quillmark_content/ops.rs
1//! Island, line and mark op channels: structural edits separate from text
2//! splices.
3//!
4//! [`IslandOp`], [`LineOp`] and [`MarkOp`] apply after
5//! [`Content::apply_text_delta`] in one [`ChangeBundle`], in that order. Each
6//! channel reaches a part of the model no splice can: an island's payload, a
7//! line's block role, a mark's range. Mark ranges are in **final-text
8//! coordinates**: mark ops run last and validate against the length every
9//! earlier stage left, so a producer
10//! computes them in the only frame it can, the text as it stands once the
11//! delta and line ops have landed. Line split/join splice a `\n` in `text` and
12//! rebase marks through that one-char change with
13//! [`Delta::map_pos`](crate::delta::Delta::map_pos), the same mapping the
14//! text-delta channel uses, so a mark's coordinates track the splice rather
15//! than drifting.
16
17use crate::delta::{Assoc, Delta, Op};
18use crate::model::{
19 line_kind_mismatch, Container, Island, Line, LineKind, LineKindMismatch, Mark, MarkKind,
20 Content, Usv, ISLAND_SLOT,
21};
22use crate::normalize::is_bidi_char;
23use crate::usv::char_to_byte;
24use std::borrow::Cow;
25
26/// A mark edit in final-text coordinates (post-delta, post-line-op).
27#[derive(Debug, Clone, PartialEq)]
28#[non_exhaustive]
29pub enum MarkOp {
30 /// Add a mark over `[start, end)`. An anchor `kind` must carry a non-empty
31 /// `id` not already live in the field: ids are caller-supplied and unique
32 /// per `Content` (`DOCUMENT_STORAGE.md` § Anchor-id identity); a collision or
33 /// the empty id is rejected ([`ApplyError::AnchorIdCollision`] /
34 /// [`ApplyError::EmptyAnchorId`]), never replaced or coexisted.
35 Add {
36 start: Usv,
37 end: Usv,
38 kind: MarkKind,
39 },
40 /// Un-format `kind` over `[start, end)`: subtract the range from each
41 /// overlapping same-kind *formatting* mark, keeping the non-overlapping
42 /// fragments (a mid-run removal punches a hole; `normalize` drops any
43 /// zero-width fragment an edge-aligned removal leaves). Non-formatting
44 /// (identity/unknown) handles can't be range-fragmented, so an overlapping
45 /// one is dropped whole: anchors normally go through [`MarkOp::RemoveAnchor`].
46 Remove {
47 start: Usv,
48 end: Usv,
49 kind: MarkKind,
50 },
51 /// Drop one identity anchor by id.
52 RemoveAnchor { id: String },
53}
54
55/// A line/block edit. Split/join splice `\n` in `text`; set ops touch metadata
56/// only.
57#[derive(Debug, Clone, PartialEq)]
58#[non_exhaustive]
59pub enum LineOp {
60 /// Paragraph break at `at`: insert `\n` and split the line metadata.
61 Split { at: Usv },
62 /// Join line `line` with the next: remove the `\n` between them.
63 Join { line: usize },
64 /// Replace a line's block role.
65 SetKind { line: usize, kind: LineKind },
66 /// Replace a line's container path.
67 SetContainers {
68 line: usize,
69 containers: Vec<Container>,
70 },
71 /// Set (or clear) a line's `continues` flag: whether it continues the
72 /// previous line's block across a within-block hard break (a markdown hard
73 /// break, a code fence's interior line) rather than starting a new block.
74 /// The op-grained twin of the value that `install` already round-trips:
75 /// `split`/`join`/text-delta `\n` insertion all mint `continues: false`
76 /// lines, so without this a hard break or a new code-fence interior line is
77 /// unreachable op-wise and falls back to a whole-`install` (losing that
78 /// edit's identity anchors). Setting `continues: true` on line 0 is
79 /// [`ApplyError::FirstLineContinues`] (nothing precedes it to continue).
80 SetContinues { line: usize, continues: bool },
81}
82
83/// An island edit: the channel that reaches [`Island`] payloads, which no other
84/// channel carries (`text` holds one [`ISLAND_SLOT`] per island and nothing
85/// more, `lines` the [`LineKind::Island`] tag, `marks` neither).
86///
87/// Both ops are **value semantics over one island entry**, not over the field:
88/// the slot stays put, so every identity anchor in the field's text survives an
89/// island edit. Without them a table edit lowers to a whole-field `install`,
90/// which drops every anchor in the field: [`LineOp::SetContinues`]'s argument at
91/// the scale of a table.
