leaf_core/wysiwyg.rs
1//! The WYSIWYG view: render the document with its markup *resolved*, not shown —
2//! headings and code tagged with a typographic role (a frontend sizes or colours
3//! them; see [`crate::style`]), `**bold**` as real bold, `# ` / `**` / `` ` ``
4//! delimiters hidden — while keeping every visible glyph tied back to the source
5//! byte it came from.
6//!
7//! That back-reference (`Glyph::src`) is what lets a caret still work: the caret
8//! stays a source offset (shared with the source view), but the [`VisualMap`]
9//! converts between an offset and a screen `(row, col)`, so cursor drawing,
10//! mouse clicks, and vertical motion all operate in *visible* space.
11//!
12//! Left and Right instead walk the map's caret *stops* in document order. On
13//! ordinary prose that's the same journey — the stops are laid out left to right
14//! — and it steps over the hidden delimiters either way. They part company only
15//! in a table, where the text is arranged in two dimensions and a cell wrapped
16//! within its column continues *below* rather than to the right. Following the
17//! document is what a caret means there.
18//!
19//! Text is walked from the AST (`str` nodes carry exact spans, and their text is
20//! the verbatim source slice), so a Markdown and a Djot file that parse alike
21//! render — and map — identically.
22
23use std::cell::Cell;
24use std::collections::HashMap;
25use std::ops::Range;
26
27use twig::{Alignment, ContainerOrigin, DirectiveForm, Editor, FlatNode, Kind, QueryMatch};
28use unicode_segmentation::UnicodeSegmentation;
29use unicode_width::UnicodeWidthStr;
30
31use crate::style::{Baseline, Role, Style};
32
33/// One rendered character plus the source byte offset it originates from.
34/// Synthetic glyphs (a list bullet, a quote gutter) point at their block's
35/// start, so clicking one lands the caret at the start of that block.
36#[derive(Clone)]
37pub struct Glyph {
38 pub ch: char,
39 pub style: Style,
40 pub src: usize,
41 /// Whether the caret may *rest* on this glyph. Decoration — a table border
42 /// or a cell's alignment padding — is visible but isn't text, so the caret
43 /// steps over it instead of into it. It also can't be a stop even in
44 /// principle: a run of decoration shares one `src`, and a caret can only
45 /// move by changing offset, so resting on it would pin horizontal motion.
46 /// A click still maps through `src`, which is why decoration points at the
47 /// text it decorates.
48 ///
49 /// Real text is a stop once per *grapheme cluster*, on the glyph that opens
50 /// it: the continuation glyphs of an emoji or an accented letter are drawn,
51 /// but standing between them is standing inside a character.
52 pub stop: bool,
53}
54
55/// One visual line. `end_src` is the source offset a caret sits at when placed
56/// at the line's end (past its last glyph) — the anchor for end-of-line and
57/// click-past-content.
58///
59/// `Clone` so a block's rows can be cached and re-emitted at a shifted offset
60/// across an edit — see [`BlockCache`].
61#[derive(Clone)]
62pub struct VRow {
63 pub glyphs: Vec<Glyph>,
64 pub end_src: usize,
65 /// A row that is drawn but holds no caret: a table's `├───┼───┤` rules, and
66 /// the blank gap a block boundary is spelled with. Vertical motion steps
67 /// over it, `pos_of_offset` never resolves onto it, and its stops (it has
68 /// none) and `end_src` stay out of the map's stop table.
69 ///
70 /// Emptiness isn't the test — an empty paragraph is a blank row too, and a
71 /// real caret stop. The test is whether the row is somewhere text can go.
72 pub decoration: bool,
73 /// This row is one line of a fenced or indented code block. Set on every row
74 /// the `"code_block"` arm emits — including its blank lines, which carry no
75 /// glyph to tell them apart otherwise. A frontend draws its own chrome (a
76 /// border and a tinted background) around each maximal run of these, and
77 /// scrolls them horizontally instead of wrapping; see
78 /// [`VisualMap::code_blocks`]. Survives the row shuffling of [`BlockCache`]
79 /// reuse and [`build_spliced`] because it rides on the row, not on a
80 /// row-index span the way a table's picture does.
81 pub code: bool,
82 /// A fenced code block's info string (its language), carried on the *first*
83 /// row of the block so it survives row reuse the way [`code`](Self::code)
84 /// does. `None` on every other row, and on an indented block (which has no
85 /// fence to label). A frontend paints it as a small label on the block's box
86 /// and edits it through a prompt — see [`CodeBlockInfo::lang`]. It's a plain
87 /// display string, not a source slice, so it needs no offset shifting; the
88 /// label re-derives from twig on the next build.
89 pub code_lang: Option<String>,
90 /// This row belongs to a `:::name{.class}` directive container — twig's
91 /// generic fenced-div block, whose meaning is entirely up to the host app
92 /// (diaryx's `:::vis{.audience}` visibility blocks, say). Set on every row
93 /// the `"directive"` arm emits, the same way [`code`](Self::code) marks a
94 /// code block's rows, so a frontend can draw a tinted panel around each
95 /// maximal run of these.
96 pub directive: bool,
97 /// A directive container's space-joined attrs — dot-prefixed classes
98 /// (`.public .family` → `"public family"`) unioned with bare pandoc-style
99 /// words (`public family`, no leading dot — diaryx's other `:::vis{...}`
100 /// convention), carried on the block's *first* row only — the
101 /// [`code_lang`](Self::code_lang) pattern. `None` on every other row, and
102 /// when the directive carries no such attrs. A frontend paints it as a
103 /// small label on the block's panel; it's a plain display string, not a
104 /// source slice, so it rides row reuse untouched.
105 pub directive_label: Option<String>,
106 /// Set on the single placeholder row a block-level image renders to, carrying
107 /// the image's destination and alt text; `None` on every other row. The row's
108 /// glyphs are the default `🖼 alt` label (which a plain surface paints as-is);
109 /// an image-capable frontend reads this to paint the real picture instead,
110 /// skipping the row named by [`MediaInfo::rows_span`]. Like
111 /// [`code_lang`](Self::code_lang) it's plain display strings, not source
112 /// slices, so it rides row reuse and needs no offset shifting; the map's
113 /// [`media`](VisualMap::media) side-table is derived from it once the rows
114 /// are final, the same way [`code_blocks`](VisualMap::code_blocks) is.
115 pub media: Option<MediaMark>,
116 /// Set on the **first** row of a task list item, carrying whether its box is
117 /// ticked; `None` on every other row, including a plain `list_item`'s. The
118 /// row's glyphs already draw the box as `☐ `/`☑ ` in the marker's place, so a
119 /// plain surface needs nothing further; a GUI reads this to paint a real
120 /// checkbox widget and to know which way it is facing.
121 ///
122 /// A `bool` rather than a source span, for the reason
123 /// [`code_lang`](Self::code_lang) is a plain string: it rides [`BlockCache`]
124 /// reuse and [`build_spliced`] untouched, needing no offset shifting. To
125 /// *toggle* the box, a frontend maps its click to a source offset the way it
126 /// maps any other — the marker's glyphs carry the item's own `src` — and
127 /// hands that to [`crate::Doc::toggle_task_at`].
128 pub task: Option<bool>,
129 /// Set on the single placeholder row a **leaf** directive (`::name{…}`)
130 /// renders to, carrying its name and attributes; `None` on every other row.
131 /// The container form isn't this — it wraps real blocks and marks each of
132 /// them [`directive`](Self::directive) instead. Like [`media`](Self::media)
133 /// it's plain display strings, so it rides row reuse untouched, and the map's
134 /// [`directives`](VisualMap::directives) side-table is derived from it once
135 /// the rows are final.
136 pub leaf_directive: Option<DirectiveMark>,
137 /// The heading level (1–6) of the block this row belongs to, on every row a
138 /// `heading` emits (a long one wraps to several) and `None` everywhere else.
139 ///
140 /// A frontend that sizes a whole line — a proportional renderer giving the
141 /// row a bigger line box — needs the level *per row*, and the glyphs can't
142 /// always supply it: an empty heading (`# ` with nothing typed after it,
143 /// which is what the toolbar's H1 leaves on a blank line) has no glyph to
144 /// carry a [`Role::Heading`] at all, so a glyph scan called it body text and
145 /// the line drew at body height until the first character landed. Riding the
146 /// row says it once, for the empty case and the wrapped case alike.
147 ///
148 /// Per-*glyph* styling still comes from [`Role::Heading`] on the glyphs; this
149 /// is the row-level fact, and the two agree wherever a heading has content —
150 /// same `u8` level, clamped the same way [`heading_style`] clamps it.
151 pub heading: Option<u8>,
152 /// What this row divides, on the blank rows a block boundary is *drawn* with
153 /// and `None` on every other row — including the navigable blank lines of
154 /// preserve-soft flow, which are somewhere text can go rather than a gap
155 /// between blocks. So `boundary.is_some()` is exactly "this row is a drawn
156 /// block boundary", the [`decoration`](Self::decoration) rows that come from
157 /// [`Builder::emit_separators_before`].
158 ///
159 /// It exists because a boundary's *height* is a frontend decision but its
160 /// *kind* is not. Typography spaces a boundary by what it separates — the
161 /// margin above a heading is wider than the one between two paragraphs, so
162 /// the heading groups with the text it introduces — and a frontend that has
163 /// only rows to look at has to re-derive the structure by sniffing glyph
164 /// roles. Three frontends sniffing separately is three chances to disagree
165 /// about the same document. Core already knows, having just walked the AST
166 /// to emit this row, so it says so once here and each frontend multiplies by
167 /// its own spacing.
168 pub boundary: Option<Boundary>,
169}
170
171/// What a drawn block boundary separates: the kinds of the blocks it falls
172/// between — the pair a frontend spaces by.
173#[derive(Clone, Copy, Debug, PartialEq, Eq)]
174pub struct Boundary {
175 pub above: BlockClass,
176 pub below: BlockClass,
177}
178
179/// The block kinds core tells apart when it walks a document — the vocabulary
180/// [`Boundary`] is spelled in. A statement about *structure*, not about how any
181/// of it should look: what a frontend does with "this gap sits above a heading"
182/// is entirely the frontend's.
183///
184/// `Class` rather than `Kind` because [`twig::BlockKind`] already means
185/// something else in this crate's public surface — the *command* vocabulary
186/// (`Paragraph | Heading(n)`) a toolbar passes to [`Doc::set_block`](crate::Doc::set_block).
187/// This is the reverse direction: what a block already *is*, read back off a
188/// rendered row.
189///
190/// [`BlockClass::Other`] is the honest answer for a node kind core doesn't
191/// separate out, so adding one here is additive for every frontend: nothing has
192/// to change until it wants to space that kind differently.
193#[derive(Clone, Copy, Debug, PartialEq, Eq)]
194pub enum BlockClass {
195 Paragraph,
196 Heading,
197 /// A whole list. Its *items* are [`BlockClass::ListItem`]; note that core
198 /// draws no boundary row between two items of one list, tight or loose, so
199 /// an item↔item pair never reaches a frontend.
200 List,
201 ListItem,
202 Quote,
203 Code,
204 Table,
205 /// A block-level image, video, or audio.
206 ///
207 /// Never reached through [`from_node_kind`](BlockClass::from_node_kind): a
208 /// block picture is not a node of its own — [`Builder::media_only`] promotes
209 /// the paragraph (or `<picture>`/`<video>` container) wrapping it — so the
210 /// walk only ever sees the wrapper's kind. [`label_media_boundaries`] reads
211 /// it back off the finished rows instead, after the fact.
212 Media,
213 /// A `:::name{.class}` directive container.
214 Directive,
215 Rule,
216 Footnote,
217 Other,
218}
219
220impl BlockClass {
221 /// Classify a twig node kind — the same vocabulary [`Builder::block`]
222 /// matches on, so the two can't drift about what a block is. Both the
223 /// whole-arena walk (which has [`FlatNode`]s) and the incremental top-level
224 /// walk (which has only a query match's kind) reach it by this one door.
225 pub fn from_node_kind(kind: &Kind) -> BlockClass {
226 match kind {
227 Kind::Para => BlockClass::Paragraph,
228 Kind::Heading => BlockClass::Heading,
229 Kind::BulletList | Kind::OrderedList | Kind::TaskList => BlockClass::List,
230 Kind::ListItem | Kind::TaskListItem => BlockClass::ListItem,
231 Kind::BlockQuote => BlockClass::Quote,
232 Kind::CodeBlock => BlockClass::Code,
233 Kind::Table => BlockClass::Table,
234 Kind::Image => BlockClass::Media,
235 // twig 2.8 folded `div`/`span`/`directive`/`element` into one
236 // `container` kind, so a `:::note` panel and a promoted `<video>`
237 // arrive here indistinguishable — telling them apart needs the
238 // node's `origin`, and the incremental walk has only this kind.
239 // `Directive` is the right answer for the case that motivates the
240 // class (nothing else draws a tinted panel) and a harmless one for
241 // the rest: `BlockClass` is descriptive and core never branches on
242 // it. The one case where it was actively wrong — a promoted
243 // `<video>`, which would have been handed to a frontend as something
244 // to draw a fenced-div panel around — is corrected by
245 // [`label_media_boundaries`] once the rows are final, along the same
246 // door as a block image. Anything else that must be exact reads
247 // [`container_is_directive`] off a real node.
248 Kind::Container => BlockClass::Directive,
249 Kind::ThematicBreak => BlockClass::Rule,
250 Kind::Footnote => BlockClass::Footnote,
251 _ => BlockClass::Other,
252 }
253 }
254}
255
256/// The name and attributes a leaf directive's placeholder row carries, so a
257/// frontend that knows the host app's vocabulary can paint the real thing —
258/// an embedded page for diaryx's `::embed{src=…}`, a generated table of
259/// contents for a `::toc`, and the plain `⧉ name` label for one it doesn't
260/// know. The peer of [`MediaMark`], and plain strings for the same reason: they
261/// survive the row shuffling of [`BlockCache`] reuse and [`build_spliced`].
262#[derive(Clone, Debug, PartialEq, Eq)]
263pub struct DirectiveMark {
264 /// The directive's type — `embed`, `toc`, `vis` — with no leading colons.
265 /// Core is agnostic of what it means: the vocabulary is the host app's.
266 pub name: String,
267 /// Its `{…}` attributes as `(key, value)` pairs in source order. A bare
268 /// attribute (`{public}`) has a `None` value, the way twig reports it.
269 pub attrs: Vec<(String, Option<String>)>,
270 /// The directive's `[label]` text, flattened from its inline children, or
271 /// empty when it has none. Also what the placeholder label shows.
272 pub label: String,
273 /// How many visual rows this directive reserves — the label row plus blank
274 /// filler rows below it, so a frontend painting something real has the
275 /// vertical room. `1` is the bare placeholder, and the only value core
276 /// produces today: unlike an image (whose height a terminal frontend
277 /// measures and reports back), nothing has told core how tall an embed is.
278 /// A pixel-laid-out GUI sets its own height regardless.
279 pub rows: usize,
280}
281
282/// What a block-level media placeholder actually is, so a frontend knows which
283/// widget to build over the reserved rows: a raster, a movie player, or a
284/// transport with no picture at all. Core classifies and stops there — it opens
285/// nothing, so this is a statement about the *markup*, not about a file it has
286/// verified exists or can decode.
287#[derive(Clone, Copy, Debug, PartialEq, Eq)]
288pub enum MediaKind {
289 /// A `` / `<img>` / `<picture>` — a still picture.
290 Image,
291 /// An HTML `<video>`. Markdown and Djot spell no video of their own, so this
292 /// only ever arrives through `html_elements` promotion (or a `::video{…}`
293 /// directive a host app maps itself, which core reports as a directive).
294 Video,
295 /// An HTML `<audio>` — a transport with no picture, so a frontend gives it a
296 /// fixed control height rather than measuring an aspect ratio.
297 Audio,
298}
299
300/// Which of the two caret homes a block media has — see
301/// [`VisualMap::block_media_stop`].
302#[derive(Clone, Copy, Debug, PartialEq, Eq)]
303pub enum MediaStop {
304 /// The stop in front of the picture. What is typed here belongs above it.
305 Before,
306 /// The stop just past it. What is typed here belongs below it.
307 After,
308}
309
310impl MediaKind {
311 /// The emoji a plain surface prefixes the placeholder label with — the
312 /// `🖼`/`🎬`/`🔊` that makes the row read as *a thing* rather than as text.
313 fn sigil(self) -> char {
314 match self {
315 MediaKind::Image => '🖼',
316 MediaKind::Video => '🎬',
317 MediaKind::Audio => '🔊',
318 }
319 }
320}
321
322/// The destination and label a block-level media placeholder row carries, so a
323/// capable frontend can resolve and paint the real thing. Plain strings (no
324/// source offsets), so they survive the row shuffling of [`BlockCache`] reuse
325/// and [`build_spliced`] untouched — see [`VRow::media`].
326#[derive(Clone, Debug, PartialEq, Eq)]
327pub struct MediaMark {
328 /// Whether this is a picture, a movie, or a sound — which widget the
329 /// frontend builds over the reserved rows.
330 pub kind: MediaKind,
331 /// The media's link destination — a path, URL, or `data:` URI, verbatim from
332 /// the AST. A frontend resolves a relative path against the document's
333 /// directory itself; core holds no I/O.
334 ///
335 /// Empty is possible and legal for a `<video>`/`<audio>`, which may carry no
336 /// `src` of its own and name its candidates in child `<source>`s instead —
337 /// unlike an `<img>`, whose `src` *is* the picture. A frontend with an empty
338 /// destination takes its URL from [`sources`](MediaMark::sources).
339 pub destination: String,
340 /// A `<picture>`'s theme/media alternatives, in document order, when this
341 /// block image came from one; empty for a plain `` / bare `<img>`. Each
342 /// is a `<source>`'s media query + candidate URL(s); a frontend that knows its
343 /// theme picks the first whose media matches and falls back to [`destination`]
344 /// (the `<img>`). Core keeps them verbatim and picks nothing — it has no theme.
345 ///
346 /// [`destination`]: MediaMark::destination
347 pub sources: Vec<MediaSource>,
348 /// The media's alt text (its rendered inline children, flattened), or empty
349 /// when it has none. Also what the placeholder label shows. For a `<video>`/
350 /// `<audio>` this is the element's own text content — the "your browser does
351 /// not support…" fallback, which doubles as its accessible name.
352 pub alt: String,
353 /// A `<video poster="…">`'s still frame, verbatim, or empty when there is
354 /// none (and always empty for an image or audio). It is an *image*
355 /// destination, so a frontend already able to draw a picture can show it
356 /// before the movie loads — or in place of one it can't play at all.
357 pub poster: String,
358 /// How many visual rows this media reserves — the placeholder label row plus
359 /// the blank filler rows below it, so a frontend that paints a real raster has
360 /// the vertical room to draw it. `1` is the bare placeholder (a frontend that
361 /// can't draw pictures, or an image it couldn't resolve). A terminal frontend
362 /// asks for as many rows as the fitted picture is tall; the pixel-laid-out GUI
363 /// ignores this and sets its own row height, so it always leaves it `1`. The
364 /// count comes from the frontend (via [`crate::Doc::set_media_rows`]) because
365 /// core does no I/O and can't measure the image itself. See [`VRow::media`].
366 pub rows: usize,
367}
368
369/// One `<source>` under a `<picture>`, `<video>`, or `<audio>`: a candidate URL
370/// plus whichever of the two things HTML lets a `<source>` be chosen by — a
371/// media query (`<picture>`) or a MIME type (`<video>`/`<audio>`). Verbatim from
372/// the AST: core carries the alternatives and resolves none of them, having
373/// neither a theme nor a codec list to judge them by.
374///
375/// The two spellings are normalised onto one field. `<picture>` writes
376/// `srcset`, `<video>`/`<audio>` write `src`; both land in
377/// [`srcset`](MediaSource::srcset), since a frontend wants the URL either way
378/// and only `<picture>` ever uses the descriptor syntax.
379#[derive(Clone, Debug, PartialEq, Eq)]
380pub struct MediaSource {
381 /// The `<source media="…">` query, verbatim (`"(prefers-color-scheme: dark)"`),
382 /// or empty for a `<source>` with no `media` (an unconditional override, and
383 /// the norm for `<video>`/`<audio>`, which pick by codec rather than theme).
384 pub media: String,
385 /// The candidate URL(s): a `<picture>`'s `srcset` verbatim — one URL, or a
386 /// comma-separated candidate list with `1x`/`2x`/width descriptors — or a
387 /// `<video>`/`<audio>` `<source>`'s plain `src`. A frontend takes the first
388 /// URL token; the theme and codec cases both only ever need that.
389 pub srcset: String,
390 /// The `<source type="…">` MIME type (`"video/webm"`), verbatim, or empty
391 /// when the `<source>` declares none. How a `<video>`/`<audio>` frontend
392 /// picks a candidate it can actually decode; a `<picture>`'s sources
393 /// normally leave it empty and are chosen by [`media`](MediaSource::media).
394 pub mime: String,
395}
396
397/// The rendered document plus the offset⇄position mapping the caret rides on.
398#[derive(Clone, Default)]
399pub struct VisualMap {
400 /// The document's **default monospace rendering** — one [`VRow`] of glyphs
401 /// per visual line, tables spelled with box-drawing borders (`│ ─ ┌┬┐…`) and
402 /// cells padded to whole character-cell columns. Any monospace surface can
403 /// draw these verbatim, so a consumer gets a working view for free: the TUI
404 /// paints them as-is, and a five-line plain-text dump would too.
405 ///
406 /// It's a *default*, not the only truth. A frontend with its own geometry —
407 /// a proportional GUI — lays text out in its own units, and for a table
408 /// skips the box-drawn rows named by [`TableInfo::rows_span`] and draws from
409 /// the structural [`TableInfo`] instead. The box glyphs live here rather than
410 /// in a frontend precisely because they *are* a renderable default: unlike a
411 /// colour (a role each surface must map to its own palette — see
412 /// [`crate::style`]), `┌─┐` is finished text that needs no interpretation.
413 pub rows: Vec<VRow>,
414 /// The first source offset that is actually rendered — the caret floor for
415 /// the WYSIWYG view. Non-zero when a leading `metadata` block (YAML/TOML
416 /// frontmatter) is skipped: the frontmatter is preserved in the source and
417 /// editable in the source view, but hidden and unreachable here, so the
418 /// caret and selection can't wander into it (and copy won't grab it).
419 pub content_start: usize,
420 /// Every offset the caret may rest at, ascending and deduplicated: each
421 /// row's stop glyphs plus the row's own end (the "after the last character"
422 /// spot every line needs). Decoration contributes nothing.
423 ///
424 /// Left/Right read this instead of walking the grid, because the grid isn't
425 /// laid out in offset order: a table with wrapped cells puts column 1's
426 /// second line *below* column 2's first, so "the next stop rightward" and
427 /// "the next stop in the document" part ways. Following the document is what
428 /// a caret means — and on every row that *is* in order the two agree anyway,
429 /// so nothing else has to change.
430 stops: Vec<usize>,
431 /// Every table in the document, in order, described structurally rather than
432 /// drawn — see [`TableInfo`] for why both exist.
433 pub tables: Vec<TableInfo>,
434 /// Every fenced/indented code block, in order, as the range of [`rows`] it
435 /// occupies — a frontend draws one bordered, tinted box around each and
436 /// scrolls it horizontally rather than wrapping. Derived from the per-row
437 /// [`VRow::code`] flag once the rows are final (so it survives incremental
438 /// row reuse), the same way [`collect_stops`] derives the stop table.
439 ///
440 /// [`rows`]: VisualMap::rows
441 pub code_blocks: Vec<CodeBlockInfo>,
442 /// Every block-level image in the document, in order — one per placeholder
443 /// row a frontend replaces with a real picture. Derived from the per-row
444 /// [`VRow::media`] mark once the rows are final (so it survives incremental
445 /// row reuse), the same way [`code_blocks`](VisualMap::code_blocks) is
446 /// derived from [`VRow::code`].
447 pub media: Vec<MediaInfo>,
448 /// Every **leaf** directive in the document, in order — one per placeholder
449 /// row a frontend may replace with whatever the host app's vocabulary makes
450 /// of it. Derived from the per-row [`VRow::leaf_directive`] mark once the
451 /// rows are final, exactly as [`media`](VisualMap::media) is.
452 pub directives: Vec<DirectiveInfo>,
453}
454
455impl VisualMap {
456 pub fn num_rows(&self) -> usize {
457 self.rows.len()
458 }
459
460 /// The width of `row` in display columns — the rightmost column its caret
461 /// can occupy, and so what a goal column is clamped to on the way in.
462 pub fn row_width(&self, row: usize) -> usize {
463 self.rows.get(row).map_or(0, |r| r.width())
464 }
465
466 /// The screen `(row, col)` for a source offset — where to draw the caret:
467 /// the *nearest* stop at or past `off`. Snaps a hidden offset (inside a
468 /// delimiter) to the next visible glyph, and never resolves onto decoration
469 /// (a table border, a cell's padding), which is drawn but holds no caret.
470 ///
471 /// "Nearest" rather than "the first one found" because a table's wrapped
472 /// cells put rows slightly out of offset order: scanning top to bottom, the
473 /// second line of column 1 comes *after* the first line of column 2 but
474 /// holds smaller offsets. Where rows are in order the two rules agree.
475 ///
476 /// A soft wrap is the one place two rows want the same offset: the row above
477 /// ends where the row below opens, the space the wrap ate being drawn on the
478 /// row above and the offset past it being the row below's first character.
479 /// It resolves *downstream*, to the row that character is on — the row
480 /// above's last column is a phantom, a place the caret can be drawn but
481 /// never sent, and resolving upstream into it is what pinned Down at the
482 /// first wrap of a paragraph: it aimed at the row below's column 0, landed
483 /// on the offset it already had, and read that back as the row above's end.
484 pub fn pos_of_offset(&self, off: usize) -> (usize, usize) {
485 let mut best: Option<(usize, usize, usize)> = None; // (src, row, col)
486 for (r, row) in self.rows.iter().enumerate() {
487 if row.decoration {
488 continue;
489 }
490 // Offsets ascend *within* a row, so its first stop at or past `off`
491 // is the best this row has to offer.
492 let cand = row
493 .glyphs
494 .iter()
495 .enumerate()
496 .find(|(_, g)| g.stop && g.src >= off)
497 .map(|(i, g)| (g.src, r, row.col_of_glyph(i)))
498 .or_else(|| (row.end_src >= off).then_some((row.end_src, r, row.width())));
499 if let Some(c) = cand {
500 // `<=`, so a tie goes to the later row: the only offset two rows
501 // both hold is a wrap boundary, and it belongs to the row below.
502 if best.is_none_or(|b| c.0 <= b.0) {
503 best = Some(c);
504 }
505 }
506 // A row's *first* stop never decreases from one row to the next —
507 // true even across a table's wrapped cells, since a cell's lines run
508 // downward. So once a row opens past the best found so far, no later
509 // row can beat it and the scan stays proportional to `off`.
510 if let (Some(b), Some(first)) = (best, row.glyphs.iter().find(|g| g.stop))
511 && first.src > b.0
512 {
513 break;
514 }
515 }
516 match best {
517 Some((_, r, c)) => (r, c),
518 None => {
519 let r = self.last_stop_row();
520 (r, self.row_width(r))
521 }
522 }
523 }
524
525 /// The rows a source range occupies, inclusive: `(first, last)`.
526 ///
527 /// A *different question* from [`pos_of_offset`](Self::pos_of_offset), which
528 /// is why it can't be spelled with two calls to it. That one answers "where
529 /// does the caret go", and for a caret its forward snap is right — an offset
530 /// inside a hidden delimiter has no column of its own, so the caret belongs
531 /// at the next visible glyph, wherever that turns out to be. This one asks
532 /// "which rows does this block cover", and there the snap is a trap: a
533 /// footnote whose body *ends* in a link (`[^2]: [title](url)`) has a last
534 /// byte inside the hidden destination, so `pos_of_offset(end - 1)` walked
535 /// clean off the note's row and landed on the next note's — and a peek
536 /// slicing `first..=last` out of the frame drew two notes where the reader
537 /// asked for one. Every block ending in a link, an image, or any trailing
538 /// hidden markup had the same fault; only a block ending in visible text
539 /// (which is what the tests happened to use) did not.
540 ///
541 /// `row.end_src` is no help either: it is where the *rendered* text of a row
542 /// ends, not how far into the source the block reaches, and redefining it
543 /// would move every end-of-line caret.
544 ///
545 /// So the last row is found by asking which rows *open* before the range
546 /// does, rather than by mapping its last byte: a row belongs to the range
547 /// when its first caret stop lies before `range.end`. Decoration is skipped
548 /// (a drawn gap between blocks is not part of either), and the answer is
549 /// never shorter than one row — a range whose every byte is hidden still
550 /// covers the row it started on.
551 pub fn row_range_for(&self, range: Range<usize>) -> (usize, usize) {
552 if self.rows.is_empty() {
553 return (0, 0);
554 }
555 let first = self.pos_of_offset(range.start).0;
556 let mut last = first;
557 for (r, row) in self.rows.iter().enumerate().skip(first) {
558 if row.decoration {
559 continue;
560 }
561 let open = row
562 .glyphs
563 .iter()
564 .find(|g| g.stop)
565 .map_or(row.end_src, |g| g.src);
566 if open >= range.end {
567 // A row's first stop never decreases from one row to the next —
568 // the invariant `pos_of_offset` breaks on, true even across a
569 // table's wrapped cells — so nothing below can be in range.
570 break;
571 }
572 last = r;
573 }
574 (first, last)
575 }
576
577 /// The source offset of the task checkbox drawn at `(row, col)`, or `None`
578 /// when that cell holds no box — the hit-test a frontend runs on a click
579 /// before treating it as a tick rather than a caret placement.
580 ///
581 /// Only the box's own cells answer. Clicking an item's *text* places the
582 /// caret like any other click, so the box is a target aimed at rather than
583 /// something tripped over while editing — which is also why this is a
584 /// separate question from [`offset_of_pos`](Self::offset_of_pos) instead of
585 /// a flag on the offset it returns.
586 pub fn task_box_at(&self, row: usize, col: usize) -> Option<usize> {
587 let r = self.rows.get(row)?;
588 self.task_box_at_glyph(row, r.glyph_at_col(col)?)
589 }
590
591 /// [`task_box_at`](Self::task_box_at) keyed by glyph index rather than
592 /// display column — for a frontend that shapes its own rows (the GUI) and so
593 /// resolves a click to a glyph before it ever has a column.
594 pub fn task_box_at_glyph(&self, row: usize, glyph: usize) -> Option<usize> {
595 let r = self.rows.get(row)?;
596 r.task?;
597 let g = r.glyphs.get(glyph)?;
598 (g.style.role == Role::ListMarker).then_some(g.src)
599 }
600
601 /// The source offset for a screen `(row, col)` — where a click or a
602 /// visual-space move lands the caret. Clicking decoration maps through its
603 /// `src`, which points at the text it decorates, so a click on a border or
604 /// on a cell's padding lands in that cell.
605 ///
606 /// The inverse of [`pos_of_offset`](Self::pos_of_offset), which it has to
607 /// agree with: `col` is a display column, and the one it names may be the
608 /// far cell of a wide glyph — [`VRow::glyph_at_col`] is where that lands.
609 pub fn offset_of_pos(&self, row: usize, col: usize) -> usize {
610 let Some(r) = self.rows.get(row) else {
611 // A click or drag below the last row — a short document with empty
612 // space under it, dragged into to extend a selection. Land on the
613 // document's last caret stop (its end), not offset 0: jumping the
614 // caret to the top is the wrong direction, and 0 isn't even a stop
615 // when the document opens on hidden frontmatter or a `# ` marker, so
616 // returning it would leave the caret where it draws in one place and
617 // types in another (`move_to` would then clamp it onto the unhomeable
618 // frontmatter floor). `None` only for a document with no stops at all
619 // (empty), where the caret has nowhere to be but 0.
620 return self.stops.last().copied().unwrap_or(0);
621 };
622 match r.glyph_at_col(col).and_then(|i| r.glyphs.get(i)) {
623 // A glyph that holds no caret is clickable, but where it points
624 // isn't always somewhere the caret can be: the blank gap between two
625 // paragraphs stands at an offset that belongs to neither of them,
626 // and the tail of a grapheme cluster stands inside a character.
627 // Land on the nearest real stop instead of handing back an offset
628 // that looks like the gap but types into the paragraph above.
629 Some(g) if !g.stop => self.nearest_stop(g.src),
630 Some(g) => g.src,
631 // A row's end is a stop by construction — unless the row is
632 // decoration, which contributes none.
633 None if r.decoration => self.nearest_stop(r.end_src),
634 None => r.end_src,
635 }
636 }
637
638 /// Which of a block media's two caret homes `off` is, or `None` for every
639 /// other offset in the document.
640 ///
641 /// [`block_media`](Builder::block_media) gives a block-level image, video, or
642 /// audio exactly two stops — one in front of it and one just past it — and
643 /// nothing inside the markup. Both are ordinary offsets to everything else in
644 /// core, but they are the two places where inserting text would *dissolve the
645 /// picture*: `` with anything typed against it is no longer a block
646 /// image but a paragraph with an inline one, and the frontend that was
647 /// painting a photo there paints a text run instead. A caller that is about to
648 /// insert asks this so it can open a paragraph first — see
649 /// [`Doc::insert`](crate::Doc::insert).
650 ///
651 /// An *inline* image reports `None`: it has no placeholder row and no stops of
652 /// its own, and typing beside one is ordinary editing.
653 ///
654 /// Answers with the media's own source span as well, since a caller that has
655 /// to keep the picture whole usually has to address it — [`Doc::backspace`]
656 /// takes the picture out in one piece rather than nibbling a byte off its
657 /// markup, which is the same dissolution from the other side.
658 ///
659 /// [`Doc::backspace`]: crate::Doc::backspace
660 pub fn block_media_stop(&self, off: usize) -> Option<(MediaStop, Range<usize>)> {
661 for m in &self.media {
662 let Some(row) = self.rows.get(m.rows_span.start) else {
663 continue;
664 };
665 // Every glyph of the `🖼 alt` label maps to the media's start offset;
666 // the row's end is past its markup. Read the start off the label
667 // rather than the first glyph, which on a quoted or listed picture is
668 // the block prefix and points at the gutter.
669 let Some(start) = row
670 .glyphs
671 .iter()
672 .find(|g| g.style.role == Role::Image)
673 .map(|g| g.src)
674 else {
675 continue;
676 };
677 if off == start {
678 return Some((MediaStop::Before, start..row.end_src));
679 }
680 if off == row.end_src {
681 return Some((MediaStop::After, start..row.end_src));
682 }
683 }
684 None
685 }
686
687 /// Snap `off` to the nearest caret stop — the funnel a frontend that
688 /// hit-tests pixels straight to a source offset must run its result through.
689 /// A click or drag can land in the blank gap a paragraph break is drawn with,
690 /// or inside a hidden delimiter; both are offsets the caret can't rest at, so
691 /// resting there would draw the caret in one place and type in another. This
692 /// settles it on a real caret home instead. Idempotent on an offset that is
693 /// already a stop — the `(row, col)` click path already snaps this way inside
694 /// [`offset_of_pos`](Self::offset_of_pos), and this gives the pixel path the
695 /// same guarantee. Returns `off` unchanged only for an empty document (no
696 /// stops at all).
697 pub fn snap_to_stop(&self, off: usize) -> usize {
698 self.nearest_stop(off)
699 }
700
701 /// The caret stop nearest `off`, preferring the one before it when `off`
702 /// falls exactly between two. Returns `off` unchanged if there are no stops
703 /// at all (an empty document).
704 fn nearest_stop(&self, off: usize) -> usize {
705 let i = self.stops.partition_point(|&s| s < off);
706 let after = self.stops.get(i).copied();
707 let before = i.checked_sub(1).map(|j| self.stops[j]);
708 match (before, after) {
709 (Some(b), Some(a)) if off - b <= a - off => b,
710 (_, Some(a)) => a,
711 (Some(b), None) => b,
712 (None, None) => off,
713 }
714 }
715
716 /// Whether the caret can occupy `row` at all: decoration rows (a table's
717 /// border rules) are stepped over by vertical motion.
718 pub fn row_is_navigable(&self, row: usize) -> bool {
719 self.rows.get(row).is_some_and(|r| !r.decoration)
720 }
721
722 /// The first offset the caret can rest at on `row` — its first stop, or the
723 /// row's own end when it holds no text (an empty paragraph). `None` for a
724 /// decoration row, which holds no caret at all.
725 ///
726 /// Not `offset_of_pos(row, 0)`: column 0 of a quoted or listed row is the
727 /// gutter, and a gutter's `src` points at the *block* it opens, so the stop
728 /// nearest it is the one on the block's first row rather than on this one.
729 /// Which is right for a click — the gutter decorates the whole block — and
730 /// wrong for Home, whose whole question is where *this* row starts.
731 pub fn row_start(&self, row: usize) -> Option<usize> {
732 let r = self.rows.get(row).filter(|r| !r.decoration)?;
733 Some(
734 r.glyphs
735 .iter()
736 .find(|g| g.stop)
737 .map_or(r.end_src, |g| g.src),
738 )
739 }
740
741 /// The last row the caret can rest on — the fallback when an offset is past
742 /// everything rendered (a table's bottom border must not swallow the caret).
743 fn last_stop_row(&self) -> usize {
744 (0..self.rows.len())
745 .rev()
746 .find(|&r| self.row_is_navigable(r))
747 .unwrap_or(0)
748 }
749
750 /// The nearest row above `row` the caret can occupy, skipping decoration.
751 pub fn navigable_above(&self, row: usize) -> Option<usize> {
752 (0..row.min(self.rows.len()))
753 .rev()
754 .find(|&r| self.row_is_navigable(r))
755 }
756
757 /// The nearest row below `row` the caret can occupy, skipping decoration.
758 pub fn navigable_below(&self, row: usize) -> Option<usize> {
759 ((row + 1)..self.rows.len()).find(|&r| self.row_is_navigable(r))
760 }
761
762 /// The caret stop just before `off` — one press of Left. `None` at the
763 /// first stop in the document.
764 ///
765 /// Runs of decoration (a table border, a cell's alignment padding) are
766 /// stepped over in a single press: they hold no stop, so they aren't in the
767 /// table to land on.
768 pub fn stop_before(&self, off: usize) -> Option<usize> {
769 let i = self.stops.partition_point(|&s| s < off);
770 i.checked_sub(1).map(|i| self.stops[i])
771 }
772
773 /// The caret stop just after `off` — one press of Right. `None` at the last
774 /// stop in the document.
775 pub fn stop_after(&self, off: usize) -> Option<usize> {
776 let i = self.stops.partition_point(|&s| s <= off);
777 self.stops.get(i).copied()
778 }
779
780 /// The first caret stop at or past `off` — where the caret at a hidden
781 /// offset is *drawn*, and so where a rightward walk over the rendered text
782 /// starts from.
783 pub fn stop_at_or_after(&self, off: usize) -> Option<usize> {
784 let i = self.stops.partition_point(|&s| s < off);
785 self.stops.get(i).copied()
786 }
787
788 /// The last caret stop at or before `off` — where a leftward walk starts
789 /// from. Snapping the way the walk is headed, rather than always forward,
790 /// is what keeps a leftward motion from ever moving the caret right.
791 pub fn stop_at_or_before(&self, off: usize) -> Option<usize> {
792 let i = self.stops.partition_point(|&s| s <= off);
793 i.checked_sub(1).map(|i| self.stops[i])
794 }
795
796 /// Whether the caret may rest at `off` — the invariant every motion in this
797 /// view has to leave standing.
798 pub fn is_stop(&self, off: usize) -> bool {
799 self.stops.binary_search(&off).is_ok()
800 }
801
802 /// The visible text a caret crosses walking rightward from `from` up to
803 /// (but not including) `to` — `UITextInput.text(in:)`'s `[from, to)` in
804 /// *this* view. A hidden inline-mark delimiter (`**`, `` ` ``, `_`, an
805 /// escape backslash) never got a glyph in the first place — see
806 /// [`push_text`]/[`synth`] — so it contributes nothing; what's left is
807 /// exactly what's drawn on screen for that span.
808 ///
809 /// Built from the same stop glyphs [`stop_after`](Self::stop_after) steps
810 /// across (every glyph with [`Glyph::stop`] set, i.e. one per grapheme
811 /// cluster, decoration excluded) — **plus one inserted `'\n'` for every
812 /// genuine block boundary strictly inside `[from, to)`**: a run of whole
813 /// [`decoration`] rows sitting between two content rows — a paragraph
814 /// gap, a table rule, an image's reserved filler rows — never an ordinary
815 /// soft wrap, which puts no decoration *row* between the two halves of
816 /// its one paragraph (only inline decoration glyphs, e.g. a table's `│`,
817 /// live inside a single content row, and never split one).
818 ///
819 /// [`decoration`]: VRow::decoration
820 ///
821 /// Without that inserted break, two blocks abutting in this string were
822 /// indistinguishable from one run of text: [`collect_stops`] gives a
823 /// block boundary *zero* stops of its own (crossing one is a single,
824 /// free hop — see `the_caret_skips_the_gap_between_two_paragraphs` in
825 /// `doc.rs`'s tests, which pins that as intentional caret behaviour, a
826 /// paragraph gap costing no extra Right presses, not a bug to fix here).
