633 let g = both_views;
11634 assert_eq!(
11635 g("par_wl", "hello wor|ld", |d| d.move_word_left(false)),
11636 "hello |world"
11637 );
11638 assert_eq!(
11639 g("par_wl2", "hello| world", |d| d.move_word_left(false)),
11640 "|hello world"
11641 );
11642 assert_eq!(
11643 g("par_wr", "hel|lo world", |d| d.move_word_right(false)),
11644 "hello| world"
11645 );
11646 assert_eq!(
11647 g("par_wr2", "hello| world", |d| d.move_word_right(false)),
11648 "hello world|"
11649 );
11650 assert_eq!(
11651 g("par_punct", "|foo.bar", |d| d.move_word_right(false)),
11652 "foo|.bar"
11653 );
11654 assert_eq!(
11655 g("par_ext", "hello |world", |d| d.move_word_right(true)),
11656 "hello [world|]"
11657 );
11658 }
11659
11660 #[test]
11661 fn word_deletion_agrees_across_the_views_on_plain_prose() {
11662 let g = both_views;
11663 assert_eq!(
11664 g("par_db", "hello world|", |d| d.delete_word_back()),
11665 "hello |"
11666 );
11667 assert_eq!(
11668 g("par_df", "hello |world", |d| d.delete_word_forward()),
11669 "hello |"
11670 );
11671 assert_eq!(
11672 g("par_db2", "foo |bar baz", |d| d.delete_word_back()),
11673 "|bar baz"
11674 );
11675 assert_eq!(g("par_utf8", "café |ok", |d| d.delete_word_back()), "|ok");
11676 }
11677
11678 #[test]
11679 fn character_motion_and_deletion_agree_across_the_views_on_plain_prose() {
11680 let g = both_views;
11681 assert_eq!(g("par_r", "he|llo", |d| d.move_right(false)), "hel|lo");
11682 assert_eq!(g("par_l", "he|llo", |d| d.move_left(false)), "h|ello");
11683 assert_eq!(g("par_bs", "hel|lo", |d| d.backspace()), "he|lo");
11684 assert_eq!(g("par_del", "hel|lo", |d| d.delete_forward()), "hel|o");
11685 }
11686
11687 #[test]
11688 fn wysiwyg_motion_steps_a_grapheme_cluster_the_way_the_source_view_does() {
11689 // The reproduction: the stop table was built one stop per `char`, so
11690 // Right parked the caret 4 bytes into a ZWJ sequence — a place the
11691 // source view, which steps by grapheme, can't reach and backspace can't
11692 // survive. The two views must land on the same offset.
11693 let family = "👨👩👧"; // three emoji strung together with joiners: one cluster
11694 for (view, tag) in VIEWS {
11695 let mut d = doc_in(view, &format!("cluster_{tag}"), &format!("a{family}b\n"));
11696 d.caret = 1;
11697 d.move_right(false);
11698 assert_eq!(d.caret, 1 + family.len(), "{tag} parked inside the cluster");
11699
11700 // ...and the edit that used to sever a joiner off the front of it.
11701 d.backspace();
11702 assert_eq!(d.source, "ab\n", "{tag} split the cluster");
11703 assert_eq!(d.caret, 1);
11704 }
11705 }
11706
11707 #[test]
11708 fn wysiwyg_motion_treats_a_combining_accent_as_one_character() {
11709 for (view, tag) in VIEWS {
11710 let mut d = doc_in(view, &format!("combining_{tag}"), "e\u{0301}x\n");
11711 d.caret = 0;
11712 d.move_right(false);
11713 assert_eq!(
11714 d.caret,
11715 "e\u{0301}".len(),
11716 "{tag} stopped on the combining mark"
11717 );
11718 }
11719 }
11720
11721 #[test]
11722 fn no_wysiwyg_motion_can_park_the_caret_inside_a_cluster() {
11723 // The general form: whatever route the caret takes through a document
11724 // full of clusters, it never lands between the codepoints of one — so no
11725 // motion-then-backspace sequence can leave a dangling joiner behind.
11726 use unicode_segmentation::UnicodeSegmentation;
11727
11728 let src = "a👨👩👧b e\u{0301}mo👨👩👧ji\n\nnext 👩🚀 line\n";
11729 let mut d = wysiwyg_doc("cluster_walk", src);
11730 d.caret = 0;
11731 let boundaries: Vec<usize> = src
11732 .grapheme_indices(true)
11733 .map(|(i, _)| i)
11734 .chain(std::iter::once(src.len()))
11735 .collect();
11736 for off in walk_right(&mut d) {
11737 assert!(
11738 boundaries.contains(&off),
11739 "Right stopped at {off}, inside a grapheme cluster"
11740 );
11741 }
11742 }
11743
11744 #[test]
11745 fn wysiwyg_word_motion_stays_out_of_hidden_delimiters() {
11746 // The reproduction: ⌥→ from inside the opening `**` computed its
11747 // boundary over the raw source and landed on byte 8 — inside the
11748 // *closing* `**`, which `caret_pos` draws at column 6, immediately after
11749 // "bold". The caret drew past the bold word and sat inside it.
11750 let mut d = wysiwyg_doc("wys_word_delim", "a **bold** c\n");
11751 d.caret = 2;
11752 d.move_word_right(false);
11753 assert!(
11754 d.vmap.is_stop(d.caret),
11755 "landed at {}, not a caret stop",
11756 d.caret
11757 );
11758 assert_eq!(d.caret, 10, "should land on the space after \"bold\"");
11759 // The rendered row is "a bold c": column 6 is the space just past "bold",
11760 // and now the caret is really there rather than only drawn there.
