markdown/doc.rs
1//! The document model.
2//!
3//! A [`Doc`] is a **flat** list of [`Block`]s with an indent level, not a
4//! nested tree. That is Notion's model rather than CommonMark's, and it is the
5//! decision the rest of this crate hangs off: editing a flat list means Enter
6//! splits, Backspace merges, and Tab indents — all list operations. On a
7//! nested tree "the previous block" is a traversal and every edit is a
8//! restructure.
9//!
10//! The trade is that arbitrarily nested CommonMark does not survive a round
11//! trip: a list inside a quote inside a list flattens. Notion has the same
12//! limitation. What is guaranteed is [`crate::serialize`]'s fixed point —
13//! parse, serialize, parse again, and the document is unchanged — so an
14//! edit/save cycle never drifts.
15
16use std::ops::Range;
17
18/// A markdown document: blocks in document order.
19#[derive(Debug, Clone, Default, PartialEq, Eq)]
20pub struct Doc {
21 pub blocks: Vec<Block>,
22}
23
24impl Doc {
25 /// Make each run of ordered items consecutive.
26 ///
27 /// Markdown honours only the *first* number in a list — `1.` followed by `9.`
28 /// renders as 1, 2. Two source lists that used different delimiters (`1.` then
29 /// `9)`) are separate lists, but a flat document has no list identity to
30 /// preserve, so they serialize as one and the second item's number would move
31 /// on the next read. Deciding it here means the document already holds what the
32 /// next parse would produce.
33 pub(crate) fn renumber(&mut self) {
34 // The number owed to the next ordered item at each indent level. A run
35 // survives blocks nested under it and ends at anything else.
36 let mut expected: Vec<Option<u64>> = Vec::new();
37 for block in &mut self.blocks {
38 let indent = block.indent as usize;
39 expected.truncate(indent + 1);
40 expected.resize(indent + 1, None);
41
42 if let BlockKind::Ordered { number, .. } = &mut block.kind {
43 if let Some(next) = expected[indent] {
44 *number = next;
45 }
46 expected[indent] = Some(number.saturating_add(1));
47 } else {
48 expected[indent] = None;
49 }
50 }
51 }
52}
53
54/// One block, and how deeply it is nested.
55///
56/// `indent` obeys one invariant, established by the parser and relied on by
57/// the serializer: the first block is at 0, and no block is more than one
58/// level deeper than the block before it. A document that satisfies it always
59/// serializes to markdown that parses back to the same indents.
60#[derive(Debug, Clone, PartialEq, Eq)]
61pub struct Block {
62 pub kind: BlockKind,
63 pub indent: u8,
64}
65
66impl From<BlockKind> for Block {
67 fn from(kind: BlockKind) -> Self {
68 Self::new(kind)
69 }
70}
71
72impl Block {
73 pub fn new(kind: BlockKind) -> Self {
74 Self { kind, indent: 0 }
75 }
76
77 pub fn at(kind: BlockKind, indent: u8) -> Self {
78 Self { kind, indent }
79 }
80
81 /// One of the block's editable texts. `None` when the block has no such
82 /// part — every block is atomic to some [`Part`], and bookmarks and rules
83 /// are atomic to all of them.