92///
93/// Removal needs no op: a text delta that deletes a slot drops the backing
94/// entry ([`Content::apply_text_delta`]'s cascade). The drop is whole, so
95/// re-landing that island is an [`IslandOp::Insert`] carrying the [`Island`]
96/// itself, which only the producer that deleted it still holds. A *block*
97/// island's line demotes to `Para` when its slot goes, so re-landing one
98/// re-tags the line too.
99#[derive(Debug, Clone, PartialEq)]
100#[non_exhaustive]
101pub enum IslandOp {
102 /// Replace the entry `island.id` names, in place. The id is the target *and*
103 /// the stored value, so an island cannot be renamed through this op: ids are
104 /// hash input and stable across edits by contract
105 /// (`DOCUMENT_STORAGE.md` § Island-id determinism). An id no island carries
106 /// is [`ApplyError::UnknownIslandId`], never a silent no-op: swallowing it
107 /// leaves the store on the old value with the caller believing it committed.
108 ///
109 /// `props`, `island_type` and `loss` all come from the op. Nothing derives
110 /// `loss` from the props: like `install`, the op stores what the caller hands
111 /// it, so a write that changes what markdown can carry restates the class or
112 /// carries the stale one forward.
113 Set { island: Island },
114 /// Insert an island: the [`ISLAND_SLOT`] at `at` and its backing entry in
115 /// one op, so a slot never exists without the [`Island`] behind it (the
116 /// orphan [`ApplyError::IslandSlotInInsert`] guards against on the text
117 /// channel is unrepresentable here rather than rejected after the fact).
118 ///
119 /// `at` is a USV position in the text the delta and this bundle's earlier
120 /// island ops left: each insert splices its slot before the next op reads
121 /// the text. Slots after `a` and `b` of `abc` therefore go in at 1 and 3,
122 /// and of two inserts at one position the later one lands first. The entry
123 /// files at its slot-order index in that same frame, so text and island
124 /// list agree whatever the emission order; a stale frame misplaces slots
125 /// and never errors.
126 ///
127 /// The id is caller-supplied, non-empty, and unique in the field, on an
128 /// anchor id's terms ([`ApplyError::EmptyIslandId`],
129 /// [`ApplyError::IslandIdCollision`]). `Set` addresses by id, so a
130 /// degenerate or shared id is an island that cannot be edited, or cannot be
131 /// told from another. Minting follows `DOCUMENT_STORAGE.md` § Island-id
132 /// determinism: a new island continues the field's positional sequence,
133 /// while re-landing a dropped one carries its original id back. A delete
134 /// earlier in the same bundle frees its id here, which is how a
135 /// replace-in-place lands as one bundle.
136 ///
137 /// **Block islands.** The slot alone is an *inline* island (a slot in a
138 /// `Para`). A block island is that slot alone on its own line under
139 /// [`LineKind::Island`], which takes three channels in one bundle: the text
140 /// delta inserts the `\n`, this op inserts the slot, and
141 /// [`LineOp::SetKind`] tags the line. That order is why island ops run
142 /// *before* line ops: `SetKind` validates the kind against the text already
143 /// on the line, so the slot has to be there first. `LineOp::Split` cannot
144 /// stand in for the delta's `\n`: it runs in the later stage.
145 ///
146 /// A slot inserted onto a line whose kind names its content (`Code`, `Rule`)
147 /// contradicts that kind; `normalize` demotes the line to `Para` at the end
148 /// of the bundle rather than failing it.
149 Insert { at: Usv, island: Island },
150}
151
152/// One committed field edit: a text delta and the three op channels, applied in
153/// field order (delta → islands → lines → marks) by
154/// [`Content::apply_field_change`].
155///
156/// A struct rather than four positional arguments so a fifth channel is an
157/// additive change at every call site. [`Default`] is the identity bundle (no
158/// text change, no ops), so a caller names only the channels it uses:
159/// `ChangeBundle { delta, ..Default::default() }`.
160///
161/// Within a channel ops apply in sequence: op *n*'s coordinates read the state
162/// ops `0..n` left, not the frame the channel opens in. That is USV positions
163/// for an island insert (its `at` counts the slots earlier inserts spliced) and
164/// line indices for [`LineOp::Split`] / [`LineOp::Join`], which renumber every
165/// later line. A stale frame stays in range, so the bundle applies cleanly and
166/// lands the wrong document.
167#[derive(Debug, Clone, PartialEq)]
168pub struct ChangeBundle {
169 /// The text splice; the identity delta (no ops) is no text change.