11761 assert_eq!(d.caret_pos(), (0, 6));
11762
11763 // ...and back again: ⌥← returns to the "b", not into the opening `**`.
11764 d.move_word_left(false);
11765 assert_eq!(d.caret, 4);
11766 assert_eq!(d.caret_pos(), (0, 2));
11767 }
11768
11769 #[test]
11770 fn wysiwyg_word_delete_takes_the_markup_with_the_word() {
11771 // The reproduction: ⌥⌫ from after "bold" walked the raw source, stopped
11772 // inside the closing `**`, and left "a ** c\n" — delimiters with no
11773 // opener. Glyph space covers the word alone, which would leave
11774 // "a **** c": markup wrapped around nothing. The word and the styling
11775 // that was only ever the word's go together.
11776 let mut d = wysiwyg_doc("wys_word_del_back", "a **bold** c\n");
11777 d.caret = 10;
11778 d.delete_word_back();
11779 assert_eq!(d.source, "a c\n");
11780 assert_eq!(d.caret, 2);
11781
11782 let mut d = wysiwyg_doc("wys_word_del_fwd", "a **bold** c\n");
11783 d.caret = 4; // the "b"
11784 d.delete_word_forward();
11785 assert_eq!(d.source, "a c\n");
11786 }
11787
11788 #[test]
11789 fn wysiwyg_word_delete_empties_a_nested_mark_and_a_code_span_too() {
11790 let src = "a ***bold*** c\n";
11791 let mut d = wysiwyg_doc("wys_word_del_nest", src);
11792 d.caret = src.find(" c").unwrap();
11793 d.delete_word_back();
11794 assert_eq!(
11795 d.source, "a c\n",
11796 "the emph inside the strong empties it too"
11797 );
11798
11799 let src = "a `code` c\n";
11800 let mut d = wysiwyg_doc("wys_word_del_code", src);
11801 d.caret = src.find(" c").unwrap();
11802 d.delete_word_back();
11803 assert_eq!(d.source, "a c\n");
11804 }
11805
11806 #[test]
11807 fn wysiwyg_word_delete_keeps_a_mark_that_still_has_text() {
11808 // Only an *emptied* node goes. Take one word of two and the `**` still
11809 // has a job to do — over the word that's left, with the space the delete
11810 // pushed against the opening delimiter moved out in front of it, or the
11811 // run would be no run at all (`** words**` is literal asterisks — see
11812 // the mark-edge rule on `splice`).
11813 let src = "a **two words** c\n";
11814 let mut d = wysiwyg_doc("wys_word_del_partial", src);
11815 d.caret = src.find(" words").unwrap();
11816 d.delete_word_back();
11817 assert_eq!(d.source, "a **words** c\n");
11818 }
11819
11820 #[test]
11821 fn source_view_word_motion_still_walks_the_markup() {
11822 // The other half of the decision: in the source view the `**` are
11823 // characters like any other — they're on the screen, so word motion has
11824 // to stop at them and a word-delete has to leave them behind. Only
11825 // WYSIWYG hides them, so only WYSIWYG steps over them.
11826 let g = |n, m, f: fn(&mut Doc)| golden(n, m, f);
11827 assert_eq!(
11828 g("src_word_motion", "a |**bold** c\n", |d| d
11829 .move_word_right(false)),
11830 "a **bold|** c\n"
11831 );
11832 // The same caret as the WYSIWYG reproduction, and the opposite outcome:
11833 // here "a ** c\n" is right, because `bold**` is what's to the left of it.
11834 assert_eq!(
11835 g("src_word_del", "a **bold**| c\n", |d| d.delete_word_back()),
11836 "a **| c\n"
11837 );
11838 }
11839
11840 #[test]
11841 fn every_wysiwyg_motion_lands_on_a_caret_stop() {
11842 // The single invariant both bugs violated: the caret draws and edits at
11843 // the same place only when it's on a stop. `debug_assert_on_a_stop`
11844 // makes the same claim in-place; this pins it from the outside, over a
11845 // document with every kind of thing the map has to be careful about.
11846 // At two widths: the wide one every other test builds at, where no
11847 // fixture folds, and one narrow enough that they all do. A soft wrap is
11848 // where an offset stops being on exactly one row, and testing only the
11849 // width that never wraps is how the caret came to be pinned at the first
11850 // one Down reached.
11851 let src = "# Title\n\na **bold** e\u{0301}mo👨👩👧ji `x` c\n\n\
11852 - item one\n\n| A | B |\n|---|---|\n| x | y |\n";
11853 // A table of named operations, which is what it looks like.
11854 #[allow(clippy::type_complexity)]
11855 let motions: [(&str, fn(&mut Doc)); 8] = [
11856 ("right", |d| d.move_right(false)),
11857 ("left", |d| d.move_left(false)),
11858 ("word_right", |d| d.move_word_right(false)),
11859 ("word_left", |d| d.move_word_left(false)),
11860 ("down", |d| d.move_down(false)),
11861 ("up", |d| d.move_up(false)),
11862 ("home", |d| d.move_home(false)),
11863 ("end", |d| d.move_end(false)),
11864 ];
11865 for width in [80, 12] {
11866 let mut d = wysiwyg_doc("stop_invariant", src);
11867 d.build_visual(width);
11868 let stops: Vec<usize> = (0..=src.len()).filter(|&o| d.vmap.is_stop(o)).collect();
11869 assert!(stops.len() > 20, "fixture should have plenty of stops");
11870 for start in stops {
11871 for (name, motion) in &motions {
11872 d.caret = start;
11873 d.anchor = None;
11874 motion(&mut d);
11875 assert!(
11876 d.vmap.is_stop(d.caret),
11877 "{name} from {start} at width {width} landed at {} — not a caret stop",
11878 d.caret
11879 );
11880 }
11881 }
11882 }
11883 }
11884
11885 #[test]
11886 fn no_wysiwyg_motion_is_a_dead_end() {
11887 // Down held to the bottom of a document reaches the bottom, and Up held
11888 // to the top reaches the top — from anywhere, at a width that wraps. The
11889 // invariant above says a motion lands somewhere legal; this one says it
11890 // gets somewhere at all, which is what a caret pinned at a wrap boundary
11891 // was quietly failing to do while every assertion around it held.