84 pub fn text_at(&self, part: Part) -> Option<&Text> {
85 match (&self.kind, part) {
86 (
87 BlockKind::Paragraph(text)
88 | BlockKind::Heading { text, .. }
89 | BlockKind::Bullet(text)
90 | BlockKind::Ordered { text, .. }
91 | BlockKind::Task { text, .. }
92 | BlockKind::Quote { text, .. },
93 Part::Body,
94 ) => Some(text),
95 (BlockKind::Code { code, .. }, Part::Code) => Some(code),
96 (BlockKind::Image { alt, .. }, Part::Caption) => Some(alt),
97 (BlockKind::Table { header, .. }, Part::Cell { row: 0, column }) => header.get(column),
98 (BlockKind::Table { rows, .. }, Part::Cell { row, column }) => {
99 rows.get(row - 1)?.get(column)
100 }
101 _ => None,
102 }
103 }
104
105 pub fn text_at_mut(&mut self, part: Part) -> Option<&mut Text> {
106 match (&mut self.kind, part) {
107 (
108 BlockKind::Paragraph(text)
109 | BlockKind::Heading { text, .. }
110 | BlockKind::Bullet(text)
111 | BlockKind::Ordered { text, .. }
112 | BlockKind::Task { text, .. }
113 | BlockKind::Quote { text, .. },
114 Part::Body,
115 ) => Some(text),
116 (BlockKind::Code { code, .. }, Part::Code) => Some(code),
117 (BlockKind::Image { alt, .. }, Part::Caption) => Some(alt),
118 (BlockKind::Table { header, .. }, Part::Cell { row: 0, column }) => {
119 header.get_mut(column)
120 }
121 (BlockKind::Table { rows, .. }, Part::Cell { row, column }) => {
122 rows.get_mut(row - 1)?.get_mut(column)
123 }
124 _ => None,
125 }
126 }
127
128 /// Every part a caret can sit in, in document order.
129 pub fn parts(&self) -> Vec<Part> {
130 match &self.kind {
131 BlockKind::Paragraph(_)
132 | BlockKind::Heading { .. }
133 | BlockKind::Bullet(_)
134 | BlockKind::Ordered { .. }
135 | BlockKind::Task { .. }
136 | BlockKind::Quote { .. } => vec![Part::Body],
137 BlockKind::Code { .. } => vec![Part::Code],
138 BlockKind::Image { .. } => vec![Part::Caption],
139 BlockKind::Table { header, rows, .. } => {
140 let mut parts = Vec::new();
141 if !header.is_empty() {
142 parts.extend((0..header.len()).map(|column| Part::Cell { row: 0, column }));
143 }
144 for (ix, row) in rows.iter().enumerate() {
145 parts.extend((0..row.len()).map(|column| Part::Cell {
146 row: ix + 1,
147 column,
148 }));
149 }
150 parts
151 }
152 BlockKind::Bookmark { .. } | BlockKind::Rule => Vec::new(),
153 }
154 }
155
156 /// Whether what the block paints past its parts holds no caret — a picture,
157 /// a card, a line. What a selection has to wash for itself, since there is
158 /// no text under it to carry the highlight.
159 pub fn opaque(&self) -> bool {
160 matches!(
161 self.kind,
162 BlockKind::Image { .. } | BlockKind::Bookmark { .. } | BlockKind::Rule
163 )
164 }
165}
166
167/// Which of a block's texts a caret sits in.
168///
169/// A block has one kind of part and never a mix — prose blocks have a body, a
170/// code block has its code, a table has cells — so this is a coordinate rather
171/// than a path, and the model stays flat. The ordering is document order, which
172/// is what makes a [`crate::Cursor`] comparable and therefore what makes a
173/// selection a range.
174#[derive(Debug, Clone, Copy, PartialEq, Eq, PartialOrd, Ord, Default)]
175pub enum Part {
176 #[default]
177 Body,
178 Code,
179 /// An image's caption, which is also its alt text.
180 Caption,
181 /// Row 0 is the header row; row `n` is `rows[n - 1]`.
182 Cell {
183 row: usize,
184 column: usize,
185 },
186}
187
188/// The block vocabulary. Closed by design — a consumer that needs a block of
189/// its own is a reason to widen this enum rather than to grow an extension
190/// system.
191#[derive(Debug, Clone, PartialEq, Eq)]
192pub enum BlockKind {
193 Paragraph(Text),
194 Heading {
195 /// 1–6.
196 level: u8,
197 text: Text,
198 },
199 Bullet(Text),
200 Ordered {
201 /// The rendered number. Stored rather than derived so a list starting
202 /// at 3 survives the round trip.
203 number: u64,
204 text: Text,
205 },
206 Task {
207 checked: bool,
208 text: Text,
209 },
210 /// A blockquote. `kind` is the GFM alert it opens with — see [`QuoteKind`].