170 pub delta: Delta,
171 /// Island edits, opening in post-delta coordinates and sequenced.
172 pub island_ops: Vec<IslandOp>,
173 /// Line edits, opening in post-delta, post-island-op coordinates and
174 /// sequenced.
175 pub line_ops: Vec<LineOp>,
176 /// Mark edits, in final-text coordinates (every earlier stage applied).
177 pub mark_ops: Vec<MarkOp>,
178}
179
180impl Default for ChangeBundle {
181 fn default() -> Self {
182 ChangeBundle {
183 delta: Delta { ops: Vec::new() },
184 island_ops: Vec::new(),
185 line_ops: Vec::new(),
186 mark_ops: Vec::new(),
187 }
188 }
189}
190
191impl ChangeBundle {
192 /// A bundle carrying `delta` and no ops: the per-keystroke splice.
193 pub fn from_delta(delta: Delta) -> Self {
194 ChangeBundle {
195 delta,
196 ..Default::default()
197 }
198 }
199
200 fn is_delta_only(&self) -> bool {
201 self.island_ops.is_empty() && self.line_ops.is_empty() && self.mark_ops.is_empty()
202 }
203}
204
205// ── Change-bundle wire (mark / line op ⇄ JSON) ──────────────────────────────
206//
207// [`Delta`] serializes through serde derive; [`MarkOp`] and [`LineOp`] carry
208// [`MarkKind`] / [`LineKind`] / [`Container`], whose canonical JSON is the
209// hand-written `serial` encoding (the `{type, …}` / `{kind, …}` discriminants a
210// `ContentMark` / `ContentLine` already uses). These converters reuse that
211// exact vocabulary so the `applyChange` bundle speaks the same shapes the
212// content read surface does, rather than a second serde-derived dialect. The
213// language bindings call them to lower a JS/Python bundle to core ops.
214
215use crate::serial::{
216 container_from_authored_value, container_to_value, island_from_value, island_to_value,
217 line_kind_from_authored_value, line_kind_to_value, mark_from_authored_value, mark_to_value,
218 usv_from, ParseError,
219};
220use serde_json::{Map, Value};
221
222/// Encode a [`MarkOp`] to its wire object. `Add`/`Remove` carry the mark
223/// vocabulary (`{op, start, end, type, …}`); `RemoveAnchor` is `{op, id}`.
224pub fn mark_op_to_value(op: &MarkOp) -> Value {
225 let mut m = Map::new();
226 match op {
227 MarkOp::Add { start, end, kind } => {
228 m.insert("op".into(), "add".into());
229 merge_mark(&mut m, *start, *end, kind);
230 }
231 MarkOp::Remove { start, end, kind } => {
232 m.insert("op".into(), "remove".into());
233 merge_mark(&mut m, *start, *end, kind);
234 }
235 MarkOp::RemoveAnchor { id } => {
236 m.insert("op".into(), "removeAnchor".into());
237 m.insert("id".into(), Value::String(id.clone()));
238 }
239 }
240 Value::Object(m)
241}
242
243/// Merge a mark's `{start, end, type, …}` fields into an op object, reusing the
244/// canonical `serial` mark encoding.
245fn merge_mark(m: &mut Map<String, Value>, start: Usv, end: Usv, kind: &MarkKind) {
246 let mark = Mark {
247 start,
248 end,
249 kind: kind.clone(),
250 };
251 if let Value::Object(fields) = mark_to_value(&mark) {
252 m.extend(fields);
253 }
254}
255
256/// Decode a [`MarkOp`] from its wire object. Dispatches on `op`; `add`/`remove`
257/// read the mark vocabulary on the authored lane, which refuses `attrs` beside a
258/// built-in `type` rather than resolving to the built-in and dropping them.
259pub fn mark_op_from_value(v: &Value) -> Result<MarkOp, ParseError> {
260 let o = v.as_object().ok_or(ParseError::Shape("mark op"))?;
261 match o.get("op").and_then(Value::as_str) {
262 Some("add") => {
263 let mark = mark_from_authored_value(v)?;
264 Ok(MarkOp::Add {
265 start: mark.start,
266 end: mark.end,
267 kind: mark.kind,
268 })
269 }
270 Some("remove") => {
271 let mark = mark_from_authored_value(v)?;
272 Ok(MarkOp::Remove {
273 start: mark.start,
274 end: mark.end,
275 kind: mark.kind,
276 })
277 }
278 Some("removeAnchor") => Ok(MarkOp::RemoveAnchor {
279 id: o
280 .get("id")
281 .and_then(Value::as_str)
282 .ok_or(ParseError::Shape("removeAnchor id"))?