11892 let src = "# Title\n\none two three four five six seven eight nine ten\n\n\
11893 - item one two three four five\n\nlast\n";
11894 for width in [80, 12] {
11895 let mut d = wysiwyg_doc("no_dead_end", src);
11896 d.build_visual(width);
11897 let stops: Vec<usize> = (0..=src.len()).filter(|&o| d.vmap.is_stop(o)).collect();
11898 let (first, last) = (stops[0], stops[stops.len() - 1]);
11899 for &start in &stops {
11900 for (name, motion, want) in [
11901 (
11902 "down",
11903 (|d: &mut Doc| d.move_down(false)) as fn(&mut Doc),
11904 last,
11905 ),
11906 ("up", |d: &mut Doc| d.move_up(false), first),
11907 ] {
11908 d.caret = start;
11909 d.anchor = None;
11910 d.goal_col = None;
11911 // Every row, plus the presses the edges take, plus slack.
11912 for _ in 0..d.vmap.num_rows() + 4 {
11913 motion(&mut d);
11914 }
11915 assert_eq!(
11916 d.caret, want,
11917 "{name} held from {start} at width {width} never arrived"
11918 );
11919 }
11920 }
11921 }
11922 }
11923 // ── display columns ──────────────────────────────────────────────────────
11924 // A `col` is a terminal cell, not a character. The two are the same number
11925 // for the ASCII the fixtures above are written in, which is how they came
11926 // apart in the first place: `你` is one character drawn in two cells, so a
11927 // column counted in characters names a cell the text isn't in — one earlier
11928 // for every wide character to its left.
11929
11930 #[test]
11931 fn a_wide_character_is_two_columns_wide() {
11932 // The reproduction: `你` is one char and two cells, so the caret just
11933 // past it drew at column 1 — inside the character it had already left.
11934 for (view, tag) in VIEWS {
11935 let mut d = doc_in(view, &format!("wide_col_{tag}"), "你好\n");
11936 d.caret = "你".len();
11937 assert_eq!(d.caret_pos(), (0, 2), "{tag}: caret drew inside 你");
11938 d.caret = "你好".len();
11939 assert_eq!(d.caret_pos(), (0, 4), "{tag}");
11940 }
11941 }
11942
11943 #[test]
11944 fn a_cluster_is_as_wide_as_it_is_drawn_not_as_its_codepoints_measure() {
11945 // `👨👩👧` is five codepoints — two-cell, joiner, two-cell, joiner,
11946 // two-cell — measuring six cells one at a time, but the character they
11947 // spell is drawn in two. Width belongs to the cluster, not the glyph,
11948 // and the frontends measure it the same way.
11949 let family = "👨👩👧";
11950 for (view, tag) in VIEWS {
11951 let src = format!("a{family}b\n");
11952 let mut d = doc_in(view, &format!("wide_cluster_{tag}"), &src);
11953 d.caret = 1 + family.len();
11954 assert_eq!(
11955 d.caret_pos(),
11956 (0, 3),
11957 "{tag}: 'a' is one cell, the family two"
11958 );
11959 }
11960 }
11961
11962 #[test]
11963 fn both_cells_of_a_wide_character_mean_the_character() {
11964 // Clicking the far half of `好` is still clicking `好`: half a character
11965 // is not a place the caret can be, so it comes to rest at the
11966 // character's start — the column it would have been drawn at anyway.
11967 for (view, tag) in VIEWS {
11968 let mut d = doc_in(view, &format!("wide_click_{tag}"), "你好\n");
11969 for col in [2, 3] {
11970 d.caret = 0;
11971 d.click(0, col, false);
11972 assert_eq!(d.caret, "你".len(), "{tag}: click at col {col}");
11973 assert_eq!(d.caret_pos(), (0, 2), "{tag}: click at col {col}");
11974 }
11975 // Past the last cell is the line's end, as it is for ASCII.
11976 d.click(0, 9, false);
11977 assert_eq!(d.caret, "你好".len(), "{tag}: click past the end");
11978 }
11979 }
11980
11981 #[test]
11982 fn every_offset_survives_the_trip_out_to_a_column_and_back() {
11983 // The mapping is only a mapping if it inverts: the cell the caret is
11984 // drawn in has to be the cell that brings it back to the same offset.
11985 // Over a fixture where a character may be one cell or two, and one
11986 // codepoint or five.
11987 use unicode_segmentation::UnicodeSegmentation;
11988
11989 let src = "ab 你好 c\n\n👨👩👧 e\u{0301}x 漢字\n\nplain ascii\n";
11990
11991 let mut d = doc_in(View::Source, "roundtrip_source", src);
11992 // Every offset the source view's caret can occupy: it steps by grapheme
11993 // cluster, so those are its boundaries.