211 Quote {
212 kind: Option<QuoteKind>,
213 text: Text,
214 },
215 /// The code carries a [`Text`] like every other editable region, so one
216 /// accessor and one edit path cover the whole document. Its marks are
217 /// unreachable rather than forbidden: nothing that writes here creates one.
218 Code {
219 language: Option<String>,
220 code: Text,
221 },
222 /// The caption is the alt text — markdown has one slot, and a reader that
223 /// cannot see the picture reads the same words. Like [`BlockKind::Code`]'s,
224 /// its marks are unreachable rather than forbidden.
225 Image {
226 url: String,
227 alt: Text,
228 /// A drag off the handle in `bezel-editor`, in whole pixels — `None` is
229 /// the natural width. `u32` rather than a float: this derives `Eq`, and
230 /// a `f32` cannot because of NaN. Spelled ``, which
231 /// leaves the title slot to say what a title says.
232 width: Option<u32>,
233 },
234 /// A link with a block to itself, painted richly. Atomic on purpose:
235 /// everything it shows past the URL comes from [`crate::preview`], so there
236 /// is nothing here for a caret to edit.
237 ///
238 /// [`Form`] picks which of the three — a chip, a card, or a card with its
239 /// picture across the width. Off a line of its own the same link is a
240 /// [`Mark::Mention`], which is the same three minus what shaped text cannot
241 /// hold.
242 Bookmark {
243 url: String,
244 form: Form,
245 },
246 Table {
247 align: Vec<Align>,
248 header: Vec<Text>,
249 rows: Vec<Vec<Text>>,
250 },
251 Rule,
252}
253
254/// A GFM alert's kind — the `[!NOTE]` marker a blockquote opens with.
255///
256/// Only recognised when the marker is alone on the quote's first line and
257/// names one of these five; anything else stays the text it was written as.
258/// A blockquote holding two paragraphs becomes two [`BlockKind::Quote`] blocks
259/// and each carries the kind, so writing the document back gives two alerts.
260#[derive(Debug, Clone, Copy, PartialEq, Eq)]
261pub enum QuoteKind {
262 Note,
263 Tip,
264 Important,
265 Warning,
266 Caution,
267}
268
269impl QuoteKind {
270 /// The marker line, bracket to bracket.
271 pub fn marker(self) -> &'static str {
272 match self {
273 Self::Note => "[!NOTE]",
274 Self::Tip => "[!TIP]",
275 Self::Important => "[!IMPORTANT]",
276 Self::Warning => "[!WARNING]",
277 Self::Caution => "[!CAUTION]",
278 }
279 }
280
281 /// What the alert calls itself where it is painted.
282 pub fn label(self) -> &'static str {
283 match self {
284 Self::Note => "Note",
285 Self::Tip => "Tip",
286 Self::Important => "Important",
287 Self::Warning => "Warning",
288 Self::Caution => "Caution",
289 }
290 }
291}
292
293/// GFM column alignment.
294#[derive(Debug, Clone, Copy, PartialEq, Eq, Default)]
295pub enum Align {
296 #[default]
297 Left,
298 Center,
299 Right,
300}
301
302/// Inline content: a string, plus marks over byte ranges of it.
303///
304/// Marks are a separate list rather than flags on a run because an editor has
305/// to *map* them through insertions and deletions, and because run flags lose
306/// nesting order — under flags `**_x_**` and `_**x**_` are the same value.
307/// Here they differ by the order of the two spans, and both survive a round
308/// trip.
309///
310/// A newline in `text` is a line break within the block (markdown's soft or
311/// hard break, which this model does not distinguish — neither does Notion).
312/// Whether it paints as a break or a space is a rendering decision.
313#[derive(Debug, Clone, Default, PartialEq, Eq)]
314pub struct Text {
315 pub text: String,
316 /// Outermost first. Ranges may overlap and may be identical.