283 .to_string(),
284 }),
285 _ => Err(ParseError::Shape("mark op kind")),
286 }
287}
288
289/// Encode a [`LineOp`] to its wire object. `SetKind` flattens the line-kind
290/// discriminant (`kind`/`level`/`lang`) alongside `op`/`line`.
291pub fn line_op_to_value(op: &LineOp) -> Value {
292 let mut m = Map::new();
293 match op {
294 LineOp::Split { at } => {
295 m.insert("op".into(), "split".into());
296 m.insert("at".into(), Value::from(*at));
297 }
298 LineOp::Join { line } => {
299 m.insert("op".into(), "join".into());
300 m.insert("line".into(), Value::from(*line));
301 }
302 LineOp::SetKind { line, kind } => {
303 m.insert("op".into(), "setKind".into());
304 m.insert("line".into(), Value::from(*line));
305 if let Value::Object(fields) = line_kind_to_value(kind) {
306 m.extend(fields);
307 }
308 }
309 LineOp::SetContainers { line, containers } => {
310 m.insert("op".into(), "setContainers".into());
311 m.insert("line".into(), Value::from(*line));
312 m.insert(
313 "containers".into(),
314 Value::Array(containers.iter().map(container_to_value).collect()),
315 );
316 }
317 LineOp::SetContinues { line, continues } => {
318 m.insert("op".into(), "setContinues".into());
319 m.insert("line".into(), Value::from(*line));
320 m.insert("continues".into(), Value::Bool(*continues));
321 }
322 }
323 Value::Object(m)
324}
325
326/// Decode a [`LineOp`] from its wire object. Dispatches on `op`.
327pub fn line_op_from_value(v: &Value) -> Result<LineOp, ParseError> {
328 let o = v.as_object().ok_or(ParseError::Shape("line op"))?;
329 let line = || usv_from(o.get("line"), "line op line");
330 match o.get("op").and_then(Value::as_str) {
331 Some("split") => Ok(LineOp::Split {
332 at: usv_from(o.get("at"), "split at")?,
333 }),
334 Some("join") => Ok(LineOp::Join { line: line()? }),
335 Some("setKind") => Ok(LineOp::SetKind {
336 line: line()?,
337 kind: line_kind_from_authored_value(v)?,
338 }),
339 Some("setContainers") => Ok(LineOp::SetContainers {
340 line: line()?,
341 containers: o
342 .get("containers")
343 .and_then(Value::as_array)
344 .ok_or(ParseError::Shape("setContainers containers"))?
345 .iter()
346 .map(container_from_authored_value)
347 .collect::<Result<_, _>>()?,
348 }),
349 Some("setContinues") => Ok(LineOp::SetContinues {
350 line: line()?,
351 continues: o
352 .get("continues")
353 .and_then(Value::as_bool)
354 .ok_or(ParseError::Shape("setContinues continues"))?,
355 }),
356 _ => Err(ParseError::Shape("line op kind")),
357 }
358}
359
360/// Encode an [`IslandOp`] to its wire object. Both arms flatten the island
361/// vocabulary (`{id, type, props, loss}`) alongside `op`, as [`LineOp::SetKind`]
362/// flattens the line-kind discriminant.
363pub fn island_op_to_value(op: &IslandOp) -> Value {
364 let (verb, at, island) = match op {
365 IslandOp::Set { island } => ("set", None, island),
366 IslandOp::Insert { at, island } => ("insert", Some(*at), island),
367 };
368 let mut m = Map::new();
369 m.insert("op".into(), verb.into());
370 if let Some(at) = at {
371 m.insert("at".into(), Value::from(at));
372 }
373 if let Value::Object(fields) = island_to_value(island) {
374 m.extend(fields);
375 }
376 Value::Object(m)
377}
378
379/// Decode an [`IslandOp`] from its wire object. Dispatches on `op`; both arms
380/// read the island vocabulary, so an op carries the same `{id, type, props,
381/// loss}` shape a `ContentIsland` does.