11994 for (off, _) in src
11995 .grapheme_indices(true)
11996 .chain(std::iter::once((src.len(), "")))
11997 {
11998 d.caret = off;
11999 let (row, col) = d.caret_pos();
12000 d.click(row, col, false);
12001 assert_eq!(d.caret, off, "source: {off} → ({row}, {col}) → {}", d.caret);
12002 }
12003
12004 // And in WYSIWYG, where the offsets the caret can occupy are the map's
12005 // stops rather than every boundary.
12006 let mut d = doc_in(View::Wysiwyg, "roundtrip_wysiwyg", src);
12007 let stops: Vec<usize> = (0..=src.len()).filter(|&o| d.vmap.is_stop(o)).collect();
12008 assert!(stops.len() > 20, "fixture should have plenty of stops");
12009 for off in stops {
12010 d.caret = off;
12011 let (row, col) = d.caret_pos();
12012 d.click(row, col, false);
12013 assert_eq!(
12014 d.caret, off,
12015 "wysiwyg: {off} → ({row}, {col}) → {}",
12016 d.caret
12017 );
12018 }
12019 }
12020
12021 #[test]
12022 fn vertical_motion_aims_at_a_column_the_reader_can_see() {
12023 // Down from under `世` lands under the glyph in that cell, not two
12024 // characters further along the line. The goal is a column, so a line of
12025 // wide characters and a line of ASCII line up the way they're drawn.
12026 //
12027 // The gap differs by view: a bare newline inside a paragraph is a soft
12028 // break, which WYSIWYG draws as a space on a single row. The views share
12029 // a grid only where the source's lines are the renderer's rows too.
12030 for (view, tag) in VIEWS {
12031 let gap = if view == View::Source { "\n" } else { "\n\n" };
12032 let src = format!("你好世{gap}abcdef\n");
12033 let mut d = doc_in(view, &format!("goal_wide_{tag}"), &src);
12034 d.caret = "你好".len();
12035 assert_eq!(d.caret_pos().1, 4, "{tag}: `世` is drawn at column 4");
12036 d.move_down(false);
12037 assert_eq!(d.caret_pos().1, 4, "{tag}: goal column lost");
12038 assert!(
12039 d.source[d.caret..].starts_with('e'),
12040 "{tag}: landed on the wrong glyph"
12041 );
12042 }
12043 }
12044
12045 #[test]
12046 fn a_goal_column_landing_inside_a_wide_character_lands_on_it() {
12047 // Down from column 3 onto `你好`, whose characters start at columns 0
12048 // and 2: column 3 is the *second* cell of `好`. There is nowhere to be
12049 // between the cells of one character, so the caret rests on it — and on
12050 // its start, which is the only offset there that is a caret stop.
12051 for (view, tag) in VIEWS {
12052 let gap = if view == View::Source { "\n" } else { "\n\n" };
12053 let src = format!("abcdef{gap}你好\n");
12054 let mut d = doc_in(view, &format!("goal_inside_{tag}"), &src);
12055 let line = src.find('你').unwrap();
12056 d.caret = 3;
12057 d.move_down(false);
12058 assert_eq!(d.caret, line + "你".len(), "{tag}: landed off `好`'s start");
12059 assert_eq!(d.caret_pos().1, 2, "{tag}: drew between `好`'s cells");
12060 }
12061 }
12062
12063 #[test]
12064 fn a_caret_in_a_table_cell_of_wide_text_draws_where_the_text_is() {
12065 // The column the cell's text is laid out in is measured in cells, so the
12066 // caret walking that text has to be too — the two agreeing is the whole
12067 // point of the grid staying square.
12068 let mut d = wysiwyg_doc("table_wide", "| A | B |\n|---|---|\n| 你好 | y |\n");
12069 let at = d.source.find("你").unwrap();
12070 d.caret = at;
12071 let (row, col) = d.caret_pos();
12072 // `│ ` opens the row, so the cell's text starts at column 2; `好` is two
12073 // cells further along.
12074 assert_eq!(col, 2, "the cell's first character");
12075 d.move_right(false);
12076 assert_eq!(
12077 d.caret_pos(),
12078 (row, 4),
12079 "`好` is drawn past `你`'s two cells"
12080 );
12081 assert_eq!(d.caret, at + "你".len());
12082 }
12083
12084 // ── active inline marks ───────────────────────────────────────────────────
12085
12086 /// The marks at a `|`-marked fixture's caret, in `InlineMarks::iter` order.
12087 fn marks(view: View, name: &str, marked: &str) -> Vec<InlineKind> {
12088 let (src, caret) = parse_caret(marked);
12089 let mut d = doc_in(view, name, &src);
12090 d.caret = caret;
12091 d.active_inline_marks().iter().collect()
12092 }
12093
12094 /// The marks over the selection `[start, end)`.
12095 fn marks_over(view: View, name: &str, src: &str, start: usize, end: usize) -> Vec<InlineKind> {
12096 let mut d = doc_in(view, name, src);
12097 d.anchor = Some(start);
12098 d.caret = end;
12099 d.active_inline_marks().iter().collect()
12100 }
12101
12102 #[test]
12103 fn a_caret_in_a_mark_reports_it() {
12104 for (view, tag) in VIEWS {
12105 let m = |marked| marks(view, &format!("marks_in_{tag}"), marked);
12106 assert_eq!(m("a **bo|ld** b"), [InlineKind::Strong], "{tag}");
12107 assert_eq!(m("a *it|alic* b"), [InlineKind::Emph], "{tag}");
12108 assert_eq!(m("a `co|de` b"), [InlineKind::Verbatim], "{tag}");
12109 // Plain text under no mark lights nothing — the toolbar's resting state.