317 pub marks: Vec<MarkSpan>,
318}
319
320impl From<&str> for Text {
321 fn from(text: &str) -> Self {
322 Self::plain(text)
323 }
324}
325
326impl From<String> for Text {
327 fn from(text: String) -> Self {
328 Self::plain(text)
329 }
330}
331
332impl Text {
333 /// Unmarked text.
334 pub fn plain(text: impl Into<String>) -> Self {
335 Self {
336 text: text.into(),
337 marks: Vec::new(),
338 }
339 }
340
341 /// A URL that links to itself — what a pasted link is, and what a bookmark
342 /// hands back when it turns into prose.
343 pub fn link(url: &str) -> Self {
344 Self {
345 text: url.to_string(),
346 marks: vec![MarkSpan {
347 range: 0..url.len(),
348 mark: Mark::Link(url.to_string()),
349 }],
350 }
351 }
352
353 pub fn is_empty(&self) -> bool {
354 self.text.is_empty()
355 }
356
357 /// Whether no other mark overlaps the one at `ix`.
358 ///
359 /// A mark written whole — a code span, a mention — leaves no room inside
360 /// itself for another mark's boundary, so this is what decides whether it
361 /// can be spelled its own way at all.
362 pub(crate) fn alone(&self, ix: usize) -> bool {
363 let span = &self.marks[ix].range;
364 self.marks.iter().enumerate().all(|(other, mark)| {
365 other == ix || mark.range.end <= span.start || mark.range.start >= span.end
366 })
367 }
368}
369
370#[derive(Debug, Clone, PartialEq, Eq)]
371pub struct MarkSpan {
372 pub range: Range<usize>,
373 pub mark: Mark,
374}
375
376/// How a [`Mark::Mention`] was written down, which is also how it paints.
377///
378/// `Auto` is the shorthand `<https://x>`: a chip in a sentence, a card on a
379/// line of its own. It is a variant rather than a resolved form because the
380/// spelling is what has to survive the round trip — resolve it at parse and
381/// every `<url>` grows brackets the first time the file is saved.
382///
383/// The other two are CommonMark's title slot, `[url](url "chip")`, which is
384/// core, ignored by every other renderer, and the only place left to say what
385/// the shorthand cannot: a chip alone on a line, and the bigger card.
386#[derive(Debug, Clone, Copy, PartialEq, Eq)]
387pub enum Form {
388 Auto,
389 Chip,
390 Embed,
391}
392
393impl Form {
394 /// The title that spells this form, and `None` for the shorthand.
395 pub(crate) fn title(self) -> Option<&'static str> {
396 match self {
397 Self::Auto => None,
398 Self::Chip => Some("chip"),
399 Self::Embed => Some("embed"),
400 }
401 }
402
403 pub(crate) fn from_title(title: &str) -> Option<Self> {
404 match title {
405 "chip" => Some(Self::Chip),
406 "embed" => Some(Self::Embed),
407 _ => None,
408 }
409 }
410}
411
412#[derive(Debug, Clone, PartialEq, Eq)]
413pub enum Mark {
414 Bold,
415 Italic,
416 Strike,
417 Code,
418 Link(String),
419 /// A link painted richly rather than as underlined text — a chip inline, a
420 /// [`BlockKind::Bookmark`] with a block to itself.
421 ///
422 /// The chip shows the URL, because a [`Text`] is one string and every caret
423 /// offset is a byte into it: an inline atom painted wider or narrower than
424 /// the text under it has nowhere to put the offsets in between.
425 Mention {
426 url: String,
427 form: Form,
428 },
429 /// A mark the app spells itself — underline, a highlight, a colour. The
430 /// name is [`crate::Marks`]'s, and the delimiter that writes it comes from
431 /// the same registry.
432 Custom(String),
433 /// An image among text. [`BlockKind::Image`] is the shape an editor offers;
434 /// this is what keeps `see  here` from silently becoming a link when
435 /// the document is saved. No width: one among text has no box of its own to
436 /// resize, and the `|480` that would say so stays the ordinary text it is.
437 Image(String),
438}