382pub fn island_op_from_value(v: &Value) -> Result<IslandOp, ParseError> {
383 let o = v.as_object().ok_or(ParseError::Shape("island op"))?;
384 let island = || island_from_value(v);
385 match o.get("op").and_then(Value::as_str) {
386 Some("set") => Ok(IslandOp::Set { island: island()? }),
387 Some("insert") => Ok(IslandOp::Insert {
388 at: usv_from(o.get("at"), "island insert at")?,
389 island: island()?,
390 }),
391 _ => Err(ParseError::Shape("island op kind")),
392 }
393}
394
395/// Lower a committed change **bundle** object (`{delta?, islandOps?, lineOps?,
396/// markOps?}`) to core ops: the whole-bundle reader the `applyChange` verb needs,
397/// so each binding lowers a JS/Python bundle in one call instead of re-deriving
398/// the delta/op extraction. A missing `delta` is the identity (no text change); a
399/// missing/`null` op array is empty. Both camelCase (`lineOps`) and snake_case
400/// (`line_ops`) keys are accepted, so the one reader serves the wasm (camelCase)
401/// and Python (either) surfaces. The error is a message string the binding wraps
402/// in its own error type.
403pub fn change_bundle_from_value(v: &Value) -> Result<ChangeBundle, String> {
404 let obj = v
405 .as_object()
406 .ok_or("bundle must be an object { delta?, islandOps?, lineOps?, markOps? }")?;
407 let get = |snake: &str, camel: &str| obj.get(snake).or_else(|| obj.get(camel));
408 let delta = match get("delta", "delta") {
409 Some(Value::Null) | None => Delta { ops: Vec::new() },
410 Some(d) => serde_json::from_value(d.clone()).map_err(|e| format!("invalid delta: {e}"))?,
411 };
412 Ok(ChangeBundle {
413 delta,
414 island_ops: op_array(
415 get("island_ops", "islandOps"),
416 island_op_from_value,
417 "islandOps",
418 )?,
419 line_ops: op_array(get("line_ops", "lineOps"), line_op_from_value, "lineOps")?,
420 mark_ops: op_array(get("mark_ops", "markOps"), mark_op_from_value, "markOps")?,
421 })
422}
423
424/// Lower an optional JSON array of op objects through `convert` (missing/`null`
425/// → empty), naming `what` in any shape-error message. The list twin shared by
426/// [`change_bundle_from_value`]'s line- and mark-op channels.
427fn op_array<T>(
428 value: Option<&Value>,
429 convert: impl Fn(&Value) -> Result<T, ParseError>,
430 what: &str,
431) -> Result<Vec<T>, String> {
432 let Some(value) = value else {
433 return Ok(Vec::new());
434 };
435 if value.is_null() {
436 return Ok(Vec::new());
437 }
438 let arr = value
439 .as_array()
440 .ok_or_else(|| format!("{what} must be an array"))?;
441 arr.iter()
442 .map(|v| convert(v).map_err(|e| format!("invalid {what}: {e}")))
443 .collect()
444}
445
446/// Why an apply failed: range or line index out of bounds, or invariants
447/// broken before normalization could repair them.
448#[derive(Debug, Clone, PartialEq, Eq)]
449#[non_exhaustive]
450pub enum ApplyError {
451 MarkOutOfRange {
452 start: Usv,
453 end: Usv,
454 len: Usv,
455 },
456 LineOutOfRange {
457 line: usize,
458 lines: usize,
459 },
460 SplitPositionOutOfRange {
461 at: Usv,
462 len: Usv,
463 },
464 SplitAtNewline {
465 at: Usv,
466 },
467 LineCountMismatch {
468 lines: usize,
469 segments: usize,
470 },
471 /// A [`LineOp::SetContinues`] tried to set `continues: true` on line 0, which
472 /// has nothing before it to continue, the apply-time twin of the
473 /// [`Invariant::FirstLineContinues`](crate::model::Invariant::FirstLineContinues)
474 /// validation error, refused here because `normalize` does not repair it.
475 FirstLineContinues,
476 /// The text delta's expected base length disagreed with the content:
477 /// it was built against a different revision.
478 DeltaBaseMismatch {
479 expected: usize,
480 actual: usize,
481 },
482 /// An `Op::Insert` carried a raw [`ISLAND_SLOT`]. Islands are structurally
483 /// uneditable through the text channel: a slot inserted here would have no
484 /// backing [`Island`], an orphaned-slot invariant violation. Islands are
485 /// created through [`IslandOp::Insert`], which carries the slot and its
486 /// entry in one op, never a text splice.
487 ///
488 /// A producer that computes one splice over the whole field text carries
489 /// slots in it whenever the user pastes an island or undoes a deletion that
490 /// removed one. Such a splice splits: the delta with its slots stripped,
491 /// plus one [`IslandOp::Insert`] per slot at the frame that delta leaves.