12110 assert_eq!(m("a| **bold** b"), [], "{tag}");
12111 assert!(m("plain t|ext").is_empty(), "{tag}");
12112 }
12113 }
12114
12115 #[test]
12116 fn nested_marks_all_report() {
12117 // Bold *and* italic: a toolbar lights both buttons, so the set has both —
12118 // the ancestor chain is a chain, and every mark on it is in force.
12119 for (view, tag) in VIEWS {
12120 assert_eq!(
12121 marks(
12122 view,
12123 &format!("marks_nested_{tag}"),
12124 "**bold and *bo|th*** end"
12125 ),
12126 [InlineKind::Strong, InlineKind::Emph],
12127 "{tag}"
12128 );
12129 }
12130 }
12131
12132 #[test]
12133 fn the_caret_at_a_marks_edge_reports_it_where_typing_would_extend_it() {
12134 // The offsets a WYSIWYG caret actually reaches at a bold run's edges are
12135 // the first byte of its text and the byte after its last — both inside
12136 // the mark's span, both places typing lands inside the bold. The offset
12137 // past the closing delimiter is the next text, and reports nothing.
12138 let src = "a **bold** b";
12139 let inner_start = src.find("bold").unwrap(); // 4
12140 let inner_end = inner_start + "bold".len(); // 8, on the closing `**`
12141 for (view, tag) in VIEWS {
12142 let mut d = doc_in(view, &format!("marks_edge_{tag}"), src);
12143 for off in [2, 3, inner_start, inner_end, 9] {
12144 d.caret = off;
12145 assert!(
12146 d.active_inline_marks().contains(InlineKind::Strong),
12147 "{tag}: offset {off} is inside the strong span"
12148 );
12149 }
12150 for off in [0, 1, 10, 11, 12] {
12151 d.caret = off;
12152 assert!(
12153 !d.active_inline_marks().contains(InlineKind::Strong),
12154 "{tag}: offset {off} is outside the strong run"
12155 );
12156 }
12157 }
12158 }
12159
12160 #[test]
12161 fn a_mark_ends_the_same_way_at_the_end_of_the_buffer_as_in_the_middle() {
12162 // Regression: twig resolves an offset that is one node's end and the
12163 // next one's start to the node that *starts* there, so `**bold**|\n`
12164 // isn't bold. With nothing following there's no tie to break and the
12165 // chain still ended at the mark, which made a trailing `\n` — not the
12166 // text — decide whether the caret after a bold word reported bold. It's
12167 // the offset past the mark either way, and typing there is plain either
12168 // way. A blank document typed into is exactly this shape.
12169 for (view, tag) in VIEWS {
12170 let m = |name: String, marked| marks(view, &name, marked);
12171 assert_eq!(
12172 m(format!("marks_eob_{tag}"), "**bold**|"),
12173 [],
12174 "{tag}: no trailing newline"
12175 );
12176 assert_eq!(
12177 m(format!("marks_eol_{tag}"), "**bold**|\n"),
12178 [],
12179 "{tag}: with one"
12180 );
12181 // And the last offset that *is* in the mark still is.
12182 assert_eq!(
12183 m(format!("marks_eob_in_{tag}"), "**bold*|*"),
12184 [InlineKind::Strong],
12185 "{tag}"
12186 );
12187 }
12188 }
12189
12190 #[test]
12191 fn a_selection_reports_a_mark_only_when_it_covers_the_whole_thing() {
12192 let src = "a **bold** b";
12193 let (b, d_) = (src.find("bold").unwrap(), src.find("bold").unwrap() + 4);
12194 for (view, tag) in VIEWS {
12195 let m = |s, e| marks_over(view, &format!("marks_sel_{tag}"), src, s, e);
12196 // The whole bold word, and a slice of it.
12197 assert_eq!(m(b, d_), [InlineKind::Strong], "{tag}: the whole word");
12198 assert_eq!(m(b + 1, d_ - 1), [InlineKind::Strong], "{tag}: a slice");
12199 // Ending exactly at the closing delimiter's start is still all-bold:
12200 // an exclusive end sits *past* the last selected character, so the
12201 // question is asked of the character, not the boundary.
12202 assert_eq!(
12203 m(b, d_ + 2),
12204 [InlineKind::Strong],
12205 "{tag}: through the close"
12206 );
12207 // Half in, half out: Bold lit here would claim a press turns it off.
12208 assert_eq!(m(0, d_), [], "{tag}: leading plain text");
12209 assert_eq!(m(b, src.len()), [], "{tag}: trailing plain text");
12210 }
12211 }
12212
12213 #[test]
12214 fn a_selection_across_two_runs_of_the_same_mark_reports_nothing() {
12215 // Both ends are bold, but the space between them isn't — two runs are two
12216 // nodes, which is exactly what the node id catches and a kind-only
12217 // comparison would not.
12218 let src = "**one** **two**";
12219 for (view, tag) in VIEWS {
12220 let m = marks_over(view, &format!("marks_runs_{tag}"), src, 2, 13);
12221 assert_eq!(m, [], "{tag}: `one** **two` is not all bold");
12222 }
12223 }
12224
12225 #[test]
12226 fn marks_read_the_document_as_it_is_edited() {
12227 // The point of asking twig every frame instead of caching: the answer has
12228 // to follow the toggle that changed it.