827 /// So the last word of one paragraph and the first word of the next used
828 /// to land directly adjacent with *nothing* between them in this string
829 /// (`"...edb\n\nhello\n"` read back as `"edbhello"`), and `UITextInput`'s
830 /// default word tokenizer then saw one unbroken run of letters and
831 /// selected across the boundary — reported as double-tapping the last
832 /// word on a line expanding the selection into the following
833 /// paragraph(s).
834 ///
835 /// This means the once-strict equality with `distance_offset`/
836 /// `step_offset` (`leaf-ffi`) no longer always holds: those intentionally
837 /// keep costing a block boundary *zero* stops, while this text now
838 /// spends one *character* on it that is never itself a stop. So the
839 /// relationship is `visible_text(a, b).chars().count() >=
840 /// distance_offset(a, b)`, equality holding whenever `(a, b)` spans no
841 /// block boundary (the common case, and the only case the previous
842 /// equality was ever tested against). It can only ever be *greater*,
843 /// never less: every character this function omits relative to a plain
844 /// stop count is a stop with no glyph of its own (a hidden delimiter, or
845 /// a block's own trailing "end of row" stop), and every such omission at
846 /// a block's end is exactly paired with the one inserted separator that
847 /// follows it, so nothing this function returns is ever short of what a
848 /// consumer walking stops one at a time would need. That inequality is
849 /// still exactly what `UITextInput`'s tokenizer needs: it only ever reads
850 /// this string to find a boundary and converts the character index it
851 /// finds back to a position with `position(from:offset:)`, which walks
852 /// stops — an inserted separator is never handed back as one, it only
853 /// keeps two paragraphs' words apart for the tokenizer's letter-run scan.
854 ///
855 /// `from` is snapped to its nearest stop first, exactly as a caret asked
856 /// to stand at a hidden offset is drawn at the next stop instead; `to` is
857 /// left as given, so a stop landing exactly on it is still the walk's
858 /// last step — the same asymmetry `distance_offset`'s own loop has.
859 pub fn visible_text(&self, from: usize, to: usize) -> String {
860 let from = self.nearest_stop(from);
861
862 // Real content: every stop glyph in range, keyed by its own source
863 // offset (`None` tags it as a genuine character, versus the
864 // synthetic separators below).
865 let mut items: Vec<(usize, Option<char>)> = self
866 .rows
867 .iter()
868 .filter(|r| !r.decoration)
869 .flat_map(|r| r.glyphs.iter())
870 .filter(|g| g.stop && g.src >= from && g.src < to)
871 .map(|g| (g.src, Some(g.ch)))
872 .collect();
873
874 // Every whole decoration row is a candidate block boundary; its
875 // `end_src` is the gap offset itself (never a stop — see
876 // `place_caret_snaps_out_of_the_blank_gap_between_paragraphs` in
877 // `doc.rs`) — a source offset like any glyph's, so it merges into the
878 // same ordering. `None` marks it a synthetic separator rather than a
879 // real character, tagged distinctly so a query landing exactly on the
880 // gap offset still opens with its break even with no glyph on either
881 // side to anchor it to (a range spanning nothing but a bare gap).
882 let mut boundaries: Vec<usize> = self
883 .rows
884 .iter()
885 .filter(|r| r.decoration)
886 .map(|r| r.end_src)
887 .filter(|&src| src >= from && src < to)
888 .collect();
889 boundaries.sort_unstable();
890 boundaries.dedup();
891 items.extend(boundaries.into_iter().map(|src| (src, None)));
892
893 // Row order matches source order except across a table's wrapped
894 // cells (see `pos_of_offset`), so sort rather than trust it here too.
895 // A boundary can't share an offset with a glyph (it's the undrawn gap
896 // between two blocks' real content), so tie-breaking never arises.
897 items.sort_by_key(|&(src, _)| src);
898 items
899 .into_iter()
900 .map(|(_, ch)| ch.unwrap_or('\n'))
901 .collect()
902 }
903}
904
905/// Collect the caret stops of a laid-out grid: every stop glyph's offset plus
906/// every row's end, ascending and deduplicated. Duplicates are the norm rather
907/// than the exception — a wrapped line's end is the same offset as the next
908/// line's first glyph — and collapsing them is what makes one press of Left or
909/// Right cross exactly one stop.
910fn collect_stops(rows: &[VRow]) -> Vec<usize> {
911 let mut stops: Vec<usize> = rows
912 .iter()
913 .filter(|r| !r.decoration)
914 .flat_map(|r| {
915 r.glyphs
916 .iter()
917 .filter(|g| g.stop)
918 .map(|g| g.src)
919 .chain(std::iter::once(r.end_src))
920 })
921 .collect();
922 stops.sort_unstable();
923 stops.dedup();
924 stops
925}
926
927/// Group the rows tagged [`VRow::code`] into one [`CodeBlockInfo`] per maximal
928/// run — the block-level view a frontend needs to box and scroll each code
929/// block. Two code blocks are always parted by the blank separator row a block
930/// boundary is spelled with (never itself a code row), so a contiguous run is
931/// exactly one block. Derived from the final rows rather than tracked through
932/// the builder so it comes out right no matter how [`build_cached`] and
933/// [`build_spliced`] shuffle rows around.
934fn code_block_spans(rows: &[VRow]) -> Vec<CodeBlockInfo> {
935 let mut blocks = Vec::new();
936 let mut start: Option<usize> = None;
937 for (i, row) in rows.iter().enumerate() {
938 match (row.code, start) {
939 (true, None) => start = Some(i),
940 (false, Some(s)) => {
941 blocks.push(CodeBlockInfo {
942 rows_span: s..i,
943 lang: rows[s].code_lang.clone(),
944 });
945 start = None;
946 }
947 _ => {}
948 }
949 }
950 if let Some(s) = start {
951 blocks.push(CodeBlockInfo {
952 rows_span: s..rows.len(),
953 lang: rows[s].code_lang.clone(),
954 });
955 }
956 blocks
957}
958
959/// The value of `node`'s `key` attribute, if it carries one *with* a value. A
960/// bare attribute (`controls`, `muted`) has a `None` value and so reads as
961/// absent here — a caller wanting presence-not-value tests the list directly.
962/// Shared by the media element and `<source>` readers.
963fn attr_of(node: &FlatNode, key: &str) -> Option<String> {
964 node.attrs
965 .iter()
966 .find(|(k, _)| k == key)
967 .and_then(|(_, v)| v.clone())
968}
969
970/// Collect one [`MediaInfo`] per row carrying a [`VRow::media`] mark — the
971/// block-level view a frontend needs to replace each placeholder row with a real
972/// picture. The mark rides the block's *first* row and names how many rows the
973/// media reserves ([`MediaMark::rows`]); the rows below it are blank
974/// [`decoration`](VRow::decoration) fillers that hold the vertical space and no
975/// caret. So the span runs from the marked row across those fillers. Derived from
976/// the final rows rather than tracked through the builder so it survives however
977/// [`build_cached`] and [`build_spliced`] shuffle rows around.
978fn media_spans(rows: &[VRow]) -> Vec<MediaInfo> {
979 rows.iter()
980 .enumerate()
981 .filter_map(|(i, row)| {
982 row.media.as_ref().map(|m| MediaInfo {
983 rows_span: i..i + m.rows.max(1),
984 kind: m.kind,
985 destination: m.destination.clone(),
986 sources: m.sources.clone(),
987 alt: m.alt.clone(),
988 poster: m.poster.clone(),
989 })
990 })
991 .collect()
992}
993
994/// Re-label the drawn block boundaries either side of a block-level media
995/// placeholder, so the pair a frontend spaces by names the picture.
996///
997/// [`BlockClass::from_node_kind`] classifies the node the walk is standing on,
998/// and a block image is never a node of its own: [`Builder::media_only`] promotes
999/// its *wrapper* — a `paragraph`, or the `container` a `<video>`/`<audio>`
1000/// arrives as — so the gap above a picture reported [`BlockClass::Paragraph`] and
1001/// the gap above a movie reported [`BlockClass::Directive`], the class a frontend
1002/// paints a tinted panel for. [`BlockClass::Media`] was unreachable in
1003/// consequence: the vocabulary named a kind no frontend could ever be told about.
1004///
1005/// Done as a pass over the finished rows rather than inside the walk because
1006/// only the rows know. The incremental top-level walk carries no node arena at
1007/// all (`nodes: &[]`) and can classify by kind alone, so teaching the wrapper
1008/// promotion to the whole-arena walk would label the full and incremental builds
1009/// differently — the exact drift that walk's own comment forbids. Both builds
1010/// emit the same [`VRow::media`] marks, so both reach the same answer here. The
1011/// [`media_spans`] / [`code_block_spans`] pattern.
1012///
1013/// One gap can be spelled with several rows — [`Builder::emit_separators_before`]
1014/// draws the row that closes the block above and the row that opens the block
1015/// below, with any extra blank source lines navigable between them — and gives
1016/// every one of them the same [`Boundary`]. So the walk crosses those navigable
1017/// blanks and relabels the whole run, stopping at the first row that is neither.
1018fn label_media_boundaries(rows: &mut [VRow]) {
1019 let spans: Vec<Range<usize>> = rows
1020 .iter()
1021 .enumerate()
1022 .filter_map(|(i, row)| row.media.as_ref().map(|m| i..i + m.rows.max(1)))
1023 .collect();
1024 // A row inside one gap: a drawn boundary to relabel, or one of the navigable
1025 // blank lines sitting between two drawn ones. Anything else ends the run.
1026 fn in_gap(row: &VRow) -> bool {
1027 row.boundary.is_some() || (!row.decoration && row.glyphs.is_empty())
1028 }
1029 for span in spans {
1030 for i in (0..span.start).rev() {
1031 if !in_gap(&rows[i]) {
1032 break;
1033 }
1034 if let Some(b) = rows[i].boundary.as_mut() {
1035 b.below = BlockClass::Media;
1036 }
1037 }
1038 for row in rows.iter_mut().skip(span.end) {
1039 if !in_gap(row) {
1040 break;
1041 }
1042 if let Some(b) = row.boundary.as_mut() {
1043 b.above = BlockClass::Media;
1044 }
1045 }
1046 }
1047}
1048
1049/// Collect one [`DirectiveInfo`] per row carrying a [`VRow::leaf_directive`]
1050/// mark — the block-level view a frontend needs to replace each placeholder row
1051/// with whatever the directive means to it. The peer of [`media_spans`], derived
1052/// from the final rows for the same reason: it survives however [`build_cached`]
1053/// and [`build_spliced`] shuffle rows around.
1054fn directive_spans(rows: &[VRow]) -> Vec<DirectiveInfo> {
1055 rows.iter()
1056 .enumerate()
1057 .filter_map(|(i, row)| {
1058 row.leaf_directive.as_ref().map(|m| DirectiveInfo {
1059 rows_span: i..i + m.rows.max(1),
1060 name: m.name.clone(),
1061 attrs: m.attrs.clone(),
1062 label: m.label.clone(),
1063 })
1064 })
1065 .collect()
1066}
1067
1068/// The source range of a fenced code block's info string — everything on the
1069/// opening line past the fence (`` ```rust `` → the `rust`). `block_start` is the
1070/// code block node's `span.start`. `None` for an indented code block, which
1071/// opens with no fence to carry one. The range is empty for a fence written
1072/// bare (`` ``` `` alone), which is exactly where a language would be inserted.
1073///
1074/// Shared by the WYSIWYG builder (to label the box) and [`crate::Doc`] (to edit
1075/// the label through a prompt), so the two agree on where the language lives.
1076pub fn code_info_span(source: &str, block_start: usize) -> Option<Range<usize>> {
1077 let rest = source.get(block_start..)?;
1078 let line_len = rest.find('\n').unwrap_or(rest.len());
1079 let line = &rest[..line_len];
1080 // A fence may be indented up to three spaces; past that it opens with a run
1081 // of the same fence character.
1082 let indent = line.len() - line.trim_start().len();
1083 if indent > 3 {
1084 return None;
1085 }
1086 let fence = line[indent..].chars().next()?;
1087 if fence != '`' && fence != '~' {
1088 return None; // an indented block, not a fenced one
1089 }
1090 let fence_len = line[indent..].chars().take_while(|&c| c == fence).count();
1091 let info_start = block_start + indent + fence_len;
1092 Some(info_start..block_start + line_len)
1093}
1094
1095/// A fenced code block's language for display: its info string, trimmed, or
1096/// `None` when there's no fence or the fence carries no language. The trimmed
1097/// text is what a frontend labels the box with; [`code_info_span`] is what an
1098/// edit replaces.
1099pub fn code_language(source: &str, block_start: usize) -> Option<String> {
1100 let span = code_info_span(source, block_start)?;
1101 let text = source.get(span)?.trim();
1102 (!text.is_empty()).then(|| text.to_string())
1103}
1104
1105/// A horizontal rule's dash count when the map isn't wrapping to a column grid
1106/// (the GUI, which wraps at pixel width): a fixed, sane width the frontend can
1107/// paint or re-wrap, instead of a runaway count from an unbounded wrap width.
1108const UNWRAPPED_RULE_WIDTH: usize = 40;
1109
1110/// Render the document to a [`VisualMap`]. `wrap` is the column budget for
1111/// word-wrapping (`Some` for the monospace TUI), or `None` to emit one row per
1112/// block — the GUI does its own proportional pixel wrapping over these rows.
1113/// Text and offsets come from the AST (`str` nodes carry the verbatim source
1114/// slice and an exact span), so the original source string isn't needed here.
1115pub fn build(
1116 nodes: &[FlatNode],
1117 source: &str,
1118 wrap: Option<usize>,
1119 preserve_soft: bool,
1120 media_rows: &HashMap<String, usize>,
1121 reveal: Option<Range<usize>>,
1122) -> VisualMap {
1123 let Some(doc) = nodes.iter().position(|n| n.kind == Kind::Doc) else {
1124 return VisualMap::default();
1125 };
1126 let top = top_level(nodes, doc);
1127 let mut b = Builder {
1128 nodes,
1129 source,
1130 wrap: wrap.map(|w| w.max(8)),
1131 rows: Vec::new(),
1132 tables: Vec::new(),
1133 last_off: 0,
1134 media_rows,
1135 break_glyph: Cell::new(' '),
1136 preserve_soft,
1137 reveal: reveal.clone(),
1138 };
1139 b.top_blocks(&top);
1140 // The hidden frontmatter's end is the baseline for both the trailing blank
1141 // rows and the caret floor — see [`hidden_prefix_end`]. `top_level` has
1142 // already dropped every `metadata` child, so read it off the arena.
1143 let hidden_end = hidden_prefix_end(source, metadata_end_of(nodes, doc));
1144 b.emit_trailing_blank_lines(
1145 top.last().map_or(BlockClass::Paragraph, |&i| {
1146 BlockClass::from_node_kind(&nodes[i].kind)
1147 }),
1148 hidden_end,
1149 );
1150 let content_start = top.first().map_or(hidden_end, |&i| nodes[i].span.start);
1151 let stops = collect_stops(&b.rows);
1152 label_media_boundaries(&mut b.rows);
1153 let code_blocks = code_block_spans(&b.rows);
1154 let media = media_spans(&b.rows);
1155 let directives = directive_spans(&b.rows);
1156 VisualMap {
1157 rows: b.rows,
1158 content_start,
1159 stops,
1160 tables: b.tables,
1161 code_blocks,
1162 media,
1163 directives,
1164 }
1165}
1166
1167/// Like [`build`], but reuses a persistent [`BlockCache`] so an edit re-renders
1168/// only the top-level blocks whose source bytes changed *and* marshals only
1169/// those blocks from twig instead of the whole arena.
1170///
1171/// `top` is the document's top-level blocks — twig's `child_spans` of the doc
1172/// root: `(node_id, kind, span)` for each, in order. `fetch_subtree(node_id)`
1173/// marshals one block's subtree (local-indexed, root at 0) and is called *only*
1174/// for a block that missed the cache, i.e. one that actually changed. So a
1175/// keystroke marshals one small subtree, not ~20k nodes. The result is
1176/// byte-for-byte identical to [`build`] on the same document (the
1177/// `build_cached_matches_build` test pins this); [`build`] stays the cache-free,
1178/// whole-arena reference. This is the entry point [`crate::Doc`] uses.
1179// One builder, and every one of these is a distinct input to the same layout
1180// pass — a struct of them would be built at the one call site and unpacked
1181// here, which is the same arguments with an extra name in the way.
1182#[allow(clippy::too_many_arguments)]
1183pub fn build_cached(
1184 top: &[QueryMatch],
1185 source: &str,
1186 wrap: Option<usize>,
1187 preserve_soft: bool,
1188 media_rows: &HashMap<String, usize>,
1189 reveal: Option<Range<usize>>,
1190 cache: &mut BlockCache,
1191 mut fetch_subtree: impl FnMut(u32) -> Vec<FlatNode>,
1192) -> VisualMap {
1193 let wrap = wrap.map(|w| w.max(8));
1194
1195 // Wrapping is a function of the width, so a width change makes every cached
1196 // row's wrap wrong: start the cache over.
1197 if cache.wrap != Some(wrap) {
1198 cache.entries.clear();
1199 cache.wrap = Some(wrap);
1200 }
1201 cache.generation = cache.generation.wrapping_add(1);
1202
1203 // Frontmatter (a leading `metadata` block) is document metadata, not prose:
1204 // hidden in the rich view exactly as [`Builder::blocks`] skips it.
1205 let blocks: Vec<&QueryMatch> = top.iter().filter(|m| m.kind != Kind::Metadata).collect();
1206
1207 // The outer builder only accumulates rows/tables and spells block boundaries
1208 // — both a function of the source and `last_off`, never of a node array — so
1209 // it carries an empty `nodes`. Each changed block is rendered by a *fresh*
1210 // builder over that block's subtree.
1211 let mut b = Builder {
1212 nodes: &[],
1213 source,
1214 wrap,
1215 rows: Vec::new(),
1216 tables: Vec::new(),
1217 last_off: 0,
1218 media_rows,
1219 break_glyph: Cell::new(' '),
1220 preserve_soft,
1221 reveal: reveal.clone(),
1222 };
1223
1224 // Record the per-block row decomposition as we go, so a later
1225 // [`build_spliced`] can patch one block without rebuilding the map.
1226 let mut layout_blocks: Vec<BlockLayout> = Vec::with_capacity(blocks.len());
1227 let mut all_shift_safe = true;
1228 for (i, block) in blocks.iter().enumerate() {
1229 let start = block.span.start;
1230 let before_sep = b.rows.len();
1231 if i > 0 {
1232 // This walker has no node arena at all (see the `nodes: &[]` above),
1233 // but a top-level query match carries its kind — the same string
1234 // `BlockClass::from_node_kind` classifies for the whole-arena walk, so
1235 // the incremental and full builds label a boundary identically.
1236 b.emit_separators_before(
1237 start,
1238 &[],
1239 true,
1240 Boundary {
1241 above: BlockClass::from_node_kind(&blocks[i - 1].kind),
1242 below: BlockClass::from_node_kind(&block.kind),
1243 },
1244 );
1245 }
1246 let sep_rows = b.rows.len() - before_sep;
1247 let after_sep = b.rows.len();
1248 let bytes = block_bytes(source, &block.span);
1249 let hash = block_hash(bytes);
1250 // How this block meets the reveal line, if at all — part of its cache
1251 // key, since the same bytes render differently on the caret's line.
1252 let rkey = reveal_key(&reveal, &block.span);
1253
1254 // Hit: clone the block's rows shifted to its current offset and restore
1255 // the (shifted) `last_off` so the next separator lands right — no marshal.
1256 // Only shift-safe blocks are ever cached, so a hit is safe by construction.
1257 if let Some(hit) = cache.reuse(hash, bytes, &rkey) {
1258 let delta = start as isize - hit.built_start as isize;
1259 for row in &hit.rows {
1260 b.rows.push(shift_row(row, delta));
1261 }
1262 b.last_off = (hit.last_off as isize + delta) as usize;
1263 } else {
1264 // Miss: marshal just this block's subtree and render it. A subtree is
1265 // self-contained with local ids (root at 0) and absolute spans, so a
1266 // fresh builder over it produces the same rows the whole-arena path
1267 // would. An empty subtree (twig couldn't hand it back) renders nothing.
1268 let subtree = fetch_subtree(block.node_id);
1269 if !subtree.is_empty() {
1270 let mut sub = Builder {
1271 nodes: &subtree,
1272 source,
1273 wrap,
1274 rows: Vec::new(),
1275 tables: Vec::new(),
1276 last_off: 0,
1277 media_rows,
1278 break_glyph: Cell::new(' '),
1279 preserve_soft,
1280 reveal: reveal.clone(),
1281 };
1282 sub.block(0, &[], &[]);
1283 let last_off = sub.last_off;
1284 // Cache only a block that is table-free AND renders inside its own
1285 // span: those two are the conditions for reuse-by-shift to be
1286 // correct. A block failing either is re-rendered every build (a
1287 // fresh render always matches a fresh whole-document build).
1288 if sub.tables.is_empty() {
1289 if rows_within(&sub.rows, &block.span) {
1290 cache.store(hash, bytes, start, sub.rows.clone(), last_off, rkey);
1291 }
1292 b.rows.extend(sub.rows);
1293 } else {
1294 // A table block is never cached; rebase its row-index
1295 // bookkeeping onto the combined row vector and append.
1296 let base = b.rows.len();
1297 for t in &mut sub.tables {
1298 t.rows_span = (t.rows_span.start + base)..(t.rows_span.end + base);
1299 }
1300 b.rows.extend(sub.rows);
1301 b.tables.extend(sub.tables);
1302 }
1303 b.last_off = last_off;
1304 }
1305 }
1306 let content_rows = b.rows.len() - after_sep;
1307 all_shift_safe &= rows_within(&b.rows[after_sep..], &block.span);
1308 layout_blocks.push(BlockLayout {
1309 span: block.span.clone(),
1310 kind: block.kind.clone(),
1311 sep_rows,
1312 content_rows,
1313 });
1314 }
1315
1316 let before_trailing = b.rows.len();
1317 let hidden_end = hidden_prefix_end(
1318 source,
1319 top.iter()
1320 .filter(|m| m.kind == Kind::Metadata)
1321 .map(|m| m.span.end)
1322 .next_back(),
1323 );
1324 b.emit_trailing_blank_lines(
1325 blocks.last().map_or(BlockClass::Paragraph, |m| {
1326 BlockClass::from_node_kind(&m.kind)
1327 }),
1328 hidden_end,
1329 );
1330 let trailing_rows = b.rows.len() - before_trailing;
1331
1332 // Evict every entry no block reused this build, so the cache tracks the
1333 // current document instead of growing without bound over a session.
1334 let g = cache.generation;
1335 cache.entries.retain(|_, bucket| {
1336 bucket.retain(|e| e.generation == g);
1337 !bucket.is_empty()
1338 });
1339
1340 cache.layout = Layout {
1341 blocks: layout_blocks,
1342 trailing_rows,
1343 built_len: source.len(),
1344 has_tables: !b.tables.is_empty(),
1345 all_shift_safe,
1346 reveal: reveal.clone(),
1347 };
1348
1349 // The first rendered offset is the first non-metadata block's start — the
1350 // analogue of [`first_content_offset`] for the top-level list. With nothing
1351 // but frontmatter it's the end of that frontmatter, and 0 for an empty
1352 // document ([`hidden_prefix_end`]).
1353 let content_start = blocks.first().map_or(hidden_end, |m| m.span.start);
1354 let stops = collect_stops(&b.rows);
1355 label_media_boundaries(&mut b.rows);
1356 let code_blocks = code_block_spans(&b.rows);
1357 let media = media_spans(&b.rows);
1358 let directives = directive_spans(&b.rows);
1359 VisualMap {
1360 rows: b.rows,
1361 content_start,
1362 stops,
1363 tables: b.tables,
1364 code_blocks,
1365 media,
1366 directives,
1367 }
1368}
1369
1370/// The fast path for a single-block edit: patch the previous [`VisualMap`] in
1371/// place rather than reassembling it. Returns `Some(new_map)` when it applies,
1372/// or `None` to tell the caller to fall back to [`build_cached`] (always
1373/// correct). Consumes `prev` either way — on `None` the caller rebuilds from
1374/// scratch and doesn't need it.
1375///
1376/// It applies only when `dirty` (twig's dirty byte range) falls inside exactly
1377/// one top-level block AND the block structure around it is unchanged — verified
1378/// by matching the new `top` list against the previous [`Layout`] block for
1379/// block: kinds unchanged, spans before the edit identical, spans after it
1380/// shifted by the byte delta, count unchanged. Any deviation — a block split or
1381/// merged, a fence opened to swallow later blocks, a table anywhere, a
1382/// multi-block edit — fails the match and returns `None`. That check is what
1383/// makes the byte-range trustworthy: twig's dirty range is exact about *bytes*
1384/// but silent about *reparse*, and the structural match catches the reparse
1385/// effects it can't see.
1386///
1387/// When it applies, the unchanged prefix rows move verbatim, the suffix rows
1388/// shift by the delta *in place* (integer adds, no glyph copy), and only the one
1389/// dirty block is re-marshalled and re-rendered; stops splice the same way by
1390/// offset. So the cost is O(rows after the edit), and nothing before the edit is
1391/// touched. The hash-keyed entry cache is left alone — a later [`build_cached`]
1392/// will miss on the changed block, re-render it, and evict the stale entry, so
1393/// chained splices neither corrupt nor grow it.
1394// One builder, and every one of these is a distinct input to the same layout
1395// pass — a struct of them would be built at the one call site and unpacked
1396// here, which is the same arguments with an extra name in the way.
1397#[allow(clippy::too_many_arguments)]
1398pub fn build_spliced(
1399 prev: VisualMap,
1400 source: &str,
1401 wrap: Option<usize>,
1402 preserve_soft: bool,
1403 top: &[QueryMatch],
1404 dirty: Range<usize>,
1405 media_rows: &HashMap<String, usize>,
1406 reveal: Option<Range<usize>>,
1407 cache: &mut BlockCache,
1408 mut fetch_subtree: impl FnMut(u32) -> Vec<FlatNode>,
1409) -> Option<VisualMap> {
1410 let wrap = wrap.map(|w| w.max(8));
1411 // A width change invalidates every cached row — a full rebuild's job.
1412 if cache.wrap != Some(wrap) {
1413 return None;
1414 }
1415 // So does a moved reveal line, and for the same reason: this path reuses
1416 // every row outside the dirty block, and those rows encode which line was
1417 // showing its raw markup when they were built. Typing almost always moves
1418 // the caret, so under `MarkupMode::Full` this bails to `build_cached` on
1419 // most keystrokes — still block-cached, so only the edited block and the
1420 // revealed one actually re-render.
1421 if cache.layout.reveal != reveal {
1422 return None;
1423 }
1424 // Take the previous layout; on any bail below the caller rebuilds it (and the
1425 // map) via `build_cached`, so leaving it empty is fine. A table or a block
1426 // that renders outside its span (a degenerate inline span) makes shifting
1427 // unsound, so those force the full-rebuild path.
1428 let prev_layout = std::mem::take(&mut cache.layout);
1429 if prev_layout.built_len == 0 || prev_layout.has_tables || !prev_layout.all_shift_safe {
1430 return None;
1431 }
1432
1433 let blocks: Vec<&QueryMatch> = top.iter().filter(|m| m.kind != Kind::Metadata).collect();
1434 if blocks.is_empty() || blocks.len() != prev_layout.blocks.len() {
1435 return None;
1436 }
1437 let delta = source.len() as isize - prev_layout.built_len as isize;
1438
1439 // The single block whose NEW span contains the whole dirty range. A dirty
1440 // range straddling a block boundary (or a separator) finds none → bail.
1441 let k = blocks
1442 .iter()
1443 .position(|m| m.span.start <= dirty.start && dirty.end <= m.span.end)?;
1444
1445 // Structural match: every OTHER block is unchanged — same kind throughout,
1446 // span identical before the edit and shifted by `delta` after it. A mismatch
1447 // means the reparse reshaped the block structure, which only a full rebuild
1448 // renders correctly.
1449 for (i, (m, pl)) in blocks.iter().zip(&prev_layout.blocks).enumerate() {
1450 if m.kind != pl.kind {
1451 return None;
1452 }
1453 if i == k {
1454 continue;
1455 }
1456 let want = if i < k {
1457 pl.span.clone()
1458 } else {
1459 (pl.span.start as isize + delta) as usize..(pl.span.end as isize + delta) as usize
1460 };
1461 if m.span != want {
1462 return None;
1463 }
1464 }
1465 // The dirty block itself: start unchanged (the edit is inside it, past its
1466 // start), end moved by exactly the delta.
1467 let pk_start = prev_layout.blocks[k].span.start;
1468 let pk_end = prev_layout.blocks[k].span.end;
1469 let pk_sep = prev_layout.blocks[k].sep_rows;
1470 let pk_content = prev_layout.blocks[k].content_rows;
1471 if blocks[k].span.start != pk_start || blocks[k].span.end != (pk_end as isize + delta) as usize
1472 {
1473 return None;
1474 }
1475
1476 // Re-render the dirty block from its subtree. A table makes the splice
1477 // bookkeeping unsafe, so bail if one appears.
1478 let subtree = fetch_subtree(blocks[k].node_id);
1479 if subtree.is_empty() {
1480 return None;
1481 }
1482 let mut sub = Builder {
1483 nodes: &subtree,
1484 source,
1485 wrap,
1486 rows: Vec::new(),
1487 tables: Vec::new(),
1488 last_off: 0,
1489 media_rows,
1490 break_glyph: Cell::new(' '),
1491 preserve_soft,
1492 reveal: reveal.clone(),
1493 };
1494 sub.block(0, &[], &[]);
1495 // A table, or content that renders outside the block's span (a degenerate
1496 // inline span), makes the shift bookkeeping unsound — fall back.
1497 if !sub.tables.is_empty() || !rows_within(&sub.rows, &blocks[k].span) {
1498 return None;
1499 }
1500 let new_content = sub.rows;
1501 let new_content_len = new_content.len();
1502 let new_stops = collect_stops(&new_content);
1503
1504 // Row span of the dirty block's CONTENT. Its leading separator stays in the
1505 // prefix: the gap before block k is unchanged, since k's start didn't move.
1506 let content_start_row: usize = prev_layout.blocks[..k]
1507 .iter()
1508 .map(|pl| pl.sep_rows + pl.content_rows)
1509 .sum::<usize>()
1510 + pk_sep;
1511 let content_end_row = content_start_row + pk_content;
1512
1513 // Splice rows: [prefix | new content | suffix + delta]. The prefix moves
1514 // untouched; the suffix shifts in place — integer adds, no glyph copy.
1515 let mut rows = prev.rows;
1516 let mut suffix = rows.split_off(content_end_row);
1517 rows.truncate(content_start_row);
1518 for row in &mut suffix {
1519 shift_row_in_place(row, delta);
1520 }
1521 rows.reserve(new_content_len + suffix.len());
1522 rows.extend(new_content);
1523 rows.extend(suffix);
1524
1525 // Splice stops by offset. The old dirty block covered `[pk_start, pk_end]`:
1526 // prefix stops fall below it, suffix stops above it (shift by delta), the new
1527 // content supplies the middle. The three ranges stay disjoint and ascending,
1528 // so the result needs no re-sort.
1529 let p1 = prev.stops.partition_point(|&s| s < pk_start);
1530 let p2 = prev.stops.partition_point(|&s| s <= pk_end);
1531 let mut stops = Vec::with_capacity(p1 + new_stops.len() + (prev.stops.len() - p2));
1532 stops.extend_from_slice(&prev.stops[..p1]);
1533 stops.extend(new_stops);
1534 for &s in &prev.stops[p2..] {
1535 stops.push((s as isize + delta) as usize);
1536 }
1537
1538 // Record the patched layout for the next splice: spans move to the new
1539 // coordinates, and the dirty block takes its new content-row count.
1540 let mut new_blocks = prev_layout.blocks;
1541 for (pl, m) in new_blocks.iter_mut().zip(&blocks) {
1542 pl.span = m.span.clone();
1543 }
1544 new_blocks[k].content_rows = new_content_len;
1545 cache.layout = Layout {
1546 blocks: new_blocks,
1547 trailing_rows: prev_layout.trailing_rows,
1548 built_len: source.len(),
1549 has_tables: false,
1550 // Every prefix/suffix block was shift-safe last build (we bailed
1551 // otherwise) and the re-rendered block was just checked, so the patched
1552 // document is still entirely shift-safe.
1553 all_shift_safe: true,
1554 reveal,
1555 };
1556
1557 label_media_boundaries(&mut rows);
1558 let code_blocks = code_block_spans(&rows);
1559 let media = media_spans(&rows);
1560 let directives = directive_spans(&rows);
1561 Some(VisualMap {
1562 rows,
1563 content_start: blocks[0].span.start,
1564 stops,
1565 tables: Vec::new(),
1566 code_blocks,
1567 media,
1568 directives,
1569 })
1570}
1571
1572/// A persistent, content-keyed cache of the rows each top-level block renders
1573/// to — the [`VisualMap`] analogue of the GUI's ShapedLine cache, one level
1574/// down. Held by a [`crate::Doc`] and threaded into [`build_cached`], it is what
1575/// makes a rebuild after a keystroke cost "re-render the edited block + shift
1576/// the rest" instead of re-rendering the whole document.
1577///
1578/// A top-level block's rows are a pure function of its source bytes and the wrap
1579/// width, so an unchanged block's rows are cloned and their source offsets
1580/// shifted by the edit's byte delta rather than rebuilt glyph by glyph. Two
1581/// things make that purity hold: at the top level the render prefix is always
1582/// empty (nesting prefixes — a quote gutter, a list indent — exist only *inside*
1583/// a top-level block, within its cached unit), and a block's output never reads
1584/// the incoming `last_off` (it writes `last_off` from its own content before any
1585/// nested separator reads it). So the only thing that differs between two
1586/// positions of an unchanged block is a uniform offset shift. Keyed by a fast
1587/// hash of the block's bytes with the bytes kept for a verify-on-hit — exactly
1588/// the shape cache's weak-hash-then-compare, so a collision costs a re-render,
1589/// never a wrong row.
1590///
1591/// Tables are never cached (a block that emits any table row is always rebuilt):
1592/// their rows are cross-referenced from the map's `tables` side-table by row
1593/// index, which a blind offset-shift wouldn't fix up, and they are rare enough
1594/// that the simplicity beats the reuse.
1595#[derive(Default)]
1596pub struct BlockCache {
1597 /// The wrap width every entry was built at; a change invalidates all of
1598 /// them. `None` before the first build (distinct from `Some(None)`, the
1599 /// unwrapped GUI width).
1600 wrap: Option<Option<usize>>,
1601 /// Bumped once per [`build_cached`]. An entry reused or inserted this build
1602 /// carries the current value; stale entries are dropped at the end of it.
1603 generation: u64,
1604 /// `hash(bytes)` → the block(s) sharing that hash — a bucket because
1605 /// distinct blocks can collide, while two *identical* blocks share one entry
1606 /// (free dedup).
1607 entries: HashMap<u64, Vec<CachedBlock>>,
1608 /// The row/stop decomposition of the last build, which [`build_spliced`]
1609 /// patches in place for a single-block edit. Kept in step with whatever
1610 /// [`VisualMap`] was last produced; empty before the first build.
1611 layout: Layout,
1612}
1613
1614/// How the last build's [`VisualMap`] decomposes into top-level blocks — the
1615/// bookkeeping [`build_spliced`] needs to splice one block's rows and stops
1616/// without rebuilding the whole map. Every field describes the *previous* build,
1617/// in that build's coordinates.
1618#[derive(Default)]
1619struct Layout {
1620 /// One entry per rendered (metadata-filtered) top-level block, in order.
1621 blocks: Vec<BlockLayout>,
1622 /// Trailing blank rows past the last block (from `emit_trailing_blank_lines`).
1623 trailing_rows: usize,
1624 /// The source length this layout was built at — the reference for the edit's
1625 /// byte delta.
1626 built_len: usize,
1627 /// Whether the last build drew any table. A table's cross-referenced row
1628 /// indices don't survive a blind splice, so their presence makes
1629 /// [`build_spliced`] bail to a full rebuild.
1630 has_tables: bool,
1631 /// Whether every block rendered strictly inside its own span (see
1632 /// [`rows_within`]). A block that doesn't — a malformed Markdown inline node
1633 /// that twig leaves with a degenerate `0..0` span renders at a fixed offset
1634 /// outside its block — can't be shifted correctly, so its presence makes
1635 /// [`build_spliced`] bail to a full rebuild.
1636 all_shift_safe: bool,
1637 /// The reveal line this layout was built under (see [`Builder::reveal`]).
1638 /// A splice reuses every row it isn't re-rendering, so a reveal line that
1639 /// has moved would leave the old line still showing its delimiters and the
1640 /// new one still hiding them — [`build_spliced`] bails when this changes.
1641 reveal: Option<Range<usize>>,
1642}
1643
1644/// One top-level block's contribution to the last build: its span and kind (for
1645/// the structural match that proves only one block changed) and how many
1646/// separator and content rows it emitted (to locate its slice of the row
1647/// vector).
1648struct BlockLayout {
1649 span: Range<usize>,
1650 kind: Kind,
1651 sep_rows: usize,
1652 content_rows: usize,
1653}
1654
1655/// One cached block: the rows it rendered to, plus what a reuse at a new
1656/// position needs to shift them. Offsets are stored absolute (as built) and
1657/// shifted by `new_start - built_start` on reuse.
1658struct CachedBlock {
1659 /// The block's exact source bytes, compared on a hash hit so a collision
1660 /// can never hand back another block's rows.
1661 bytes: Box<[u8]>,
1662 /// The offset the rows were built at (the block's `span.start`).
1663 built_start: usize,
1664 /// The block's rows, offsets absolute as built.
1665 rows: Vec<VRow>,
1666 /// `last_off` after this block was emitted, absolute as built — restored
1667 /// (shifted) on reuse so the following separator lands correctly.
1668 last_off: usize,
1669 /// Where the reveal line fell *within this block* when the rows were built,
1670 /// as a block-relative byte range — see [`reveal_key`]. Compared alongside
1671 /// `bytes` on a hit, because identical source renders to different rows
1672 /// depending on whether the caret's line is inside it: the same `*em*`
1673 /// shows its asterisks on the revealed line and hides them everywhere else.
1674 ///
1675 /// Block-relative rather than absolute so an unaffected block still hits
1676 /// after an edit shifts it, and `None` for the overwhelmingly common
1677 /// no-reveal case — which is why an entry stored under `MarkupMode::None`
1678 /// keeps hitting for every block that isn't the caret's.
1679 reveal: Option<Range<usize>>,
1680 /// The build that last reused or inserted this entry (see `generation`).
1681 generation: u64,
1682}
1683
1684/// Where `reveal` falls inside a block, in block-relative bytes — the extra key
1685/// a cached block is stored and matched under.
1686///
1687/// `None` when the block doesn't meet the reveal line at all, which is every
1688/// block on every build in the two hidden modes, and all but one of them under
1689/// [`crate::MarkupMode::Full`]. So the cache keeps its hit rate as the caret
1690/// moves: only the line the caret leaves and the line it arrives at re-render.
1691fn reveal_key(reveal: &Option<Range<usize>>, span: &Range<usize>) -> Option<Range<usize>> {
1692 let r = reveal.as_ref()?;
1693 // The same generous intersection test `Builder::revealed` uses, so a block
1694 // is keyed as revealed exactly when its glyphs will be built that way.
1695 (span.start <= r.end && r.start <= span.end).then(|| {
1696 let start = r.start.max(span.start) - span.start;
1697 let end = r.end.min(span.end) - span.start;
1698 start..end
1699 })
1700}
1701
1702impl BlockCache {
1703 /// Look up a block by hash, verify its bytes and reveal key, and on a hit
1704 /// stamp it used this build and hand back a borrow to shift-and-clone from.
1705 /// `None` on a miss (unknown hash, a collision whose bytes differ, or the
1706 /// same bytes built under a different reveal).
1707 fn reuse(
1708 &mut self,
1709 hash: u64,
1710 bytes: &[u8],
1711 reveal: &Option<Range<usize>>,
1712 ) -> Option<&CachedBlock> {
1713 let g = self.generation;
1714 let bucket = self.entries.get_mut(&hash)?;
1715 let e = bucket
1716 .iter_mut()
1717 .find(|e| &*e.bytes == bytes && &e.reveal == reveal)?;
1718 e.generation = g;
1719 Some(&*e)
1720 }
1721
1722 /// Cache the rows a freshly-rendered block produced (or refresh an existing
1723 /// entry for the same bytes and reveal — an identical block elsewhere, or a
1724 /// re-render).
1725 fn store(
1726 &mut self,
1727 hash: u64,
1728 bytes: &[u8],
1729 built_start: usize,
1730 rows: Vec<VRow>,
1731 last_off: usize,
1732 reveal: Option<Range<usize>>,
1733 ) {
1734 let g = self.generation;
1735 let bucket = self.entries.entry(hash).or_default();
1736 if let Some(e) = bucket
1737 .iter_mut()
1738 .find(|e| &*e.bytes == bytes && e.reveal == reveal)
1739 {
1740 e.built_start = built_start;
1741 e.rows = rows;
1742 e.last_off = last_off;
1743 e.generation = g;
1744 } else {
1745 bucket.push(CachedBlock {
1746 bytes: bytes.into(),
1747 built_start,
1748 rows,
1749 last_off,
1750 reveal,
1751 generation: g,
1752 });
1753 }
1754 }
1755}
1756
1757/// The source bytes a top-level block covers — the block cache's key material.