492 IslandSlotInInsert,
493 /// A [`MarkOp::Add`] of an anchor whose `id` is already live in the field.
494 /// An anchor id is a caller-supplied handle, unique per `Content`
495 /// (`DOCUMENT_STORAGE.md` § Anchor-id identity); `add` rejects a collision
496 /// rather than replace (which would silently retarget a live thread) or
497 /// coexist (which `RemoveAnchor` cannot disambiguate). The op-time twin of
498 /// [`Invariant::AnchorIdCollision`](crate::model::Invariant::AnchorIdCollision).
499 AnchorIdCollision { id: String },
500 /// A [`MarkOp::Add`] of an anchor with the empty `id`: a degenerate handle,
501 /// refused so every anchor carries a usable referent.
502 EmptyAnchorId,
503 /// An [`IslandOp::Set`] naming an `id` no island in the field carries,
504 /// refused rather than ignored ([`IslandOp::Set`] states why).
505 UnknownIslandId { id: String },
506 /// An [`IslandOp::Insert`] whose `id` is already live in the field. Island
507 /// ids are unique per `Content` (the
508 /// [`Invariant::IslandIdCollision`](crate::model::Invariant::IslandIdCollision)
509 /// this is the op-time twin of); `Set` addresses by id, so a duplicate is an
510 /// island neither op can name unambiguously.
511 IslandIdCollision { id: String },
512 /// An [`IslandOp::Insert`] carrying the empty `id`: an island `Set` could
513 /// never address, refused on the same terms as [`Self::EmptyAnchorId`].
514 EmptyIslandId,
515 /// An [`IslandOp::Insert`] whose `at` is past the end of the text the delta
516 /// and this bundle's earlier island ops left.
517 IslandInsertOutOfRange { at: Usv, len: Usv },
518 /// A [`LineOp::SetKind`] whose kind contradicts the line's text: tagging
519 /// prose `Island` or `Rule`, or a slot-bearing line `Code`. Export trusts the
520 /// kind over the text, so the write would silently drop the line's content;
521 /// the op-time twin of
522 /// [`Invariant::LineKindMismatch`](crate::model::Invariant::LineKindMismatch),
523 /// refused here because `normalize` does not repair it.
524 LineKindMismatch {
525 line: usize,
526 mismatch: LineKindMismatch,
527 },
528 /// A [`LineOp::SetContainers`] nested a line deeper than
529 /// [`MAX_NESTING_DEPTH`](crate::MAX_NESTING_DEPTH), the op-time twin of
530 /// [`Invariant::NestingTooDeep`](crate::model::Invariant::NestingTooDeep).
531 NestingTooDeep {
532 line: usize,
533 depth: usize,
534 max: usize,
535 },
536}
537
538impl Content {
539 /// Splice `text` via `delta`, rebase marks, sync `lines` to `\n` changes,
540 /// cascade island removal for any deleted slot, then normalize.
541 ///
542 /// Islands stay in lockstep with their [`ISLAND_SLOT`] chars: a delta that
543 /// *deletes* a slot drops the corresponding [`Island`] (the content goes
544 /// away with its slot); a delta that *inserts* a raw slot is rejected
545 /// ([`ApplyError::IslandSlotInInsert`]), islands are created through
546 /// [`IslandOp::Insert`], never a text splice, so a slot arriving here would
547 /// orphan.
548 ///
549 /// Inserted text is sanitized first: `\r` and Unicode bidi controls (the
550 /// chars [`Content::validate`] forbids) are stripped, mirroring the
551 /// normalization `import` applies at the string boundary. The text-delta
552 /// channel is the *other* way text enters the content, so without this an
553 /// insert of `\r` or a bidi control returned `Ok` while leaving a content
554 /// that fails `validate()`.
555 pub fn apply_text_delta(&mut self, delta: &Delta) -> Result<(), ApplyError> {
556 self.apply_text_delta_inner(delta)?;
557 self.normalize();
558 Ok(())
559 }
560
561 /// [`apply_text_delta`](Self::apply_text_delta) without the terminal
562 /// normalize: the stage [`apply_field_change`](Self::apply_field_change)
563 /// runs so a committed bundle canonicalizes once at the end, not after each
564 /// op.
565 fn apply_text_delta_inner(&mut self, delta: &Delta) -> Result<(), ApplyError> {
566 // Reject before mutating: a raw slot in an insert would create a slot
567 // with no backing island. Checked up front so the content is untouched
568 // on this error.