12229 let mut d = wysiwyg_doc("marks_live", "one two\n");
12230 d.anchor = Some(0);
12231 d.caret = 3;
12232 assert!(d.active_inline_marks().is_empty(), "plain to start");
12233 d.toggle(InlineKind::Strong);
12234 assert_eq!(d.source, "**one** two\n");
12235 // `toggle` leaves the bolded text selected, so the button it lit stays lit.
12236 assert!(d.active_inline_marks().contains(InlineKind::Strong));
12237 d.toggle(InlineKind::Strong);
12238 assert!(d.active_inline_marks().is_empty(), "and off again");
12239 }
12240
12241 #[test]
12242 fn a_link_is_not_an_inline_mark() {
12243 // `link`/`str` are inline nodes, but nothing on the inline toolbar
12244 // toggles them — a set with a "link mark" in it would have no button.
12245 for (view, tag) in VIEWS {
12246 assert_eq!(
12247 marks(view, &format!("marks_link_{tag}"), "a [te|xt](u) b"),
12248 [],
12249 "{tag}"
12250 );
12251 }
12252 }
12253
12254 // ── blank documents ───────────────────────────────────────────────────────
12255
12256 #[test]
12257 fn a_blank_document_is_untitled_empty_and_markdown() {
12258 let mut d = Doc::blank().unwrap();
12259 assert!(d.is_untitled());
12260 assert_eq!(d.path, PathBuf::new());
12261 assert_eq!(
12262 d.file_name(),
12263 "untitled",
12264 "the header has to show something"
12265 );
12266 assert_eq!(d.format_name(), "markdown");
12267 assert_eq!(d.source, "");
12268 assert!(!d.dirty, "nothing typed yet is nothing to lose");
12269 assert_eq!(d.disk_state(), DiskState::Untitled);
12270 // And it's a document you can be in: the default view renders it.
12271 d.build_visual(80);
12272 assert_eq!(d.caret, 0);
12273 }
12274
12275 #[test]
12276 fn saving_an_untitled_document_asks_for_a_name_instead_of_writing() {
12277 let mut d = Doc::blank().unwrap();
12278 d.insert("hello");
12279 assert!(d.dirty);
12280 d.save();
12281 assert_eq!(d.status.as_deref(), Some("untitled — save as…"));
12282 assert!(d.dirty, "it must not come away believing it saved");
12283 assert!(d.is_untitled(), "and it still has no file");
12284 }
12285
12286 #[test]
12287 fn a_blank_document_becomes_a_real_one_at_the_first_save_as() {
12288 let p = temp_path("blank_save_as");
12289 let mut d = Doc::blank().unwrap();
12290 // Plain text — a blank doc opens in Hidden mode, where a typed `#` would
12291 // be kept literal (`\#`); this test is about save-as, not escaping (which
12292 // has its own test), so it types nothing that escaping would touch.
12293 d.insert("hi");
12294 d.save_as(p.clone());
12295 assert_eq!(std::fs::read_to_string(&p).unwrap(), "hi");
12296 assert!(!d.is_untitled());
12297 assert!(!d.dirty);
12298 assert_eq!(d.file_name(), p.file_name().unwrap().to_string_lossy());
12299 assert_eq!(
12300 d.disk_state(),
12301 DiskState::Unchanged,
12302 "the watermark is stamped"
12303 );
12304 // And ⌘S is a plain save from here on.
12305 d.insert("!");
12306 d.save();
12307 assert_eq!(std::fs::read_to_string(&p).unwrap(), "hi!");
12308 let _ = std::fs::remove_file(&p);
12309 }
12310
12311 // ── save as ───────────────────────────────────────────────────────────────
12312
12313 /// A unique path in the temp dir that no fixture wrote — a Save As target.
12314 fn temp_path(name: &str) -> PathBuf {
12315 static SEQ: std::sync::atomic::AtomicUsize = std::sync::atomic::AtomicUsize::new(0);
12316 let seq = SEQ.fetch_add(1, std::sync::atomic::Ordering::Relaxed);
12317 let mut p = std::env::temp_dir();
12318 p.push(format!("leaf_test_target_{name}_{seq}.md"));
12319 let _ = std::fs::remove_file(&p);
12320 p
12321 }
12322
12323 #[test]
12324 fn save_as_moves_the_document_and_leaves_the_old_file_alone() {
12325 let mut d = doc_with("save_as_move", "original\n");
12326 let old = d.path.clone();
12327 let new = temp_path("save_as_move");
12328 d.insert("edited: ");
12329 d.save_as(new.clone());
12330
12331 assert_eq!(std::fs::read_to_string(&new).unwrap(), "edited: original\n");
12332 assert_eq!(
12333 std::fs::read_to_string(&old).unwrap(),
12334 "original\n",
12335 "Save As doesn't touch the file it came from"
12336 );
12337 assert_eq!(d.path, new, "the document moved");
12338 assert!(!d.dirty);
12339 assert_eq!(
12340 d.status.as_deref(),
12341 Some(&*format!("saved {}", d.file_name()))
12342 );
12343
12344 // Every later save follows it, which is the whole difference from a copy.
12345 d.caret = 0;
12346 d.insert("re-");
12347 d.save();
12348 assert_eq!(
12349 std::fs::read_to_string(&new).unwrap(),
12350 "re-edited: original\n"
12351 );
12352 assert_eq!(std::fs::read_to_string(&old).unwrap(), "original\n");
12353 let _ = std::fs::remove_file(&new);
12354 }
12355
12356 #[test]
12357 fn save_as_overwrites_an_existing_target() {
12358 // The picker already asked; asking again down here is the same question
12359 // twice, and the second one has no way to be answered.