1758///
1759/// Clamped to the source rather than sliced by the span as twig gives it,
1760/// because that span can end *past* the last byte: the final block of a document
1761/// with no trailing newline is closed on the virtual newline the parser supplies
1762/// at EOF, so its `span.end` is `source.len() + 1`. Slicing by such a range
1763/// yields `None`, and the obvious `unwrap_or(&[])` reads that as *this block has
1764/// no bytes* — the wrong answer twice over.
1765///
1766/// Two blocks whose spans both overrun then key alike, and the second is served
1767/// the first one's rows. That is not hypothetical: a footnote definition is a
1768/// root beside `doc` merged back into the top level by [`top_blocks`], while the
1769/// `section` above it spans the definition's bytes too, so both end at EOF —
1770/// and a document ending in `[^note]: …` renders that definition as a second
1771/// copy of the heading. Even alone, a block that keeps hashing empty as the user
1772/// types in it is served the stale rows built before the edit.
1773///
1774/// Clamping hands back the bytes the block really covers, which tells both cases
1775/// apart, and costs nothing for a span that was in range to begin with.
1776fn block_bytes<'a>(source: &'a str, span: &Range<usize>) -> &'a [u8] {
1777 let bytes = source.as_bytes();
1778 let start = span.start.min(bytes.len());
1779 &bytes[start..span.end.clamp(start, bytes.len())]
1780}
1781
1782/// A fast, allocation-free content hash (FNV-1a) for a block's bytes. Weak by
1783/// design — the bytes are compared on a hit — so its only job is to spread
1784/// blocks across buckets cheaply. SipHash over every block's bytes on every
1785/// keystroke would cost more than it saves, the same lesson the shape cache
1786/// learned when it stopped hashing through the standard hasher.
1787fn block_hash(bytes: &[u8]) -> u64 {
1788 let mut h: u64 = 0xcbf2_9ce4_8422_2325;
1789 for &x in bytes {
1790 h ^= x as u64;
1791 h = h.wrapping_mul(0x0000_0100_0000_01b3);
1792 }
1793 h
1794}
1795
1796/// Clone a cached row with every source offset advanced by `delta` — the whole
1797/// cost of reusing an unchanged block: integer adds where a rebuild would
1798/// re-shape every glyph.
1799fn shift_row(row: &VRow, delta: isize) -> VRow {
1800 let shift = |off: usize| (off as isize + delta) as usize;
1801 VRow {
1802 glyphs: row
1803 .glyphs
1804 .iter()
1805 .map(|g| Glyph {
1806 ch: g.ch,
1807 style: g.style,
1808 src: shift(g.src),
1809 stop: g.stop,
1810 })
1811 .collect(),
1812 end_src: shift(row.end_src),
1813 decoration: row.decoration,
1814 code: row.code,
1815 code_lang: row.code_lang.clone(),
1816 directive: row.directive,
1817 directive_label: row.directive_label.clone(),
1818 media: row.media.clone(),
1819 // A tick, not an offset — reuse carries it as-is, like `code_lang`.
1820 task: row.task,
1821 leaf_directive: row.leaf_directive.clone(),
1822 heading: row.heading,
1823 // Structure, not offsets: a reused block's rows divide the same blocks
1824 // wherever the edit above moved them to.
1825 boundary: row.boundary,
1826 }
1827}
1828
1829/// Advance a row's source offsets by `delta` in place — the suffix half of
1830/// [`build_spliced`], where the rows are already owned and only need shifting,
1831/// not copying.
1832fn shift_row_in_place(row: &mut VRow, delta: isize) {
1833 for g in &mut row.glyphs {
1834 g.src = (g.src as isize + delta) as usize;
1835 }
1836 row.end_src = (row.end_src as isize + delta) as usize;
1837}
1838
1839/// Whether every source offset a block's rows carry falls inside the block's own
1840/// span — the precondition for reusing the block by a uniform offset shift. It
1841/// holds for well-formed blocks (their glyphs and row ends address bytes within
1842/// the block, synthetic glyphs point at the block start). It fails when a node
1843/// renders *outside* its block, which today means a malformed Markdown inline
1844/// node twig leaves with a degenerate `0..0` span: that content lands at a fixed
1845/// offset that doesn't move with the block. Such a block is re-rendered every
1846/// build instead of shifted, so the incremental map still matches a fresh one —
1847/// see [`build_cached`] and [`build_spliced`].
1848fn rows_within(rows: &[VRow], span: &Range<usize>) -> bool {
1849 rows.iter().all(|r| {
1850 r.end_src >= span.start
1851 && r.end_src <= span.end
1852 && r.glyphs
1853 .iter()
1854 .all(|g| g.src >= span.start && g.src <= span.end)
1855 })
1856}
1857
1858/// Where the rendered document begins when a leading `metadata` block is all
1859/// there is — the end of that hidden frontmatter, past the newline that closes
1860/// its last line so the floor sits at the start of the (empty) body rather than
1861/// on the closing `---`.
1862///
1863/// With a real block after it the frontmatter's end is never needed: the floor
1864/// is that block's start, and the rows begin there. With nothing after it, both
1865/// the caret floor and the trailing-blank-line count would otherwise fall back
1866/// to offset 0 — inside the hidden frontmatter — which put the caret *before*
1867/// the metadata and made typing land ahead of the opening `---`.
1868fn hidden_prefix_end(source: &str, meta_end: Option<usize>) -> usize {
1869 let Some(end) = meta_end else { return 0 };
1870 let end = end.min(source.len());
1871 let rest = &source[end..];
1872 if rest.starts_with("\r\n") {
1873 end + 2
1874 } else if rest.starts_with('\n') {
1875 end + 1
1876 } else {
1877 end
1878 }
1879}
1880
1881/// The end of the document's hidden frontmatter: the last `metadata` child of
1882/// `doc`, which is what [`top_level`] and [`Builder::blocks`] drop. `None` when
1883/// there is none.
1884fn metadata_end_of(nodes: &[FlatNode], doc: usize) -> Option<usize> {
1885 let mut end = None;
1886 let mut child = nodes[doc].first_child;
1887 while let Some(cid) = child {
1888 let n = &nodes[cid.0 as usize];
1889 if n.kind == Kind::Metadata {
1890 end = Some(n.span.end);
1891 }
1892 child = n.next_sibling;
1893 }
1894 end
1895}
1896
1897/// The document's rendered top-level blocks, as node indices in source order.
1898///
1899/// Not simply `doc`'s children, for two reasons. Frontmatter (a leading
1900/// `metadata` block) is document metadata rather than prose and is dropped, the
1901/// way [`Builder::blocks`] drops it. And a **footnote definition** (`[^1]: …`)
1902/// is not a child of `doc` at all: twig parses it as a root of its own, a
1903/// *sibling* of the document node with `parent == None`. A walk that starts at
1904/// `doc` therefore never reaches one, which is why a definition — and every
1905/// byte of its body — used to render as nothing at all. Merging the roots back
1906/// in by `span.start` puts each definition on screen exactly where it was
1907/// written, which is what keeps rows, stops, and offsets monotonic.
1908///
1909/// Only `footnote` roots are merged. twig also leaves stray orphan `str` nodes
1910/// parented to nothing (the `*` of an emphasis run, for one); those are already
1911/// rendered as part of the subtree that owns their bytes, and re-emitting them
1912/// here would double them.
1913fn top_level(nodes: &[FlatNode], doc: usize) -> Vec<usize> {
1914 let mut out = Vec::new();
1915 let mut child = nodes[doc].first_child;
1916 while let Some(cid) = child {
1917 let n = &nodes[cid.0 as usize];
1918 if n.kind != Kind::Metadata {
1919 out.push(cid.0 as usize);
1920 }
1921 child = n.next_sibling;
1922 }
1923 out.extend(
1924 nodes
1925 .iter()
1926 .enumerate()
1927 .filter(|(_, n)| n.kind == Kind::Footnote && n.parent.is_none())
1928 .map(|(i, _)| i),
1929 );
1930 out.sort_by_key(|&i| nodes[i].span.start);
1931 out
1932}
1933
1934/// The top-level blocks to hand [`build_cached`] / [`build_spliced`] — the
1935/// incremental path's twin of [`top_level`], which the two must agree with block
1936/// for block or the render paths diverge.
1937///
1938/// `child_spans(None)` gives `doc`'s children, which is all of them for an
1939/// ordinary document. A **footnote definition** is not one: twig parses `[^1]: …`
1940/// as a root beside `doc` with no parent, and indexes it at no offset either —
1941/// `node_at` inside its bytes answers `doc`, and a `query("footnote")` selector
1942/// finds nothing. Leaf used to discover them by marshalling the whole arena with
1943/// `nodes()` — the very cost the incremental path exists to avoid — behind a
1944/// byte-scan gate that gave documents with no `[^…]:` line a substring search
1945/// instead. twig 3.0's `definitions()` asks the library the question directly,
1946/// so both the marshal and the gate are gone.
1947///
1948/// Filtered to [`Kind::Footnote`]: `definitions()` also reports the *link*
1949/// reference definitions (`[foo]: /url`), which leaf has never rendered as
1950/// blocks and which are not this change's business to start rendering.
1951///
1952/// This is the one part of the render that needs an [`Editor`] rather than a
1953/// marshalled node array. The builders themselves stay editor-free; this only
1954/// prepares their input.
1955pub(crate) fn top_blocks(editor: &mut Editor) -> Vec<QueryMatch> {
1956 let mut top = editor.child_spans(None).unwrap_or_default();
1957 let notes = footnote_definitions(editor);
1958 if notes.is_empty() {
1959 return top;
1960 }
1961 top.extend(notes);
1962 // Source order — what every offset-keyed thing downstream (rows, stops, the
1963 // splice path's block-for-block match) is built to assume.
1964 top.sort_by_key(|m| m.span.start);
1965 top
1966}
1967
1968/// Every `[^label]: …` definition in the document, in whatever order twig
1969/// reports them.
1970///
1971/// Filtered to [`Kind::Footnote`]: `definitions()` also reports the *link*
1972/// reference definitions (`[foo]: /url`), which leaf has never rendered as
1973/// blocks and which are not this function's business.
1974///
1975/// Empty when the document can't be walked, which leaves [`top_blocks`] with
1976/// the ordinary top-level children and [`crate::Doc::footnote_at_caret`] with an
1977/// undefined reference — in both cases the same answer as a document that has
1978/// no definitions, which is the right way to degrade.
1979pub(crate) fn footnote_definitions(editor: &mut Editor) -> Vec<QueryMatch> {
1980 let Ok(mut doc) = editor.document() else {
1981 return Vec::new();
1982 };
1983 doc.definitions()
1984 .unwrap_or_default()
1985 .into_iter()
1986 .filter(|m| m.kind == Kind::Footnote)
1987 .collect()
1988}
1989
1990/// The label of the footnote definition starting at `start` — the `1` in
1991/// `[^1]: …`. twig gives the `footnote` node no label of its own (no `text`, no
1992/// `name`), and the bytes that spell it belong to no child node either — the
1993/// body `para` starts its *content* past them — so the source is the only place
1994/// to read it from. `None` when what's there isn't a definition after all.
1995pub(crate) fn footnote_label(source: &str, start: usize) -> Option<&str> {
1996 let rest = source.get(start..)?.strip_prefix("[^")?;
1997 let end = rest.find("]:")?;
1998 Some(&rest[..end])
1999}
2000
2001/// Where the body of the footnote definition spanning `span` sits in `source` —
2002/// everything past the `[^1]:` marker, which is the part a reader actually wants
2003/// when they follow a reference.
2004///
2005/// Source bytes, verbatim but for the whitespace trimmed off each end: a note
2006/// that says `see *later*` answers with the asterisks in. Rendering that body is
2007/// a frontend's business the same way painting a [`Role`] is, and a caller that
2008/// wants it laid out already has the definition on screen where it was written.
2009///
2010/// The trim is what makes the common case read right — `[^1]: text` has a space
2011/// after the colon that belongs to the marker, not the note, and a definition's
2012/// span runs to the newline ending it.
2013///
2014/// The span is taken at its word, which it has only been safe to do since twig
2015/// 3.1: a djot definition's span used to run *past* its own last line, through
2016/// the blank line separating it from the next block and into that block's first
2017/// byte, so `[^2a]: a note.` came back as `"a note.\n\n["` and the offsets named
2018/// the following note's rows as well as this one's — a reader asking about one
2019/// footnote was shown two. leaf measured the body itself to get around that, and
2020/// paid for it: the scan stopped at the first blank line, so a note with a second
2021/// indented paragraph lost it. Both halves go away with the fix, since a blank
2022/// line *inside* a definition was always interior to the span and still is.
2023///
2024/// A range rather than a slice because "go to note" needs the *position* as much
2025/// as the text, and it needs the position of the body specifically: a
2026/// definition's `[^1]:` marker is decoration the caret can't occupy (the rich
2027/// view draws it as `[1] ` and gives it no stop), so aiming a caret at the
2028/// definition's first byte lands it on the nearest real stop instead — which is
2029/// up in the paragraph *above* the note. The body's first byte is a stop, and is
2030/// where a reader following a reference wants to arrive anyway.
2031pub(crate) fn footnote_body_span(source: &str, span: Range<usize>) -> Option<Range<usize>> {
2032 let rest = source.get(span.clone())?.strip_prefix("[^")?;
2033 let marker = rest.find("]:")?;
2034 // `span.start` + `[^` + the label + `]:`.
2035 let after_marker = span.start + 2 + marker + 2;
2036 let raw = source.get(after_marker..span.end)?;
2037 // Written as a start plus a length so an all-whitespace body lands on an
2038 // empty range at the end rather than an inverted one.
2039 let start = after_marker + (raw.len() - raw.trim_start().len());
2040 Some(start..start + raw.trim().len())
2041}
2042
2043/// The label of the footnote *reference* spanning `span` — the `1` in `[^1]`.
2044///
2045/// The peer of [`footnote_label`] for the other half of the pair, and needed for
2046/// the same reason: a reference whose node carries neither a `content_span` nor
2047/// a `text` still spells its label plainly in the source. `None` when the bytes
2048/// aren't a reference after all.
2049pub(crate) fn footnote_reference_label(source: &str, span: Range<usize>) -> Option<&str> {
2050 let rest = source.get(span)?.strip_prefix("[^")?;
2051 let end = rest.find(']')?;
2052 Some(&rest[..end])
2053}
2054
2055/// Where a heading's *content* starts — past the `#`s and the space the rich
2056/// view hides, for an ATX heading; the block's own start for a setext one (which
2057/// has no leading marker) and for a format that spells headings some other way.
2058///
2059/// Only an empty heading needs asking: with any content at all, the row ends on
2060/// its last glyph. Bounded to the heading's own first line so a marker-less
2061/// heading can't scan into the text under it.
2062fn heading_content_start(source: &str, span: &Range<usize>) -> usize {
2063 let end = span.end.min(source.len());
2064 let Some(line) = source.get(span.start..end) else {
2065 return span.start;
2066 };
2067 let line = line.split('\n').next().unwrap_or("");
2068 let hashes = line.len() - line.trim_start_matches('#').len();
2069 if hashes == 0 {
2070 return span.start;
2071 }
2072 let after = &line[hashes..];
2073 span.start + hashes + (after.len() - after.trim_start_matches([' ', '\t']).len())
2074}
2075
2076struct Builder<'a> {
2077 nodes: &'a [FlatNode],
2078 /// The document source, consulted to place blank-line rows at the source
2079 /// offsets the caret should occupy on them (the AST drops blank lines).
2080 source: &'a str,
2081 /// The word-wrap column budget, or `None` to emit each block as a single
2082 /// unwrapped row (the frontend wraps).
2083 wrap: Option<usize>,
2084 rows: Vec<VRow>,
2085 /// Built alongside `rows`, never instead of them — see [`TableInfo`].
2086 tables: Vec<TableInfo>,
2087 /// The end offset of the last content emitted — the anchor for blank
2088 /// separator rows so the caret never snaps onto one.
2089 last_off: usize,
2090 /// How many rows each block image reserves, keyed by its destination — the
2091 /// frontend's per-image height, threaded in from [`crate::Doc::set_media_rows`]
2092 /// so [`Builder::block_media`] can size the placeholder without core doing any
2093 /// I/O. A destination absent from the map (or a `0`/`1` entry) reserves the
2094 /// bare one-row placeholder, which is the whole-document default and what
2095 /// every existing test — passing an empty map — still gets.
2096 media_rows: &'a HashMap<String, usize>,
2097 /// The glyph a hard break renders as while the current inline run is built:
2098 /// a space in prose (a break folds into the flow the frontend wraps), but a
2099 /// newline (`\n`) inside a table cell, where a row is one source line and the
2100 /// only break it can carry is an explicit one that must show as a line of its
2101 /// own. Set around [`Builder::row_cells`] and otherwise left at `' '`.
2102 break_glyph: Cell<char>,
2103 /// Render a soft break (a bare newline inside a paragraph) as a line break
2104 /// where it was written, rather than folding it into the reflowed paragraph
2105 /// — the `LineFlow::Preserve` behaviour. A soft break emits a `'\n'` glyph
2106 /// (like a hard break in a cell), which [`Builder::emit_wrapped`] turns into
2107 /// a fresh visual row. `false` is the flowing-prose default. Inside a table
2108 /// cell (where `break_glyph` is already `'\n'`) it has no effect: a cell is
2109 /// one line and folds its own soft breaks regardless.
2110 preserve_soft: bool,
2111 /// The source byte range of the one line that should render its markup
2112 /// *raw* — the caret's line under `MarkupMode::Full` (see
2113 /// [`crate::Doc::reveal_line`]). `None` in every other mode and view, which
2114 /// is the delimiters-always-hidden behaviour every build had before the
2115 /// preference existed.
2116 ///
2117 /// Read only by [`Builder::revealed`], which every delimiter-bearing arm of
2118 /// [`Builder::inline`] consults. A range rather than a bare caret offset
2119 /// because the decision is per-*node*, not per-caret: a node is revealed
2120 /// when its span meets this line, so `*em*` shows both its asterisks even
2121 /// with the caret at one end of it.
2122 reveal: Option<Range<usize>>,
2123}
2124
2125impl Builder<'_> {
2126 /// Whether `span` belongs to the line that is showing its raw markup. True
2127 /// only when a reveal line is set (`MarkupMode::Full`) and the two ranges
2128 /// actually meet.
2129 ///
2130 /// Touching at an endpoint counts: an emphasis ending exactly where the line
2131 /// does is on that line, and a zero-length reveal range (the caret alone on
2132 /// a blank line) still meets a node that starts there. The test is
2133 /// deliberately generous — the failure it avoids is revealing one delimiter
2134 /// of a pair while hiding the other, which looks like corruption rather than
2135 /// like markup.
2136 fn revealed(&self, span: &Range<usize>) -> bool {
2137 self.reveal
2138 .as_ref()
2139 .is_some_and(|r| span.start <= r.end && r.start <= span.end)
2140 }
2141
2142 /// The `(opening, closing)` source byte ranges of a node's delimiters — the
2143 /// bytes its `span` holds that its `content_span` doesn't.
2144 ///
2145 /// This is how *every* inline delimiter is recovered, rather than a table of
2146 /// spellings per kind: twig gives `*em*` a span of `13..17` and a content
2147 /// span of `14..16`, so the gaps at each end are the delimiters, whatever
2148 /// they happen to be. That matters because one kind has many spellings —
2149 /// `*em*` and `_em_` are both emphasis, `` `x` `` and ``` ``x`` ``` both
2150 /// verbatim — and re-deriving the text from the source is the only way to
2151 /// show back what the author actually typed. It also gets a link's
2152 /// asymmetric `[` / `](dest)` right for free.
2153 ///
2154 /// `None` when the node has no content span, or when content and span
2155 /// coincide (nothing was elided, so there is nothing to reveal).
2156 fn delims(&self, id: usize) -> Option<(Range<usize>, Range<usize>)> {
2157 let node = &self.nodes[id];
2158 let content = node.content_span.clone()?;
2159 let span = node.span.clone();
2160 // A content span that escapes its own node's span means the two are
2161 // describing different things; reveal nothing rather than slice wildly.
2162 if content.start < span.start || content.end > span.end {
2163 return None;
2164 }
2165 let (open, close) = (span.start..content.start, content.end..span.end);
2166 // A delimiter that spans a newline isn't this line's to reveal — a setext
2167 // heading's `\n=====` underline is the case that arises in practice. It
2168 // would also inject a `'\n'` glyph, which `emit_wrapped` reads as a hard
2169 // row break, so the row would split where the author wrote no break.
2170 let multiline =
2171 |r: &Range<usize>| self.source.get(r.clone()).is_some_and(|s| s.contains('\n'));
2172 if multiline(&open) || multiline(&close) {
2173 return None;
2174 }
2175 (!open.is_empty() || !close.is_empty()).then_some((open, close))
2176 }
2177
2178 /// Emit the source bytes of `range` as revealed markup — real glyphs, each
2179 /// mapped to its own source byte and each a caret stop, so a delimiter shown
2180 /// is a delimiter that can be selected, edited and deleted like any other
2181 /// text. Styled [`Role::Delimiter`] on top of the run's own style, which is
2182 /// how a frontend tells scaffolding from prose and dims it.
2183 ///
2184 /// Deliberately *not* [`push_escaped_text`]: this is raw source, not parsed
2185 /// text, so there is no escape-driven drift between the two to correct.
2186 fn push_delim(&self, out: &mut Vec<Glyph>, range: &Range<usize>, base: Style) {
2187 let Some(text) = self.source.get(range.clone()) else {
2188 return;
2189 };
2190 push_text(out, text, range.start, base.role(Role::Delimiter));
2191 }
2192
2193 /// Render an inline node's children wrapped in its raw delimiters when the
2194 /// node is on the revealed line, and bare (delimiters resolved away) when it
2195 /// isn't — the shared body of every delimiter-bearing arm of
2196 /// [`inline`](Self::inline).
2197 ///
2198 /// `style` is the resolved styling the content still gets in *both* modes:
2199 /// revealing `*em*` shows the asterisks *and* keeps the text italic, the
2200 /// live-preview behaviour. Showing the markup is not the same as turning the
2201 /// rendering off — that is what [`crate::View::Source`] is for.
2202 fn inline_delimited(&self, id: usize, style: Style, out: &mut Vec<Glyph>) {
2203 let show = self
2204 .revealed(&self.nodes[id].span)
2205 .then(|| self.delims(id))
2206 .flatten();
2207 if let Some((open, _)) = &show {
2208 self.push_delim(out, open, style);
2209 }
2210 self.recurse(id, style, out);
2211 if let Some((_, close)) = &show {
2212 self.push_delim(out, close, style);
2213 }
2214 }
2215
2216 fn children(&self, id: usize) -> Vec<usize> {
2217 let mut out = Vec::new();
2218 let mut c = self.nodes[id].first_child;
2219 while let Some(cid) = c {
2220 out.push(cid.0 as usize);
2221 c = self.nodes[cid.0 as usize].next_sibling;
2222 }
2223 out
2224 }
2225
2226 /// Render a node's block children, a blank separator between each. `tight`
2227 /// suppresses the *fabricated* separator between adjacent children that share
2228 /// a source line boundary — a tight list item and the sub-list nested in it —
2229 /// while a real blank source line between them still opens a gap.
2230 fn blocks(&mut self, id: usize, pf: &[Glyph], pc: &[Glyph], tight: bool) {
2231 // Frontmatter (a leading `metadata` block) is document metadata, not
2232 // prose: hide it entirely in the rich-text view. Skipping it here means
2233 // no phantom blank rows for its lines and no separator before the first
2234 // real block — the document opens straight into its content.
2235 let kids: Vec<usize> = self
2236 .children(id)
2237 .into_iter()
2238 .filter(|&c| self.nodes[c].kind != Kind::Metadata)
2239 .collect();
2240 let mut above: Option<BlockClass> = None;
2241 for (i, child) in kids.into_iter().enumerate() {
2242 let below = BlockClass::from_node_kind(&self.nodes[child].kind);
2243 if let Some(above) = above {
2244 self.emit_separators_before(
2245 self.nodes[child].span.start,
2246 pc,
2247 !tight,
2248 Boundary { above, below },
2249 );
2250 }
2251 let first = if i == 0 { pf } else { pc };
2252 self.block(child, first, pc);
2253 above = Some(below);
2254 }
2255 }
2256
2257 /// Render an explicit, ordered list of top-level blocks — [`Builder::blocks`]
2258 /// for a walk that isn't "the children of one node". The document's top level
2259 /// no longer is: a footnote definition is a root beside `doc`, not under it,
2260 /// and [`top_level`] merges it into this list by source position.
2261 ///
2262 /// The separator between blocks is spelled by the same
2263 /// [`Builder::emit_separators_before`] the incremental top-level walk in
2264 /// [`build_cached`] uses, so the two paths can't drift on how a boundary
2265 /// looks.
2266 fn top_blocks(&mut self, ids: &[usize]) {
2267 for (i, &child) in ids.iter().enumerate() {
2268 let below = BlockClass::from_node_kind(&self.nodes[child].kind);
2269 if i > 0 {
2270 let above = BlockClass::from_node_kind(&self.nodes[ids[i - 1]].kind);
2271 self.emit_separators_before(
2272 self.nodes[child].span.start,
2273 &[],
2274 true,
2275 Boundary { above, below },
2276 );
2277 }
2278 self.block(child, &[], &[]);
2279 }
2280 }
2281
2282 /// Emit the blank separator row(s) that sit between a block ending at the
2283 /// current `last_off` and the next block starting at `next_start`, wearing
2284 /// the continuation prefix `pc`. Shared by [`Builder::blocks`] and the
2285 /// incremental top-level walk so the two can't drift on how a boundary is
2286 /// spelled.
2287 ///
2288 /// The blank line(s) between two blocks are real caret stops, each needing
2289 /// its *own* source offset — one strictly past the previous block's content,
2290 /// else it collides with that block's last row and `pos_of_offset`
2291 /// (first-match-wins) would resolve the caret onto the wrong row, pinning
2292 /// downward motion there.
2293 ///
2294 /// One row *per* blank source line, not a single collapsed separator: an
2295 /// empty paragraph opened between two blocks (Enter in the gap,
2296 /// `…\n\n\n\n…`) must be a navigable empty row, not vanish — else the caret
2297 /// in it snaps onto the *next* block's start and Enter looks like it did
2298 /// nothing.
2299 fn emit_separators_before(
2300 &mut self,
2301 next_start: usize,
2302 pc: &[Glyph],
2303 synthetic: bool,
2304 boundary: Boundary,
2305 ) {
2306 let mut offs = self.blank_rows_between(self.last_off, next_start);
2307 if offs.is_empty() {
2308 if !synthetic {
2309 // A tight list item's own text sits directly above the sub-list
2310 // nested in it — no fabricated gap. The "breathe" row belongs
2311 // between free-standing blocks, not between an item and its
2312 // child list, which the source writes on the very next line. A
2313 // real blank source line (a loose list) still lands a gap below,
2314 // because `blank_rows_between` found it and we never reach here.
2315 return;
2316 }
2317 // A tight gap with no blank line (e.g. a heading directly above its
2318 // text): keep the one conventional separator row so blocks still
2319 // breathe, as they always have.
2320 offs.push(self.blank_line_offset(self.last_off, next_start));
2321 }
2322 let last = offs.len() - 1;
2323 for (k, end_src) in offs.into_iter().enumerate() {
2324 // Only the drawn-only rows carry the boundary: the navigable blank
2325 // lines between them (and every blank line under preserve-soft flow)
2326 // are somewhere text can go, not a gap between blocks, and a frontend
2327 // that shrank one would be shrinking a line the author is typing on.
2328 let drawn = !self.preserve_soft && (k == 0 || k == last);
2329 // The blank line a boundary is *drawn* with isn't a place text can
2330 // go. The first one closes the block above and the last one opens the
2331 // block below — with a single blank line, the usual case, doing both
2332 // at once. Typing on either just continues the paragraph it abuts,
2333 // since the blank line it would need to be a paragraph of its own is
2334 // the very line being typed on. So they're a gap, like a table's
2335 // border: drawn, clickable, never a caret's home.
2336 //
2337 // The lines *between* them are the real ones. That's what Enter
2338 // opens: it inserts a paragraph break (`\n\n`), which leaves a blank
2339 // line spare on each side and the caret on the navigable line
2340 // between them.
2341 //
2342 // Preserve flow is the exception: there a bare `\n` is a visible line
2343 // break the author edits directly, so a lone blank line *is* a caret
2344 // home — typing on it makes the soft break the mode exists to show,
2345 // and Enter at a line's end lands the caret on exactly this row. So no
2346 // separator is drawn-only; every blank line is navigable.
2347 self.rows.push(VRow {
2348 glyphs: pc.to_vec(),
2349 end_src,
2350 decoration: drawn,
2351 code: false,
2352 code_lang: None,
2353 directive: false,
2354 directive_label: None,
2355 media: None,
2356 task: None,
2357 leaf_directive: None,
2358 heading: None,
2359 boundary: drawn.then_some(boundary),
2360 });
2361 }
2362 }
2363
2364 fn block(&mut self, id: usize, pf: &[Glyph], pc: &[Glyph]) {
2365 let node = &self.nodes[id];
2366 match node.kind.as_str() {
2367 "doc" | "section" => self.blocks(id, pf, pc, false),
2368 "heading" => {
2369 // A heading whose only visible content is a single image — a
2370 // banner set in an `<h1>` (`<h1><picture><img></picture></h1>`),
2371 // or `# ` — is a block picture, not text. Render
2372 // it as one; anything with real heading text falls through.
2373 if let Some((m, kind)) = self.media_only(id) {
2374 self.block_media(m, kind, id, pf);
2375 return;
2376 }
2377 let level = node.level.unwrap_or(1);
2378 let style = heading_style(level);
2379 let mut glyphs = Vec::new();
2380 // On the revealed line the `# ` comes back as real, editable
2381 // text in front of the heading. Only the opening marker: a
2382 // closing `#`-run (`## title ##`) is covered by the same
2383 // `delims` pair, and a setext underline is excluded there for
2384 // being on another line entirely.
2385 if let Some((open, close)) =
2386 self.revealed(&node.span).then(|| self.delims(id)).flatten()
2387 {
2388 self.push_delim(&mut glyphs, &open, style);
2389 glyphs.extend(self.inline_children_with_trailing(id, style));
2390 self.push_delim(&mut glyphs, &close, style);
2391 } else {
2392 glyphs = self.inline_children_with_trailing(id, style);
2393 }
2394 // An *empty* heading — `# ` with nothing typed after it, which is
2395 // what the toolbar's H1 leaves on a blank line — has no glyph for
2396 // its row to end on, so the fallback below is the row's whole
2397 // extent: its only caret stop, and the offset every row after it
2398 // is measured from. The block's start is the wrong answer for
2399 // both, because it sits *in front of* the `# ` the rich view
2400 // hides: the caret drew (and typed) before the hashes, and the
2401 // rows below inherited an offset short by the marker's length,
2402 // which put the caret on one of them the moment the heading grew
2403 // text. Its content's start is where the caret belongs.
2404 let home = heading_content_start(self.source, &node.span);
2405 let first = self.rows.len();
2406 self.emit_wrapped(glyphs, home, pf, pc);
2407 // Stamp the level on every row the heading just emitted — a
2408 // wrapped heading's continuation rows as much as its first, and
2409 // an empty one's single glyphless row, which is the whole point
2410 // (see [`VRow::heading`]).
2411 for row in &mut self.rows[first..] {
2412 row.heading = Some(level.min(255) as u8);
2413 }
2414 }
2415 "block_quote" => {
2416 let (start, end) = (node.span.start, node.span.end);
2417 let gutter = synth("│ ", Role::QuoteGutter, start);
2418 let f = concat(pf, &gutter);
2419 let c = concat(pc, &gutter);
2420 // A childless quote — a bare `> ` on an otherwise blank line,
2421 // which is what the toolbar's Quote button leaves there — has no
2422 // inner block to carry the gutter or a caret home, so `blocks`
2423 // emitted *nothing at all*: the quote didn't merely draw
2424 // unstyled, it disappeared, and a document that was only `> `
2425 // rendered zero rows with the caret nowhere to stand. Emit the
2426 // gutter row itself, ending just past the marker, exactly as an
2427 // empty `list_item` emits its bare bullet.
2428 if self.children(id).is_empty() {
2429 self.push_row_at(f, end.min(self.source.len()));
2430 } else {
2431 self.blocks(id, &f, &c, false);
2432 self.emit_quote_trailing_lines(&c, end);
2433 }
2434 }
2435 // A generic `:::name{.class}` fenced-div container (twig's
2436 // `directive`, container form). Core is agnostic of `name` — it's
2437 // the host app's vocabulary (diaryx's `vis` for audience
2438 // visibility, say) and isn't available here regardless: twig only
2439 // threads an `element`'s tag name through `FlatNode::name`, not a
2440 // directive's own identifier. Every row gets marked `directive` (a
2441 // frontend draws a tinted panel around each maximal run, the
2442 // `code`/`code_block` recipe) and the first row carries a label —
2443 // the way a code fence's language rides only its first row.
2444 //
2445 // The label reads BOTH attribute conventions diaryx content
2446 // actually uses: twig's own dot-prefixed classes (`{.public
2447 // .family}`, one combined `class` attr) and bare pandoc-style
2448 // words with no leading dot (`{public family}` — the syntax
2449 // `diaryx_core::visibility`'s hand-rolled publish-time filter and
2450 // apps/web's directive serializer both write; twig parses each
2451 // bare word as its own attribute with an empty value, per
2452 // `languages/markdown/attributes.zig`). Reading only `.class`
2453 // would leave every *existing* diaryx `:::vis{...}` block
2454 // unlabeled.
2455 // Only the *container* form is the panel below. A `text` directive
2456 // is inline and never reaches the block walker (see `is_inline`); a
2457 // `leaf` one is a standalone block with no body, drawn as a
2458 // placeholder the way an image is.
2459 "container"
2460 if container_is_directive(node)
2461 && node.directive_form == Some(DirectiveForm::Leaf) =>
2462 {
2463 self.block_directive(id, pf);
2464 }
2465 "container" if container_is_directive(node) => {
2466 let label = directive_attr_label(&node.attrs);
2467 let start_row = self.rows.len();
2468 self.blocks(id, pf, pc, false);
2469 for (i, row) in self.rows[start_row..].iter_mut().enumerate() {
2470 row.directive = true;
2471 if i == 0 {
2472 row.directive_label = label.clone();
2473 }
2474 }
2475 // Anchor the block's end past its closing `:::` fence, exactly as
2476 // the code-block arm anchors past its ```` ``` ````. A container's
2477 // last content row ends at its last *child*, before the fence and
2478 // the blank line under it, so the separator logic counted the
2479 // fence line as a blank row of its own and drew a second boundary
2480 // — one gap's worth of margin twice, under every fenced div.
2481 self.last_off = node.span.end;
2482 }
2483 "bullet_list" | "ordered_list" | "task_list" => {
2484 let ordered = node.kind == Kind::OrderedList;
2485 let mut item_no = 0usize;
2486 let kids = self.children(id);
2487 for (i, child) in kids.iter().copied().enumerate() {
2488 let kind = &self.nodes[child].kind;
2489 if *kind == Kind::ListItem || *kind == Kind::TaskListItem {
2490 let start = self.nodes[child].span.start;
2491 item_no += 1;
2492 // A task item's box replaces the bullet rather than
2493 // joining it. The `[ ] ` that spells it is markup twig
2494 // has already consumed — the item's paragraph *content*
2495 // starts past it — so without a drawn box a task item
2496 // was indistinguishable from a plain bullet, ticked or
2497 // not. `☐`/`☑` is the marker for the same reason `•` is:
2498 // it stands where the source's own marker stands. Which
2499 // way it faces is `checked`, straight off the node.
2500 let checked = self.nodes[child].checked;
2501 let marker = match (checked, ordered) {
2502 (Some(true), _) => "☑ ".to_string(),
2503 (Some(false), _) => "☐ ".to_string(),
2504 (None, true) => format!("{item_no}. "),
2505 (None, false) => "• ".to_string(),
2506 };
2507 let bullet = synth(&marker, Role::ListMarker, start);
2508 let indent = synth(&" ".repeat(text_width(&marker)), Role::Body, start);
2509 let first_row = self.rows.len();
2510 self.block(child, &concat(pc, &bullet), &concat(pc, &indent));
2511 // On the item's first row, the way `code_lang` rides the
2512 // first row of its block.
2513 if let (Some(c), Some(row)) = (checked, self.rows.get_mut(first_row)) {
2514 row.task = Some(c);
2515 }
2516 } else {
2517 // twig can nest a *following* top-level block as a direct
2518 // child of the list rather than a sibling of it — e.g.
2519 // `- item\n\n> quote` parses the block quote under the
2520 // `bullet_list`. It isn't a list item, so render it de-nested:
2521 // no bullet, at the list's own prefix, with the usual block
2522 // separator — never `• │ quote`.
2523 if i > 0 {
2524 self.emit_separators_before(
2525 self.nodes[child].span.start,
2526 pc,
2527 true,
2528 Boundary {
2529 above: BlockClass::from_node_kind(
2530 &self.nodes[kids[i - 1]].kind,
2531 ),
2532 below: BlockClass::from_node_kind(&self.nodes[child].kind),
2533 },
2534 );
2535 }
2536 self.block(child, pc, pc);
2537 }
2538 }
2539 }
2540 "list_item" | "task_list_item" => {
2541 // A childless item — the empty bullet you get the instant you
2542 // press Enter to open a new one — has no inner block to carry the
2543 // marker prefix or a caret home, so `blocks` would emit nothing
2544 // and the new bullet simply wouldn't appear until something was
2545 // typed into it. Emit the prefixed row itself, ending at a caret
2546 // stop just past the marker (the item's `span.end`), the way an
2547 // empty paragraph emits its one prefixed row via `emit_wrapped`.
2548 if self.children(id).is_empty() {
2549 let home = self.nodes[id].span.end.min(self.source.len());
2550 self.push_row_at(pf.to_vec(), home);
2551 } else {
2552 // Tight: an item's text and the list nested under it butt
2553 // together (`• a` / ` • b`), no fabricated blank row between —
2554 // a loose item's real blank line still parts them.
2555 self.blocks(id, pf, pc, true);
2556 }
2557 }
2558 // A footnote *definition* (`[^1]: the note`). It reaches this walker
2559 // only because [`top_level`] merges it back in — twig hangs it off no
2560 // parent at all, so a walk from `doc` never sees one and every byte
2561 // of its body used to render as nothing.
2562 //
2563 // Drawn as a hanging-indent item, the way a list item is: the marker
2564 // reads `[1] `, matching the `[1]` its references render as, so the
2565 // two can be paired by eye, and the body wraps under it. The marker
2566 // is synthetic decoration (one shared offset, never a caret stop) —
2567 // the `[^1]: ` that spells it in the source is markup, hidden like a
2568 // heading's `# `.
2569 "footnote" => {
2570 let (start, end) = (node.span.start, node.span.end);
2571 let source = self.source;
2572 let marker = format!("[{}] ", footnote_label(source, start).unwrap_or(""));
2573 let indent = " ".repeat(text_width(&marker));
2574 let f = concat(pf, &synth(&marker, Role::ListMarker, start));
2575 let c = concat(pc, &synth(&indent, Role::Body, start));
2576 if self.children(id).is_empty() {
2577 // A definition with no body yet — the instant `[^1]: ` has
2578 // been typed and nothing after it. `blocks` would emit
2579 // nothing and the definition simply wouldn't appear, so emit
2580 // the marker row itself with a caret home just past it,
2581 // exactly as an empty list item does.
2582 self.push_row_at(f, end.min(source.len()));
2583 } else {
2584 self.blocks(id, &f, &c, false);
2585 }
2586 }
2587 "table" => self.table(id, pf, pc),
2588 "code_block" => {
2589 let style = Style::default().role(Role::Code);
2590 let text = node.text.clone().unwrap_or_default();
2591 let lines: Vec<&str> = text.trim_end_matches('\n').split('\n').collect();
2592 // Each line at its own source offset, so the caret can walk the
2593 // code a character at a time like any other text. Where the
2594 // lines can't be lined up with the source there's no honest
2595 // offset to give, so the block maps coarsely to its start (and
2596 // stays a source-view job, as all of it once was).
2597 let offs = node
2598 .content_span
2599 .as_ref()
2600 .and_then(|c| self.code_line_offsets(c, &lines));
2601 // The fence's info string, carried on the block's first row as
2602 // its language label (`None` for an indented block or a bare
2603 // fence). Kept on the row so it rides the block cache.
2604 let lang = code_language(self.source, node.span.start);
2605 for (i, raw) in lines.iter().enumerate() {
2606 let at = offs.as_ref().map_or(node.span.start, |o| o[i]);
2607 // No gutter glyph: the block is set apart by the border and
2608 // tint a frontend draws around the whole run of `code` rows,
2609 // not by a per-line mark. Just the block prefix (a list
2610 // indent, a quote gutter) and the code text.
2611 let mut glyphs: Vec<Glyph> = pf.to_vec();
2612 push_text(&mut glyphs, raw, at, style);
2613 // Explicitly past the line's *text*: a blank code line has no
2614 // glyph, and any prefix's offset would put the row's end
2615 // inside the next line.