569 for op in &delta.ops {
570 if let Op::Insert(s) = op {
571 if s.contains(ISLAND_SLOT) {
572 return Err(ApplyError::IslandSlotInInsert);
573 }
574 }
575 }
576
577 // Strip the chars `validate()` forbids (`\r`, bidi controls) from every
578 // insert before they reach the content. Stripping (not rejecting)
579 // mirrors `import`: these are content to normalize away, unlike a raw
580 // slot, which has no backing island and must be refused. Sanitizing the
581 // whole delta up front keeps `try_apply` / `map_pos` / line+island sync
582 // in agreement on one cleaned op stream; a clean delta (every keystroke)
583 // is borrowed through untouched, so the hot path skips the clone.
584 let sanitized = sanitize_inserts(delta);
585 let delta = sanitized.as_ref();
586
587 let old_chars: Vec<char> = self.text.chars().collect();
588 let old_lines = self.lines.clone();
589 // A splice may name only the region it changes: `try_apply` retains the
590 // untouched remainder implicitly, so a bare prepend applies against the
591 // whole content. An over-long delta (consuming more base than exists)
592 // still fails the base-length check.
593 let new_text = delta
594 .try_apply(&self.text)
595 .map_err(|e| ApplyError::DeltaBaseMismatch {
596 expected: e.expected,
597 actual: e.actual,
598 })?;
599
600 self.rebase_marks(delta);
601 let new_len = new_text.chars().count();
602 self.marks.retain(|m| {
603 m.start <= m.end
604 && m.end <= new_len
605 && (m.start < m.end || !m.kind.is_formatting())
606 });
607
608 self.text = new_text;
609 self.lines = sync_lines_for_delta(&old_chars, old_lines, delta);
610 let old_islands = std::mem::take(&mut self.islands);
611 self.islands = sync_islands_for_delta(&old_chars, old_islands, delta);
612 if self.lines.len() != self.segment_count() {
613 return Err(ApplyError::LineCountMismatch {
614 lines: self.lines.len(),
615 segments: self.segment_count(),
616 });
617 }
618 Ok(())
619 }
620
621 /// Rebase every mark's range through `delta`'s
622 /// [`map_pos`](crate::delta::Delta::map_pos): a range mark's start biases
623 /// `After` and its end `Before` (an insertion at either edge grows text
624 /// *outside* the span), a point (zero-width) mark biases `Before`. The one
625 /// mapping the text-delta channel and line split/join both rebase marks by.
626 fn rebase_marks(&mut self, delta: &Delta) {
627 for m in &mut self.marks {
628 if m.start == m.end {
629 let p = delta.map_pos(m.start, Assoc::Before);
630 m.start = p;
631 m.end = p;
632 } else {
633 m.start = delta.map_pos(m.start, Assoc::After);
634 m.end = delta.map_pos(m.end, Assoc::Before);
635 }
636 }
637 }
638
639 /// Apply mark ops in final-text coordinates, then normalize.
640 pub fn apply_mark_ops(&mut self, ops: &[MarkOp]) -> Result<(), ApplyError> {
641 self.apply_mark_ops_inner(ops)?;
642 self.normalize();
643 Ok(())
644 }
645
646 /// [`apply_mark_ops`](Self::apply_mark_ops) without the terminal normalize:
647 /// the bundle's final stage, canonicalized once by
648 /// [`apply_field_change`](Self::apply_field_change).
649 fn apply_mark_ops_inner(&mut self, ops: &[MarkOp]) -> Result<(), ApplyError> {
650 let len = self.len_usv();
651 for op in ops {
652 match op {
653 MarkOp::Add { start, end, kind } => {
654 if *start > *end || *end > len {
655 return Err(ApplyError::MarkOutOfRange {
656 start: *start,
657 end: *end,
658 len,
659 });
660 }
661 if kind.is_formatting() && start == end {
662 return Err(ApplyError::MarkOutOfRange {
663 start: *start,
664 end: *end,
665 len,
666 });
667 }
668 // Anchor id: caller-supplied, unique per `Content`, non-empty
669 // (`DOCUMENT_STORAGE.md` § Anchor-id identity). Reject a live
670 // collision (`RemoveAnchor` cannot tell two same-id anchors
671 // apart) and the empty degenerate handle. Ops apply in
672 // sequence, so a `RemoveAnchor` earlier in the bundle frees
673 // the id for re-add here.