12360 let new = temp_path("save_as_over");
12361 std::fs::write(&new, "theirs\n").unwrap();
12362 let mut d = doc_with("save_as_over", "ours\n");
12363 d.save_as(new.clone());
12364 assert_eq!(std::fs::read_to_string(&new).unwrap(), "ours\n");
12365 let _ = std::fs::remove_file(&new);
12366 }
12367
12368 #[test]
12369 fn a_save_as_that_fails_leaves_the_document_where_it_was() {
12370 let mut d = doc_with("save_as_fail", "body\n");
12371 let old = d.path.clone();
12372 d.insert("x");
12373 // A directory that doesn't exist: the write can't land.
12374 let bad = std::env::temp_dir().join("leaf_test_no_such_dir_9f2/doc.md");
12375 d.save_as(bad);
12376
12377 assert_eq!(
12378 d.path, old,
12379 "the document must not move to a file that isn't there"
12380 );
12381 assert!(d.dirty, "and must not believe it saved");
12382 assert!(
12383 d.status.as_deref().unwrap().starts_with("save failed:"),
12384 "the same failure a plain save reports, got {:?}",
12385 d.status
12386 );
12387 // The original is still the document's file, and still saveable.
12388 d.save();
12389 assert_eq!(std::fs::read_to_string(&old).unwrap(), "xbody\n");
12390 assert!(!d.dirty);
12391 }
12392
12393 #[test]
12394 fn save_as_renames_without_reparsing_the_format() {
12395 // `.dj` on the name doesn't make the buffer djot: it was parsed as
12396 // Markdown and still is, and saying otherwise would be a conversion the
12397 // user never asked for (and an undo history thrown away to do it).
12398 let mut d = doc_with("save_as_format", "**b**\n");
12399 let mut new = temp_path("save_as_format");
12400 new.set_extension("dj");
12401 d.save_as(new.clone());
12402 assert_eq!(d.format_name(), "markdown");
12403 let _ = std::fs::remove_file(&new);
12404 }
12405
12406 // ── external change / reload ──────────────────────────────────────────────
12407
12408 #[test]
12409 fn an_untouched_file_reports_unchanged() {
12410 let mut d = doc_with("disk_clean", "body\n");
12411 assert_eq!(d.disk_state(), DiskState::Unchanged);
12412 // Editing the buffer is not editing the file.
12413 d.insert("x");
12414 assert_eq!(d.disk_state(), DiskState::Unchanged);
12415 assert!(d.dirty);
12416 // Saving re-stamps the watermark rather than reporting our own bytes back.
12417 d.save();
12418 assert_eq!(d.disk_state(), DiskState::Unchanged);
12419 }
12420
12421 #[test]
12422 fn a_file_written_underneath_reports_changed() {
12423 let mut d = doc_with("disk_changed", "body\n");
12424 std::fs::write(&d.path, "someone else\n").unwrap();
12425 assert_eq!(d.disk_state(), DiskState::Changed);
12426 // Dirty *and* changed is the clobber: both halves are readable, and
12427 // leaf-core takes neither side.
12428 d.insert("x");
12429 assert!(d.dirty && d.disk_state() == DiskState::Changed);
12430 // Saving anyway is allowed — the frontend asked, or chose not to.
12431 d.save();
12432 assert_eq!(std::fs::read_to_string(&d.path).unwrap(), "xbody\n");
12433 assert_eq!(d.disk_state(), DiskState::Unchanged);
12434 }
12435
12436 #[test]
12437 fn a_file_rewritten_with_the_same_bytes_is_unchanged() {
12438 // The hash is what makes this honest: the file was written (a fresh
12439 // mtime), and nothing about the document is stale.
12440 let d = doc_with("disk_same_bytes", "body\n");
12441 std::fs::write(&d.path, "body\n").unwrap();
12442 assert_eq!(d.disk_state(), DiskState::Unchanged);
12443 }
12444
12445 #[test]
12446 fn a_deleted_file_reports_missing() {
12447 let mut d = doc_with("disk_missing", "body\n");
12448 std::fs::remove_file(&d.path).unwrap();
12449 assert_eq!(d.disk_state(), DiskState::Missing);
12450 // A save recreates it, and the document is whole again.
12451 d.save();
12452 assert_eq!(d.disk_state(), DiskState::Unchanged);
12453 assert_eq!(std::fs::read_to_string(&d.path).unwrap(), "body\n");
12454 }
12455
12456 #[test]
12457 fn reload_replaces_the_document_with_the_file() {
12458 for (view, tag) in VIEWS {
12459 let mut d = doc_in(view, &format!("reload_{tag}"), "one\n\ntwo\n");
12460 d.insert("edited ");
12461 assert!(d.dirty);
12462 std::fs::write(&d.path, "one\n\ntwo\n\nthree\n").unwrap();
12463 d.reload();
12464
12465 assert_eq!(d.source, "one\n\ntwo\n\nthree\n", "{tag}");
12466 assert!(!d.dirty, "{tag}: the file is what we have");
12467 assert_eq!(d.disk_state(), DiskState::Unchanged, "{tag}");
12468 assert_eq!(
12469 d.status.as_deref(),
12470 Some(&*format!("reloaded {}", d.file_name()))
12471 );
12472 // The reloaded tree is live, not the old parse.