2616 self.push_row_at(glyphs, at + raw.len());
2617 if let Some(row) = self.rows.last_mut() {
2618 row.code = true;
2619 if i == 0 {
2620 row.code_lang = lang.clone();
2621 }
2622 }
2623 }
2624 // Anchor the block's end past its closing fence. Its last content
2625 // row ends at the last code line, before the ``` and the blank
2626 // line under it; without this the separator logic would count the
2627 // closing-fence line as its own blank row and open a phantom
2628 // second gap below the block.
2629 self.last_off = node.span.end;
2630 }
2631 "thematic_break" => {
2632 let full = self.wrap.unwrap_or(UNWRAPPED_RULE_WIDTH);
2633 let w = full.saturating_sub(prefix_width(pf)).max(4);
2634 let mut glyphs = pf.to_vec();
2635 for _ in 0..w {
2636 glyphs.push(Glyph {
2637 ch: '─',
2638 style: Style::default().role(Role::Rule),
2639 src: node.span.start,
2640 // A rule is a block the caret can sit on, as it always
2641 // has; it maps coarsely to the block's start.
2642 stop: true,
2643 });
2644 }
2645 // The dashes share one caret home in front of the atomic block,
2646 // while the row's end is the second home just past its source.
2647 // Without that trailing stop a final rule made the document end
2648 // unreachable: Right could not cross it and a click in the
2649 // empty space below it snapped back before the rule.
2650 let after_line = node.span.end
2651 + self.source[node.span.end..]
2652 .strip_prefix("\r\n")
2653 .map_or_else(
2654 || usize::from(self.source[node.span.end..].starts_with('\n')),
2655 |_| 2,
2656 );
2657 self.push_row_at(glyphs, after_line);
2658 }
2659 // A block-level image node with no wrapping paragraph — a promoted
2660 // top-level HTML `<img>` lands as a direct `doc` child like this
2661 // (a Markdown `` comes wrapped in a `para`, handled below).
2662 "image" => self.block_media(id, MediaKind::Image, id, pf),
2663 // The same case for a promoted top-level `<video>`/`<audio>`, which
2664 // arrives as a generic `container` rather than a node kind of its
2665 // own. It can't be found by the `media_only` scan below the way a
2666 // wrapped one is: that scan looks at a wrapper's *children*, and here
2667 // the media element is itself the block.
2668 "container" if matches!(element_tag(node), Some("video") | Some("audio")) => {
2669 let kind = match element_tag(node) {
2670 Some("audio") => MediaKind::Audio,
2671 _ => MediaKind::Video,
2672 };
2673 self.block_media(id, kind, id, pf);
2674 }
2675 _ => {
2676 // A container of blocks, or an inline-bearing paragraph.
2677 let kids = self.children(id);
2678 // A block-level image: a paragraph (or other wrapper — a
2679 // `<picture>`, an `<h1>` banner) whose only visible content is a
2680 // single `image` node. Render it as a placeholder row + record an
2681 // [`MediaInfo`] a capable frontend replaces. An image mixed with
2682 // real text or other images on the line isn't block-level and
2683 // falls through to the inline path below, still as its alt text.
2684 if let Some((m, kind)) = self.media_only(id) {
2685 self.block_media(m, kind, id, pf);
2686 return;
2687 }
2688 let inline = !kids.is_empty() && kids.iter().all(|&c| is_inline(&self.nodes[c]));
2689 if inline || kids.is_empty() {
2690 let glyphs = self.inline_children_with_trailing(id, Style::default());
2691 if !glyphs.is_empty() {
2692 self.emit_wrapped(glyphs, node.span.start, pf, pc);
2693 }
2694 } else {
2695 self.blocks(id, pf, pc, false);
2696 }
2697 }
2698 }
2699 }
2700
2701 /// Render a table as a box-drawn grid: every column as wide as its widest
2702 /// cell, the header bold and ruled off, each cell padded to its column's
2703 /// alignment. This is the *default* monospace rendering (see
2704 /// [`VisualMap::rows`]); the same cells are also published structurally as
2705 /// [`TableInfo`], so a frontend that lays the grid out in its own units draws
2706 /// from there and skips the picture built here.
2707 ///
2708 /// The alignment comes from twig's `cell.alignment` — the delimiter row
2709 /// (`|:--|--:|`) that spells it out is consumed by the parser and leaves no
2710 /// node, so the snapshot is the only source for it.
2711 ///
2712 /// Borders and padding are *decoration*: they carry the source offset of the
2713 /// text they surround, so a click lands in that cell, but they're never
2714 /// caret stops — the caret steps cell-to-cell instead of into the box art.
2715 fn table(&mut self, id: usize, pf: &[Glyph], pc: &[Glyph]) {
2716 let node_end = self.nodes[id].span.end;
2717 // twig's shape is `[caption, row, row, …]`: the caption is always
2718 // present (usually empty in Markdown) and is not part of the grid.
2719 let row_ids: Vec<usize> = self
2720 .children(id)
2721 .into_iter()
2722 .filter(|&c| self.nodes[c].kind == Kind::Row)
2723 .collect();
2724 if row_ids.is_empty() {
2725 return;
2726 }
2727 // Lay every cell out first — the column widths depend on all of them.
2728 let grid: Vec<Vec<TableCell>> = row_ids.iter().map(|&r| self.row_cells(r)).collect();
2729 let heads: Vec<bool> = row_ids
2730 .iter()
2731 .map(|&r| self.nodes[r].head.unwrap_or(false))
2732 .collect();
2733 let cols = grid.iter().map(|r| r.len()).max().unwrap_or(0);
2734 if cols == 0 {
2735 return;
2736 }
2737 let mut widths = vec![0usize; cols];
2738 for row in &grid {
2739 for (c, cell) in row.iter().enumerate() {
2740 widths[c] = widths[c].max(cell_width(&cell.glyphs));
2741 }
2742 }
2743 // Every column at its widest cell is only the *wish*; a grid wider than
2744 // the surface has its far side hanging off the edge where no amount of
2745 // caret motion can reach it. Cut it down to what's actually there, and
2746 // let the cells wrap into the space they're given.
2747 if let Some(w) = self.wrap {
2748 fit_widths(&mut widths, w.saturating_sub(prefix_width(pc)));
2749 }
2750
2751 // Where the picture starts, so a frontend drawing its own grid knows
2752 // which rows to skip. Recorded before the first border goes down.
2753 let rows_start = self.rows.len();
2754
2755 let anchor = grid[0].first().map(|c| c.start).unwrap_or(node_end);
2756 self.push_rule(&rule_text(&widths, '┌', '┬', '┐'), anchor, pf);
2757 for (ri, row) in grid.iter().enumerate() {
2758 self.push_table_row(row, &widths, pc);
2759 // The rule under the header: only where the head actually ends.
2760 let ends_head = heads[ri] && heads.get(ri + 1) == Some(&false);
2761 if ends_head {
2762 let next = grid[ri + 1].first().map(|c| c.start).unwrap_or(node_end);
2763 self.push_rule(&rule_text(&widths, '├', '┼', '┤'), next, pc);
2764 }
2765 }
2766 self.push_rule(&rule_text(&widths, '└', '┴', '┘'), node_end, pc);
2767
2768 // The same cells the picture above was drawn from, published unwrapped
2769 // and unpadded for a frontend that lays them out in pixels.
2770 self.tables.push(TableInfo {
2771 rows_span: rows_start..self.rows.len(),
2772 end_src: node_end,
2773 // The *continuation* prefix: `pf` opens the block and only its first
2774 // row wears it, but every row of a grid is a continuation of the
2775 // block the table sits in.
2776 prefix: pc.to_vec(),
2777 grid: grid
2778 .into_iter()
2779 .zip(heads)
2780 .map(|(cells, head)| TableRow { head, cells })
2781 .collect(),
2782 });
2783 // The table's own end anchors whatever separator follows it; the border
2784 // rows deliberately don't move `last_off` (they hold no content).
2785 self.last_off = node_end;
2786 }
2787
2788 /// One row of laid-out cells, in column order.
2789 fn row_cells(&self, row: usize) -> Vec<TableCell> {
2790 // A cell is one source line, so a break within it is an explicit line
2791 // break (an inline `<br>`) that must render as a line of its own — not the
2792 // flow-folding space a break is in prose.
2793 self.break_glyph.set('\n');
2794 let cells = self
2795 .children(row)
2796 .into_iter()
2797 .filter(|&c| self.nodes[c].kind == Kind::Cell)
2798 .enumerate()
2799 .map(|(col, c)| {
2800 let n = &self.nodes[c];
2801 let style = if n.head.unwrap_or(false) {
2802 Style::default().bold()
2803 } else {
2804 Style::default()
2805 };
2806 // A cell's own `span` is the whole row; only `content_span` bounds
2807 // its text. An EMPTY cell has no `content_span` at all — twig
2808 // records no interior for it — so both offsets would fall back to
2809 // the row's start (before its first `│`), where every empty cell
2810 // in the row collapses onto the same spot and a click or caret
2811 // there types *before* the table. Derive the cell's own interior
2812 // from the row source and this cell's column instead, so each
2813 // empty cell has a distinct, editable caret home.
2814 let span = n.content_span.clone().unwrap_or_else(|| {
2815 let off = empty_cell_offset(
2816 &self.source[n.span.start.min(self.source.len())
2817 ..n.span.end.min(self.source.len())],
2818 n.span.start,
2819 col,
2820 );
2821 off..off
2822 });
2823 TableCell {
2824 glyphs: self.inline_children(c, style),
2825 start: span.start,
2826 end: span.end,
2827 align: n.alignment.unwrap_or(Alignment::Default),
2828 }
2829 })
2830 .collect();
2831 self.break_glyph.set(' ');
2832 cells
2833 }
2834
2835 /// A horizontal rule between/around rows — entirely decoration.
2836 fn push_rule(&mut self, text: &str, src: usize, prefix: &[Glyph]) {
2837 let glyphs = concat(prefix, &synth(text, Role::Rule, src));
2838 self.rows.push(VRow {
2839 glyphs,
2840 end_src: src,
2841 decoration: true,
2842 code: false,
2843 code_lang: None,
2844 directive: false,
2845 directive_label: None,
2846 media: None,
2847 task: None,
2848 leaf_directive: None,
2849 heading: None,
2850 boundary: None,
2851 });
2852 }
2853
2854 /// One `│ a │ b │` row of the grid: real cell text between decoration.
2855 ///
2856 /// A row of cells is not a row of the screen — a cell wrapped to its column
2857 /// spans several, each one `│`-divided across the full width so the grid
2858 /// stays square. Cells in the same row are laid out independently and run
2859 /// out at their own heights; a column that has run dry pads out as
2860 /// decoration while its neighbours keep going.
2861 fn push_table_row(&mut self, cells: &[TableCell], widths: &[usize], prefix: &[Glyph]) {
2862 let fallback = cells.last().map(|c| c.end).unwrap_or(0);
2863 let laid: Vec<Vec<Vec<Glyph>>> = cells
2864 .iter()
2865 .enumerate()
2866 .map(|(ci, c)| wrap_glyphs(&c.glyphs, widths.get(ci).copied().unwrap_or(0)))
2867 .collect();
2868 let height = laid.iter().map(|l| l.len()).max().unwrap_or(1).max(1);
2869
2870 for j in 0..height {
2871 let mut glyphs = prefix.to_vec();
2872 for (ci, &w) in widths.iter().enumerate() {
2873 let cell = cells.get(ci);
2874 let line = laid.get(ci).and_then(|l| l.get(j));
2875 // The divider before this column belongs to the cell it
2876 // introduces, so clicking it lands in that cell — on this line
2877 // of it, which is what's next to the divider being clicked.
2878 let at = line
2879 .and_then(|l| l.first().map(|g| g.src))
2880 .or_else(|| cell.map(|c| c.start))
2881 .unwrap_or(fallback);
2882 glyphs.extend(synth("│", Role::Rule, at));
2883 match (cell, line) {
2884 (Some(cell), Some(line)) => {
2885 let pad = w.saturating_sub(glyphs_width(line));
2886 let (lead, trail) = match cell.align {
2887 Alignment::Right => (pad, 0),
2888 Alignment::Center => (pad / 2, pad - pad / 2),
2889 Alignment::Left | Alignment::Default => (0, pad),
2890 };
2891 // Every line renders at least one space after its text
2892 // (the gutter before `│`), so there is always somewhere
2893 // to put the "after the last character" caret a line
2894 // needs. It's the one padding glyph that is a stop: on
2895 // the cell's last line that's the cell's end, and on any
2896 // other it's the space the wrap consumed.
2897 let last = laid[ci].len() == j + 1;
2898 let end = match last {
2899 true => cell.end,
2900 false => line
2901 .last()
2902 .map(|g| g.src + g.ch.len_utf8())
2903 .unwrap_or(cell.end),
2904 };
2905 glyphs.extend(synth(&" ".repeat(lead + 1), Role::Body, at));
2906 glyphs.extend(line.iter().cloned());
2907 glyphs.push(Glyph {
2908 ch: ' ',
2909 style: Style::default(),
2910 src: end,
2911 stop: true,
2912 });
2913 glyphs.extend(synth(&" ".repeat(trail), Role::Body, end));
2914 }
2915 // A ragged row, or a column whose cell ended higher up: pad
2916 // it out so the grid stays square.
2917 _ => {
2918 let at = cell.map(|c| c.end).unwrap_or(fallback);
2919 glyphs.extend(synth(&" ".repeat(w + 2), Role::Body, at));
2920 }
2921 }
2922 }
2923 glyphs.extend(synth("│", Role::Rule, fallback));
2924 // The row ends where its last stop does. A table row has no gap
2925 // between its final cell and the border, so inventing an end past
2926 // that would be a stop with nothing under it.
2927 let end_src = glyphs
2928 .iter()
2929 .rev()
2930 .find(|g| g.stop)
2931 .map_or(fallback, |g| g.src);
2932 self.rows.push(VRow {
2933 glyphs,
2934 end_src,
2935 decoration: false,
2936 code: false,
2937 code_lang: None,
2938 directive: false,
2939 directive_label: None,
2940 media: None,
2941 task: None,
2942 leaf_directive: None,
2943 heading: None,
2944 boundary: None,
2945 });
2946 }
2947 }
2948
2949 /// Render a block-level image, video, or audio as one placeholder row: the
2950 /// `🖼 alt` / `🎬 alt` / `🔊 alt` label styled [`Role::Image`], every glyph
2951 /// mapped to the media's start offset and a caret stop there (they share the
2952 /// offset, so the stop table dedups them to a single home in front of it, as
2953 /// a rule's dashes do), and the row's end stop set past it so the caret can
2954 /// also rest after it. The row carries a [`MediaMark`] so [`media_spans`]
2955 /// publishes it as a [`MediaInfo`] a capable frontend replaces with the real
2956 /// picture or player; a plain surface paints the label as-is. `pf` is the
2957 /// block prefix (a list indent, a quote gutter) the row opens with, exactly
2958 /// as every other block honours it.
2959 fn block_media(&mut self, img: usize, kind: MediaKind, wrapper: usize, pf: &[Glyph]) {
2960 let node = &self.nodes[img];
2961 let start = node.span.start;
2962 let end = node.span.end;
2963 // An `image`'s URL is twig's `destination`; a `<video>`/`<audio>` is a
2964 // generic element, so its URL is the `src` attribute — and may be absent
2965 // entirely, the element naming its candidates in child `<source>`s.
2966 let destination = match kind {
2967 MediaKind::Image => node.destination.clone().unwrap_or_default(),
2968 MediaKind::Video | MediaKind::Audio => attr_of(node, "src").unwrap_or_default(),
2969 };
2970 let poster = match kind {
2971 MediaKind::Video => attr_of(node, "poster").unwrap_or_default(),
2972 MediaKind::Image | MediaKind::Audio => String::new(),
2973 };
2974 // The `<source>`s under the media element itself, not under `wrapper`: a
2975 // `<video>` is its own container, unlike an `<img>`, whose `<picture>`
2976 // alternatives are its *siblings* and so only reachable from the wrapper.
2977 let sources = match kind {
2978 MediaKind::Image => self.media_sources(wrapper),
2979 MediaKind::Video | MediaKind::Audio => self.media_sources(img),
2980 };
2981 let alt = self.image_alt(img);
2982 let sigil = kind.sigil();
2983 let label = if alt.is_empty() {
2984 // With no alt, name the file — but a `<video>` with neither `src` nor
2985 // alt has only its `<source>`s to be named by, so fall back to the
2986 // first candidate rather than labelling the row a bare sigil.
2987 let named = if destination.is_empty() {
2988 sources
2989 .first()
2990 .map(|s| s.srcset.as_str())
2991 .unwrap_or_default()
2992 } else {
2993 &destination
2994 };
2995 format!("{sigil} {}", media_label(named))
2996 } else {
2997 format!("{sigil} {alt}")
2998 };
2999 let style = Style::default().role(Role::Image);
3000 let mut glyphs = pf.to_vec();
3001 for ch in label.chars() {
3002 glyphs.push(Glyph {
3003 ch,
3004 style,
3005 src: start,
3006 stop: true,
3007 });
3008 }
3009 // How many rows the frontend wants for this picture: the label row plus
3010 // the blank fillers below it. Absent (a GUI that lays images out in
3011 // pixels, an image that didn't resolve, or a plain surface) means the
3012 // bare one-row placeholder.
3013 let rows = self
3014 .media_rows
3015 .get(&destination)
3016 .copied()
3017 .unwrap_or(1)
3018 .max(1);
3019 // End past the image so the caret has a stop after it: the last glyph's
3020 // offset is the image *start*, not its extent, so `push_row`'s
3021 // last-glyph rule would strand the end stop inside the markup.
3022 self.push_row_at(glyphs, end);
3023 if let Some(row) = self.rows.last_mut() {
3024 row.media = Some(MediaMark {
3025 kind,
3026 destination,
3027 sources,
3028 alt,
3029 poster,
3030 rows,
3031 });
3032 }
3033 // Reserve the picture's remaining height as blank `decoration` rows: drawn
3034 // (so the frontend has the vertical room to paint the raster over them),
3035 // but holding no caret and contributing no stops — vertical motion steps
3036 // over them and the caret's only homes stay the stop in front of the image
3037 // and the one just past it, both on the label row above. They anchor at the
3038 // image's end offset so a click on the picture's lower half lands after it,
3039 // the nearest caret home. Mirrors how a table's box-rule rows reserve space
3040 // without ever holding the caret.
3041 for _ in 1..rows {
3042 self.rows.push(VRow {
3043 glyphs: Vec::new(),
3044 end_src: end,
3045 decoration: true,
3046 code: false,
3047 code_lang: None,
3048 directive: false,
3049 directive_label: None,
3050 media: None,
3051 task: None,
3052 leaf_directive: None,
3053 heading: None,
3054 boundary: None,
3055 });
3056 }
3057 self.last_off = end;
3058 }
3059
3060 /// The `<picture>` alternatives inside block-image `wrapper`, in document
3061 /// order — every `<source>` element in its subtree. Empty when there's no
3062 /// `<picture>`. Each is a `<source>`'s `media` + `srcset`; core keeps them
3063 /// verbatim and picks none (see [`MediaSource`]). A `<source>` with no
3064 /// `srcset` is dropped (nothing to load); its `media` may be empty (an
3065 /// unconditional override), which a frontend treats as always-matching.
3066 ///
3067 /// It scans the wrapper's whole subtree (via the forward `first_child` /
3068 /// `next_sibling` links, the reliable ones) rather than the `<img>`'s parent,
3069 /// for two reasons. A `<picture>` reaches core in two shapes: twig promotes a
3070 /// block `<picture>` to an `element(picture)` wrapping `[source, img]`, but
3071 /// leaves an inline one's tags as raw siblings — `[raw "<picture>", source,
3072 /// img, raw "</picture>"]` — so the `<source>`s sit at different depths in
3073 /// the two. And the editor's flat arena leaves a promoted inline node's
3074 /// `parent` back-pointer dangling on a phantom root, so only the wrapper
3075 /// (known at the call site) is a trustworthy anchor. A block image is the
3076 /// sole visible content of its wrapper, so every `<source>` under it is its
3077 /// picture's.
3078 fn media_sources(&self, wrapper: usize) -> Vec<MediaSource> {
3079 let mut out = Vec::new();
3080 self.collect_sources(wrapper, &mut out);
3081 out
3082 }
3083
3084 fn collect_sources(&self, id: usize, out: &mut Vec<MediaSource>) {
3085 for c in self.children(id) {
3086 let node = &self.nodes[c];
3087 if node.name.as_deref() == Some("source") {
3088 // `<picture>` spells its candidate `srcset`, `<video>`/`<audio>`
3089 // spell it `src`. Both mean "the URL to load", so they normalise
3090 // onto one field; `srcset` wins where (illegally) both appear.
3091 let url = attr_of(node, "srcset").or_else(|| attr_of(node, "src"));
3092 if let Some(srcset) = url {
3093 out.push(MediaSource {
3094 media: attr_of(node, "media").unwrap_or_default(),
3095 srcset,
3096 mime: attr_of(node, "type").unwrap_or_default(),
3097 });
3098 }
3099 }
3100 self.collect_sources(c, out);
3101 }
3102 }
3103
3104 /// The single block-level media `id`'s subtree resolves to, or `None`.
3105 ///
3106 /// A wrapper is a block picture when the only *visible* thing under it is one
3107 /// image: whitespace-only text and structure-only elements (a `<picture>`'s
3108 /// `<source>`, which declares an alternate but paints nothing) don't count,
3109 /// and the search descends through wrapping elements (`<picture>`, a linking
3110 /// `<a>`). This is what makes `<p><img></p>`, a bare `<img>`, and
3111 /// `<h1><picture>…<img></picture></h1>` all render as one framed picture.
3112 /// Any real text, or a second image, means it isn't image-only — it falls
3113 /// back to inline rendering, where the image still shows as its alt text.
3114 ///
3115 /// [`FlatNode`]'s snapshot doesn't carry an element's tag name, so a
3116 /// `<source>` can't be skipped by name — but it needs no special case:
3117 /// contributing no image and no text, it's simply invisible to the scan.
3118 fn media_only(&self, id: usize) -> Option<(usize, MediaKind)> {
3119 let mut found = None;
3120 let mut count = 0usize;
3121 let mut has_text = false;
3122 self.scan_visual(id, &mut found, &mut count, &mut has_text);
3123 (count == 1 && !has_text).then(|| found.unwrap())
3124 }
3125
3126 /// Walk `id`'s subtree tallying visible leaves for [`media_only`]: each
3127 /// image, `<video>`, or `<audio>` (remembering the last, counting the total)
3128 /// and whether any non-whitespace text appears. Media isn't descended into —
3129 /// an image's inline children are alt text, and a `<video>`'s are its
3130 /// no-support fallback and its `<source>` declarations, none of which is
3131 /// document content.
3132 ///
3133 /// [`media_only`]: Self::media_only
3134 fn scan_visual(
3135 &self,
3136 id: usize,
3137 found: &mut Option<(usize, MediaKind)>,
3138 count: &mut usize,
3139 has_text: &mut bool,
3140 ) {
3141 for c in self.children(id) {
3142 let node = &self.nodes[c];
3143 match node.kind.as_str() {
3144 "image" => {
3145 *found = Some((c, MediaKind::Image));
3146 *count += 1;
3147 }
3148 // A `<video>`/`<audio>` reaches core as a generic `container`
3149 // (twig gives neither a semantic node, so `html_elements`
3150 // promotion leaves the tag name on `name`). Counted as media and
3151 // *not* descended into, so its `<source>` children and its
3152 // "your browser does not support…" fallback text neither add a
3153 // second count nor make the block look like text.
3154 "container" if matches!(element_tag(node), Some("video") | Some("audio")) => {
3155 let kind = match element_tag(node) {
3156 Some("audio") => MediaKind::Audio,
3157 _ => MediaKind::Video,
3158 };
3159 *found = Some((c, kind));
3160 *count += 1;
3161 }
3162 // Text leaves: only non-whitespace counts as visible content.
3163 // (Twig keeps the whitespace `str`s between HTML tags — the
3164 // newlines and indentation inside a `<picture>` — as real nodes.)
3165 "str" | "smart_punctuation" | "verbatim" | "inline_math" => {
3166 if node.text.as_deref().is_some_and(|t| !t.trim().is_empty()) {
3167 *has_text = true;
3168 }
3169 }
3170 // Structural breaks carry no visible glyph of their own.
3171 "soft_break" | "hard_break" | "non_breaking_space" => {}
3172 // Any other wrapper (emphasis, a link, a `<picture>`) is
3173 // transparent to the scan — descend into it.
3174 _ => self.scan_visual(c, found, count, has_text),
3175 }
3176 }
3177 }
3178
3179 /// A leaf directive (`::name{…}`) as one placeholder row — the
3180 /// [`block_media`](Self::block_media) recipe, for the same reason: it is a
3181 /// block that renders as *a thing*, not as text, and the frontend paints
3182 /// whatever the host app's vocabulary makes of it.
3183 ///
3184 /// The row's glyphs are a `⧉ label` (or `⧉ name`) stand-in a plain surface
3185 /// paints as-is, every glyph anchored at the directive's start with a caret
3186 /// stop there, and the row ending past it so the caret can also rest after
3187 /// it. It carries a [`DirectiveMark`] for [`directive_spans`], and is marked
3188 /// [`directive`](VRow::directive) so a frontend already drawing the
3189 /// container form's panel frames this one identically for free.
3190 ///
3191 /// Before this, a leaf directive emitted no rows at all: it was invisible,
3192 /// held no caret, and vertical motion crossed a void where it stood.
3193 fn block_directive(&mut self, id: usize, pf: &[Glyph]) {
3194 let node = &self.nodes[id];
3195 let (start, end) = (node.span.start, node.span.end);
3196 let name = node.name.clone().unwrap_or_default();
3197 let attrs = node.attrs.clone();
3198 let label = self.image_alt(id); // its `[label]` children, flattened
3199 let shown = if label.is_empty() { &name } else { &label };
3200 let style = Style::default().role(Role::Image);
3201 let mut glyphs = pf.to_vec();
3202 for ch in format!("⧉ {shown}").chars() {
3203 glyphs.push(Glyph {
3204 ch,
3205 style,
3206 src: start,
3207 stop: true,
3208 });
3209 }
3210 // End past the directive so the caret has a stop after it — the same
3211 // reason `block_media` anchors its row at the image's end.
3212 self.push_row_at(glyphs, end);
3213 if let Some(row) = self.rows.last_mut() {
3214 row.directive = true;
3215 row.leaf_directive = Some(DirectiveMark {
3216 name,
3217 attrs,
3218 label,
3219 rows: 1,
3220 });
3221 }
3222 self.last_off = end;
3223 }
3224
3225 /// An image's alt text: the flattened text of its inline descendants (an
3226 /// image's children *are* its alt content), empty when it has none. Also a
3227 /// leaf directive's `[label]`, which is the same shape — inline children
3228 /// standing for the block.
3229 fn image_alt(&self, id: usize) -> String {
3230 let mut out = String::new();
3231 self.collect_text(id, &mut out);
3232 out
3233 }
3234
3235 /// Append every descendant's `text` to `out`, in document order. Inline text
3236 /// (`str`) nodes are leaves, so a node never contributes both its own text and
3237 /// a child's — no double counting.
3238 fn collect_text(&self, id: usize, out: &mut String) {
3239 for c in self.children(id) {
3240 if let Some(t) = &self.nodes[c].text {
3241 out.push_str(t);
3242 }
3243 self.collect_text(c, out);
3244 }
3245 }
3246
3247 fn inline_children(&self, id: usize, base: Style) -> Vec<Glyph> {
3248 let mut out = Vec::new();
3249 for c in self.children(id) {
3250 self.inline(c, base, &mut out);
3251 }
3252 out
3253 }
3254
3255 /// [`inline_children`](Self::inline_children) plus any trailing whitespace the
3256 /// block carries past its inline content (see [`trailing_ws_glyphs`]). Used
3257 /// for the leaf inline blocks — paragraphs and headings — whose own `span`
3258 /// bounds exactly one line of text, so the trailing gap is theirs. *Not* for
3259 /// a table cell, whose `span` is the whole row and would swallow the
3260 /// delimiters and neighbours between it and the row's end.
3261 ///
3262 /// [`trailing_ws_glyphs`]: Self::trailing_ws_glyphs
3263 fn inline_children_with_trailing(&self, id: usize, base: Style) -> Vec<Glyph> {
3264 let mut out = self.inline_children(id, base);
3265 out.extend(self.trailing_ws_glyphs(id, base));
3266 out
3267 }
3268
3269 /// Glyphs for whatever trailing whitespace a block's source carries past its
3270 /// last inline node — the space(s) at the end of `hello ` that Markdown and
3271 /// Djot drop from the `str` node as insignificant. twig still records them:
3272 /// a block's `content_span` ends at its last meaningful character while its
3273 /// `span` runs to the end of the line's text (before the terminating
3274 /// newline), so the gap between the two *is* that trailing whitespace.
3275 ///
3276 /// Emitting it as real caret-stop glyphs is what lets the caret be drawn
3277 /// past the last visible character. Without it, typing a space at the end of
3278 /// a paragraph moved the caret in the source but not on screen — the caret
3279 /// stuck on the last glyph until the next visible character reparsed the
3280 /// space into an interior `str` node that finally carried it.
3281 ///
3282 /// Restricted to spaces: only they are safe to synthesize one-cell-per-byte,
3283 /// and only they are what the parser silently strips. Anything else in the
3284 /// gap means the span accounting isn't what this assumes, so it's left alone.
3285 fn trailing_ws_glyphs(&self, id: usize, style: Style) -> Vec<Glyph> {
3286 let node = &self.nodes[id];
3287 let Some(content) = &node.content_span else {
3288 return Vec::new();
3289 };
3290 let (from, to) = (content.end, node.span.end);
3291 let Some(slice) = (from < to).then(|| self.source.get(from..to)).flatten() else {
3292 return Vec::new();
3293 };
3294 if slice.is_empty() || slice.bytes().any(|b| b != b' ') {
3295 return Vec::new();
3296 }
3297 slice
3298 .bytes()
3299 .enumerate()
3300 .map(|(i, _)| Glyph {
3301 ch: ' ',
3302 style,
3303 src: from + i,
3304 stop: true,
3305 })
3306 .collect()
3307 }
3308
3309 fn inline(&self, id: usize, base: Style, out: &mut Vec<Glyph>) {
3310 let node = &self.nodes[id];
3311 match node.kind.as_str() {
3312 "str" | "smart_punctuation" => push_escaped_text(
3313 out,
3314 node.text.as_deref().unwrap_or(""),
3315 node.span.clone(),
3316 self.source,
3317 base,
3318 ),
3319 "soft_break" | "hard_break" | "non_breaking_space" => {
3320 // A break renders as a real, caret-navigable glyph — but twig
3321 // gives it no span of its own (`0..0`), so the offset comes from
3322 // the text in front of it: one *past* the last glyph, which is
3323 // the newline the break stands for. Past, not on: sharing the
3324 // previous glyph's offset would put two stops on one byte, and a
3325 // caret that can't change offset can't move.
3326 let src = if node.span.start != 0 {
3327 node.span.start
3328 } else {
3329 out.last().map(|g| g.src + g.ch.len_utf8()).unwrap_or(0)
3330 };
3331 // A *hard* break renders as this run's break glyph — a newline
3332 // inside a table cell (its own line), the same space in prose the
3333 // frontend re-wraps. A soft break normally folds into a space;
3334 // under `LineFlow::Preserve` it renders as a `'\n'` too, so the
3335 // author's line break shows where it was written. Never inside a
3336 // cell (`break_glyph` is `'\n'` there): a cell is one line and
3337 // folds its own soft breaks regardless.
3338 let ch = if node.kind == Kind::HardBreak {
3339 self.break_glyph.get()
3340 } else if node.kind == Kind::SoftBreak
3341 && self.preserve_soft
3342 && self.break_glyph.get() == ' '
3343 {
3344 '\n'
3345 } else {
3346 ' '
3347 };
3348 out.push(Glyph {
3349 ch,
3350 style: base,
3351 src,
3352 stop: true,
3353 });
3354 }
3355 // A cell's only spelling for an in-line break is a raw `<br>`; read it
3356 // back as one (outside a cell it stays the literal text it falls to
3357 // below). The tag's bytes carry no stop of their own — the line it
3358 // ends stops just before it, the next just after.
3359 "raw_inline" if self.break_glyph.get() == '\n' && is_br(node.text.as_deref()) => {
3360 out.push(Glyph {
3361 ch: '\n',
3362 style: base,
3363 src: node.span.start,
3364 stop: true,
3365 });
3366 }
3367 "emph" => self.inline_delimited(id, base.italic(), out),
3368 "strong" => self.inline_delimited(id, base.bold(), out),
3369 "mark" => self.inline_delimited(id, base.role(Role::Mark), out),
3370 "insert" => self.inline_delimited(id, base.underline(), out),
3371 "delete" => self.inline_delimited(id, base.strikethrough(), out),
3372 // The one pair whose whole meaning is *where the glyphs sit*. Drawn
3373 // in the surrounding style otherwise, so `^**2**^` stays bold and a
3374 // superscript inside a heading keeps the heading's role — which is
3375 // exactly why this is a `Baseline` and not a `Role`.
3376 "superscript" => self.inline_delimited(id, base.baseline(Baseline::Super), out),
3377 "subscript" => self.inline_delimited(id, base.baseline(Baseline::Sub), out),
3378 "verbatim" | "inline_math" => {
3379 // The interior begins at `content_span.start` — past however many
3380 // backticks the fence used, which `span.start + 1` only guessed
3381 // right for a single one. Fall back to that guess if it's absent.
3382 let at = node
3383 .content_span
3384 .as_ref()
3385 .map_or(node.span.start + 1, |c| c.start);
3386 let style = base.role(Role::Code);
3387 // Not `inline_delimited`: verbatim has no child nodes to recurse
3388 // into — its content is its own `text` — so the fences bracket a
3389 // `push_text` instead. The fences themselves keep `Role::Code`'s
3390 // sibling treatment via `push_delim`'s role override.
3391 let show = self.revealed(&node.span).then(|| self.delims(id)).flatten();
3392 if let Some((open, _)) = &show {
3393 self.push_delim(out, open, style);
3394 }
3395 push_text(out, node.text.as_deref().unwrap_or(""), at, style);
3396 if let Some((_, close)) = &show {
3397 self.push_delim(out, close, style);
3398 }
3399 }
3400 // A text directive (`:name[label]{…}`) — the inline form of a generic
3401 // directive. Its `[label]` children are the visible text; the name and
3402 // the `{…}` attributes are the host app's vocabulary (diaryx's
3403 // `:vis[…]`) and stay hidden markup, exactly as a link's `](dest)` is.
3404 // Drawn in the surrounding style: a role of its own would need one
3405 // every frontend maps, and the bug this fixes is that the text was
3406 // invisible, not that it was unstyled.
3407 "container" if container_is_directive(node) && !self.children(id).is_empty() => {
3408 self.recurse(id, base, out)
3409 }
3410 // No `[label]`, so there are no children to render and recursing
3411 // emitted *nothing*: the directive's bytes vanished from the document
3412 // and left no caret stop behind. What to draw instead turns on
3413 // whether the syntax looks deliberate.
3414 //
3415 // Bare `:word` almost never is. twig matches a colon followed by any
3416 // letter-led word (`scanTextDirective`, deliberately matching remark),
3417 // so ordinary prose is full of them — `:see below`, a `:smile:`
3418 // shortcode, a stray colon before a word. Those are prose, and prose
3419 // renders as itself: every byte visible, every byte a caret stop, so a
3420 // colon typed by accident can be seen and deleted. Hiding them behind
3421 // a placeholder would be the invisible-and-unreachable failure this
3422 // arm exists to fix, just wearing a nicer glyph.
3423 "container" if container_is_directive(node) && node.attrs.is_empty() => {
3424 let span = node.span.clone();
3425 push_text(
3426 out,
3427 self.source.get(span.clone()).unwrap_or(""),
3428 span.start,
3429 base,
3430 );
3431 }
3432 // `{…}` attributes, though, are unmistakably deliberate — nobody
3433 // types `:vis{.family}` by accident, and diaryx writes exactly that
3434 // inline. So an attribute-bearing directive with no label draws as a
3435 // chip on `block_directive`'s recipe (`⧉ name attrs`, `Role::Image`),
3436 // the inline peer of the leaf form's placeholder row.
3437 //
3438 // Only the first glyph is a caret stop, and the whole chip shares the
3439 // directive's start offset: the caret treats it as one atomic thing
3440 // rather than walking hidden markup a byte at a time, and a paragraph
3441 // holding nothing but a chip still has a stop to be navigated to.
3442 "container" if container_is_directive(node) => {
3443 let start = node.span.start;
3444 let name = node.name.clone().unwrap_or_default();
3445 let shown = match directive_attr_label(&node.attrs) {
3446 Some(attrs) if !name.is_empty() => format!("⧉ {name} {attrs}"),
3447 Some(attrs) => format!("⧉ {attrs}"),
3448 None => format!("⧉ {name}"),
3449 };
3450 let style = base.role(Role::Image);
3451 for (i, ch) in shown.chars().enumerate() {
3452 out.push(Glyph {
3453 ch,
3454 style,
3455 src: start,
3456 stop: i == 0,
3457 });
3458 }
3459 }
3460 // A footnote reference (`[^1]`). The label bracketed is what a reader
3461 // needs — bare, `note1` reads as a typo rather than a reference — so
3462 // the `^` is hidden as the spelling artefact it is (a link's
3463 // `](dest)` goes the same way) and the brackets are kept as
3464 // decoration: one shared offset, never a caret stop, like a table's
3465 // borders, so the caret walks the label alone.
3466 //
3467 // Styled `Role::Link`: a reference *is* a link to its definition, and
3468 // every frontend already paints that role. A role of its own would
3469 // need one in each of them, and what a frontend needs to tell the two
3470 // apart is not a paint colour but an answer to "what does clicking
3471 // here do" — which is [`Doc::footnote_at_caret`]'s job, not a glyph's.
3472 //
3473 // Raised, though, because that a reference is *set* differently from
3474 // the prose it interrupts is exactly what makes it read as a
3475 // reference. `[1]` at body size reads as bracketed text.
3476 "footnote_reference" => {
3477 let style = base.role(Role::Link);
3478 // Revealed, the reference is just its source bytes: the `^` that
3479 // is normally elided comes back and every byte becomes a real
3480 // stop, so the brackets stop being decoration and start being
3481 // text. That's the whole point of the mode, and it replaces the
3482 // hand-built chip below rather than decorating it — including the
3483 // raised baseline, since what's on screen there is source, and
3484 // source is set as prose.
3485 if self.revealed(&node.span) {
3486 self.push_delim(out, &node.span, style);
3487 return;
3488 }
3489 let style = style.baseline(Baseline::Super);
3490 // The label's own span, so its glyphs map to their true bytes.
3491 // Absent one, it starts past the `[^` that opens the reference.
3492 let (label, at) = match &node.content_span {
3493 Some(c) => (self.source.get(c.clone()).unwrap_or(""), c.start),
3494 None => (node.text.as_deref().unwrap_or(""), node.span.start + 2),
3495 };
3496 out.push(Glyph {
3497 ch: '[',
3498 style,
3499 src: node.span.start,
3500 stop: false,
3501 });
3502 push_text(out, label, at, style);
3503 out.push(Glyph {
3504 ch: ']',
3505 style,
3506 src: node.span.end.saturating_sub(1),
3507 stop: false,
3508 });
3509 }
3510 "link" | "url" | "email" => {
3511 let style = base.role(Role::Link);
3512 if self.children(id).is_empty() {
3513 // A bare autolink (`<a@b.c>`, a naked URL): the destination
3514 // *is* the visible text, so there is nothing elided to
3515 // reveal and both modes draw the same thing.
3516 push_text(
3517 out,
3518 node.destination
3519 .as_deref()
3520 .or(node.text.as_deref())
3521 .unwrap_or("link"),
3522 node.span.start,
3523 style,
3524 );
3525 } else {
3526 // An inline link reveals asymmetrically — `[` before the
3527 // label, `](dest)` after it — which the generic
3528 // span-minus-content derivation already produces.
3529 self.inline_delimited(id, style, out);
3530 }
3531 }
3532 _ => {
3533 if self.children(id).is_empty() {
3534 if let Some(t) = &node.text {
3535 push_text(out, t, node.span.start, base);
3536 }
3537 } else {
3538 self.recurse(id, base, out);
3539 }
3540 }
3541 }
3542 }
3543
3544 fn recurse(&self, id: usize, style: Style, out: &mut Vec<Glyph>) {
3545 for c in self.children(id) {
3546 self.inline(c, style, out);
3547 }
3548 }
3549
3550 /// Lay a block's inline `glyphs` into visual rows, prefixing the first with
3551 /// `pf` and the rest with `pc`. A preserved soft break arrives as a `'\n'`
3552 /// glyph (see the `soft_break` arm): a hard row boundary that splits the
3553 /// glyphs so each run lays out on its own and the author's line structure
3554 /// shows on screen. The `'\n'` is dropped from the row it closes and its
3555 /// source offset becomes that row's end stop — exactly how a table cell's
3556 /// in-line `<br>` is handled — so the caret can rest at the line's end
3557 /// without a zero-width control char leaking into what the frontends render.
3558 /// With no `'\n'` present (the folding default, and every build that isn't
3559 /// `LineFlow::Preserve`) there is one run and this is byte-identical to
3560 /// laying the glyphs out directly.
3561 fn emit_wrapped(&mut self, glyphs: Vec<Glyph>, block_start: usize, pf: &[Glyph], pc: &[Glyph]) {
3562 if !glyphs.iter().any(|g| g.ch == '\n') {
3563 self.emit_line(glyphs, block_start, pf, pc, None);
3564 return;
3565 }
3566 // Each run up to a '\n' is a line of its own: the first wears the block's
3567 // opening prefix, every later one the continuation prefix, and the break's
3568 // own offset ends the run's last row. The break glyph is dropped. A
3569 // trailing '\n' flushes its run and leaves nothing behind, so no spurious
3570 // blank row follows it.
3571 let mut run: Vec<Glyph> = Vec::new();
3572 let mut first = true;
3573 for g in glyphs {
3574 if g.ch == '\n' {
3575 let lead = if first { pf } else { pc };
3576 self.emit_line(std::mem::take(&mut run), block_start, lead, pc, Some(g.src));
3577 first = false;
3578 } else {
3579 run.push(g);
3580 }
3581 }
3582 if !run.is_empty() {
3583 let lead = if first { pf } else { pc };
3584 self.emit_line(run, block_start, lead, pc, None);
3585 }
3586 }
3587
3588 /// Word-wrap a single line of `glyphs` (no interior line breaks) to the
3589 /// available width and push the visual rows, prefixing the first with `pf`
3590 /// and the rest with `pc`. `end`, when set, is the source offset that ends
3591 /// the line's final row — the offset of the break that terminated it, which
3592 /// the caller has already stripped from `glyphs`; when `None` the row ends
3593 /// just past its last glyph, as an unbroken block's does.
3594 fn emit_line(
3595 &mut self,
3596 glyphs: Vec<Glyph>,
3597 block_start: usize,
3598 pf: &[Glyph],
3599 pc: &[Glyph],
3600 end: Option<usize>,
3601 ) {
3602 // The line's final row ends at `end` when a break gave one, else just
3603 // past its last glyph (`push_row`'s default).
3604 let push_last = |b: &mut Self, row: Vec<Glyph>| match end {
3605 Some(e) => b.push_row_at(row, e),
3606 None => b.push_row(row, block_start),
3607 };
3608
3609 // No column budget: emit the whole line as one row and let the frontend
3610 // wrap it at its own (pixel) width.
3611 let Some(width) = self.wrap else {
3612 let row = if glyphs.is_empty() {
3613 pf.to_vec()
3614 } else {
3615 concat(pf, &glyphs)
3616 };
3617 push_last(self, row);
3618 return;
3619 };
3620
3621 // Split into words (maximal non-space runs), each carrying the space
3622 // glyph that followed it (so its source offset is preserved).
3623 let mut words: Vec<(Vec<Glyph>, Option<Glyph>)> = Vec::new();
3624 let mut word: Vec<Glyph> = Vec::new();
3625 for g in glyphs {
3626 if g.ch == ' ' {
3627 words.push((std::mem::take(&mut word), Some(g)));
3628 } else {
3629 word.push(g);
3630 }
3631 }
3632 if !word.is_empty() {
3633 words.push((word, None));
3634 }
3635 if words.is_empty() {
3636 // An empty block (or an empty preserved line) still occupies one
3637 // (prefixed) row.
3638 push_last(self, pf.to_vec());
3639 return;
3640 }
3641
3642 let mut line: Vec<Glyph> = Vec::new();
3643 let mut used = 0usize;
3644 let mut first = true;
3645 for (w, space) in words {
3646 let avail = width
3647 .saturating_sub(prefix_width(if first { pf } else { pc }))
3648 .max(1);
3649 let cells = glyphs_width(&w);
3650 if used > 0 && used + cells > avail {
3651 let row = concat(if first { pf } else { pc }, &line);
3652 self.push_row(row, block_start);
3653 line = Vec::new();
3654 used = 0;
3655 first = false;
3656 }
3657 used += cells;
3658 line.extend(w);
3659 if let Some(sp) = space {
3660 used += 1;
3661 line.push(sp);
3662 }
3663 }
3664 let row = concat(if first { pf } else { pc }, &line);
3665 push_last(self, row);
3666 }
3667
3668 /// The source offset of each line of a code block's `text`.
3669 ///
3670 /// `content` is the block's `content_span` — where twig says the body lives
3671 /// in the source, fences already excluded. Its lines run 1:1 with the
3672 /// rendered `text` lines, so no search is needed; each is anchored at the
3673 /// *end* of its source line, which places it past whatever indent `text` had
3674 /// stripped (a fenced block's fences, an indented one's leading spaces)
3675 /// without having to know how much there was.
3676 ///
3677 /// `None` when the body and the rendered lines don't line up — a coarse
3678 /// fallback the caller turns into the block's start offset.
3679 fn code_line_offsets(&self, content: &Range<usize>, lines: &[&str]) -> Option<Vec<usize>> {
3680 let mut src_lines: Vec<(usize, &str)> = Vec::new();
3681 let mut at = content.start;
3682 for l in self.source.get(content.start..content.end)?.split('\n') {
3683 src_lines.push((at, l));
3684 at += l.len() + 1;
3685 }
3686 if src_lines.len() != lines.len() {
3687 return None;
3688 }
3689 Some(
3690 lines
3691 .iter()
3692 .zip(&src_lines)
3693 .map(|(l, (start, sl))| start + sl.len().saturating_sub(l.len()))
3694 .collect(),
3695 )
3696 }
3697
3698 fn push_row(&mut self, glyphs: Vec<Glyph>, fallback: usize) {
3699 // Step past the character the *source* holds at the last glyph's offset,
3700 // not past the glyph's own `ch`. The two agree for ordinary text, but a
3701 // glyph is not always the character it stands on: `synth` decoration and
3702 // a substituted run (an image's `⧉ label`) share one offset by design.
3703 // Trusting `ch` there yields an offset inside a multi-byte character,
3704 // which every later slice of `source` panics on.
3705 let end_src = glyphs
3706 .last()
3707 .map(|g| {
3708 let at = g.src.min(self.source.len());
3709 at + self.source[at..].chars().next().map_or(0, char::len_utf8)
3710 })
3711 .unwrap_or(fallback);
3712 self.push_row_at(glyphs, end_src);
3713 }
3714
3715 /// Push a row with an explicit end stop, for content that knows its own
3716 /// extent better than its last glyph does.
3717 fn push_row_at(&mut self, glyphs: Vec<Glyph>, end_src: usize) {
3718 self.last_off = end_src;
3719 self.rows.push(VRow {
3720 glyphs,
3721 end_src,
3722 decoration: false,
3723 code: false,
3724 code_lang: None,
3725 directive: false,
3726 directive_label: None,
3727 media: None,
3728 task: None,
3729 leaf_directive: None,
3730 heading: None,
3731 boundary: None,
3732 });
3733 }
3734
3735 /// The quote's own trailing marker lines: the `>` / `> ` lines that lie past
3736 /// its last child but inside its span, one gutter row each.
3737 ///
3738 /// Pressing Enter at the end of `> a` writes `> a\n>\n> \n` — twig's
3739 /// spelling, and the right one. Those last two lines hold no block (a
3740 /// `block_quote`'s `content_span` still stops at its last child) so the
3741 /// children walk never reaches them, and they used to fall all the way to
3742 /// the document-level [`Builder::emit_trailing_blank_lines`], which knows no
3743 /// prefix: the gutter simply stopped, and a writer adding a line to a quote
3744 /// watched it draw as plain prose.
3745 ///
3746 /// This is only answerable since twig 3.2.0, where a Markdown `block_quote`'s
3747 /// span covers its own trailing marker lines (it reported `0..3` for that
3748 /// source and now reports `0..8`). Before that the lines belonged to no node
3749 /// at any level, and the only way to draw them was to sniff `>` off the raw
3750 /// source and re-derive the nesting depth by counting markers — format
3751 /// inference this crate exists to keep out of the render path.
3752 ///
3753 /// Each row is a real caret home rather than a decoration gap: the writer
3754 /// spelled every one of these lines with a marker of its own, so each is a
3755 /// line of the quote to stand on, not the spacing between two blocks (which
3756 /// is [`Builder::emit_separators_before`]'s, and falls *between* children
3757 /// where this never looks).
3758 fn emit_quote_trailing_lines(&mut self, pc: &[Glyph], end: usize) {
3759 let end = end.min(self.source.len());
3760 let mut at = self.rows.last().map_or(0, |r| r.end_src);
3761 // Walk line by line from the last child's end to the quote's, taking each
3762 // line's *end* as the row's offset — the caret home at the end of a line
3763 // is where one on an empty quoted line belongs, and it keeps every row's
3764 // offset distinct from its neighbours'.
3765 while at < end {
3766 let Some(k) = self.source[at..end].find('\n') else {
3767 break;
3768 };
3769 let line_start = at + k + 1;
3770 let line_end = self.source[line_start..end]
3771 .find('\n')
3772 .map_or(end, |i| line_start + i);
3773 self.push_row_at(pc.to_vec(), line_end);
3774 at = line_end;
3775 }
3776 }
3777
3778 /// The source offset the caret rests at on the blank line separating a block
3779 /// that ends at `prev_end` from the next block starting at `next_start`:
3780 /// just past the newline that terminates the previous block, but kept
3781 /// strictly before the next block so the offset is unique to this row.
3782 fn blank_line_offset(&self, prev_end: usize, next_start: usize) -> usize {
3783 let after_nl = self.source[prev_end..]
3784 .find('\n')
3785 .map_or(prev_end, |p| prev_end + p + 1);
3786 after_nl.min(next_start.saturating_sub(1)).max(prev_end)
3787 }
3788
3789 /// The source offset of each blank row between a block ending at `prev_end`
3790 /// and content starting at `next_start` — one per blank source line. The
3791 /// first newline terminates the previous block's line; every line it opens up
3792 /// to (but not including) the line that holds `next_start` is a blank row the
3793 /// caret can occupy. Offsets are unique and ascending so `pos_of_offset`
3794 /// resolves each to its own row. Empty when the two blocks are tight (no
3795 /// blank line between them).
3796 fn blank_rows_between(&self, prev_end: usize, next_start: usize) -> Vec<usize> {
3797 // Spans aren't always in tidy source order (e.g. a block after
3798 // frontmatter can start *before* the previous block's rendered content
3799 // ends). There's no blank line to place then — fall back to the clamped
3800 // single separator (an empty return) rather than slicing an inverted
3801 // range.
3802 if next_start <= prev_end {
3803 return Vec::new();
3804 }
3805 let gap = &self.source[prev_end..next_start];
3806 let Some(nl) = gap.find('\n') else {
3807 return Vec::new();
3808 };
3809 // The line holding `next_start` belongs to the next block; blank rows
3810 // stop before it.
3811 let next_line_start = self.source[..next_start].rfind('\n').map_or(0, |p| p + 1);
3812 let mut offs = Vec::new();
3813 let mut start = prev_end + nl + 1;
3814 while start < next_line_start {
3815 offs.push(start);
3816 match self.source[start..next_start].find('\n') {
3817 Some(k) => start += k + 1,
3818 None => break,
3819 }
3820 }
3821 offs
3822 }
3823
3824 /// Blank lines the user typed past the end of the last block (e.g. two
3825 /// `Enter`s to open a fresh paragraph) leave no AST node, so nothing renders
3826 /// and the caret appears stuck on the old line. Reconstruct one empty row
3827 /// per extra trailing newline from the source, each at its own offset, so
3828 /// the caret rides down onto the new line the moment it's created.
3829 ///
3830 /// `above` is the class of the last block in the document — the one this gap
3831 /// closes. A document with no blocks at all has nothing above these rows, and
3832 /// [`BlockClass::Paragraph`] is the honest answer there too: what they are is
3833 /// empty paragraphs, on both sides of the gap.
3834 fn emit_trailing_blank_lines(&mut self, above: BlockClass, hidden_end: usize) {
3835 // With no rows at all the count starts past any hidden frontmatter, not
3836 // at 0: its newlines are not trailing blank lines, and counting them
3837 // opened phantom rows *inside* the metadata for a frontmatter-only file.
3838 let last_end = self.rows.last().map_or(hidden_end, |r| r.end_src);
3839 if last_end >= self.source.len() {
3840 return;
3841 }
3842 // The first newline after the last content just terminates that line, so
3843 // a lone trailing `\n` (an ordinary file ending) opens no blank row. A
3844 // *second* newline opens an empty paragraph: render it the way a block
3845 // boundary is rendered — a blank spacer row, then the empty paragraph row
3846 // the caret rests on — so the just-pressed-Enter view already shows the
3847 // gap it will keep once text is typed, and typing doesn't shift the line
3848 // down. One row per trailing newline (each its own caret offset), the
3849 // last landing at the document end where the caret sits.
3850 let extra = self.source[last_end..].matches('\n').count();
3851 if extra < 2 {
3852 return;
3853 }
3854 for k in 1..=extra {
3855 self.rows.push(VRow {
3856 glyphs: Vec::new(),
3857 end_src: last_end + k,
3858 // As between two blocks: the first blank row is the gap that
3859 // closes the block above, not somewhere to type. Nothing follows
3860 // to need a gap of its own, though, so every row after it is a
3861 // real empty paragraph — the end of the document bounds the last
3862 // one the way a following block would. Preserve flow makes even
3863 // that first row navigable, as it does every blank line.
3864 decoration: !self.preserve_soft && k == 1,
3865 code: false,
3866 code_lang: None,
3867 directive: false,
3868 directive_label: None,
3869 media: None,
3870 task: None,
3871 leaf_directive: None,
3872 heading: None,
3873 // The one drawn row here is a block boundary like any other —
3874 // "rendered the way a block boundary is rendered" is the whole
3875 // point of it — so it says so, and a frontend spacing boundaries
3876 // spaces this one the same. The rows below it are navigable empty
3877 // paragraphs, not gaps.
3878 boundary: (!self.preserve_soft && k == 1).then_some(Boundary {
3879 above,
3880 below: BlockClass::Paragraph,
3881 }),
3882 });
3883 }
3884 }
3885}
3886
3887// ── display width ────────────────────────────────────────────────────────────
3888//
3889// Two things a row can be counted in, and they are not the same number:
3890//
3891// *glyphs*, one per codepoint — how the text is stored here, and what an
3892// index into `VRow::glyphs` means; and
3893// *columns*, one per terminal cell — where the text is drawn, and what every
3894// `col` in this crate means.
3895//
3896// `你` is one glyph in two columns. Counting columns with `glyphs.len()` (or,
3897// in the source view, `chars().count()`) is the same number only for the ASCII
3898// that most fixtures are written in, and drifts one cell per wide character
3899// everywhere else — the caret drawn a column short of the text it types into.
3900// Everything below converts between the two; nothing else should have to.
3901
3902/// The display width of `s` in terminal cells.
3903///
3904/// Measured per grapheme cluster, because that is the unit a surface advances
3905/// by: `👨👩👧` is five codepoints measuring 2 + 0 + 2 + 0 + 2 cells one at a
3906/// time, but the character they spell is drawn in 2. Both frontends already
3907/// measure it that way — ratatui asks `unicode-width` per cluster, and the GUI
3908/// asks its own text system — so the caret only lands where the text is if this
3909/// agrees with them.
3910pub fn text_width(s: &str) -> usize {
3911 UnicodeWidthStr::width(s)
3912}
3913
3914/// One grapheme cluster of a laid-out row: the glyphs that spell it, and the
3915/// cells it is drawn in.
3916///
3917/// The cluster, not the glyph, is what has a width. A row's glyphs are one per
3918/// codepoint, so an accented letter or an emoji is several of them drawn in one
3919/// character's worth of cells — the glyph that opens the cluster claims those
3920/// cells, and the ones continuing it are drawn *inside* them rather than beside
3921/// them. It's the same cluster the stop table is built on: the opening glyph is
3922/// the one a caret can rest on, and so the only one whose column it can be
3923/// drawn at.
3924struct Cluster {
3925 /// Index of the glyph that opens it.
3926 glyph: usize,
3927 /// The display column it starts at.
3928 col: usize,
3929 /// How many cells it is drawn in. Zero for a cluster with no width of its
3930 /// own (a lone joiner), which therefore sits at no column at all.
3931 cells: usize,
3932}
3933
3934/// Walk a row's glyphs as the clusters they spell, in column order.
3935fn clusters(glyphs: &[Glyph]) -> Vec<Cluster> {
3936 let text: String = glyphs.iter().map(|g| g.ch).collect();
3937 let mut out = Vec::new();
3938 let (mut glyph, mut col) = (0, 0);
3939 for cluster in text.graphemes(true) {
3940 let cells = text_width(cluster);
3941 out.push(Cluster { glyph, col, cells });
3942 // One glyph per codepoint, so a cluster spans exactly its own.
3943 glyph += cluster.chars().count();
3944 col += cells;
3945 }
3946 out
3947}
3948
3949/// The display width of a run of glyphs.
3950fn glyphs_width(glyphs: &[Glyph]) -> usize {
3951 clusters(glyphs).last().map_or(0, |c| c.col + c.cells)
3952}
3953
3954/// A cell's display width — the widest of its lines, since an in-cell `\n` break
3955/// splits it into several. Sizes the column that must hold every line.
3956fn cell_width(glyphs: &[Glyph]) -> usize {
3957 glyphs
3958 .split(|g| g.ch == '\n')
3959 .map(glyphs_width)
3960 .max()
3961 .unwrap_or(0)
3962}
3963
3964/// Whether a raw inline HTML tag is a line break (`<br>`, `<br/>`, `<br />`,
3965/// case-insensitively) — the one tag a table cell reads as an in-cell break.
3966fn is_br(text: Option<&str>) -> bool {
3967 let Some(t) = text else { return false };
3968 matches!(
3969 t.trim().to_ascii_lowercase().replace(' ', "").as_str(),
3970 "<br>" | "<br/>"
3971 )
3972}
3973
3974impl VRow {
3975 /// The row's width in display columns — and so the column of the caret
3976 /// placed past its last glyph, which is the rightmost column it can occupy.
3977 fn width(&self) -> usize {
3978 glyphs_width(&self.glyphs)
3979 }
3980
3981 /// The display column glyph `i` is drawn at. Glyphs continuing a cluster
3982 /// report the column of the glyph that opened it, since that is where they
3983 /// are drawn; none of them is ever a stop, so no caret is placed by it.
3984 fn col_of_glyph(&self, i: usize) -> usize {
3985 clusters(&self.glyphs)
3986 .iter()
3987 .rev()
3988 .find(|c| c.glyph <= i)
3989 .map_or(0, |c| c.col)
3990 }
3991
3992 /// The glyph drawn at display column `col`, or `None` past the row's last
3993 /// cell.
3994 ///
3995 /// A column landing on the *second* cell of a wide glyph resolves to that
3996 /// glyph: half a character is not a place to be, so clicking either cell of
3997 /// `你` means `你`, and the caret comes to rest at its start — the column it
3998 /// would be drawn at anyway. That rule is what makes the mapping invertible:
3999 /// every offset has one column, and every column has one offset.
4000 fn glyph_at_col(&self, col: usize) -> Option<usize> {
4001 clusters(&self.glyphs)
4002 .into_iter()
4003 .find(|c| col < c.col + c.cells)
4004 .map(|c| c.glyph)
4005 }
4006}
4007
4008// ── helpers ──────────────────────────────────────────────────────────────────
4009
4010/// The caret home inside an *empty* table cell (`col`, 0-based) of a row whose
4011/// source is `row_src` starting at byte `row_start`. twig gives an empty cell no
4012/// `content_span`, so its interior is read from the pipes: cell `col` lies
4013/// between the `col`-th and `col+1`-th unescaped `│`/`|`, and the home is one
4014/// space past the opening one — mimicking the `| ` padding a filled cell has,
4015/// and never at or past the closing pipe. So `| | |` gives the two cells
4016/// distinct, editable homes instead of both collapsing onto the row's start.
4017fn empty_cell_offset(row_src: &str, row_start: usize, col: usize) -> usize {
4018 let bytes = row_src.as_bytes();
4019 let mut pipes = Vec::new();
4020 for (i, &b) in bytes.iter().enumerate() {
4021 if b == b'|' && (i == 0 || bytes[i - 1] != b'\\') {
4022 pipes.push(i);
4023 }
4024 }
4025 match (pipes.get(col).copied(), pipes.get(col + 1).copied()) {
4026 (Some(open), Some(close)) => {
4027 let lo = open + 1; // just inside the opening pipe
4028 let hi = close.saturating_sub(1); // just inside the closing pipe
4029 let inside = if hi < lo {
4030 lo
4031 } else {
4032 (open + 2).clamp(lo, hi)
4033 };
4034 row_start + inside
4035 }
4036 (Some(open), None) => row_start + open + 1,
4037 _ => row_start,
4038 }
4039}
4040
4041/// One laid-out table cell: its rendered text, the source range that text
4042/// occupies (`start`/`end` are the caret anchors decoration points at), and the
4043/// column alignment its padding honours.
4044///
4045/// `glyphs` is the cell's inline content *unwrapped* — the box-drawn rows wrap
4046/// it to a column width, but a frontend laying the grid out itself needs the
4047/// text before that decision was made.
4048#[derive(Clone)]
4049pub struct TableCell {
4050 pub glyphs: Vec<Glyph>,
4051 pub start: usize,
4052 pub end: usize,
4053 pub align: Alignment,
4054}
4055
4056/// One row of a table's grid, as the document spells it — not as it's drawn.
4057#[derive(Clone)]
4058pub struct TableRow {
4059 /// A header row: drawn bold, and ruled off from the body below it.
4060 pub head: bool,
4061 pub cells: Vec<TableCell>,
4062}
4063
4064/// A table's structure, published alongside the box-drawn rows that spell it.
4065///
4066/// The rows in [`VisualMap::rows`] are the *default monospace* picture of a
4067/// table: every border a `│`, every column a whole number of character cells.
4068/// That picture is exactly right on any monospace surface, and unfixable off one
4069/// — in a proportional font the `│`s of two rows land at different x and the grid
4070/// shears. So a frontend that draws its own geometry reads this instead: the
4071/// cells, their alignment, and which rows are the head, with no opinion about
4072/// how wide a column is or what a border looks like.
4073///
4074/// Both are always built. The TUI paints `rows` and ignores this; the GUI skips
4075/// `rows` for the span in `rows_span` and draws from here. They describe the
4076/// same cells, so the caret lands on the same offsets either way.
4077#[derive(Clone)]
4078pub struct TableInfo {
4079 /// The `VisualMap::rows` this table's picture occupies, borders included —
4080 /// what a frontend drawing its own table skips over.
4081 pub rows_span: Range<usize>,
4082 /// The source span of the table node, and the offset its trailing caret
4083 /// stop sits at.
4084 pub end_src: usize,
4085 /// The block prefix every row of this table carries — a blockquote's `│ `
4086 /// gutter, a list item's indent. Empty for a table at the top level.
4087 ///
4088 /// A frontend drawing its own grid has to render this and start the table
4089 /// past it, exactly as the picture does; a table nested in a quote that
4090 /// draws flush at the left margin has left the quote.
4091 pub prefix: Vec<Glyph>,
4092 pub grid: Vec<TableRow>,
4093}
4094
4095/// A fenced or indented code block, named by the [`VisualMap::rows`] it occupies.
4096///
4097/// Unlike a table, the rows *are* the block's content — a frontend still paints
4098/// them, it just draws a border and a tinted background around the whole span
4099/// and lets the code inside scroll horizontally instead of wrapping. So this
4100/// carries only the row range; there's no structural alternative to the picture
4101/// the way [`TableInfo`] is one. Derived from [`VRow::code`] — see
4102/// [`code_block_spans`].
4103#[derive(Clone, Debug, PartialEq, Eq)]
4104pub struct CodeBlockInfo {
4105 /// The contiguous run of [`VisualMap::rows`] this code block spans, blank
4106 /// code lines included.
4107 pub rows_span: Range<usize>,
4108 /// The block's language, from a fenced block's info string — what a frontend
4109 /// paints as a small label on the box (`` ```rust `` → `Some("rust")`).
4110 /// `None` for a fence written without one, or an indented block. Editing it
4111 /// goes through [`crate::Doc::set_code_language`], which re-finds the fence
4112 /// in the AST, so this stays a display string.
4113 pub lang: Option<String>,
4114}
4115
4116/// A block-level image (`` on its own line), named by the single
4117/// [`VisualMap::rows`] row it occupies.
4118///
4119/// Like [`CodeBlockInfo`], the row *is* the block's default rendering — a plain
4120/// surface paints the `🖼 alt` placeholder glyphs as-is. An image-capable
4121/// frontend instead **skips the row in `rows_span`** and paints the resolved
4122/// picture there, exactly as it skips a [`TableInfo`]'s box-drawn rows. Derived
4123/// from [`VRow::media`] by [`media_spans`], so it survives the row reuse of
4124/// [`BlockCache`] and [`build_spliced`].
4125#[derive(Clone, Debug, PartialEq, Eq)]
4126pub struct MediaInfo {
4127 /// The [`VisualMap::rows`] rows this media's placeholder occupies — what a
4128 /// capable frontend replaces with the picture or player.
4129 pub rows_span: Range<usize>,
4130 /// Whether this is a picture, a movie, or a sound — which widget the
4131 /// frontend builds over [`rows_span`](MediaInfo::rows_span). A frontend that
4132 /// handles only some kinds leaves the rest as core's placeholder rows, which
4133 /// already read sensibly on their own.
4134 pub kind: MediaKind,
4135 /// The media's link destination — a path, URL, or `data:` URI, verbatim from
4136 /// the AST. A frontend resolves a relative path against the document's own
4137 /// directory; core does no I/O. For a `<picture>` this is the `<img>`
4138 /// fallback — the source used when no [`sources`](MediaInfo::sources) media
4139 /// query matches (or the frontend has no theme). Empty when a `<video>`/
4140 /// `<audio>` carries no `src` and names its candidates in `<source>`s
4141 /// instead; [`resolve`](MediaInfo::resolve) already accounts for that.
4142 pub destination: String,
4143 /// The `<source>` alternatives in document order, or empty for a plain
4144 /// image. See [`MediaSource`]; a theme- or codec-aware frontend picks one and
4145 /// otherwise loads [`destination`](MediaInfo::destination).
4146 pub sources: Vec<MediaSource>,
4147 /// The media's alt text, flattened from its inline children (empty when it
4148 /// has none).
4149 pub alt: String,
4150 /// A `<video poster="…">`'s still frame, or empty when there is none — an
4151 /// image destination, resolved exactly as [`destination`] is.
4152 ///
4153 /// [`destination`]: MediaInfo::destination
4154 pub poster: String,
4155}
4156
4157/// One leaf directive (`::name{…}`) as a frontend sees it: which rows its
4158/// placeholder occupies, its type, and its attributes. A plain surface paints
4159/// the `⧉ name` placeholder glyphs as-is; a frontend that knows the host app's
4160/// vocabulary **skips the rows in `rows_span`** and paints the real thing there,
4161/// exactly as an image-capable one does with [`MediaInfo`]. Derived from
4162/// [`VRow::leaf_directive`] by [`directive_spans`].
4163///
4164/// Core resolves nothing here — it has no idea what an `embed` or a `toc` is,
4165/// and deliberately so: the directive vocabulary belongs to the app on top.
4166#[derive(Clone, Debug, PartialEq, Eq)]
4167pub struct DirectiveInfo {
4168 /// The [`VisualMap::rows`] rows this directive's placeholder occupies — the
4169 /// label row plus any blank fillers under it.
4170 pub rows_span: Range<usize>,
4171 /// The directive's type (`embed`, `toc`, `vis`), no leading colons.
4172 pub name: String,
4173 /// Its `{…}` attributes in source order; a bare one has a `None` value.
4174 pub attrs: Vec<(String, Option<String>)>,
4175 /// Its `[label]` text, flattened from its inline children (empty when it has
4176 /// none) — what the placeholder row shows.
4177 pub label: String,
4178}
4179
4180impl DirectiveInfo {
4181 /// The value of attribute `key`, if it has one with a value. The convenience
4182 /// a frontend reaches for first (`info.attr("src")`), since almost every
4183 /// directive that draws as something real is pointed at by one attribute.
4184 pub fn attr(&self, key: &str) -> Option<&str> {
4185 self.attrs
4186 .iter()
4187 .find(|(k, _)| k == key)
4188 .and_then(|(_, v)| v.as_deref())
4189 }
4190}
4191
4192impl MediaInfo {
4193 /// The image URL to load under `scheme`: the first [`sources`] `<source>`
4194 /// whose media query matches, else the [`destination`] `<img>` fallback. The
4195 /// pick is a `<source>`'s first `srcset` URL or the destination — a frontend
4196 /// resolves whichever it gets against the document directory exactly as it
4197 /// resolves `destination`, and reserves/keys the picture under `destination`
4198 /// regardless, so a theme switch just re-picks without disturbing the layout.
4199 ///
4200 /// Only `prefers-color-scheme` is understood (that's what a light/dark banner
4201 /// uses); a `<source>` with any other media query is skipped, and one with no
4202 /// media at all always matches (an unconditional override). With no matching
4203 /// source — including every frontend that can't/doesn't theme and passes
4204 /// [`ColorScheme::Light`] to a dark-only picture — it's the plain `<img>`.
4205 ///
4206 /// [`sources`]: MediaInfo::sources
4207 /// [`destination`]: MediaInfo::destination
4208 pub fn resolve(&self, scheme: ColorScheme) -> &str {
4209 if let Some(url) = self
4210 .sources
4211 .iter()
4212 .find(|s| media_matches(&s.media, scheme))
4213 .and_then(|s| first_srcset_url(&s.srcset))
4214 {
4215 return url;
4216 }
4217 // A `<video>`/`<audio>` may carry no `src` of its own, naming its
4218 // candidates only in child `<source>`s — none of which matched above,
4219 // because a codec-typed `<source>` has no media query and core judges no
4220 // MIME types. Falling through to an empty destination would hand the
4221 // frontend nothing to load, so take the first candidate URL instead and
4222 // let the frontend reject it if it can't decode it. An `<img>` never
4223 // reaches this: its `src` is the picture.
4224 if self.destination.is_empty()
4225 && let Some(url) = self
4226 .sources
4227 .iter()
4228 .find_map(|s| first_srcset_url(&s.srcset))
4229 {
4230 return url;
4231 }
4232 &self.destination
4233 }
4234
4235 /// The **still picture** that stands for this media under `scheme`, for a
4236 /// frontend that can rasterize an image but not play a movie — a terminal, or
4237 /// a GUI still growing its player. `None` when there is no picture to draw,
4238 /// which is the honest answer for audio and for a poster-less video: the
4239 /// caller leaves core's labelled placeholder row, which already reads as
4240 /// *a thing that isn't text*.
4241 ///
4242 /// This exists so those frontends never hand a `.mp4` to an image decoder.
4243 /// That fails harmlessly today (a failed decode falls back to the same
4244 /// placeholder), but it spends a file read and a decode attempt per frame to
4245 /// arrive where this gets in one match.
4246 pub fn still(&self, scheme: ColorScheme) -> Option<&str> {
4247 match self.kind {
4248 MediaKind::Image => Some(self.resolve(scheme)),
4249 // A `poster` is an image destination, so it resolves the same way —
4250 // but it is named directly and has no `<source>` alternatives of its
4251 // own, so it needs no theme matching.
4252 MediaKind::Video if !self.poster.is_empty() => Some(&self.poster),
4253 MediaKind::Video | MediaKind::Audio => None,
4254 }
4255 }
4256}
4257
4258/// A frontend's active color scheme — what a `<picture>`'s `prefers-color-scheme`
4259/// `<source>`s are matched against by [`MediaInfo::resolve`]. A frontend with no
4260/// notion of theme passes [`Light`](ColorScheme::Light), the web's own default.
4261#[derive(Clone, Copy, Debug, PartialEq, Eq)]
4262pub enum ColorScheme {
4263 Light,
4264 Dark,
4265}
4266
4267/// Whether a `<source media="…">` query applies under `scheme`. Empty media is
4268/// an unconditional `<source>` (always matches); otherwise only a
4269/// `prefers-color-scheme: dark|light` feature is understood — anything else
4270/// (a width query, `print`, …) doesn't match, so resolution falls through to the
4271/// next source or the `<img>`. Deliberately lax about the surrounding syntax
4272/// (`(prefers-color-scheme: dark)`, `screen and (prefers-color-scheme:dark)`):
4273/// it keys off the feature and its value, which is all the theme case needs.
4274fn media_matches(media: &str, scheme: ColorScheme) -> bool {
4275 let media = media.trim();
4276 if media.is_empty() {
4277 return true;
4278 }
4279 let lower = media.to_ascii_lowercase();
4280 let Some(after) = lower
4281 .split_once("prefers-color-scheme")
4282 .map(|(_, rest)| rest)
4283 else {
4284 return false;
4285 };
4286 // Skip the `:` and any spaces to reach the value word.
4287 let value = after.trim_start_matches([':', ' ', '\t']);
4288 let wanted = match scheme {
4289 ColorScheme::Light => "light",
4290 ColorScheme::Dark => "dark",
4291 };
4292 value.starts_with(wanted)
4293}
4294
4295/// The first URL in a `srcset`: its first comma-separated candidate, before any
4296/// `1x`/`2x`/width descriptor. The theme case only ever puts one URL per
4297/// `<source>`, so the first candidate is the picture.
4298fn first_srcset_url(srcset: &str) -> Option<&str> {
4299 let first = srcset.split(',').next()?.trim();
4300 first.split_whitespace().next().filter(|u| !u.is_empty())
4301}
4302
4303/// The narrowest a column may be squeezed. Below a few characters a column
4304/// stops carrying text and just shreds it one letter per line, which is worse
4305/// than letting the grid run wide.
4306const MIN_COL_WIDTH: usize = 3;
4307
4308/// Shrink `widths` until the grid fits `avail` screen columns, taking from the
4309/// widest column each time so the loss is shared out rather than falling on
4310/// whichever column happens to be last. No column goes below
4311/// [`MIN_COL_WIDTH`]; a table with more columns than the surface has room for
4312/// still overflows, which is the honest outcome — there's nothing left to give.
4313fn fit_widths(widths: &mut [usize], avail: usize) {
4314 // Chrome: each column is its content plus a gutter either side, and every
4315 // column is closed by a `│` — with one more opening the row.
4316 let budget = avail.saturating_sub(3 * widths.len() + 1);
4317 while widths.iter().sum::<usize>() > budget {
4318 let Some(w) = widths.iter_mut().filter(|w| **w > MIN_COL_WIDTH).max() else {
4319 return;
4320 };
4321 *w -= 1;
4322 }
4323}
4324
4325/// Word-wrap `glyphs` into lines of at most `width` columns, hard-breaking any
4326/// single word too long to fit.
4327///
4328/// Unlike a paragraph — where an overlong word just trails off the end of the
4329/// line — a table column is a hard boundary: a glyph past it lands on top of
4330/// the border, or on the next cell. So the width here is a promise, and a word
4331/// that won't keep it is broken.
4332///
4333/// The space at a break is dropped rather than hung past the edge. Its offset
4334/// isn't lost: the caller gives every line an end stop just past its last
4335/// glyph, which is exactly where that space was.
4336///
4337/// `width` is in display columns, and a break only ever falls between grapheme
4338/// clusters. Both matter to more than the picture: the caller anchors each
4339/// line's end stop just past its last glyph, so a line cut mid-cluster would
4340/// put a caret stop inside a character — reachable by Down or a click, and the
4341/// next Backspace would take the cluster apart from the middle.
4342///
4343/// An explicit in-cell break (a `\n` glyph, from a `<br>`) is a hard boundary:
4344/// each run between the breaks wraps on its own and the results stack. The break
4345/// glyphs are dropped — the caller's per-line end stop already sits exactly where
4346/// each break was, so no offset is lost.
4347fn wrap_glyphs(glyphs: &[Glyph], width: usize) -> Vec<Vec<Glyph>> {
4348 if glyphs.iter().any(|g| g.ch == '\n') {
4349 return glyphs
4350 .split(|g| g.ch == '\n')
4351 .flat_map(|seg| wrap_segment(seg, width))
4352 .collect();
4353 }
4354 wrap_segment(glyphs, width)
4355}
4356
4357/// [`wrap_glyphs`] for a run with no explicit breaks — the word-wrap proper.
4358fn wrap_segment(glyphs: &[Glyph], width: usize) -> Vec<Vec<Glyph>> {
4359 let width = width.max(1);
4360 // Words are maximal non-space runs, each carrying the space that followed it
4361 // — which survives only if the next word joins it on this line.
4362 let mut words: Vec<(Vec<Glyph>, Option<Glyph>)> = Vec::new();
4363 let mut word: Vec<Glyph> = Vec::new();
4364 for g in glyphs {
4365 if g.ch == ' ' {
4366 words.push((std::mem::take(&mut word), Some(g.clone())));
4367 } else {
4368 word.push(g.clone());
4369 }
4370 }
4371 if !word.is_empty() {
4372 words.push((word, None));
4373 }
4374
4375 let mut lines: Vec<Vec<Glyph>> = Vec::new();
4376 let mut line: Vec<Glyph> = Vec::new();
4377 let mut used = 0usize;
4378 let mut gap: Option<Glyph> = None;
4379 for (word, space) in words {
4380 for chunk in hard_break(&word, width) {
4381 let sep = gap.is_some() as usize;
4382 let cells = glyphs_width(chunk);
4383 if !line.is_empty() && used + sep + cells > width {
4384 lines.push(std::mem::take(&mut line));
4385 used = 0;
4386 gap = None; // the break swallows the space
4387 }
4388 if let Some(sp) = gap.take() {
4389 line.push(sp);
4390 used += 1;
4391 }
4392 line.extend_from_slice(chunk);
4393 used += cells;
4394 }
4395 gap = space;
4396 }
4397 // An empty cell is still one (empty) line — it has an end the caret can
4398 // sit at, which is how you type into it.
4399 if !line.is_empty() || lines.is_empty() {
4400 lines.push(line);
4401 }
4402 lines
4403}
4404
4405/// Break a single word into pieces of at most `width` columns, cutting only
4406/// between grapheme clusters — the replacement for slicing it into fixed runs
4407/// of glyphs, which measures a wide character as one column and can cut an
4408/// emoji in half.
4409///
4410/// A cluster wider than the whole column still gets a piece to itself: there is
4411/// nowhere legal to cut it, and overflowing by a cell is better than splitting a
4412/// character. An empty word yields no pieces at all, which is what keeps a
4413/// double space from opening a line of its own.
4414fn hard_break(word: &[Glyph], width: usize) -> Vec<&[Glyph]> {
4415 let mut out = Vec::new();
4416 if word.is_empty() {
4417 return out;
4418 }
4419 let (mut start, mut used) = (0usize, 0usize);
4420 for c in clusters(word) {
4421 if used > 0 && used + c.cells > width {
4422 out.push(&word[start..c.glyph]);
4423 start = c.glyph;
4424 used = 0;
4425 }
4426 used += c.cells;
4427 }
4428 out.push(&word[start..]);
4429 out
4430}
4431
4432/// A table rule spanning `widths`, e.g. `┌──────┬─────┐`. Each column is its
4433/// content width plus the one-space gutter on either side.
4434fn rule_text(widths: &[usize], left: char, mid: char, right: char) -> String {
4435 let mut s = String::new();
4436 s.push(left);
4437 for (i, w) in widths.iter().enumerate() {
4438 if i > 0 {
4439 s.push(mid);
4440 }
4441 for _ in 0..w + 2 {
4442 s.push('─');
4443 }
4444 }
4445 s.push(right);
4446 s
4447}
4448
4449/// Push real document text: each glyph maps to its own source byte, and the one
4450/// that opens a grapheme cluster is the caret stop for the whole cluster.
4451///
4452/// Per cluster rather than per codepoint because a cluster is the character the
4453/// user sees, and it's the unit backspace and delete already step by. A stop
4454/// inside 👨👩👧 — five codepoints strung together with joiners — is a caret
4455/// parked in the middle of a character: one press of Right lands there, and the
4456/// next Backspace severs a joiner from what it joined, leaving a dangling ZWJ in
4457/// the source. The rest of the cluster still gets its glyph (it has to be
4458/// drawn); it just isn't somewhere to stand.
4459fn push_text(out: &mut Vec<Glyph>, text: &str, base_src: usize, style: Style) {
4460 for (gi, cluster) in text.grapheme_indices(true) {
4461 for (ci, ch) in cluster.char_indices() {
4462 out.push(Glyph {
4463 ch,
4464 style,
4465 src: base_src + gi + ci,
4466 stop: ci == 0,
4467 });
4468 }
4469 }
4470}
4471
4472/// Emit an inline `str`/`smart_punctuation` run, mapping every visible char back
4473/// to its *true* source byte even when the source carries backslash escapes the
4474/// parsed `text` dropped (`\*` → `*`). The naive `span.start + text_offset`
4475/// mapping [`push_text`] uses drifts by one byte after each escape, so a caret or
4476/// click past an escaped `*` would land on the wrong character; walking the text
4477/// against its source keeps them aligned, and the hidden escape backslash gets no
4478/// glyph of its own (it is a spelling artefact, not something the caret lands on).
4479fn push_escaped_text(
4480 out: &mut Vec<Glyph>,
4481 text: &str,
4482 span: Range<usize>,
4483 source: &str,
4484 style: Style,
4485) {
4486 let end = span.end.min(source.len());
4487 let src = source.get(span.start..end).unwrap_or("");
4488 // Fast path — no dropped bytes, so text and source align 1:1 (the common
4489 // case: prose with no escapes). Byte lengths equal ⇒ no backslash was eaten.