12473 d.caret = d.source.find("three").unwrap();
12474 assert_eq!(d.breadcrumb(), "doc › para › str", "{tag}");
12475 }
12476 }
12477
12478 #[test]
12479 fn reload_clamps_the_caret_and_drops_the_selection() {
12480 let mut d = doc_with("reload_caret", "a long first line\n");
12481 d.caret = 12;
12482 d.anchor = Some(4);
12483 std::fs::write(&d.path, "short\n").unwrap();
12484 d.reload();
12485 assert_eq!(d.caret, d.source.len(), "clamped into the shorter file");
12486 assert_eq!(
12487 d.anchor, None,
12488 "a selection over bytes that changed is a lie"
12489 );
12490 assert!(d.selection().is_none());
12491
12492 // A caret the file still has room for stays put.
12493 let mut d = doc_with("reload_caret_keep", "one\n\ntwo\n");
12494 d.caret = 2;
12495 std::fs::write(&d.path, "one\n\ntwo\n\nthree\n").unwrap();
12496 d.reload();
12497 assert_eq!(d.caret, 2);
12498 }
12499
12500 #[test]
12501 fn reload_drops_the_undo_history() {
12502 // twig's stack belongs to the buffer, and these are different bytes:
12503 // replaying a step recorded against the old ones would corrupt the file.
12504 let mut d = doc_with("reload_undo", "body\n");
12505 d.insert("x");
12506 std::fs::write(&d.path, "replaced\n").unwrap();
12507 d.reload();
12508 d.undo();
12509 assert_eq!(
12510 d.source, "replaced\n",
12511 "an undo must not resurrect the old buffer"
12512 );
12513 assert_eq!(d.status.as_deref(), Some("nothing to undo"));
12514 }
12515
12516 #[test]
12517 fn a_reload_that_cant_read_leaves_the_document_alone() {
12518 let mut d = doc_with("reload_gone", "body\n");
12519 d.insert("x");
12520 std::fs::remove_file(&d.path).unwrap();
12521 d.reload();
12522 assert_eq!(d.source, "xbody\n", "the unsaved work is still here");
12523 assert!(d.dirty);
12524 assert!(
12525 d.status.as_deref().unwrap().starts_with("reload failed:"),
12526 "{:?}",
12527 d.status
12528 );
12529
12530 // And an untitled document has nothing to reload from.
12531 let mut d = Doc::blank().unwrap();
12532 d.insert("typed");
12533 d.reload();
12534 assert_eq!(d.source, "typed");
12535 assert_eq!(d.status.as_deref(), Some("no file to reload"));
12536 }
12537
12538 #[test]
12539 fn a_read_only_document_refuses_every_door() {
12540 let mut d = doc_with("readonly", "one two three\n");
12541 d.insert("x");
12542 assert!(d.dirty, "writable first, so the undo step exists");
12543 d.set_read_only(true);
12544 let before = d.source.clone();
12545 d.insert("y");
12546 d.backspace();
12547 d.undo();
12548 d.redo();
12549 assert_eq!(d.source, before, "no door moved a byte");
12550 d.set_read_only(false);
12551 d.undo();
12552 assert_ne!(d.source, before, "off again, the same doors work");
12553 }
12554
12555 #[test]
12556 fn a_selection_quote_carries_its_context_on_char_boundaries() {
12557 let mut d = doc_with("quote", "before 你好 exact 世界 after\n");
12558 let start = d.source.find("exact").unwrap();
12559 d.place_caret(start, false);
12560 d.place_caret(start + "exact".len(), true);
12561 let q = d.selection_quote(3).unwrap();
12562 assert_eq!(q.exact, "exact");
12563 assert_eq!(
12564 q.prefix, "你好 ",
12565 "chars, not bytes — the multibyte pair counts as two"
12566 );
12567 assert_eq!(q.suffix, " 世界");
12568 assert_eq!(&d.source[q.start..q.end], "exact");
12569 // At the edges the context clips rather than erring.
12570 d.place_caret(0, false);
12571 d.place_caret(6, true);
12572 let q = d.selection_quote(40).unwrap();
12573 assert_eq!(q.prefix, "");
12574 assert_eq!(q.exact, "before");
12575 // No selection is no quote.
12576 d.place_caret(0, false);
12577 assert!(d.selection_quote(3).is_none());
12578 }
12579
12580 #[test]
12581 fn highlights_are_kept_sorted_and_answer_point_queries() {
12582 let mut d = doc_with("hl", "one two three\n");
12583 d.set_highlights(vec![
12584 Highlight {
12585 start: 8,
12586 end: 13,
12587 id: "b".into(),
12588 color: None,
12589 marker: None,
12590 },
12591 Highlight {
12592 start: 0,
12593 end: 3,
12594 id: "a".into(),
12595 color: Some("#ffe066".into()),
12596 marker: None,
12597 },
12598 Highlight {
12599 start: 5,
12600 end: 5,
12601 id: "empty".into(),
12602 color: None,
12603 marker: None,
12604 },
12605 ]);
12606 assert_eq!(
12607 d.highlights()
12608 .iter()
12609 .map(|h| h.id.as_str())
12610 .collect::<Vec<_>>(),
12611 ["a", "b"],
12612 "sorted by start, the empty range dropped"
12613 );
12614 assert_eq!(d.highlight_at(1).map(|h| h.id.as_str()), Some("a"));
12615 assert_eq!(d.highlight_at(3), None, "end is exclusive");
12616 assert_eq!(d.highlight_at(8).map(|h| h.id.as_str()), Some("b"));
12617 d.set_highlights(Vec::new());
12618 assert!(d.highlights().is_empty(), "a replace is a replace");
12619 }
12620}