4490 if src.len() == text.len() {
4491 push_text(out, text, span.start, style);
4492 return;
4493 }
4494 // Slow path: some `\` was consumed. Walk char-by-char, skipping a backslash
4495 // in the source exactly when it escapes the next visible char (a real escape),
4496 // never when it is a literal backslash the parse kept (that case has equal
4497 // lengths and takes the fast path above).
4498 let sb = src.as_bytes();
4499 let mut si = 0usize;
4500 'text: for (_, cluster) in text.grapheme_indices(true) {
4501 for (ci, ch) in cluster.char_indices() {
4502 // The text outlasted the source it is being mapped onto. In a
4503 // consistent document that cannot happen on this path: the slow path
4504 // is only entered when the two lengths differ, and everything that
4505 // makes them differ makes the *source* the longer one — an escape
4506 // backslash the parse ate, or source folded into a neighbouring node.
4507 // A `smart_punctuation` node reports its canonical ASCII spelling
4508 // (`--`, `...`, `"`), which is never longer than what was written.
4509 //
4510 // So reaching here means `span` was measured against a document that
4511 // `source` is no longer, and there is no honest offset left to give
4512 // the remaining characters. Stop: the row comes out short, which is
4513 // a wrong picture of a document that is already inconsistent. The
4514 // alternative was `si` stepping past the end and the slice below
4515 // panicking — which is what it did, in a paint loop.
4516 if si >= sb.len() {
4517 break 'text;
4518 }
4519 // Advance to the source character this one came from, stepping over
4520 // whatever the parse dropped on the way. An escape backslash is the
4521 // common case, but not the only one: a span can cover source that
4522 // was folded into a neighbouring node (smart punctuation next to a
4523 // bracket gives `text: "]"` over a source span of `"…]"`). Advancing
4524 // by the *text* character's length assumed escapes were the only
4525 // divergence, so one dropped multi-byte character desynchronized
4526 // every glyph after it — placing `]` inside the `…` before it.
4527 while si < sb.len() && !src[si..].starts_with(ch) {
4528 si += src[si..].chars().next().map_or(1, char::len_utf8);
4529 }
4530 out.push(Glyph {
4531 ch,
4532 style,
4533 src: span.start + si.min(src.len()),
4534 stop: ci == 0,
4535 });
4536 si += src[si..]
4537 .chars()
4538 .next()
4539 .map_or(ch.len_utf8(), char::len_utf8);
4540 }
4541 }
4542}
4543
4544/// Build synthetic decoration glyphs (a bullet, a gutter) all pointing at `src`,
4545/// each carrying `role` so the frontend can style it (`Role::Body` for plain
4546/// padding). Synthetic glyphs are never caret stops — they share one offset, so
4547/// the caret steps over them (a click still lands at `src`).
4548fn synth(text: &str, role: Role, src: usize) -> Vec<Glyph> {
4549 let style = Style::default().role(role);
4550 text.chars()
4551 .map(|ch| Glyph {
4552 ch,
4553 style,
4554 src,
4555 stop: false,
4556 })
4557 .collect()
4558}
4559
4560fn concat(a: &[Glyph], b: &[Glyph]) -> Vec<Glyph> {
4561 let mut v = a.to_vec();
4562 v.extend_from_slice(b);
4563 v
4564}
4565
4566/// The columns a row's prefix (a bullet, a quote gutter, an indent) takes up
4567/// before the text it introduces — what the wrap budget has left to spend.
4568fn prefix_width(prefix: &[Glyph]) -> usize {
4569 glyphs_width(prefix)
4570}
4571
4572/// The label shown for an image with no alt text: the final path segment of its
4573/// destination (`img/cat.png` → `cat.png`), the whole destination when it has no
4574/// separator, and `"image"` when it's empty. A `data:` URI (which has no useful
4575/// tail) shows its scheme so the placeholder isn't a wall of base64.
4576fn media_label(dest: &str) -> String {
4577 if dest.is_empty() {
4578 return "image".to_string();
4579 }
4580 if dest.starts_with("data:") {
4581 return "data:…".to_string();
4582 }
4583 // Trim a query/fragment so a URL's `?v=2#frag` doesn't ride along.
4584 let clean = dest.split(['?', '#']).next().unwrap_or(dest);
4585 let tail = clean
4586 .trim_end_matches('/')
4587 .rsplit(['/', '\\'])
4588 .next()
4589 .unwrap_or(clean);
4590 if tail.is_empty() {
4591 dest.to_string()
4592 } else {
4593 tail.to_string()
4594 }
4595}
4596
4597/// A directive's attributes read as a human label — what a frontend puts on a
4598/// container's tinted panel, and what an attribute-bearing inline directive
4599/// shows in its chip.
4600///
4601/// Reads BOTH conventions diaryx content actually uses: twig's own dot-prefixed
4602/// classes (`{.public .family}`, arriving as one combined `class` attr) and bare
4603/// pandoc-style words with no leading dot (`{public family}` — what
4604/// `diaryx_core::visibility`'s publish-time filter and apps/web's directive
4605/// serializer both write, and which twig parses as one valueless attribute
4606/// each). Reading only `.class` would leave every *existing* diaryx `:::vis{…}`
4607/// block unlabeled. A `key=value` attr is configuration rather than a name, so
4608/// it contributes nothing. `None` when nothing readable is left.
4609fn directive_attr_label(attrs: &[(String, Option<String>)]) -> Option<String> {
4610 let mut parts: Vec<String> = Vec::new();
4611 for (k, v) in attrs {
4612 if k == "class" {
4613 if let Some(v) = v
4614 && !v.is_empty()
4615 {
4616 parts.push(v.clone());
4617 }
4618 } else if v.as_deref().unwrap_or("").is_empty() {
4619 parts.push(k.clone());
4620 }
4621 }
4622 (!parts.is_empty()).then(|| parts.join(" "))
4623}
4624
4625fn heading_style(level: u32) -> Style {
4626 // Just the role — a frontend decides how a heading of this level *looks*
4627 // (the terminal cycles a color and bolds it, the GUI scales the font). The
4628 // author wrote no emphasis here, so core records none. `level as u8` is safe:
4629 // Markdown/Djot cap headings at 6.
4630 Style::default().role(Role::Heading(level.min(255) as u8))
4631}
4632
4633/// Is this `container` node a *directive* (`:::note{…}`, `::embed{…}`,
4634/// `:vis[…]`) rather than an HTML element (`<video>`, `<picture>`, `<div>`)?
4635///
4636/// twig 2.8 folded `div`/`span`/`directive`/`element` into one `container` kind,
4637/// and left nothing that separated them: `kind`, `name` and `directive_form` all
4638/// agree, field for field, on an HTML `<div>` and a Markdown `:::div`. Leaf
4639/// answered it by sniffing the span for whichever of `:` or `<` came first.
4640/// twig 3.0 records the answer at parse time as [`ContainerOrigin`], so this is
4641/// now the parser's own knowledge rather than a guess rebuilt from the bytes it
4642/// consumed.
4643pub(crate) fn container_is_directive(node: &FlatNode) -> bool {
4644 node.origin == Some(ContainerOrigin::Directive)
4645}
4646
4647/// The tag a `container` node carries when it is an HTML element rather than a
4648/// directive — `Some("video")` for a promoted `<video>`, `None` for a `:::note`
4649/// or for any node that is not a container at all.
4650pub(crate) fn element_tag(node: &FlatNode) -> Option<&str> {
4651 (node.origin == Some(ContainerOrigin::Element))
4652 .then_some(node.name.as_deref())
4653 .flatten()
4654}
4655
4656pub(crate) fn is_inline(node: &FlatNode) -> bool {
4657 // A directive is inline only in its `text` form (`:name[label]{…}`); the
4658 // `leaf` and `container` forms are blocks. All three report the same `kind`,
4659 // so the form is the only thing telling them apart — and getting it wrong
4660 // costs a whole paragraph: a text directive misread as a block makes its
4661 // paragraph fail the "all children inline" test in `block`, and the line is
4662 // then walked as a container of blocks, rendering as empty rows with no
4663 // caret home at all.
4664 //
4665 // An HTML element shares the `container` kind but never the `text` form, so
4666 // it answers `false` here and is walked as the block it is.
4667 if node.kind == Kind::Container {
4668 return container_is_directive(node) && node.directive_form == Some(DirectiveForm::Text);
4669 }
4670 is_inline_kind(&node.kind)
4671}
4672
4673/// [`is_inline`] by kind alone — for the ancestor walks, whose `QueryMatch`es
4674/// carry no `directive_form`. It answers `false` for every directive, which its
4675/// callers must (and do) reconcile: they pair it with `is_block_container`,
4676/// which claims every directive, so the pair's verdict is the same one a form
4677/// would have given. Anything looking at a *directive itself* wants [`is_inline`]
4678/// and a real node.
4679pub(crate) fn is_inline_kind(kind: &Kind) -> bool {
4680 matches!(
4681 kind,
4682 Kind::Str
4683 | Kind::SoftBreak
4684 | Kind::HardBreak
4685 | Kind::NonBreakingSpace
4686 | Kind::Emph
4687 | Kind::Strong
4688 | Kind::Mark
4689 | Kind::Insert
4690 | Kind::Delete
4691 | Kind::Verbatim
4692 | Kind::InlineMath
4693 | Kind::DisplayMath
4694 | Kind::Url
4695 | Kind::Email
4696 | Kind::Link
4697 | Kind::Image
4698 | Kind::SmartPunctuation
4699 | Kind::Superscript
4700 | Kind::Subscript
4701 | Kind::FootnoteReference
4702 )
4703}
4704
4705/// Assert two maps are identical down to every glyph, stop, and table span — the
4706/// contract `build_cached` and `build_spliced` must hold against `build`. Lives
4707/// at module scope (not in `mod tests`) so the Doc-driven differential test in
4708/// `doc.rs` can reach it and the private `stops` field it compares.
4709#[cfg(test)]
4710pub(crate) fn assert_maps_eq(a: &VisualMap, b: &VisualMap, ctx: &str) {
4711 assert_eq!(a.rows.len(), b.rows.len(), "row count ({ctx})");
4712 for (i, (ra, rb)) in a.rows.iter().zip(&b.rows).enumerate() {
4713 assert_eq!(ra.end_src, rb.end_src, "row {i} end_src ({ctx})");
4714 assert_eq!(ra.decoration, rb.decoration, "row {i} decoration ({ctx})");
4715 // The incremental walk labels a boundary from a query match's kind
4716 // string and the whole-arena walk from a `FlatNode`'s; this is what says
4717 // the two doors reach the same answer.
4718 assert_eq!(ra.boundary, rb.boundary, "row {i} boundary ({ctx})");
4719 assert_eq!(ra.code, rb.code, "row {i} code ({ctx})");
4720 assert_eq!(ra.code_lang, rb.code_lang, "row {i} code_lang ({ctx})");
4721 assert_eq!(
4722 ra.glyphs.len(),
4723 rb.glyphs.len(),
4724 "row {i} glyph count ({ctx})"
4725 );
4726 for (j, (ga, gb)) in ra.glyphs.iter().zip(&rb.glyphs).enumerate() {
4727 assert_eq!(
4728 (ga.ch, ga.src, ga.stop, ga.style),
4729 (gb.ch, gb.src, gb.stop, gb.style),
4730 "row {i} glyph {j} ({ctx})"
4731 );
4732 }
4733 }
4734 assert_eq!(a.content_start, b.content_start, "content_start ({ctx})");
4735 assert_eq!(a.stops, b.stops, "stops ({ctx})");
4736 assert_eq!(a.tables.len(), b.tables.len(), "table count ({ctx})");
4737 for (i, (ta, tb)) in a.tables.iter().zip(&b.tables).enumerate() {
4738 assert_eq!(ta.rows_span, tb.rows_span, "table {i} rows_span ({ctx})");
4739 assert_eq!(ta.end_src, tb.end_src, "table {i} end_src ({ctx})");
4740 }
4741 assert_eq!(a.code_blocks, b.code_blocks, "code_blocks ({ctx})");
4742 assert_eq!(a.media, b.media, "images ({ctx})");
4743}
4744
4745#[cfg(test)]
4746mod tests {
4747 use super::*;
4748 use twig::{Editor, Format, NodeId};
4749
4750 fn map(src: &str) -> VisualMap {
4751 let mut ed = Editor::new_str(src, Format::Markdown).unwrap();
4752 build_t(&ed.nodes().unwrap(), src, Some(80))
4753 }
4754
4755 /// [`map`] over a Djot source. Djot is the format that spells superscript
4756 /// and subscript at all — Markdown has no syntax for either.
4757 fn map_djot(src: &str) -> VisualMap {
4758 let mut ed = Editor::new_str(src, Format::Djot).unwrap();
4759 build_t(&ed.nodes().unwrap(), src, Some(80))
4760 }
4761
4762 /// The baseline every glyph spelling `ch` was built with, in row order —
4763 /// how a test reads a raised or lowered run off the map without caring
4764 /// which row it landed on.
4765 fn baselines_of(m: &VisualMap, ch: char) -> Vec<Baseline> {
4766 m.rows
4767 .iter()
4768 .flat_map(|r| r.glyphs.iter())
4769 .filter(|g| g.ch == ch)
4770 .map(|g| g.style.baseline)
4771 .collect()
4772 }
4773
4774 /// [`map`] at a chosen wrap width.
4775 fn map_at(src: &str, wrap: Option<usize>) -> VisualMap {
4776 let mut ed = Editor::new_str(src, Format::Markdown).unwrap();
4777 build_t(&ed.nodes().unwrap(), src, wrap)
4778 }
4779
4780 /// [`map`], but with twig's `directives` extension on (off by twig's own
4781 /// default) — the `:::name{.class}` fenced-div containers leaf-core's
4782 /// `"directive"` wysiwyg arm renders.
4783 fn map_directives(src: &str) -> VisualMap {
4784 let mut ed = Editor::new_ext(
4785 src.as_bytes(),
4786 Format::Markdown,
4787 twig::MarkdownExtensions {
4788 directives: true,
4789 ..Default::default()
4790 },
4791 )
4792 .unwrap();
4793 build_t(&ed.nodes().unwrap(), src, Some(80))
4794 }
4795
4796 /// [`map`] with soft breaks preserved (`LineFlow::Preserve`).
4797 fn map_preserve(src: &str, wrap: Option<usize>) -> VisualMap {
4798 let mut ed = Editor::new_str(src, Format::Markdown).unwrap();
4799 build(&ed.nodes().unwrap(), src, wrap, true, &HashMap::new(), None)
4800 }
4801
4802 /// The cache-free reference [`build`], with no per-image height overrides —
4803 /// every block image stays its default one-row placeholder. The tests that
4804 /// need a taller image drive it through [`crate::Doc::set_media_rows`] instead.
4805 fn build_t(nodes: &[FlatNode], src: &str, wrap: Option<usize>) -> VisualMap {
4806 build(nodes, src, wrap, false, &HashMap::new(), None)
4807 }
4808
4809 /// An arena and a string that disagree — spans reaching past the source they
4810 /// are built against.
4811 ///
4812 /// [`crate::Doc`] keeps the two in step, so this is a "cannot happen" that
4813 /// nonetheless *did*: `examples/bench` timed a loop of `edit_range` and then
4814 /// went on handing the grown editor's spans to a builder holding the string
4815 /// from before it, and every run ended in a slice panic rather than a
4816 /// number. `push_escaped_text` was already written to survive the mismatch —
4817 /// it clamps the span's end and falls back to an empty slice — and this is
4818 /// the half of that intent it did not carry through.
4819 ///
4820 /// Rendering the wrong thing is the acceptable answer here; panicking in a
4821 /// paint loop is not.
4822 #[test]
4823 fn a_source_shorter_than_the_arena_built_over_it_renders_rather_than_panicking() {
4824 // An escape puts the run on `push_escaped_text`'s slow path — the fast
4825 // path is a length comparison that a truncated source fails anyway.
4826 let src = "alpha \\*beta\\* gamma delta epsilon\n";
4827 let mut ed = Editor::new_str(src, Format::Markdown).unwrap();
4828 let nodes = ed.nodes().unwrap();
4829
4830 // Every truncation of it, so the cut lands before, inside and after the
4831 // escaped run rather than only where one hand-picked index put it.
4832 for cut in 0..=src.len() {
4833 if !src.is_char_boundary(cut) {
4834 continue;
4835 }
4836 let map = build_t(&nodes, &src[..cut], Some(80));
4837 for row in &map.rows {
4838 for g in &row.glyphs {
4839 assert!(
4840 g.src <= src.len(),
4841 "cut {cut}: glyph {:?} points past the source at {}",
4842 g.ch,
4843 g.src
4844 );
4845 }
4846 }
4847 }
4848 }
4849
4850 fn rendered(m: &VisualMap) -> String {
4851 m.rows
4852 .iter()
4853 .map(|r| r.glyphs.iter().map(|g| g.ch).collect::<String>())
4854 .collect::<Vec<_>>()
4855 .join("\n")
4856 }
4857
4858 /// Render a source both ways: `build` over the whole marshalled arena (the
4859 /// reference), and `build_cached` driven the way [`crate::Doc`] drives it —
4860 /// top-level blocks from `child_spans`, per-block subtrees on a miss.
4861 fn render_both(
4862 ed: &mut Editor,
4863 src: &str,
4864 wrap: Option<usize>,
4865 cache: &mut BlockCache,
4866 ) -> (VisualMap, VisualMap) {
4867 let all = ed.nodes().unwrap();
4868 let media_rows = HashMap::new();
4869 let plain = build(&all, src, wrap, false, &media_rows, None);
4870 let top = top_blocks(ed);
4871 let cached = build_cached(&top, src, wrap, false, &media_rows, None, cache, |id| {
4872 ed.subtree(NodeId(id)).unwrap_or_default()
4873 });
4874 (plain, cached)
4875 }
4876
4877 /// The whole correctness claim of the block cache: `build_cached` produces a
4878 /// byte-identical map to `build`, on a fresh cache *and* — the case that
4879 /// actually exercises reuse-and-shift plus per-block subtree marshalling — on
4880 /// a warm cache after the source has been edited underneath it.
4881 /// **Every glyph must stand on the character it claims.** A row's source
4882 /// extent is computed from its last glyph's offset, so a glyph carrying an
4883 /// offset that is not its own character's start yields a row end inside a
4884 /// multi-byte character — and every later slice of the source panics on it.
4885 ///
4886 /// Reproduces a real crash from a journal entry: a bracketed elision inside
4887 /// a blockquote (`[…]`) gave the closing bracket a `text` of `"]"` over a
4888 /// source span covering `"…]"`, because the parse folded the ellipsis into a
4889 /// neighbouring node. `push_escaped_text` walked that span assuming a
4890 /// dropped backslash was the only way text and source could diverge, so the
4891 /// `]` landed on the `…`'s first byte:
4892 /// `byte index 1236 is not a char boundary; it is inside '…'`.
4893 #[test]
4894 fn a_glyph_never_lands_inside_the_character_before_it() {
4895 let src = "> engage with it rather than look away. […]\n>\n> The through-line\n";
4896 let vmap = map(src);
4897 for (r, row) in vmap.rows.iter().enumerate() {
4898 assert!(
4899 src.is_char_boundary(row.end_src.min(src.len())),
4900 "row {r} ends at {} — inside a character",
4901 row.end_src
4902 );
4903 for g in &row.glyphs {
4904 assert!(
4905 src.is_char_boundary(g.src.min(src.len())),
4906 "row {r} has {:?} at {}, which is inside a character",
4907 g.ch,
4908 g.src
4909 );
4910 }
4911 }
4912 // The elision survives, and its bracket sits on the real `]`.
4913 let text: String = vmap
4914 .rows
4915 .iter()
4916 .flat_map(|r| r.glyphs.iter().map(|g| g.ch))
4917 .collect();
4918 assert!(text.contains("[…]"), "the elision should render: {text:?}");
4919 let close = vmap
4920 .rows
4921 .iter()
4922 .flat_map(|r| r.glyphs.iter())
4923 .find(|g| g.ch == ']')
4924 .expect("a closing bracket");
4925 assert_eq!(
4926 src[close.src..].chars().next(),
4927 Some(']'),
4928 "the bracket glyph should stand on the source's own `]`"
4929 );
4930 }
4931
4932 #[test]
4933 fn build_cached_matches_build() {
4934 let docs = [
4935 "# Title\n\nThe quick brown fox.\n\nAnother paragraph here.\n",
4936 "## H\n\n- one\n- two\n- three\n\n> a quote\n> continued\n",
4937 "para one\n\n```\ncode\nlines\n```\n\nafter code\n",
4938 "| a | b |\n|---|---|\n| 1 | 2 |\n\ntext after a table\n",
4939 "line\n- \nsetext?\n\nreal para\n\n\n\ntrailing blanks\n",
4940 "> quote with **bold** and a [link](https://x.dev)\n>\n> - item\n> - item2\n\ntail\n",
4941 "intro\n\n\n\nbetween\n\n\n\nend\n",
4942 "- text item\n- \n- more text\n",
4943 // Footnotes: twig parses each definition as a root beside `doc`, so
4944 // these are the docs where the reference build and the incremental
4945 // one could disagree about what the top-level blocks even are.
4946 "A claim[^1] and another[^src].\n\n[^1]: First note.\n\n[^src]: Second.\n\ntail\n",
4947 "note[^a]\n\n[^a]: body **bold**\n wrapped on\n three lines\n\nafter\n",
4948 // No trailing newline. twig closes the document's last block on the
4949 // virtual newline it supplies at EOF, so that block's `span.end` is
4950 // `source.len() + 1` — a range that slices no bytes at all. Keying
4951 // the block cache off such a slice made every last block hash alike;
4952 // see [`block_bytes`].
4953 "# Title\n\nThe quick brown fox.\n\nA tail with no newline",
4954 "A claim[^1] and another[^src].\n\n[^1]: First note.\n[^src]: Second, ending the file.",
4955 ];
4956 for wrap in [None, Some(80usize), Some(20)] {
4957 for src in docs {
4958 let ctx = format!("wrap={wrap:?} src={src:?}");
4959 let mut ed = Editor::new_str(src, Format::Markdown).unwrap();
4960 let mut cache = BlockCache::default();
4961
4962 // 1) Fresh cache equals the cache-free build.
4963 let (plain, cached) = render_both(&mut ed, src, wrap, &mut cache);
4964 assert_maps_eq(&plain, &cached, &format!("fresh {ctx}"));
4965
4966 // 2) Type a char mid-document, reparse, rebuild with the now-warm
4967 // cache: the edited block is re-marshalled and re-rendered,
4968 // every block below it is reused shifted, and the result must
4969 // still match a from-scratch build.
4970 let at = (src.len() / 2..=src.len())
4971 .find(|&i| src.is_char_boundary(i))
4972 .unwrap();
4973 ed.edit_range(at, at, "Z").unwrap();
4974 let src2 = ed.source_str().unwrap();
4975 let (plain2, cached2) = render_both(&mut ed, &src2, wrap, &mut cache);
4976 assert_maps_eq(&plain2, &cached2, &format!("after insert {ctx}"));
4977
4978 // 3) Delete it again: offsets shift back the other way, and the
4979 // warm cache must not hand back stale shifted rows.
4980 ed.edit_range(at, at + 1, "").unwrap();
4981 let src3 = ed.source_str().unwrap();
4982 let (plain3, cached3) = render_both(&mut ed, &src3, wrap, &mut cache);
4983 assert_maps_eq(&plain3, &cached3, &format!("after delete {ctx}"));
4984 }
4985 }
4986 }
4987
4988 /// A document that does not end in a newline is the one place twig hands
4989 /// leaf a top-level span that addresses no source: the last block is closed
4990 /// on the virtual newline the parser supplies at EOF, so its `span.end` is
4991 /// `source.len() + 1`. The block cache keys on the bytes under that span, and
4992 /// reading the out-of-range slice as *no bytes* broke it two ways at once —
4993 /// [`block_bytes`] has the full account. Both ways are checked here, because
4994 /// they fail independently.
4995 #[test]
4996 fn a_block_running_past_the_last_byte_still_keys_the_cache_by_its_own_bytes() {
4997 // One: two overrunning blocks collide. A footnote definition is a root
4998 // beside `doc` that [`top_blocks`] merges into the top level, while the
4999 // `section` above it spans the definition's bytes too — so when the
5000 // definition ends the file, both blocks end past it. The second was
5001 // served the first's rows, and the definition rendered as a copy of the
5002 // heading.
5003 let src = "A claim[^1] worth checking.\n\n# A heading with a reference[^1] in it\n\n[^1]: The first note.\n[^note]: A note with a word for a label.";
5004 let mut ed = Editor::new_str(src, Format::Djot).unwrap();
5005 let (plain, cached) = render_both(&mut ed, src, Some(80), &mut BlockCache::default());
5006 assert_maps_eq(&plain, &cached, "a definition ending the file");
5007 let text = rendered(&cached);
5008 assert!(
5009 text.ends_with("[note] A note with a word for a label."),
5010 "the last definition should render itself: {text:?}"
5011 );
5012 assert_eq!(
5013 text.matches("A heading with a reference").count(),
5014 1,
5015 "the heading should render exactly once: {text:?}"
5016 );
5017
5018 // Two: one overrunning block goes stale. Its bytes are its cache key, so
5019 // a block that keeps hashing the same however it is edited is served the
5020 // rows built before the edit — the whole last line frozen as the user
5021 // types in it.
5022 let mut cache = BlockCache::default();
5023 let first = "first para\n\n# A heading\n\nlast para with no newline";
5024 let mut ed = Editor::new_str(first, Format::Djot).unwrap();
5025 let (_, warm) = render_both(&mut ed, first, Some(80), &mut cache);
5026 assert!(rendered(&warm).ends_with("last para with no newline"));
5027
5028 let second = "first para\n\n# A heading\n\nDIFFERENT text without a newline";
5029 let mut ed = Editor::new_str(second, Format::Djot).unwrap();
5030 let (plain, cached) = render_both(&mut ed, second, Some(80), &mut cache);
5031 assert_maps_eq(&plain, &cached, "edited last block, warm cache");
5032 let text = rendered(&cached);
5033 assert!(
5034 text.ends_with("DIFFERENT text without a newline"),
5035 "the warm cache served the pre-edit rows: {text:?}"
5036 );
5037 }
5038
5039 #[test]
5040 fn resolves_markup_to_plain_text() {
5041 let text = rendered(&map("# Title\n\na **bold** word\n"));
5042 assert!(!text.contains('#'), "heading marker shown: {text:?}");
5043 assert!(!text.contains("**"), "strong delimiters shown: {text:?}");
5044 assert!(text.contains("Title") && text.contains("bold word"));
5045 }
5046
5047 #[test]
5048 fn every_glyph_points_at_its_source_byte() {
5049 let src = "a **bold** c\n";
5050 let m = map(src);
5051 for row in &m.rows {
5052 for g in &row.glyphs {
5053 // A real (non-synthetic) glyph's source byte is the glyph's char.
5054 if g.src < src.len()
5055 && src.is_char_boundary(g.src)
5056 && let Some(sc) = src[g.src..].chars().next()
5057 && sc == g.ch
5058 {
5059 continue;
5060 }
5061 // Synthetic prefixes (none here) would be the only exceptions.
5062 panic!("glyph {:?} at src {} doesn't match source", g.ch, g.src);
5063 }
5064 }
5065 }
5066
5067 #[test]
5068 fn offset_and_position_round_trip_on_visible_text() {
5069 let m = map("hello world\n");
5070 let (r, c) = m.pos_of_offset(6); // the 'w'
5071 assert_eq!(m.offset_of_pos(r, c), 6);
5072 }
5073
5074 #[test]
5075 fn unwrapped_mode_emits_one_row_per_paragraph() {
5076 // A long paragraph that would wrap under a column budget stays a single
5077 // row when wrap is None (the GUI wraps it at pixel width instead).
5078 let long = "one two three four five six seven eight nine ten eleven twelve\n";
5079 let mut ed = Editor::new_str(long, Format::Markdown).unwrap();
5080 let wrapped = build_t(&ed.nodes().unwrap(), long, Some(12));
5081 let unwrapped = build_t(&ed.nodes().unwrap(), long, None);
5082 assert!(wrapped.num_rows() > 1, "narrow column should wrap");
5083 assert_eq!(unwrapped.num_rows(), 1, "no budget should keep it one row");
5084 // Every glyph's source byte is preserved in the single row.
5085 let text: String = unwrapped.rows[0].glyphs.iter().map(|g| g.ch).collect();
5086 assert_eq!(text.trim_end(), long.trim_end());
5087 }
5088
5089 fn line_texts(m: &VisualMap) -> Vec<String> {
5090 m.rows
5091 .iter()
5092 .map(|r| {
5093 // Trim the trailing whitespace a row may carry — the zero-width
5094 // '\n' that closes a preserved line, and any space glyph left at
5095 // a wrap boundary (both real caret stops, neither visible text).
5096 r.glyphs
5097 .iter()
5098 .map(|g| g.ch)
5099 .collect::<String>()
5100 .trim_end()
5101 .to_string()
5102 })
5103 .collect()
5104 }
5105
5106 #[test]
5107 fn preserve_lays_each_soft_break_on_its_own_row() {
5108 // A soft break (a bare newline inside a paragraph) folds into a space by
5109 // default — the whole paragraph is one reflowed row...
5110 let src = "one two\nthree four\n";
5111 let mut ed = Editor::new_str(src, Format::Markdown).unwrap();
5112 let folded = build_t(&ed.nodes().unwrap(), src, None);
5113 assert_eq!(folded.num_rows(), 1, "fold: one reflowed row");
5114 assert_eq!(
5115 line_texts(&folded),
5116 vec!["one two three four"],
5117 "break folded to a space"
5118 );
5119
5120 // ...and under Preserve it renders where it was written, a row per line.
5121 let kept = map_preserve(src, None);
5122 assert_eq!(
5123 line_texts(&kept),
5124 vec!["one two", "three four"],
5125 "preserve: a row per line"
5126 );
5127 }
5128
5129 #[test]
5130 fn a_preserved_break_keeps_the_newline_offset_as_a_caret_stop() {
5131 // The break must leave a caret stop at the newline byte, or the caret
5132 // could not rest at the end of the first line. The '\n' glyph is dropped
5133 // from the row (so nothing stray renders); its offset (7 here) becomes the
5134 // row's end stop instead — the same offset the folded space would carry.
5135 let src = "one two\nthree four\n";
5136 let m = map_preserve(src, None);
5137 assert!(
5138 !m.rows[0].glyphs.iter().any(|g| g.ch == '\n'),
5139 "the break glyph is dropped"
5140 );
5141 assert_eq!(
5142 m.rows[0].end_src, 7,
5143 "the first row ends at the newline byte"
5144 );
5145 assert!(m.is_stop(7), "the newline offset is a caret stop");
5146 // Row end offsets stay strictly ascending — no two rows pin one offset.
5147 let offs: Vec<usize> = m.rows.iter().map(|r| r.end_src).collect();
5148 assert!(
5149 offs.windows(2).all(|w| w[0] < w[1]),
5150 "offsets not unique: {offs:?}"
5151 );
5152 }
5153
5154 #[test]
5155 fn preserved_lines_wrap_independently() {
5156 // Each preserved line wraps to the column on its own; the break between
5157 // them is hard, so a word never crosses it — "gamma" and "delta" could
5158 // share a row on width alone but the soft break keeps them apart.
5159 let src = "alpha beta gamma\ndelta epsilon\n";
5160 let m = map_preserve(src, Some(12));
5161 assert_eq!(
5162 line_texts(&m),
5163 vec!["alpha beta", "gamma", "delta", "epsilon"],
5164 "each source line wraps on its own"
5165 );
5166 }
5167
5168 #[test]
5169 fn an_empty_paragraph_between_blocks_renders_its_own_rows() {
5170 // "A", then two blank lines (an empty paragraph opened with Enter), then
5171 // "B": the empty paragraph must be navigable rows, not collapsed onto B.
5172 // Rows: "A", spacer, empty-paragraph, spacer, "B" — each blank row a
5173 // distinct source offset.
5174 let m = map("A\n\n\n\nB\n");
5175 let text: Vec<String> = m
5176 .rows
5177 .iter()
5178 .map(|r| r.glyphs.iter().map(|g| g.ch).collect())
5179 .collect();
5180 assert_eq!(text, vec!["A", "", "", "", "B"], "got {text:?}");
5181 let offs: Vec<usize> = m.rows.iter().map(|r| r.end_src).collect();
5182 // Strictly ascending — no two rows share an offset (else the caret pins).
5183 assert!(
5184 offs.windows(2).all(|w| w[0] < w[1]),
5185 "offsets not unique: {offs:?}"
5186 );
5187 }
5188
5189 #[test]
5190 fn a_tight_block_boundary_still_gets_one_separator() {
5191 // A heading directly above text (no blank line between) keeps the single
5192 // conventional separator row, as before.
5193 let m = map("# H\ntext\n");
5194 let text: Vec<String> = m
5195 .rows
5196 .iter()
5197 .map(|r| r.glyphs.iter().map(|g| g.ch).collect())
5198 .collect();
5199 assert_eq!(text, vec!["H", "", "text"], "got {text:?}");
5200 }
5201
5202 #[test]
5203 fn an_escaped_delimiter_renders_without_its_backslash_and_maps_true_offsets() {
5204 // `a\*b` renders the three visible chars `a * b` — the escape backslash
5205 // is hidden — and every glyph points at its real source byte, so a caret
5206 // past the escape lands right (the `*` at source 2, `b` at source 3, not
5207 // the drifted 1/2 the naive text-offset mapping gave).
5208 let m = map("a\\*b\n");
5209 let row: Vec<(char, usize)> = m.rows[0].glyphs.iter().map(|g| (g.ch, g.src)).collect();
5210 assert_eq!(row, vec![('a', 0), ('*', 2), ('b', 3)], "got {row:?}");
5211 }
5212
5213 #[test]
5214 fn an_escaped_hash_stays_a_paragraph_and_shows_the_hash() {
5215 // `\# hi` is a paragraph beginning with a literal `#`, not a heading —
5216 // the backslash is hidden, the `#` shown at its true offset.
5217 let m = map("\\# hi\n");
5218 let text: String = m.rows[0].glyphs.iter().map(|g| g.ch).collect();
5219 assert_eq!(text, "# hi");
5220 assert_eq!(
5221 m.rows[0].glyphs[0].src, 1,
5222 "the # is at source byte 1, past the \\"
5223 );
5224 }
5225
5226 #[test]
5227 fn a_tight_nested_list_hangs_its_sublist_directly_under_the_item() {
5228 // A list item's own text and the sub-list nested under it are written on
5229 // adjacent source lines, so the rich view butts them together — no
5230 // fabricated blank row. Regression: the synthetic "breathe" separator
5231 // used to open a gap between `• a` and its ` • b`.
5232 assert_eq!(rendered(&map("- a\n - b\n")), "• a\n • b");
5233 }
5234
5235 #[test]
5236 fn a_loose_nested_list_keeps_its_real_blank_line() {
5237 // A genuine blank source line (a loose list) still parts the item from
5238 // its sub-list — only the *fabricated* separator is suppressed, never a
5239 // real one the author typed. The gap row wears the item's continuation
5240 // prefix (the two-space indent), so it renders as " ", not empty.
5241 assert_eq!(rendered(&map("- a\n\n - b\n")), "• a\n \n • b");
5242 }
5243
5244 #[test]
5245 fn frontmatter_is_hidden_and_the_document_opens_into_its_content() {
5246 // Leading YAML frontmatter renders nothing — no phantom blank rows for
5247 // its lines, no leading gap — and `content_start` points at the first
5248 // real block so the caret floor can keep out of the hidden metadata.
5249 let fm = "---\nconfig: prov.yaml\ncontents:\n- '[Sample](sample.md)'\n---\n";
5250 let src = format!("{fm}# leaf\n\nA line.\n");
5251 let m = map(&src);
5252 let text = rendered(&m);
5253 assert!(
5254 !text.contains("config"),
5255 "frontmatter body leaked: {text:?}"
5256 );
5257 assert!(!text.contains("prov"), "frontmatter body leaked: {text:?}");
5258 assert_eq!(
5259 m.rows[0].glyphs.iter().map(|g| g.ch).collect::<String>(),
5260 "leaf"
5261 );
5262 assert_eq!(
5263 m.content_start,
5264 fm.len(),
5265 "floor should be the first real block"
5266 );
5267 }
5268
5269 #[test]
5270 fn a_frontmatter_only_document_puts_the_floor_after_the_frontmatter() {
5271 // Nothing to render, so the caret floor is the end of the hidden
5272 // frontmatter — not 0, which is *before* the opening `---` and made the
5273 // first keystroke in a fresh metadata-only note land ahead of it. And
5274 // the frontmatter's own newlines are not trailing blank lines: they used
5275 // to open phantom rows at offsets 1..4, inside the metadata.
5276 let src = "---\ntitle: 2026-08-29\nid: f8s32cd\n---\n";
5277 let m = map(src);
5278 assert_eq!(m.content_start, src.len(), "floor must clear the metadata");
5279 assert!(
5280 m.rows.is_empty(),
5281 "frontmatter must render no rows: {:?}",
5282 rendered(&m)
5283 );
5284 assert!(
5285 m.stops.is_empty(),
5286 "no stop may sit inside the metadata: {:?}",
5287 m.stops
5288 );
5289 }
5290
5291 #[test]
5292 fn a_frontmatter_only_document_still_counts_its_real_blank_lines() {
5293 // Two blank lines after the frontmatter are the author's empty paragraph
5294 // and still render, counted from the metadata's end rather than from 0.
5295 let fm = "---\ntitle: n\n---\n";
5296 let m = map(&format!("{fm}\n\n"));
5297 assert_eq!(m.content_start, fm.len());
5298 assert_eq!(m.rows.len(), 2, "the two trailing newlines each open a row");
5299 assert!(
5300 m.rows.iter().all(|r| r.end_src > fm.len()),
5301 "rows must sit past the frontmatter"
5302 );
5303 }
5304
5305 #[test]
5306 fn a_document_without_frontmatter_has_a_zero_floor() {
5307 let m = map("# leaf\n\nbody\n");
5308 assert_eq!(m.content_start, 0);
5309 }
5310
5311 #[test]
5312 fn trailing_spaces_become_caret_stops_so_the_caret_can_be_drawn_past_them() {
5313 // Markdown/Djot drop the trailing space in `hello ` from the `str` node,
5314 // so without help the row would end at `hello` and the caret couldn't be
5315 // drawn past column 5 — typing a space at a line's end wouldn't move it
5316 // on screen until the next visible character reparsed the space into an
5317 // interior node. The builder recovers it from the block's span/content_span
5318 // gap and emits it as a real, caret-stoppable glyph.
5319 let m = map("hello \n");
5320 assert_eq!(
5321 m.rows[0].glyphs.iter().map(|g| g.ch).collect::<String>(),
5322 "hello "
5323 );
5324 assert_eq!(
5325 m.rows[0].end_src, 6,
5326 "the row now ends past the trailing space"
5327 );
5328 // The caret can rest both on and past the space.
5329 assert_eq!(m.pos_of_offset(5), (0, 5), "between 'o' and the space");
5330 assert_eq!(m.pos_of_offset(6), (0, 6), "past the space");
5331 // Two trailing spaces, both stops.
5332 let m = map("hello \n");
5333 assert_eq!(
5334 m.rows[0].glyphs.iter().map(|g| g.ch).collect::<String>(),
5335 "hello "
5336 );
5337 assert_eq!(m.pos_of_offset(7), (0, 7));
5338 }
5339
5340 #[test]
5341 fn a_headings_trailing_space_is_a_caret_stop_too() {
5342 // The hidden `# ` marker means `# hi ` renders as `hi ` in three columns;
5343 // the caret past the trailing space lands on the third.
5344 let m = map("# hi \n");
5345 assert_eq!(
5346 m.rows[0].glyphs.iter().map(|g| g.ch).collect::<String>(),
5347 "hi "
5348 );
5349 assert_eq!(m.pos_of_offset(5), (0, 3));
5350 }
5351
5352 #[test]
5353 fn a_table_cells_trailing_padding_is_not_mistaken_for_block_trailing_space() {
5354 // A cell's own `span` is the whole row, so the trailing-whitespace
5355 // recovery must not run for cells or it would swallow the `│` delimiters
5356 // and neighbours between the cell text and the row's end. The grid stays
5357 // exactly as before.
5358 let text = rendered(&map(TABLE));
5359 assert!(
5360 text.contains("│ Pear │ 3 │"),
5361 "cell padding disturbed:\n{text}"
5362 );
5363 }
5364
5365 #[test]
5366 fn a_click_below_the_last_row_lands_on_the_last_stop_not_offset_zero() {
5367 // A drag into the empty space under a short document used to resolve to
5368 // offset 0 — the wrong direction, and not even a caret stop when the
5369 // document opens on hidden frontmatter (its `content_start` floor is not
5370 // a stop), which crashed the caret invariant. It now lands on the last
5371 // stop: the end of the document, where dragging downward should reach.
5372 let fm = "---\ntitle: n\n---\n";
5373 let m = map(&format!("{fm}# Hi\n\nbody\n"));
5374 let below = m.num_rows() + 5;
5375 let off = m.offset_of_pos(below, 0);
5376 assert!(
5377 m.is_stop(off),
5378 "offset {off} from a below-content click is not a stop"
5379 );
5380 assert_eq!(
5381 off,
5382 m.stops.last().copied().unwrap(),
5383 "should be the document's last stop"
5384 );
5385 assert!(
5386 off > fm.len(),
5387 "must not fall onto the hidden frontmatter floor"
5388 );
5389 }
5390
5391 #[test]
5392 fn offset_of_pos_is_a_stop_for_every_row_including_past_the_end() {
5393 // The invariant the caret motion asserts: whatever cell a click names,
5394 // the offset it resolves to is one the caret can actually rest at.
5395 for src in [
5396 "hello \n",
5397 "# A heading here \n\nbody text goes on \n",
5398 "---\nk: v\n---\n# Title\n\nprose here that wraps a bit \n",
5399 ] {
5400 let m = map(src);
5401 for row in 0..m.num_rows() + 3 {
5402 for col in 0..30 {
5403 let off = m.offset_of_pos(row, col);
5404 assert!(
5405 m.is_stop(off),
5406 "row {row} col {col} → {off} is not a stop in {src:?}"
5407 );
5408 }
5409 }
5410 }
5411 }
5412
5413 /// `| Name | Qty |` with Name left-aligned and Qty right-aligned.
5414 const TABLE: &str = "| Name | Qty |\n|:-----|----:|\n| Pear | 3 |\n| Fig | 12 |\n";
5415
5416 #[test]
5417 fn a_table_renders_as_an_aligned_grid() {
5418 let text = rendered(&map(TABLE));
5419 assert_eq!(
5420 text,
5421 "┌──────┬─────┐\n\
5422 │ Name │ Qty │\n\
5423 ├──────┼─────┤\n\
5424 │ Pear │ 3 │\n\
5425 │ Fig │ 12 │\n\
5426 └──────┴─────┘",
5427 "got:\n{text}"
5428 );
5429 }
5430
5431 #[test]
5432 fn table_columns_honour_their_alignment() {
5433 // Centre and default(left) come straight from twig's cell.alignment —
5434 // the delimiter row it's spelled in is consumed and has no node.
5435 let text = rendered(&map("| A | Bee |\n| --- | :---: |\n| x | y |\n"));
5436 assert!(text.contains("│ x │ y │"), "centred column: {text:?}");
5437 }
5438
5439 #[test]
5440 fn table_borders_are_decoration_the_caret_never_lands_on() {
5441 let m = map(TABLE);
5442 // The rules are whole decoration rows.
5443 for r in [0, 2, 5] {
5444 assert!(m.rows[r].decoration, "row {r} should be a decoration rule");
5445 assert!(
5446 !m.rows[r].glyphs.iter().any(|g| g.stop),
5447 "row {r} has a stop"
5448 );
5449 }
5450 // A content row's `│` and padding are decoration; only the cell text
5451 // and each cell's one end-stop are stops.
5452 let header = &m.rows[1];
5453 assert!(!header.decoration);
5454 for g in &header.glyphs {
5455 if g.ch == '│' {
5456 assert!(!g.stop, "a border is not a caret stop");
5457 }
5458 }
5459 let stops: String = header
5460 .glyphs
5461 .iter()
5462 .filter(|g| g.stop)
5463 .map(|g| g.ch)
5464 .collect();
5465 assert_eq!(stops, "Name Qty ", "cell text plus one end-stop space each");
5466 }
5467
5468 #[test]
5469 fn a_cell_maps_to_its_own_source_text() {
5470 let m = map(TABLE);
5471 // "Pear" starts at byte 32 in TABLE; the caret there draws on the 'P'.
5472 let pear = TABLE.find("Pear").unwrap();
5473 let (r, c) = m.pos_of_offset(pear);
5474 assert_eq!(m.rows[r].glyphs[c].ch, 'P');
5475 assert_eq!(m.offset_of_pos(r, c), pear, "round trips");
5476 }
5477
5478 #[test]
5479 fn a_wide_table_is_cut_to_fit_and_its_cells_wrap() {
5480 // Columns wider than the surface used to run off the right edge, where
5481 // nothing could reach them. They're cut to the budget instead, and the
5482 // text wraps down inside the column — the header rule stays put, and
5483 // an alignment holds on every line of a wrapped cell, not just the first.
5484 let src = "| Ingredient | Notes |\n|---|---:|\n\
5485 | flour milled coarse | sift it twice |\n| salt | a pinch |\n";
5486 let mut ed = Editor::new_str(src, Format::Markdown).unwrap();
5487 let m = build_t(&ed.nodes().unwrap(), src, Some(30));
5488 let text = rendered(&m);
5489 assert_eq!(
5490 text,
5491 "┌──────────────┬─────────────┐\n\
5492 │ Ingredient │ Notes │\n\
5493 ├──────────────┼─────────────┤\n\
5494 │ flour milled │ sift it │\n\
5495 │ coarse │ twice │\n\
5496 │ salt │ a pinch │\n\
5497 └──────────────┴─────────────┘",
5498 "got:\n{text}"
5499 );
5500 for (r, row) in m.rows.iter().enumerate() {
5501 assert!(
5502 row.glyphs.len() <= 30,
5503 "row {r} overflows: {}",
5504 row.glyphs.len()
5505 );
5506 }
5507 }
5508
5509 #[test]
5510 fn a_column_too_narrow_for_a_word_breaks_it_rather_than_spilling() {
5511 // A paragraph lets an overlong word trail off the end of the line; a
5512 // table column can't — a glyph past the border lands on the border.
5513 let src = "| A | B |\n|---|---|\n| antidisestablishmentarianism | x |\n";
5514 let mut ed = Editor::new_str(src, Format::Markdown).unwrap();
5515 let m = build_t(&ed.nodes().unwrap(), src, Some(20));
5516 for (r, row) in m.rows.iter().enumerate() {
5517 assert!(
5518 row.glyphs.len() <= 20,
5519 "row {r} overflows: {}",
5520 row.glyphs.len()
5521 );
5522 }
5523 // Broken across lines, but whole: every letter is still drawn, at its
5524 // own source byte, where the caret can reach it.
5525 let word = "antidisestablishmentarianism";
5526 let at = src.find(word).unwrap();
5527 for (i, ch) in word.char_indices() {
5528 assert!(
5529 m.rows
5530 .iter()
5531 .flat_map(|r| r.glyphs.iter())
5532 .any(|g| g.stop && g.src == at + i && g.ch == ch),
5533 "{ch:?} at {} was lost to the break",
5534 at + i
5535 );
5536 }
5537 }
5538
5539 #[test]
5540 fn a_code_block_maps_each_line_to_its_own_source_text() {
5541 // Every glyph used to point at the block's start, which made the whole
5542 // block one offset — visible, but impossible to put a caret inside.
5543 let src = "```rust\nlet x = 1;\nfn f() {}\n```\n";
5544 let m = map(src);
5545 for row in &m.rows {
5546 for g in row.glyphs.iter().filter(|g| g.stop) {
5547 assert_eq!(
5548 src[g.src..].chars().next(),
5549 Some(g.ch),
5550 "glyph {:?} at {} isn't the source byte it claims",
5551 g.ch,
5552 g.src
5553 );
5554 }
5555 }
5556 }
5557
5558 #[test]
5559 fn an_indented_code_block_maps_past_its_stripped_indent() {
5560 // twig strips the four-space indent, so `text` isn't a source slice and
5561 // the lines have to be re-found. Offsets land on the code, not the indent.
5562 let src = " indented\n code\n";
5563 let m = map(src);
5564 let stops: Vec<(char, usize)> = m
5565 .rows
5566 .iter()
5567 .flat_map(|r| r.glyphs.iter().filter(|g| g.stop).map(|g| (g.ch, g.src)))
5568 .collect();
5569 assert_eq!(
5570 stops[0],
5571 ('i', 4),
5572 "first line should start past the indent"
5573 );
5574 assert!(
5575 stops.contains(&('c', 17)),
5576 "second line misplaced: {stops:?}"
5577 );
5578 }
5579
5580 #[test]
5581 fn a_fenced_block_whose_code_echoes_its_info_string_maps_to_the_code() {
5582 // The one case that defeats a forward search: the opening fence
5583 // ```` ```rust ```` ends with the same text as the code under it.
5584 let src = "```rust\nrust\n```\n";
5585 let m = map(src);
5586 let first = m.rows[0].glyphs.iter().find(|g| g.stop).unwrap();
5587 assert_eq!(first.src, 8, "matched the info string, not the code");
5588 }
5589
5590 #[test]
5591 fn a_code_block_carries_no_gutter_and_is_published_as_a_row_span() {
5592 // The old `▏ ` gutter is gone: a code row is the block prefix (none, at
5593 // the top level) plus the code text, and the whole run is named in
5594 // `code_blocks` so a frontend can box it.
5595 let src = "para\n\n```\ncode\nlines\n```\n\nafter\n";
5596 let m = map(src);
5597 assert_eq!(m.code_blocks.len(), 1, "one code block");
5598 let span = m.code_blocks[0].rows_span.clone();
5599 let rows: Vec<String> = m.rows[span.clone()]
5600 .iter()
5601 .map(|r| r.glyphs.iter().map(|g| g.ch).collect())
5602 .collect();
5603 assert_eq!(rows, vec!["code".to_string(), "lines".to_string()]);
5604 assert!(!rendered(&m).contains('▏'), "gutter still drawn");
5605 assert!(
5606 m.rows[span].iter().all(|r| r.code),
5607 "every row in the span is flagged code"
5608 );
5609 }
5610
5611 #[test]
5612 fn a_directive_container_is_tinted_and_labeled_on_its_first_row() {
5613 // diaryx's `:::vis{.public .family}` visibility block, and any other
5614 // `:::name{.class}` fenced div — core is agnostic of `name`.
5615 let src = ":::vis{.public .family}\nhello\n\nworld\n:::\nafter\n";
5616 let m = map_directives(src);
5617
5618 let content_rows: Vec<usize> = (0..m.rows.len()).filter(|&i| m.rows[i].directive).collect();
5619 assert!(!content_rows.is_empty(), "some row is flagged directive");
5620
5621 let after_rows: Vec<usize> = (0..m.rows.len())
5622 .filter(|&i| !content_rows.contains(&i) && !m.rows[i].glyphs.is_empty())
5623 .collect();
5624 assert!(
5625 after_rows.iter().all(|&i| !m.rows[i].directive),
5626 "content outside the fence isn't tinted"
5627 );
5628
5629 let labels: Vec<&str> = content_rows
5630 .iter()
5631 .filter_map(|&i| m.rows[i].directive_label.as_deref())
5632 .collect();
5633 assert_eq!(
5634 labels,
5635 vec!["public family"],
5636 "only the first row carries the label"
5637 );
5638
5639 assert_eq!(
5640 rendered(&m)
5641 .lines()
5642 .filter(|l| !l.is_empty())
5643 .collect::<Vec<_>>(),
5644 vec!["hello", "world", "after"],
5645 "fence markers don't leak into the rendered text"
5646 );
5647 }
5648
5649 #[test]
5650 fn a_bare_word_directive_is_labeled_same_as_dot_classes() {
5651 // diaryx_core::visibility's own `:::vis{public family}` — no leading
5652 // dots — is what apps/web's directive serializer and the native
5653 // publish-time filter both actually write today, distinct from twig's
5654 // `.class` convention. Both must label the same way so every existing
5655 // diaryx `:::vis{...}` block reads, not just newly dot-authored ones.
5656 let src = ":::vis{public family}\nhello\n:::\n";
5657 let m = map_directives(src);
5658 let label = m.rows.iter().find_map(|r| r.directive_label.clone());
5659 assert_eq!(label.as_deref(), Some("public family"));
5660 }
5661
5662 #[test]
5663 fn a_text_directive_keeps_its_paragraph_visible() {
5664 // Regression: an inline `:name[label]{…}` used to make its paragraph
5665 // fail the "all children inline" test, so the whole line was walked as
5666 // a container of blocks and rendered as empty rows with NO caret stops —
5667 // the text vanished from the editor and the caret couldn't enter it.
5668 // diaryx's inline `:vis[…]` is exactly this shape.
5669 let src = "Text with :abbr[HTML]{title=\"HyperText\"} inline.\n";
5670 let m = map_directives(src);
5671 assert_eq!(rendered(&m).trim_end(), "Text with HTML inline.");
5672 // Every character of the line is a caret home, markup excluded — the
5673 // label reads as ordinary text, the way a link's does.
5674 let stops: usize = m
5675 .rows
5676 .iter()
5677 .map(|r| r.glyphs.iter().filter(|g| g.stop).count())
5678 .sum();
5679 assert_eq!(stops, "Text with HTML inline.".chars().count());
5680 // It is inline, so it is not the container form's tinted panel.
5681 assert!(m.rows.iter().all(|r| !r.directive));
5682 }
5683
5684 #[test]
5685 fn a_text_directives_label_maps_to_its_true_source_bytes() {
5686 // Regression (needs twig-doc >= 2.5.0): twig parses a `[label]` as a
5687 // detached slice, and until it rebased the enclosing scan's segments
5688 // onto it every node inside the label reported a span of `(0,0)`. Read
5689 // by anything that trusts a span that means "byte 0", so the label's
5690 // glyphs mapped to the START OF THE DOCUMENT — a click on the label put
5691 // the caret at the top of the file, its stops collided with the real
5692 // first line's, and an edit there landed on the wrong bytes entirely.
5693 //
5694 // The sibling test `a_text_directive_keeps_its_paragraph_visible` only
5695 // counts stops, which is exactly why this went unnoticed: the right
5696 // NUMBER of stops at completely wrong offsets.
5697 let src = "x :abbr[HTML]{title=\"y\"} z\n";
5698 let m = map_directives(src);
5699 let stops: Vec<(char, usize)> = m
5700 .rows
5701 .iter()
5702 .flat_map(|r| &r.glyphs)
5703 .filter(|g| g.stop)
5704 .map(|g| (g.ch, g.src))
5705 .collect();
5706 // `HTML` sits at 8..12. The name, brackets and `{…}` are hidden markup
5707 // the caret steps over, so the line's stops run 0, 1, 8..12, then 24.
5708 assert_eq!(
5709 stops,
5710 [
5711 ('x', 0),
5712 (' ', 1),
5713 ('H', 8),
5714 ('T', 9),
5715 ('M', 10),
5716 ('L', 11),
5717 (' ', 24),
5718 ('z', 25)
5719 ]
5720 );
5721 }
5722
5723 #[test]
5724 fn every_glyph_in_a_directive_label_points_at_its_source_byte() {
5725 // The `every_glyph_points_at_its_source_byte` invariant, extended over
5726 // directive labels now that their offsets are real. Nested markup is
5727 // included: its delimiters are hidden, so the visible glyphs must skip
5728 // them and still name their own bytes.
5729 let src = "x :abbr[a *b* c] y and :vis[family only] z\n";
5730 let m = map_directives(src);
5731 for g in m.rows.iter().flat_map(|r| &r.glyphs).filter(|g| g.stop) {
5732 let at = src[g.src..].chars().next();
5733 assert_eq!(
5734 at,
5735 Some(g.ch),
5736 "glyph {:?} claims byte {}, which is {at:?}",
5737 g.ch,
5738 g.src
5739 );
5740 }
5741 assert_eq!(rendered(&m).trim_end(), "x a b c y and family only z");
5742 }
5743
5744 #[test]
5745 fn a_directive_labels_nested_emphasis_keeps_both_its_style_and_its_offsets() {
5746 let src = "x :abbr[a *b* c] y\n";
5747 let m = map_directives(src);
5748 let b = m
5749 .rows
5750 .iter()
5751 .flat_map(|r| &r.glyphs)
5752 .find(|g| g.ch == 'b')
5753 .expect("the emphasised char");
5754 assert!(b.style.italic, "the label's *b* lost its emphasis");
5755 assert_eq!(b.src, 11, "the label's *b* lost its source byte");
5756 }
5757
5758 #[test]
5759 fn a_bare_colon_word_renders_as_the_prose_it_almost_always_is() {
5760 // Regression: twig matches a colon followed by any letter-led word, so
5761 // ordinary prose is full of "text directives" nobody meant to write.
5762 // With no `[label]` there are no children, and the arm recursed into
5763 // them — rendering *nothing*. The word vanished from the document with
5764 // no caret stop left behind, so it could not even be deleted.
5765 for src in ["a :word b\n", "note :see below\n", ":smile: hi\n"] {
5766 let m = map_directives(src);
5767 assert_eq!(
5768 rendered(&m).trim_end(),
5769 src.trim_end(),
5770 "prose was eaten: {src:?}"
5771 );
5772 }
5773 }
5774
5775 #[test]
5776 fn a_bare_colon_word_keeps_every_byte_a_caret_stop() {
5777 let src = "a :word b\n";
5778 let m = map_directives(src);
5779 // Nothing here is markup, so nothing is hidden: each byte maps to
5780 // itself and can be stood on, which is what makes the colon deletable.
5781 let stops: Vec<(char, usize)> = m
5782 .rows
5783 .iter()
5784 .flat_map(|r| &r.glyphs)
5785 .filter(|g| g.stop)
5786 .map(|g| (g.ch, g.src))
5787 .collect();
5788 assert_eq!(
5789 stops,
5790 "a :word b"
5791 .chars()
5792 .enumerate()
5793 .map(|(i, c)| (c, i))
5794 .collect::<Vec<_>>()
5795 );
5796 }
5797
5798 #[test]
5799 fn an_attribute_bearing_text_directive_draws_a_chip() {
5800 // `{…}` is deliberate in a way a bare colon is not — diaryx writes
5801 // `:vis{.family}` inline — so this one reads as an embed, on the same
5802 // `⧉ label` recipe the leaf form's placeholder row uses.
5803 // Both attribute conventions label it: twig's dot-prefixed classes and
5804 // the bare pandoc-style words diaryx also writes.
5805 for src in ["a :vis{.family} b\n", "a :vis{family} b\n"] {
5806 let m = map_directives(src);
5807 assert_eq!(rendered(&m).trim_end(), "a ⧉ vis family b", "{src:?}");
5808 }
5809 // A `key=value` attr is configuration, not a name, so it adds nothing.
5810 let m = map_directives("a :foo{title=\"x\"} b\n");
5811 assert_eq!(rendered(&m).trim_end(), "a ⧉ foo b");
5812 }
5813
5814 #[test]
5815 fn a_directive_chip_is_one_atomic_caret_stop_at_its_own_offset() {
5816 let src = "a :vis{.family} b\n";
5817 let m = map_directives(src);
5818 let stops: Vec<usize> = m
5819 .rows
5820 .iter()
5821 .flat_map(|r| &r.glyphs)
5822 .filter(|g| g.stop)
5823 .map(|g| g.src)
5824 .collect();
5825 // The chip contributes exactly one stop, at the directive's start (2),
5826 // so the caret steps over it whole instead of walking hidden markup a
5827 // byte at a time. `{.family}`'s bytes (3..15) are never stood on.
5828 assert_eq!(stops, [0, 1, 2, 15, 16]);
5829 }
5830
5831 #[test]
5832 fn a_paragraph_holding_only_a_chip_is_still_navigable() {
5833 // With no stop of its own the row would be unreachable — the caret
5834 // could never be put on the line to edit or delete the directive.
5835 let m = map_directives(":vis{.family}\n");
5836 assert!(
5837 m.row_is_navigable(0),
5838 "a chip-only paragraph has no caret home"
5839 );
5840 assert_eq!(
5841 m.offset_of_pos(0, 0),
5842 0,
5843 "its caret home isn't the directive's start"
5844 );
5845 }
5846
5847 #[test]
5848 fn a_ratio_or_a_clock_time_is_never_a_directive() {
5849 // twig requires a letter after the colon, so these stay prose — the
5850 // verbatim arm must not be reached for them at all.
5851 let src = "ratio 3:4 and 10:30\n";
5852 assert_eq!(
5853 rendered(&map_directives(src)).trim_end(),
5854 "ratio 3:4 and 10:30"
5855 );
5856 }
5857
5858 #[test]
5859 fn a_leaf_directive_is_a_placeholder_row_with_its_attrs_published() {
5860 // `::name{…}` is a standalone block with no body — an embed, a table of
5861 // contents. It used to emit no rows at all: invisible, no caret home,
5862 // vertical motion crossing a void. Now it draws the image recipe's
5863 // placeholder and publishes what the host app needs to paint the real
5864 // thing.
5865 let src = "before\n\n::embed{src=\"demo.html\" height=\"400\"}\n\nafter\n";
5866 let m = map_directives(src);
5867
5868 let row = m
5869 .rows
5870 .iter()
5871 .position(|r| r.leaf_directive.is_some())
5872 .expect("a placeholder row");
5873 assert_eq!(
5874 m.rows[row].glyphs.iter().map(|g| g.ch).collect::<String>(),
5875 "⧉ embed"
5876 );
5877 assert!(
5878 m.rows[row].glyphs.iter().any(|g| g.stop),
5879 "the caret can land on it"
5880 );
5881 assert!(
5882 m.rows[row].directive,
5883 "a frontend frames it like the container form"
5884 );
5885
5886 assert_eq!(m.directives.len(), 1);
5887 let info = &m.directives[0];
5888 assert_eq!(info.name, "embed");
5889 assert_eq!(info.rows_span, row..row + 1);
5890 assert_eq!(info.attr("src"), Some("demo.html"));
5891 assert_eq!(info.attr("height"), Some("400"));
5892 assert_eq!(info.attr("nope"), None);
5893 // The prose around it is untouched.
5894 assert!(rendered(&m).contains("before") && rendered(&m).contains("after"));
5895 }
5896
5897 #[test]
5898 fn a_leaf_directive_shows_its_label_and_honours_its_prefix() {
5899 // A `[label]` names the placeholder (the way an image's alt does), and a
5900 // quoted directive keeps the quote's gutter — it is a block like any
5901 // other, not a special case that escapes its container.
5902 let m = map_directives("::embed[Audience demo]{src=\"demo.html\"}\n");
5903 assert_eq!(rendered(&m).trim_end(), "⧉ Audience demo");
5904 assert_eq!(m.directives[0].label, "Audience demo");
5905
5906 let quoted = map_directives("> ::embed{src=\"x.html\"}\n");
5907 assert_eq!(rendered("ed).trim_end(), "│ ⧉ embed");
5908 assert_eq!(quoted.directives[0].name, "embed");
5909 }
5910
5911 #[test]
5912 fn a_container_directive_is_still_a_panel_not_a_placeholder() {
5913 // The three forms must not bleed into each other: only the leaf form is
5914 // a placeholder, and only the container form tints the blocks it wraps.
5915 let m = map_directives(":::note{.warning}\nBody\n:::\n");
5916 assert!(
5917 m.directives.is_empty(),
5918 "a container publishes no placeholder"
5919 );
5920 assert!(m.rows.iter().all(|r| r.leaf_directive.is_none()));
5921 assert_eq!(rendered(&m).trim_end(), "Body");
5922 assert!(
5923 m.rows
5924 .iter()
5925 .any(|r| r.directive && r.directive_label.as_deref() == Some("warning"))
5926 );
5927 }
5928
5929 /// A production-path build with both extensions on — the only way to put a
5930 /// promoted HTML element and a directive in one document, which is what the
5931 /// `container` kind made necessary to tell apart. Returns the whole `Doc`
5932 /// because [`VisualMap`] is not `Clone`; read `doc.vmap`.
5933 fn doc_built(src: &str) -> crate::Doc {
5934 let mut doc = crate::Doc::from_source(src.to_string(), Format::Markdown).unwrap();
5935 doc.build_visual(80);
5936 doc
5937 }
5938
5939 /// Every `container` node in `src`, parsed the way production does (both
5940 /// extensions on), paired with what [`container_is_directive`] makes of it.
5941 fn containers(src: &str) -> Vec<(String, bool, Option<DirectiveForm>)> {
5942 let mut ed = Editor::new_ext(
5943 src.as_bytes(),
5944 Format::Markdown,
5945 twig::MarkdownExtensions {
5946 directives: true,
5947 html_elements: true,
5948 ..Default::default()
5949 },
5950 )
5951 .unwrap();
5952 ed.nodes()
5953 .unwrap()
5954 .iter()
5955 .filter(|n| n.kind == Kind::Container)
5956 .map(|n| {
5957 (
5958 n.name.clone().unwrap_or_default(),
5959 container_is_directive(n),
5960 n.directive_form,
5961 )
5962 })
5963 .collect()
5964 }
5965
5966 #[test]
5967 fn a_directive_and_an_html_element_are_told_apart_by_spelling_not_by_form() {
5968 // twig 2.8 folded `div`/`span`/`directive`/`element` into one `container`
5969 // kind. `directive_form` reads as though it separates them and does not:
5970 // a block-level `<div>` reports `Some(DirectiveForm::Container)` exactly
5971 // as a `:::note` does. Trusting it would draw directive chrome — a tinted
5972 // panel, a `.class` audience label — on every pasted Slack/Docs div.
5973 for (src, name, want) in [
5974 (":::note{.a}\nbody\n:::\n", "note", true),
5975 ("::embed{src=x}\n", "embed", true),
5976 ("a :vis[hi]{.b} b\n", "vis", true),
5977 ("<div class=\"x\">\nhi\n</div>\n", "div", false),
5978 ("<video src=\"v.mp4\" controls></video>\n", "video", false),
5979 ("<audio src=\"a.mp3\" controls></audio>\n", "audio", false),
5980 ("<figure>\n\nhi\n\n</figure>\n", "figure", false),
5981 // The `:` in an attribute must not read as a directive opener: the
5982 // `<` of the tag comes first, and first one wins.
5983 (
5984 "<video src=\"http://x.test/v.mp4\" controls></video>\n",
5985 "video",
5986 false,
5987 ),
5988 (
5989 "<source media=\"(prefers-color-scheme: dark)\" srcset=\"d.svg\">\n",
5990 "source",
5991 false,
5992 ),
5993 ] {
5994 let found = containers(src);
5995 let hit = found.iter().find(|(n, ..)| n == name);
5996 let Some((_, is_directive, form)) = hit else {
5997 panic!("no `{name}` container in {src:?} — found {found:?}");
5998 };
5999 assert_eq!(*is_directive, want, "{name} in {src:?} (form was {form:?})");
6000 }
6001
6002 // And the reason this can't just read the field: for the one collision
6003 // that matters, the field says the same thing for both.
6004 let div = containers("<div class=\"x\">\nhi\n</div>\n");
6005 let note = containers(":::note{.a}\nbody\n:::\n");
6006 assert_eq!(
6007 div[0].2, note[0].2,
6008 "if these ever differ, `directive_form` became usable and this rule can go"
6009 );
6010 }
6011
6012 #[test]
6013 fn a_directive_nested_in_a_quote_or_list_is_still_a_directive() {
6014 // A container's span opens with its *block prefix*, not its own markup —
6015 // `> ::embed{…}` starts at the `>`. Reading only the first byte to tell a
6016 // directive from an element (both `container` since 2.8) therefore misses
6017 // every nested one, and the placeholder silently renders as nothing.
6018 for (src, ctx) in [
6019 ("> ::embed{src=\"x\"}\n", "quoted"),
6020 ("- ::embed{src=\"x\"}\n", "listed"),
6021 (">> ::embed{src=\"x\"}\n", "twice quoted"),
6022 ] {
6023 let m = map_directives(src);
6024 assert_eq!(m.directives.len(), 1, "{ctx} directive was lost");
6025 assert_eq!(m.directives[0].name, "embed", "{ctx}");
6026 }
6027 }
6028
6029 #[test]
6030 fn a_video_is_still_media_and_not_a_directive() {
6031 // The other side of the same coin: `<video>` is a `container` too, and
6032 // must reach `block_media` rather than the directive arms.
6033 let doc = doc_built("<video src=\"clip.mp4\" controls></video>\n");
6034 assert_eq!(doc.vmap.media.len(), 1, "the video is block media");
6035 assert!(
6036 doc.vmap.rows.iter().all(|r| !r.directive),
6037 "the video drew directive chrome"
6038 );
6039 }
6040
6041 #[test]
6042 fn a_directive_needs_the_extension_flag() {
6043 // `map` (twig's default extensions) leaves `directives` off — the fence
6044 // renders as literal paragraph text, same as any other unrecognized
6045 // punctuation, never corrupting or panicking.
6046 let src = ":::vis{.public}\nhello\n:::\n";
6047 let m = map(src);
6048 assert!(m.rows.iter().all(|r| !r.directive));
6049 assert!(rendered(&m).contains(":::vis{.public}"));
6050 }
6051
6052 #[test]
6053 fn a_footnote_reference_keeps_its_paragraph_visible() {
6054 // Regression: `footnote_reference` was in neither `is_inline_kind` nor
6055 // the inline walker, so a paragraph carrying one failed the "all children
6056 // inline" test, was walked as a container of blocks, and rendered as
6057 // empty rows with no caret stop anywhere — the whole line vanished.
6058 let src = "A claim[^1] and more.\n";
6059 let m = map(src);
6060 assert_eq!(rendered(&m).trim_end(), "A claim[1] and more.");
6061 // The `^` is spelling, not text: hidden the way a link's `](dest)` is.
6062 assert!(!rendered(&m).contains('^'));
6063 }
6064
6065 #[test]
6066 fn a_footnote_reference_is_raised_and_the_prose_around_it_is_not() {
6067 // What makes `[1]` read as a reference rather than as bracketed text.
6068 // The brackets ride with the label: the chip is one raised mark.
6069 let m = map("A claim[^1] and more.\n");
6070 assert_eq!(baselines_of(&m, '1'), vec![Baseline::Super]);
6071 assert_eq!(baselines_of(&m, '['), vec![Baseline::Super]);
6072 assert_eq!(baselines_of(&m, ']'), vec![Baseline::Super]);
6073 assert_eq!(baselines_of(&m, 'A'), vec![Baseline::Normal]);
6074 }
6075
6076 #[test]
6077 fn a_footnote_reference_keeps_the_link_role_it_had() {
6078 // The raised baseline is added to the role, not swapped for it: every
6079 // frontend already paints `Role::Link`, and a reference is one.
6080 let m = map("A claim[^1].\n");
6081 let label = m
6082 .rows
6083 .iter()
6084 .flat_map(|r| &r.glyphs)
6085 .find(|g| g.ch == '1')
6086 .unwrap();
6087 assert_eq!(label.style.role, Role::Link);
6088 assert_eq!(label.style.baseline, Baseline::Super);
6089 }
6090
6091 #[test]
6092 fn a_superscript_and_a_subscript_sit_off_the_baseline() {
6093 // Regression: both rendered flat, so the toolbar's superscript button
6094 // produced markup that looked exactly like the text around it.
6095 let m = map_djot("H~2~O and x^2^\n");
6096 assert_eq!(baselines_of(&m, '2'), vec![Baseline::Sub, Baseline::Super]);
6097 assert_eq!(baselines_of(&m, 'H'), vec![Baseline::Normal]);
6098 assert_eq!(baselines_of(&m, 'O'), vec![Baseline::Normal]);
6099 }
6100
6101 #[test]
6102 fn a_raised_glyph_keeps_the_style_it_was_raised_out_of() {
6103 // Why this is a `Baseline` and not a `Role`: raising a glyph says where
6104 // it sits, and must not cost it what it already was.
6105 let m = map_djot("# Heading x^2^\n");
6106 let two = m
6107 .rows
6108 .iter()
6109 .flat_map(|r| &r.glyphs)
6110 .find(|g| g.ch == '2')
6111 .unwrap();
6112 assert_eq!(two.style.baseline, Baseline::Super);
6113 assert_eq!(two.style.role, Role::Heading(1), "still heading text");
6114 }
6115
6116 #[test]
6117 fn a_footnote_references_brackets_are_decoration_and_only_its_label_is_a_stop() {
6118 let src = "see[^note] here\n";
6119 let m = map(src);
6120 // `[^note]` spans 3..10, its label `note` 5..9. The caret walks the
6121 // label; the brackets are drawn but never stood on, as a table's are,
6122 // and the `[^`/`]` bytes are stepped over like any hidden delimiter.
6123 let stops: Vec<usize> = m
6124 .rows
6125 .iter()
6126 .flat_map(|r| &r.glyphs)
6127 .filter(|g| g.stop)
6128 .map(|g| g.src)
6129 .collect();
6130 for off in 5..9 {
6131 assert!(
6132 stops.contains(&off),
6133 "label byte {off} isn't a caret stop: {stops:?}"
6134 );
6135 }
6136 for off in [3usize, 4, 9] {
6137 assert!(
6138 !stops.contains(&off),
6139 "delimiter byte {off} is a caret stop: {stops:?}"
6140 );
6141 }
6142 }
6143
6144 #[test]
6145 fn a_task_item_draws_its_box_where_the_bullet_would_be() {
6146 // Regression: the `[ ] ` is markup twig consumes — the item's paragraph
6147 // content starts past it — so a task item used to render as `• todo`,
6148 // identical to a plain bullet and with no way to see it was ticked.
6149 let m = map("- [ ] todo\n- [x] done\n- plain\n");
6150 assert_eq!(rendered(&m), "☐ todo\n☑ done\n• plain");
6151
6152 // The tick rides the item's first row, for a GUI that paints its own box.
6153 let ticks: Vec<Option<bool>> = m.rows.iter().map(|r| r.task).collect();
6154 assert_eq!(ticks, [Some(false), Some(true), None]);
6155 }
6156
6157 #[test]
6158 fn a_task_items_box_survives_a_wrap_and_marks_only_the_first_row() {
6159 let m = map_at(
6160 "- [x] a much longer task that has to wrap somewhere\n",
6161 Some(20),
6162 );
6163 assert!(m.rows.len() > 1, "the item should wrap: {:?}", rendered(&m));
6164 assert_eq!(m.rows[0].task, Some(true));
6165 assert!(
6166 m.rows[1..].iter().all(|r| r.task.is_none()),
6167 "only the first row"
6168 );
6169 // The continuation lines hang under the box, not under column zero.
6170 assert!(
6171 rendered(&m)
6172 .lines()
6173 .nth(1)
6174 .is_some_and(|l| l.starts_with(" "))
6175 );
6176 }
6177
6178 #[test]
6179 fn a_bracket_in_an_items_prose_is_not_a_checkbox() {
6180 // `task_checked` finds the box past the list marker; a plain item whose
6181 // text merely contains a bracket has none, and must keep its bullet.
6182 let m = map("- see [1] below\n");
6183 assert_eq!(rendered(&m), "• see [1] below");
6184 assert_eq!(m.rows[0].task, None);
6185 }
6186
6187 #[test]
6188 fn a_footnote_definition_renders_where_it_was_written() {
6189 // Regression: twig parses `[^1]: …` as a root *beside* `doc` — not a
6190 // child of it — so the walk from `doc` never reached one and every byte
6191 // of the note's body rendered as nothing at all.
6192 let src = "A claim[^1].\n\n[^1]: The note body.\n\nAfter.\n";
6193 let m = map(src);
6194 let text = rendered(&m);
6195 assert!(
6196 text.contains("The note body."),
6197 "the note body is invisible: {text:?}"
6198 );
6199 // In source order — between the paragraph that cites it and the one
6200 // after — not hoisted to the end, and marked to match its reference.
6201 let lines: Vec<&str> = text.lines().filter(|l| !l.trim().is_empty()).collect();
6202 assert_eq!(lines, ["A claim[1].", "[1] The note body.", "After."]);
6203 }
6204
6205 #[test]
6206 fn a_footnote_definitions_body_maps_to_its_own_source_bytes() {
6207 let src = "x[^a].\n\n[^a]: body\n";
6208 let m = map(src);
6209 // `body` sits at 14..18. Its glyphs must map there — a marker that ate
6210 // the offsets would put the caret in the wrong place on every click.
6211 let body: Vec<(char, usize)> = m
6212 .rows
6213 .iter()
6214 .flat_map(|r| &r.glyphs)
6215 .filter(|g| g.stop && g.src >= 14)
6216 .map(|g| (g.ch, g.src))
6217 .collect();
6218 assert_eq!(body, [('b', 14), ('o', 15), ('d', 16), ('y', 17)]);
6219 }
6220
6221 #[test]
6222 fn an_empty_footnote_definition_still_shows_its_marker() {
6223 // The instant `[^1]: ` has been typed and nothing after it. `blocks`
6224 // renders no child, so without the explicit marker row the definition
6225 // wouldn't appear at all until something was typed into it.
6226 let src = "x[^1]\n\n[^1]:\n";
6227 let m = map(src);
6228 assert!(
6229 rendered(&m).contains("[1] "),
6230 "no marker row: {:?}",
6231 rendered(&m)
6232 );
6233 }
6234
6235 #[test]
6236 fn a_footnote_definition_wearing_a_long_label_indents_its_wrapped_body() {
6237 let src = "x[^src]\n\n[^src]: one two three four five six seven\n";
6238 let m = map_at(src, Some(24));
6239 let text = rendered(&m);
6240 let lines: Vec<&str> = text.lines().filter(|l| !l.trim().is_empty()).collect();
6241 // Continuation lines hang under the marker, as a list item's do — the
6242 // indent is the marker's own width, not a fixed one.
6243 assert_eq!(lines[1].trim_end(), "[src] one two three four");
6244 assert!(
6245 lines[2].starts_with(" "),
6246 "body doesn't hang: {:?}",
6247 lines[2]
6248 );
6249 assert_eq!(lines[2].trim(), "five six seven");
6250 }
6251
6252 #[test]
6253 fn a_code_block_leaves_exactly_one_blank_row_below_it() {
6254 // The closing fence line used to be miscounted as a blank separator,
6255 // opening a phantom second gap under the block. One block boundary is
6256 // one blank row, code block or not.
6257 let src = "para\n\n```\ncode\n```\n\nafter\n";
6258 let m = map(src);
6259 let code_end = m.code_blocks[0].rows_span.end;
6260 let after = m
6261 .rows
6262 .iter()
6263 .position(|r| r.glyphs.iter().map(|g| g.ch).collect::<String>() == "after")
6264 .unwrap();
6265 assert_eq!(
6266 after - code_end,
6267 1,
6268 "exactly one row between code and 'after'"
6269 );
6270 }
6271
6272 #[test]
6273 fn a_fenced_block_publishes_its_language_on_its_code_block() {
6274 // The info string becomes the block's label; a bare fence and an indented
6275 // block carry none.
6276 assert_eq!(
6277 map("```rust\nlet x = 1;\n```\n").code_blocks[0]
6278 .lang
6279 .as_deref(),
6280 Some("rust")
6281 );
6282 assert_eq!(map("```\nplain\n```\n").code_blocks[0].lang, None);
6283 assert_eq!(map(" indented\n").code_blocks[0].lang, None);
6284 }
6285
6286 #[test]
6287 fn inline_code_is_not_a_code_block() {
6288 // A `code` span inside prose is styled by role, not boxed: it's part of a
6289 // normal paragraph row, so it names no `code_blocks` entry.
6290 let m = map("a `snippet` b\n");
6291 assert!(m.code_blocks.is_empty(), "inline code wrongly boxed");
6292 assert!(
6293 m.rows.iter().all(|r| !r.code),
6294 "inline code flagged a code row"
6295 );
6296 }
6297
6298 #[test]
6299 fn caret_steps_over_hidden_delimiters() {
6300 // "a **bold** c": bytes 8,9 are the closing ** — no glyph. Moving right
6301 // from 'd' (src 7) lands on the space before 'c' (src 10), not inside **.
6302 let m = map("a **bold** c\n");
6303 let (r, c) = m.pos_of_offset(7);
6304 assert_eq!(m.offset_of_pos(r, c + 1), 10);
6305 }
6306
6307 // ── the structural view of a table ───────────────────────────────────────
6308
6309 #[test]
6310 fn a_table_is_published_structurally_beside_its_picture() {
6311 let m = map(TABLE);
6312 let t = &m.tables[0];
6313 let cell = |r: usize, c: usize| -> String {
6314 t.grid[r].cells[c].glyphs.iter().map(|g| g.ch).collect()
6315 };
6316 assert_eq!(t.grid.len(), 3, "head + two body rows");
6317 assert_eq!(
6318 (cell(0, 0), cell(0, 1), cell(1, 0), cell(2, 1)),
6319 ("Name".into(), "Qty".into(), "Pear".into(), "12".into())
6320 );
6321 assert_eq!(
6322 t.grid.iter().map(|r| r.head).collect::<Vec<_>>(),
6323 [true, false, false]
6324 );
6325 // The alignment the delimiter row spelled, carried per cell — the only
6326 // place it survives, since the parser consumes that row.
6327 assert!(matches!(t.grid[1].cells[0].align, Alignment::Left));
6328 assert!(matches!(t.grid[1].cells[1].align, Alignment::Right));
6329 }
6330
6331 #[test]
6332 fn a_block_media_is_published_structurally_beside_its_placeholder() {
6333 let m = map("intro\n\n\n\nend\n");
6334 assert_eq!(m.media.len(), 1, "one block image");
6335 let img = &m.media[0];
6336 assert_eq!(img.destination, "img/cat.png");
6337 assert_eq!(img.alt, "a cat");
6338 // The placeholder row named by `rows_span` carries the label a plain
6339 // surface paints and a capable frontend replaces.
6340 let row_text = |r: usize| -> String { m.rows[r].glyphs.iter().map(|g| g.ch).collect() };
6341 assert_eq!(
6342 img.rows_span.end - img.rows_span.start,
6343 1,
6344 "one placeholder row"
6345 );
6346 assert_eq!(row_text(img.rows_span.start), "🖼 a cat");
6347 // The row carries the mark `media_spans` derives the side-table from.
6348 assert!(m.rows[img.rows_span.start].media.is_some());
6349 }
6350
6351 #[test]
6352 fn an_image_without_alt_labels_itself_with_its_filename() {
6353 let m = map("\n");
6354 let row = &m.rows[m.media[0].rows_span.start];
6355 assert_eq!(
6356 row.glyphs.iter().map(|g| g.ch).collect::<String>(),
6357 "🖼 beach.jpg"
6358 );
6359 assert_eq!(m.media[0].alt, "");
6360 }
6361
6362 #[test]
6363 fn a_block_media_gives_the_caret_a_home_before_and_after_it() {
6364 // `` on its own line: the caret can rest in front of the image
6365 // (its start) and just past it (the row end), and nowhere inside the
6366 // markup — the same coarse mapping a thematic break uses.
6367 let src = "\n";
6368 let m = map(src);
6369 let img = &m.rows[m.media[0].rows_span.start];
6370 let start = 0; // the image opens the document
6371 let end = "".len();
6372 // Every placeholder glyph maps to the image start and is a stop there.
6373 assert!(img.glyphs.iter().all(|g| g.src == start && g.stop));
6374 assert_eq!(img.end_src, end, "the row ends past the image");
6375 assert_eq!(m.stops.first(), Some(&start));
6376 assert!(m.stops.contains(&end), "a stop sits after the image");
6377 // Nothing inside the markup is a stop.
6378 assert!(!m.stops.iter().any(|&s| s > start && s < end));
6379 }
6380
6381 #[test]
6382 fn an_inline_image_amid_text_is_not_a_block_media() {
6383 // An image sharing its line with prose isn't block-level: it stays in the
6384 // inline path (rendered as its alt text), and publishes no MediaInfo.
6385 let m = map("see  here\n");
6386 assert!(m.media.is_empty(), "not a block image");
6387 assert!(
6388 rendered(&m).contains("a cat"),
6389 "alt text still renders inline"
6390 );
6391 }
6392
6393 /// The block images `Doc` publishes for `src`, driven through the real
6394 /// production build (`build_visual` → `build_cached`) with `html_elements`
6395 /// on — the path a `<picture>` actually travels. Not the raw `build` the
6396 /// other tests use: the editor's flat whole-arena snapshot tangles the links
6397 /// of inline-promoted HTML (phantom roots, dangling `parent`s), which only
6398 /// the per-block subtree walk `build_cached` does untangles.
6399 fn doc_media(src: &str) -> Vec<MediaInfo> {
6400 let mut doc = crate::Doc::from_source(src.to_string(), Format::Markdown).unwrap();
6401 doc.build_visual(80);
6402 doc.vmap.media.clone()
6403 }
6404
6405 #[test]
6406 fn a_video_block_is_media_with_its_src_poster_and_kind() {
6407 // The load-bearing assumption of video support: twig has no `video` node
6408 // kind, so `html_elements` promotion must land a `<video>` as a generic
6409 // `element` whose tag name and attributes survive onto `FlatNode` — the
6410 // same treatment `<picture>` gets. If that ever stops holding, this is
6411 // the test that says so.
6412 let m = doc_media("<video src=\"clip.mp4\" poster=\"still.png\" controls>\n</video>\n");
6413 assert_eq!(m.len(), 1, "the video is one block media");
6414 assert_eq!(m[0].kind, MediaKind::Video);
6415 assert_eq!(m[0].destination, "clip.mp4");
6416 assert_eq!(m[0].poster, "still.png");
6417 }
6418
6419 #[test]
6420 fn a_single_line_video_is_a_block_too() {
6421 // The spelling everyone actually writes. It used to parse as a paragraph
6422 // of raw inline HTML — CommonMark opens a block on a complete tag only
6423 // when the line ends there, and its fixed tag list predates `<video>` —
6424 // so the tags never reached core as an element at all. twig 2.5.1 widened
6425 // that list under `html_elements`; this is the test that would catch the
6426 // pin sliding back.
6427 let m = doc_media("<video src=\"clip.mp4\" controls></video>\n");
6428 assert_eq!(m.len(), 1, "single-line <video> is a block");
6429 assert_eq!(m[0].kind, MediaKind::Video);
6430 assert_eq!(m[0].destination, "clip.mp4");
6431 }
6432
6433 #[test]
6434 fn a_single_line_picture_is_a_block_with_its_alternatives() {
6435 // `<picture>` had the identical gap and it went unnoticed because the
6436 // conventional spelling breaks the lines. Same twig fix covers it.
6437 let src = "<picture><source media=\"(prefers-color-scheme: dark)\" srcset=\"d.svg\">\
6438 <img src=\"l.svg\" alt=\"banner\"></picture>\n";
6439 let m = doc_media(src);
6440 assert_eq!(m.len(), 1);
6441 assert_eq!(m[0].kind, MediaKind::Image);
6442 assert_eq!(m[0].destination, "l.svg");
6443 assert_eq!(m[0].resolve(ColorScheme::Dark), "d.svg");
6444 }
6445
6446 #[test]
6447 fn an_audio_block_is_media_with_no_poster() {
6448 let m = doc_media("<audio src=\"take.mp3\" controls>\n</audio>\n");
6449 assert_eq!(m.len(), 1);
6450 assert_eq!(m[0].kind, MediaKind::Audio);
6451 assert_eq!(m[0].destination, "take.mp3");
6452 assert!(m[0].poster.is_empty(), "audio has no poster frame");
6453 }
6454
6455 #[test]
6456 fn a_videos_source_children_are_its_candidates_typed_by_mime() {
6457 // A `<video>` with no `src` of its own — the common shape, since it's how
6458 // you offer more than one codec. The candidates come from `<source src>`
6459 // (not `srcset`, which is `<picture>`'s spelling) and carry their MIME.
6460 let src = "<video controls>\n\
6461 <source src=\"a.webm\" type=\"video/webm\">\n\
6462 <source src=\"a.mp4\" type=\"video/mp4\">\n\
6463 fallback\n\
6464 </video>\n";
6465 let m = doc_media(src);
6466 assert_eq!(m.len(), 1);
6467 assert!(
6468 m[0].destination.is_empty(),
6469 "no src attribute on the element"
6470 );
6471 assert_eq!(m[0].sources.len(), 2);
6472 assert_eq!(m[0].sources[0].srcset, "a.webm");
6473 assert_eq!(m[0].sources[0].mime, "video/webm");
6474 assert_eq!(m[0].sources[1].srcset, "a.mp4");
6475 // With an empty destination, `resolve` falls through to the first
6476 // candidate rather than handing the frontend nothing to load.
6477 assert_eq!(m[0].resolve(ColorScheme::Light), "a.webm");
6478 }
6479
6480 #[test]
6481 fn a_video_placeholder_row_carries_its_own_sigil_and_mark() {
6482 // The placeholder contract images already hold, now for a video: the row
6483 // renders as a labelled stand-in a plain surface can paint as-is, and
6484 // carries the mark a capable frontend replaces it from.
6485 let src = "<video src=\"clip.mp4\" controls>\n</video>\n";
6486 let mut doc = crate::Doc::from_source(src.to_string(), Format::Markdown).unwrap();
6487 doc.build_visual(80);
6488 let row = &doc.vmap.rows[doc.vmap.media[0].rows_span.start];
6489 let text: String = row.glyphs.iter().map(|g| g.ch).collect();
6490 assert!(
6491 text.starts_with('🎬'),
6492 "video sigil, not the image one: {text:?}"
6493 );
6494 assert!(row.media.is_some(), "the mark rides the placeholder row");
6495 }
6496
6497 #[test]
6498 fn a_picture_block_carries_its_source_alternatives() {
6499 // A `<picture>` with a dark-mode `<source>`: one block image, whose
6500 // fallback destination is the `<img>` and whose `sources` carry the
6501 // `<source>`'s media + srcset for a theme-aware frontend to pick.
6502 let src = "<picture><source media=\"(prefers-color-scheme: dark)\" srcset=\"dark.svg\"><img src=\"light.svg\" alt=\"banner\"></picture>\n";
6503 let images = doc_media(src);
6504 assert_eq!(images.len(), 1, "the picture is one block image");
6505 let img = &images[0];
6506 assert_eq!(img.destination, "light.svg", "fallback is the <img>");
6507 assert_eq!(img.alt, "banner");
6508 assert_eq!(
6509 img.sources,
6510 vec![MediaSource {
6511 media: "(prefers-color-scheme: dark)".into(),
6512 srcset: "dark.svg".into(),
6513 mime: String::new(),
6514 }],
6515 );
6516 }
6517
6518 #[test]
6519 fn a_picture_inside_a_heading_is_still_a_block_media_with_sources() {
6520 // fig.md's shape: the banner is an `<h1>` wrapping the `<picture>`.
6521 let src = "<h1><picture><source media=\"(prefers-color-scheme: dark)\" srcset=\"d.svg\"><img src=\"l.svg\" alt=\"fig\"></picture></h1>\n";
6522 let images = doc_media(src);
6523 assert_eq!(images.len(), 1, "heading-wrapped picture is a block image");
6524 assert_eq!(images[0].destination, "l.svg");
6525 assert_eq!(images[0].sources.len(), 1);
6526 assert_eq!(images[0].sources[0].srcset, "d.svg");
6527 }
6528
6529 #[test]
6530 fn a_plain_image_has_no_media_sources() {
6531 // A bare Markdown image carries an empty `sources` — nothing to pick from.
6532 let images = doc_media("\n");
6533 assert_eq!(images.len(), 1);
6534 assert!(
6535 images[0].sources.is_empty(),
6536 "no <picture>, no alternatives"
6537 );
6538 }
6539
6540 #[test]
6541 fn resolve_picks_the_source_matching_the_scheme() {
6542 let src = "<picture><source media=\"(prefers-color-scheme: dark)\" srcset=\"dark.svg\"><img src=\"light.svg\" alt=\"b\"></picture>\n";
6543 let images = doc_media(src);
6544 let img = &images[0];
6545 // Dark theme takes the dark source; light falls through to the <img>.
6546 assert_eq!(img.resolve(ColorScheme::Dark), "dark.svg");
6547 assert_eq!(img.resolve(ColorScheme::Light), "light.svg");
6548 }
6549
6550 #[test]
6551 fn resolve_falls_back_for_a_plain_image_and_unknown_media() {
6552 // A plain image ignores the scheme.
6553 let plain = doc_media("\n");
6554 assert_eq!(plain[0].resolve(ColorScheme::Dark), "p.png");
6555
6556 // A <source> with an unrecognized media query is skipped; a light source
6557 // is taken under a light theme.
6558 let m = doc_media(
6559 "<picture><source media=\"print\" srcset=\"p.svg\"><source media=\"(prefers-color-scheme: light)\" srcset=\"l.svg\"><img src=\"f.svg\" alt=\"x\"></picture>\n",
6560 );
6561 assert_eq!(m[0].resolve(ColorScheme::Light), "l.svg");
6562 assert_eq!(
6563 m[0].resolve(ColorScheme::Dark),
6564 "f.svg",
6565 "no dark source → <img>"
6566 );
6567 }
6568
6569 #[test]
6570 fn resolve_reads_the_first_srcset_url_ignoring_descriptors() {
6571 // A comma/descriptor srcset resolves to its first URL.
6572 assert_eq!(first_srcset_url("a.png 1x, b.png 2x"), Some("a.png"));
6573 assert_eq!(first_srcset_url(" solo.svg "), Some("solo.svg"));
6574 assert_eq!(first_srcset_url(""), None);
6575 // An empty (unconditional) media always matches.
6576 assert!(media_matches("", ColorScheme::Light));
6577 assert!(media_matches(
6578 "(prefers-color-scheme:dark)",
6579 ColorScheme::Dark
6580 ));
6581 assert!(!media_matches(
6582 "(prefers-color-scheme: dark)",
6583 ColorScheme::Light
6584 ));
6585 }
6586
6587 #[test]
6588 fn a_block_media_carries_its_list_prefix() {
6589 // An image that is a list item's body opens past the bullet, like every
6590 // other block does.
6591 let m = map("- \n");
6592 let row = &m.rows[m.media[0].rows_span.start];
6593 let text: String = row.glyphs.iter().map(|g| g.ch).collect();
6594 assert!(
6595 text.starts_with("• "),
6596 "the list marker prefixes the image row: {text:?}"
6597 );
6598 assert!(text.contains("🖼 alt"));
6599 }
6600
6601 #[test]
6602 fn the_structural_table_spans_exactly_its_drawn_rows() {
6603 // A frontend drawing its own grid skips `rows_span` and renders from
6604 // `grid`. If the span were short the leftover border rows would be
6605 // painted as text under the real table; if long it would eat a
6606 // neighbouring paragraph. Both are silent, so pin it to the picture.
6607 let m = map(&format!("before\n\n{TABLE}\nafter\n"));
6608 let t = &m.tables[0];
6609 let row_text = |r: usize| -> String { m.rows[r].glyphs.iter().map(|g| g.ch).collect() };
6610 assert!(
6611 row_text(t.rows_span.start).starts_with('┌'),
6612 "opens on the top border"
6613 );
6614 assert!(
6615 row_text(t.rows_span.end - 1).starts_with('└'),
6616 "closes on the bottom border"
6617 );
6618 assert!(
6619 !row_text(t.rows_span.start - 1).contains('┌'),
6620 "the row before the span is not the table's"
6621 );
6622 assert_eq!(
6623 row_text(t.rows_span.end),
6624 "",
6625 "the span ends before the gap row"
6626 );
6627 }
6628
6629 #[test]
6630 fn a_nested_tables_structure_carries_the_block_prefix() {
6631 // The picture puts the quote's gutter on every row of the grid. A
6632 // frontend drawing its own table has to draw that too and start past it,
6633 // so the prefix has to travel with the structure — without it a quoted
6634 // table renders flush at the margin and leaves the quote it's in.
6635 let m = map("> | a | b |\n> |---|---|\n> | c | d |\n");
6636 let t = &m.tables[0];
6637 let prefix: String = t.prefix.iter().map(|g| g.ch).collect();
6638 assert_eq!(prefix, "│ ", "the quote's gutter should ride the structure");
6639 // And it matches what the picture actually drew.
6640 let drawn: String = m.rows[t.rows_span.start]
6641 .glyphs
6642 .iter()
6643 .map(|g| g.ch)
6644 .collect();
6645 assert!(
6646 drawn.starts_with(&prefix),
6647 "picture and structure disagree: {drawn:?}"
6648 );
6649 }
6650
6651 #[test]
6652 fn a_top_level_table_carries_no_prefix() {
6653 assert!(map(TABLE).tables[0].prefix.is_empty());
6654 }
6655
6656 #[test]
6657 fn structural_cells_are_unwrapped_even_when_the_picture_wraps_them() {
6658 // The picture wraps a cell to its column; a frontend laying the grid out
6659 // in pixels needs the text as the document spells it, before that
6660 // decision. Narrow enough that the drawn cell must break.
6661 let src = "| Name |\n|------|\n| alpha beta gamma |\n";
6662 let mut ed = Editor::new_str(src, Format::Markdown).unwrap();
6663 let m = build_t(&ed.nodes().unwrap(), src, Some(12));
6664 let drawn = rendered(&m);
6665 let cell: String = m.tables[0].grid[1].cells[0]
6666 .glyphs
6667 .iter()
6668 .map(|g| g.ch)
6669 .collect();
6670 assert_eq!(
6671 cell, "alpha beta gamma",
6672 "structure must not carry the wrap"
6673 );
6674 assert!(
6675 drawn.lines().count() > 5,
6676 "the picture should have wrapped, else this proves nothing:\n{drawn}"
6677 );
6678 }
6679
6680 // ── display columns ──────────────────────────────────────────────────────
6681
6682 #[test]
6683 fn a_table_column_is_as_wide_as_its_cells_are_drawn() {
6684 // A column sized by counting characters is drawn narrower than the text
6685 // it has to hold — `你好` is two characters in four cells — and the cell
6686 // spills over the border it is supposed to sit inside, taking the whole
6687 // grid out of square with it. Squareness is the property: every row of a
6688 // grid is drawn to the same column, whatever its cells are spelled with.
6689 for src in [
6690 "| A | B |\n|---|---|\n| 你好 | y |\n",
6691 "| A | B |\n|---|---|\n| a👨👩👧b | y |\n",
6692 "| A | 漢字 |\n|---|---|\n| x | y |\n",
6693 ] {
6694 let m = map(src);
6695 let widths: Vec<usize> = m.rows.iter().map(|r| r.width()).collect();
6696 assert!(
6697 widths.windows(2).all(|w| w[0] == w[1]),
6698 "ragged grid {widths:?} for {src:?}:\n{}",
6699 rendered(&m)
6700 );
6701 }
6702 }
6703
6704 #[test]
6705 fn a_cell_wrapped_narrow_never_breaks_inside_a_character() {
6706 // A column too narrow for its cell hard-breaks the text, and every line
6707 // of it is given an end stop just past its last glyph. Broken into runs
6708 // of four glyphs, the first line of this cell ends between `👨👩` and the
6709 // joiner holding `👧` on — so its end stop lands inside a character,
6710 // where a click or Down can reach it and the next Backspace takes the
6711 // cluster apart from the middle.
6712 let src = "| A |\n|---|\n| 👨👩👧👨👩👧 |\n";
6713 let mut ed = Editor::new_str(src, Format::Markdown).unwrap();
6714 let m = build_t(&ed.nodes().unwrap(), src, Some(8));
6715 let boundaries: Vec<usize> = src
6716 .grapheme_indices(true)
6717 .map(|(i, _)| i)
6718 .chain(std::iter::once(src.len()))
6719 .collect();
6720 for off in (0..=src.len()).filter(|&o| m.is_stop(o)) {
6721 assert!(
6722 boundaries.contains(&off),
6723 "stop at {off} is inside a character:\n{}",
6724 rendered(&m)
6725 );
6726 }
6727 }
6728
6729 #[test]
6730 fn a_wrapped_cell_keeps_every_line_inside_its_column() {
6731 // The width is a promise in a table, where a glyph past the column lands
6732 // on the border or in the next cell — and it is a promise about cells,
6733 // which is not what a count of glyphs measures.
6734 let src = "| A |\n|---|\n| 你好世界漢字 |\n";
6735 let mut ed = Editor::new_str(src, Format::Markdown).unwrap();
6736 let m = build_t(&ed.nodes().unwrap(), src, Some(14));
6737 for r in &m.rows {
6738 assert_eq!(r.width(), 14, "{:?} is not drawn to the grid", rendered(&m));
6739 }
6740 }
6741
6742 #[test]
6743 fn a_hard_break_falls_between_clusters_and_measures_in_cells() {
6744 let glyphs = |s: &str| {
6745 let mut out = Vec::new();
6746 push_text(&mut out, s, 0, Style::default());
6747 out
6748 };
6749 let piece = |p: &[Glyph]| p.iter().map(|g| g.ch).collect::<String>();
6750
6751 // Six cells of CJK broken at four: two characters, then one — never
6752 // between the two cells of `好`.
6753 let w = glyphs("你好世");
6754 let pieces: Vec<String> = hard_break(&w, 4).iter().map(|p| piece(p)).collect();
6755 assert_eq!(pieces, ["你好", "世"]);
6756
6757 // A character wider than the column has nowhere legal to break, so it
6758 // keeps its cells rather than being cut in half.
6759 let w = glyphs("你好");
6760 let pieces: Vec<String> = hard_break(&w, 1).iter().map(|p| piece(p)).collect();
6761 assert_eq!(pieces, ["你", "好"]);
6762
6763 // An empty word yields no pieces at all — a double space stays a space.
6764 assert!(hard_break(&[], 4).is_empty());
6765 }
6766
6767 #[test]
6768 fn an_empty_list_item_still_gets_a_bulleted_row_with_a_caret_home() {
6769 // Pressing Enter at the end of a list item opens a new, empty item —
6770 // a childless `list_item`. Without a row of its own the new bullet
6771 // wouldn't appear until something was typed into it (the caret would be
6772 // stranded on an offset no row draws). It now renders as one prefixed
6773 // row whose end is a caret stop, so the bullet shows and the caret lands
6774 // just past the marker.
6775 let m = map("- item\n- \n");
6776 assert_eq!(m.num_rows(), 2, "the empty second item needs its own row");
6777 assert_eq!(
6778 m.rows[1].glyphs.iter().map(|g| g.ch).collect::<String>(),
6779 "• ",
6780 "the empty item draws just its bullet",
6781 );
6782 // Its end is the caret home (past the `- ` marker), and it's a real stop.
6783 assert!(
6784 m.is_stop(m.rows[1].end_src),
6785 "the empty item's caret home is not a stop"
6786 );
6787 assert_eq!(
6788 m.pos_of_offset(m.rows[1].end_src),
6789 (1, 2),
6790 "caret sits after '• '"
6791 );
6792 }
6793
6794 #[test]
6795 fn a_notes_row_range_stops_at_the_note_even_when_it_ends_in_a_link() {
6796 // The peek bug: a note whose body ends in a link has its last byte
6797 // inside the hidden destination, so mapping `end - 1` through
6798 // `pos_of_offset` snapped *forward* — past its own row, past the drawn
6799 // gap, and onto the next note's row. The popover then drew both notes.
6800 let src = "A[^1] B[^2].\n\n[^1]: bare text\n\n[^2]: [title](https://example.com/x)\n\n[^3]: last\n";
6801 let m = map(src);
6802 let body = src.find("[title]").unwrap();
6803 let end = src.find("\n\n[^3]").unwrap();
6804
6805 let (first, last) = m.row_range_for(body..end);
6806 assert_eq!(
6807 first, last,
6808 "a one-block note is one row, not a span onto the next"
6809 );
6810
6811 // The old arithmetic, kept here as the thing that must stay wrong: it
6812 // is what this method exists instead of.
6813 assert_ne!(
6814 m.pos_of_offset(end - 1).0,
6815 last,
6816 "the forward snap still leaves the note's row — that is the whole point",
6817 );
6818
6819 // A note ending in *visible* text was never broken, and still isn't:
6820 // both readings agree there, which is why the original test missed it.
6821 let plain = src.find("bare text").unwrap();
6822 let plain_end = src.find("\n\n[^2]").unwrap();
6823 let (pf, pl) = m.row_range_for(plain..plain_end);
6824 assert_eq!(pf, pl);
6825 assert_eq!(m.pos_of_offset(plain_end - 1).0, pl);
6826 }
6827
6828 #[test]
6829 fn a_row_range_covers_every_row_of_a_block_that_spans_several() {
6830 // The range is a span, not a point: a quote of two paragraphs covers its
6831 // gap row and both of its text rows, so a peek draws the whole thing.
6832 let src = "> one\n>\n> two\n\nafter\n";
6833 let m = map(src);
6834 let (first, last) = m.row_range_for(0..src.find("\n\nafter").unwrap());
6835 assert_eq!((first, last), (0, 2));
6836
6837 // And a range with no visible byte at all still covers the row it opened
6838 // on, rather than collapsing to nothing.
6839 let (f, l) = m.row_range_for(0..1);
6840 assert_eq!((f, l), (0, 0));
6841 }
6842
6843 #[test]
6844 fn an_empty_block_quote_still_gets_a_gutter_row_with_a_caret_home() {
6845 // The peer of the empty list item, and the case that made an empty line
6846 // in a quote draw as plain body text: a childless `block_quote` — a bare
6847 // `> `, which is what the toolbar's Quote button leaves on a blank line —
6848 // has no inner block to carry the gutter, so the whole quote used to
6849 // render as *nothing*. It didn't merely lose its bar; the row went away
6850 // and the caret had no home on it.
6851 let m = map("a\n\n> \n\nb\n");
6852 assert_eq!(
6853 m.rows[2].glyphs.iter().map(|g| g.ch).collect::<String>(),
6854 "│ ",
6855 "the empty quote draws just its gutter",
6856 );
6857 assert!(
6858 m.rows[2]
6859 .glyphs
6860 .iter()
6861 .all(|g| g.style.role == Role::QuoteGutter)
6862 );
6863 assert!(
6864 !m.rows[2].decoration,
6865 "it is a line text can go on, not a drawn gap"
6866 );
6867 assert!(
6868 m.is_stop(m.rows[2].end_src),
6869 "the empty quote's caret home is not a stop"
6870 );
6871 assert_eq!(
6872 m.pos_of_offset(m.rows[2].end_src),
6873 (2, 2),
6874 "caret sits after '│ '"
6875 );
6876
6877 // And a document that is *only* an empty quote still renders a row — it
6878 // used to render none at all, leaving the caret nowhere to stand.
6879 let m = map("> \n");
6880 assert_eq!(m.num_rows(), 1);
6881 assert_eq!(
6882 m.rows[0].glyphs.iter().map(|g| g.ch).collect::<String>(),
6883 "│ "
6884 );
6885 }
6886
6887 #[test]
6888 fn a_quotes_own_trailing_marker_lines_stay_inside_the_quote() {
6889 // Enter at the end of `> a` writes `> a\n>\n> \n`. Those last two lines
6890 // hold no block — a quote's `content_span` stops at its last child — so
6891 // the children walk never reaches them, and they used to fall through to
6892 // the document-level trailing pass, which knows no prefix: the gutter
6893 // stopped and the writer's new line drew as plain prose. Fixable only
6894 // since twig 3.2.0, where the quote's *span* covers its own marker lines
6895 // (`0..3` before, `0..8` now) and there is finally a node saying they
6896 // are the quote's.
6897 let m = map("> a\n>\n> \n");
6898 assert_eq!(m.num_rows(), 3, "one row per line the quote spells");
6899 for (i, row) in m.rows.iter().enumerate() {
6900 let text = row.glyphs.iter().map(|g| g.ch).collect::<String>();
6901 assert!(text.starts_with("│ "), "row {i} lost the gutter: {text:?}");
6902 assert!(
6903 !row.decoration,
6904 "row {i} is a line to type on, not a drawn gap"
6905 );
6906 assert!(m.is_stop(row.end_src), "row {i} has no caret home");
6907 }
6908 assert_eq!(
6909 m.rows[0].glyphs.iter().map(|g| g.ch).collect::<String>(),
6910 "│ a"
6911 );
6912 // Distinct offsets, so ↑/↓ between them moves the caret rather than
6913 // landing twice on the same byte.
6914 assert!(m.rows[0].end_src < m.rows[1].end_src);
6915 assert!(m.rows[1].end_src < m.rows[2].end_src);
6916
6917 // A blank line *after* the quote is not the quote's: it is spelled with
6918 // no marker, so it stays an ordinary boundary and the gutter ends.
6919 let m = map("> a\n\nb\n");
6920 assert_eq!(m.num_rows(), 3);
6921 assert_eq!(
6922 m.rows[2].glyphs.iter().map(|g| g.ch).collect::<String>(),
6923 "b"
6924 );
6925 assert!(
6926 !m.rows[1]
6927 .glyphs
6928 .iter()
6929 .any(|g| g.style.role == Role::QuoteGutter)
6930 );
6931
6932 // Nesting is the case this could get wrong, and the depth has to come
6933 // from which quote's span the line falls in rather than from the row
6934 // above it. A trailing `>` under `> > a` matches only the OUTER quote,
6935 // so it wears one gutter; spell it `> >` and it wears two.
6936 let m = map("> > a\n>\n");
6937 assert_eq!(
6938 m.rows[0].glyphs.iter().map(|g| g.ch).collect::<String>(),
6939 "│ │ a"
6940 );
6941 assert_eq!(
6942 m.rows[1].glyphs.iter().map(|g| g.ch).collect::<String>(),
6943 "│ "
6944 );
6945 let m = map("> > a\n> >\n");
6946 assert_eq!(
6947 m.rows[1].glyphs.iter().map(|g| g.ch).collect::<String>(),
6948 "│ │ "
6949 );
6950
6951 // And a marker line BETWEEN two quoted paragraphs is untouched: that is
6952 // the boundary `emit_separators_before` spells, and it stays a drawn gap
6953 // rather than becoming a line to type on.
6954 let m = map("> a\n>\n> b\n");
6955 assert_eq!(m.num_rows(), 3);
6956 assert!(
6957 m.rows[1].decoration,
6958 "the gap between two quoted blocks is still a gap"
6959 );
6960 }
6961
6962 #[test]
6963 fn an_empty_ordered_item_gets_its_number_and_a_caret_home() {
6964 let m = map("1. item\n2. \n");
6965 assert_eq!(m.num_rows(), 2);
6966 assert_eq!(
6967 m.rows[1].glyphs.iter().map(|g| g.ch).collect::<String>(),
6968 "2. "
6969 );
6970 assert!(m.is_stop(m.rows[1].end_src));
6971 assert_eq!(
6972 m.pos_of_offset(m.rows[1].end_src),
6973 (1, 3),
6974 "caret sits after '2. '"
6975 );
6976 }
6977
6978 #[test]
6979 fn an_empty_headings_caret_home_is_past_its_hidden_marker() {
6980 // The toolbar's H1 on a blank line writes `# ` and nothing else. The row
6981 // it renders is empty (the marker is hidden), so its end *is* its only
6982 // caret stop — and it has to be the offset past the `# `, where typing
6983 // continues the heading. Anchored at the block's start instead, the caret
6984 // drew in front of the hashes and the first character typed there landed
6985 // before them (`x# `), which isn't a heading at all.
6986 let m = map("# \n");
6987 assert_eq!(m.num_rows(), 1);
6988 assert!(m.rows[0].glyphs.is_empty(), "the `# ` marker is hidden");
6989 assert_eq!(m.rows[0].end_src, 2, "the caret home is past the marker");
6990 assert!(m.is_stop(2), "the empty heading's caret home is not a stop");
6991 }
6992
6993 #[test]
6994 fn a_headings_rows_carry_its_level_even_with_nothing_typed_in_it() {
6995 // The row-level fact a proportional frontend sizes a whole line by. An
6996 // empty heading has no glyph to read a `Role::Heading` off, so a renderer
6997 // scanning glyphs drew `# ` (and its caret) at body height until the
6998 // first character landed.
6999 let m = map("# \n");
7000 assert_eq!(
7001 m.rows[0].heading,
7002 Some(1),
7003 "the empty heading knows its level"
7004 );
7005
7006 // Every row of one that wraps, not just the first — and nothing else.
7007 let m = map_at(
7008 "## a heading long enough to wrap over two rows\n\nbody\n",
7009 Some(20),
7010 );
7011 let heads: Vec<Option<u8>> = m.rows.iter().map(|r| r.heading).collect();
7012 assert!(
7013 heads.iter().filter(|h| **h == Some(2)).count() >= 2,
7014 "got {heads:?}"
7015 );
7016 assert_eq!(
7017 m.rows.last().and_then(|r| r.heading),
7018 None,
7019 "the paragraph under it is not a heading",
7020 );
7021 }
7022
7023 #[test]
7024 fn an_empty_heading_leaves_the_rows_under_it_at_their_own_offsets() {
7025 // The row's end is also what the *next* row's separator is measured from,
7026 // so an empty heading that under-reported it shifted every offset below —
7027 // and the blank line under the heading then claimed the same offset as the
7028 // heading's own end. `pos_of_offset` resolves such a tie downstream (a
7029 // soft wrap belongs to the row below), so the caret at the end of the
7030 // heading was drawn two rows lower, on the blank line.
7031 // `text\n\n# \n\n`: the heading's content opens at 8, and the two rows
7032 // under it end at 9 and 10 — the blank line and the document's end.
7033 let m = map("text\n\n# \n\n");
7034 let end = m.rows.last().expect("a trailing blank row").end_src;
7035 assert_eq!(end, 10, "the trailing rows must end at their real offsets");
7036 // The heading's caret home is its own row's, not one shared with a row
7037 // below — the tie that drew the caret two rows down.
7038 assert_eq!(m.pos_of_offset(8), (2, 0), "the empty heading's own row");
7039 assert!(
7040 m.rows[3..].iter().all(|r| r.end_src > 8),
7041 "rows below own later offsets"
7042 );
7043 }
7044
7045 // ── block boundaries ─────────────────────────────────────────────────────
7046
7047 /// Every drawn boundary in `src`, in order, as `(above, below)`.
7048 fn boundaries(m: &VisualMap) -> Vec<(BlockClass, BlockClass)> {
7049 m.rows
7050 .iter()
7051 .filter_map(|r| r.boundary)
7052 .map(|b| (b.above, b.below))
7053 .collect()
7054 }
7055
7056 #[test]
7057 fn a_boundary_says_which_blocks_it_divides() {
7058 use BlockClass::*;
7059 let m = map("one\n\ntwo\n\n# Head\n\ntail\n\n> quoted\n\n```\ncode\n```\n");
7060 assert_eq!(
7061 boundaries(&m),
7062 vec![
7063 (Paragraph, Paragraph),
7064 (Paragraph, Heading),
7065 (Heading, Paragraph),
7066 (Paragraph, Quote),
7067 (Quote, Code),
7068 // The blank the document trails off with is a boundary too — it
7069 // closes the last block above the empty paragraph the caret rests
7070 // on. See `emit_trailing_blank_lines`.
7071 (Code, Paragraph),
7072 ],
7073 "each gap names the pair it falls between, in document order"
7074 );
7075 }
7076
7077 #[test]
7078 fn the_trailing_gap_closes_the_last_block() {
7079 // Two Enters at the end of a document: a drawn gap, then the navigable
7080 // empty paragraph. Only the gap is labelled, so a frontend that shrinks
7081 // boundaries shrinks the spacer and leaves the row being typed on alone.
7082 let m = map("# Head\n\n\n");
7083 assert_eq!(
7084 boundaries(&m),
7085 vec![(BlockClass::Heading, BlockClass::Paragraph)]
7086 );
7087 }
7088
7089 #[test]
7090 fn only_the_drawn_gap_rows_carry_a_boundary() {
7091 let m = map("one\n\ntwo\n");
7092 for row in &m.rows {
7093 assert_eq!(
7094 row.boundary.is_some(),
7095 row.decoration,
7096 "a boundary is exactly a drawn gap row: {:?}",
7097 row.glyphs.iter().map(|g| g.ch).collect::<String>()
7098 );
7099 }
7100 }
7101
7102 #[test]
7103 fn preserve_flow_labels_no_boundary() {
7104 // Every blank line is a caret home there — somewhere text can go, not a
7105 // gap between blocks — so nothing is drawn-only and nothing is labelled.
7106 // A frontend keying its spacing off `boundary` can't shrink a row the
7107 // author is about to type on.
7108 let m = map_preserve("one\n\ntwo\n\n# Head\n", Some(80));
7109 assert!(boundaries(&m).is_empty());
7110 }
7111
7112 #[test]
7113 fn a_list_draws_no_boundary_between_its_items() {
7114 // Tight or loose, core puts no gap row between two items of one list —
7115 // so an item↔item boundary is a shape no frontend will ever be handed,
7116 // and spacing one is spacing something that isn't there.
7117 for src in ["- one\n- two\n", "- one\n\n- two\n"] {
7118 let m = map(src);
7119 assert!(
7120 boundaries(&m).is_empty(),
7121 "no gap row inside the list of {src:?}"
7122 );
7123 }
7124 // Leaving the list is an ordinary boundary, and the list is named as
7125 // what sits above it.
7126 let m = map("- one\n- two\n\npara\n");
7127 assert_eq!(
7128 boundaries(&m),
7129 vec![(BlockClass::List, BlockClass::Paragraph)]
7130 );
7131 }
7132
7133 #[test]
7134 fn a_nested_boundary_names_the_blocks_inside_the_container() {
7135 // Two paragraphs inside a blockquote are divided by a Paragraph↔Paragraph
7136 // boundary — the quote is the container they're both in, not what the gap
7137 // separates.
7138 let m = map("> one\n>\n> two\n");
7139 assert_eq!(
7140 boundaries(&m),
7141 vec![(BlockClass::Paragraph, BlockClass::Paragraph)]
7142 );
7143 }
7144
7145 #[test]
7146 fn a_directive_container_draws_one_boundary_like_every_other_block() {
7147 // A container's rows stop at its last *child*, so without anchoring
7148 // `last_off` past the closing `:::` the separator logic counted the fence
7149 // line as a blank row of its own and drew the gap twice — one authored
7150 // blank line, two boundaries, and a frontend spacing each of them put
7151 // double margin under every fenced div. The code-block arm anchors past
7152 // its ``` for exactly this reason; compare the two here.
7153 let fenced = map_directives(":::note\nin\n:::\n\ntwo\n");
7154 assert_eq!(
7155 boundaries(&fenced),
7156 vec![(BlockClass::Directive, BlockClass::Paragraph)],
7157 "one authored gap, one boundary row"
7158 );
7159 let code = map("```\nc\n```\n\ntwo\n");
7160 assert_eq!(
7161 boundaries(&code).len(),
7162 boundaries(&fenced).len(),
7163 "a fenced div spaces like a fenced code block"
7164 );
7165 // Nesting closes several fences at once; still one gap.
7166 let nested = map_directives(":::a\n:::b\nin\n:::\n:::\n\ntwo\n");
7167 assert_eq!(
7168 boundaries(&nested),
7169 vec![(BlockClass::Directive, BlockClass::Paragraph)]
7170 );
7171 }
7172
7173 #[test]
7174 fn a_block_media_names_itself_in_the_boundaries_either_side() {
7175 use BlockClass::*;
7176 // A block image is never a node of its own — `media_only` promotes the
7177 // *paragraph* wrapping it — so classifying the node the walk stands on
7178 // called the picture `Paragraph` and left `BlockClass::Media` unreachable:
7179 // a frontend could not give a photo more air than a line of prose.
7180 // `label_media_boundaries` reads it back off the finished rows instead.
7181 let m = map("one\n\n\n\ntwo\n");
7182 assert_eq!(boundaries(&m), vec![(Paragraph, Media), (Media, Paragraph)]);
7183 // At the edges of the document too: the leading gap has no boundary of
7184 // its own, and the trailing one is `emit_trailing_blank_lines`'.
7185 let edges = map("\n\nmid\n\n\n");
7186 assert_eq!(
7187 boundaries(&edges),
7188 vec![(Media, Paragraph), (Paragraph, Media)]
7189 );
7190 // One gap spelled with several rows — the row closing the block above and
7191 // the row opening the one below, with the author's spare blank line
7192 // navigable between them — carries the same pair on every drawn row.
7193 let roomy = map("one\n\n\n\n\n");
7194 assert_eq!(
7195 boundaries(&roomy),
7196 vec![(Paragraph, Media), (Paragraph, Media)]
7197 );
7198 }
7199
7200 #[test]
7201 fn a_block_video_is_media_at_its_boundaries_not_a_directive_panel() {
7202 // Worse than the image case before `label_media_boundaries`: a `<video>`
7203 // arrives as twig's generic `container`, which classifies `Directive` —
7204 // the one class a frontend reads as "draw a tinted panel here". A movie
7205 // got the chrome of a fenced div.
7206 let mut doc = crate::Doc::from_source(
7207 "one\n\n<video src=\"v.mp4\"></video>\n\ntwo\n".to_string(),
7208 Format::Markdown,
7209 )
7210 .unwrap();
7211 doc.build_visual(80);
7212 assert_eq!(
7213 boundaries(&doc.vmap),
7214 vec![
7215 (BlockClass::Paragraph, BlockClass::Media),
7216 (BlockClass::Media, BlockClass::Paragraph),
7217 ]
7218 );
7219 }
7220
7221 #[test]
7222 fn the_incremental_walk_labels_boundaries_like_the_full_one() {
7223 // `assert_maps_eq` compares boundaries too, so this pins the two doors
7224 // into `BlockClass::from_node_kind` — a `FlatNode`'s kind on the full
7225 // build, a query match's on the cached one — against a document with one
7226 // of every boundary in it.
7227 let src = "one\n\n# Head\n\ntwo\n\n- a\n- b\n\n> q\n\n```\nc\n```\n\npara\n";
7228 let mut ed = Editor::new_str(src, Format::Markdown).unwrap();
7229 let mut cache = BlockCache::default();
7230 let (full, cached) = render_both(&mut ed, src, Some(80), &mut cache);
7231 assert_maps_eq(&full, &cached, "boundary labelling");
7232 assert!(
7233 !boundaries(&full).is_empty(),
7234 "the fixture has boundaries to compare"
7235 );
7236 }
7237
7238 #[test]
7239 fn every_caret_stop_opens_a_cluster_of_its_row() {
7240 // The two ways of finding a cluster have to agree. `push_text` marks the
7241 // stops by segmenting one run of text; the column mapping segments the
7242 // whole row, decoration and all. A stop that came out as the *middle* of
7243 // some row-level cluster would be a caret with no column of its own —
7244 // drawn at the column of whatever swallowed it.
7245 let src = "# 標題\n\na **bold** e\u{0301}mo👨👩👧ji `x` 你好\n\n\
7246 - 項目 one\n- e\u{0301}dge\n\n> 引用 text\n\n\
7247 | A | 值 |\n|---|---|\n| 你好 | 👩🚀 |\n";
7248 let m = map(src);
7249 for (r, row) in m.rows.iter().enumerate() {
7250 let openers: Vec<usize> = clusters(&row.glyphs).iter().map(|c| c.glyph).collect();
7251 for (i, g) in row.glyphs.iter().enumerate() {
7252 assert!(
7253 !g.stop || openers.contains(&i),
7254 "row {r}: the stop at glyph {i} ({:?}) is inside a cluster, \
7255 so it is drawn at another glyph's column",
7256 g.ch
7257 );
7258 }
7259 }
7260 }
7261}