674 if let MarkKind::Anchor { id } = kind {
675 if id.is_empty() {
676 return Err(ApplyError::EmptyAnchorId);
677 }
678 if self
679 .marks
680 .iter()
681 .any(|m| matches!(&m.kind, MarkKind::Anchor { id: aid } if aid == id))
682 {
683 return Err(ApplyError::AnchorIdCollision { id: id.clone() });
684 }
685 }
686 self.marks.push(Mark {
687 start: *start,
688 end: *end,
689 kind: kind.clone(),
690 });
691 }
692 MarkOp::Remove { start, end, kind } => {
693 if *start > *end || *end > len {
694 return Err(ApplyError::MarkOutOfRange {
695 start: *start,
696 end: *end,
697 len,
698 });
699 }
700 let mut next = Vec::with_capacity(self.marks.len());
701 for m in self.marks.drain(..) {
702 // Untouched: a different kind, or no overlap with the
703 // removed range.
704 if m.kind != *kind || !ranges_overlap(m.start, m.end, *start, *end) {
705 next.push(m);
706 continue;
707 }
708 // Identity/unknown handles have no range algebra to
709 // subtract: drop the overlapping one whole.
710 if !kind.is_formatting() {
711 continue;
712 }
713 // Formatting: subtract [start, end), re-emitting the
714 // surviving fragments. An edge-aligned removal yields a
715 // zero-width fragment here; `normalize` drops it.
716 if m.start < *start {
717 next.push(Mark {
718 start: m.start,
719 end: *start,
720 kind: m.kind.clone(),
721 });
722 }
723 if *end < m.end {
724 next.push(Mark {
725 start: *end,
726 end: m.end,
727 kind: m.kind.clone(),
728 });
729 }
730 }
731 self.marks = next;
732 }
733 MarkOp::RemoveAnchor { id } => {
734 self.marks
735 .retain(|m| !matches!(&m.kind, MarkKind::Anchor { id: aid } if aid == id));
736 }
737 }
738 }
739 Ok(())
740 }
741
742 /// Apply island ops: replace an entry by id, or insert a slot and its entry
743 /// together.
744 pub fn apply_island_ops(&mut self, ops: &[IslandOp]) -> Result<(), ApplyError> {
745 self.apply_island_ops_inner(ops)?;
746 self.normalize();
747 Ok(())
748 }
749
750 /// [`apply_island_ops`](Self::apply_island_ops) without the terminal
751 /// normalize: a bundle stage canonicalized once by
752 /// [`apply_field_change`](Self::apply_field_change).
753 fn apply_island_ops_inner(&mut self, ops: &[IslandOp]) -> Result<(), ApplyError> {
754 for op in ops {
755 match op {
756 IslandOp::Set { island } => {
757 let idx = self
758 .islands
759 .iter()
760 .position(|i| i.id == island.id)
761 .ok_or_else(|| ApplyError::UnknownIslandId {
762 id: island.id.clone(),
763 })?;
764 // In place: the entry's slot-order index is its slot's, and
765 // the slot does not move. Nothing here touches `text` or
766 // `marks`: an island edit costs no anchors.
767 self.islands[idx] = island.clone();
768 }
769 IslandOp::Insert { at, island } => {
770 // Refuse before the write, as the mark channel does for an
771 // anchor id: an empty or colliding id is an island `Set`
772 // cannot address. Ops apply in sequence, so an earlier
773 // delete in the same bundle frees the id for reuse here.
774 if island.id.is_empty() {
775 return Err(ApplyError::EmptyIslandId);
776 }
777 if self.islands.iter().any(|i| i.id == island.id) {
778 return Err(ApplyError::IslandIdCollision {
779 id: island.id.clone(),
780 });
781 }
782 let chars: Vec<char> = self.text.chars().collect();
783 if *at > chars.len() {
784 return Err(ApplyError::IslandInsertOutOfRange {
785 at: *at,
786 len: chars.len(),
787 });
788 }
789 // Islands are stored in slot order, so the entry's index is
790 // the count of slots before `at`.
791 let slot_idx = chars[..*at].iter().filter(|&&c| c == ISLAND_SLOT).count();
792 let byte = char_to_byte(&self.text, *at);
793 self.text.insert(byte, ISLAND_SLOT);
794 // Rebase marks through the one-char insertion, as the text
795 // channel and line split both do: an anchor after the new
796 // island tracks the splice instead of drifting.
797 self.rebase_marks(&Delta {
798 ops: vec![Op::Retain(*at), Op::Insert(ISLAND_SLOT.to_string())],
799 });
800 self.islands.insert(slot_idx, island.clone());
801 // A slot is not a `\n`: the segment count, and so the line
802 // list, is unchanged.
803 }
804 }
805 }
806 Ok(())
807 }
808
809 /// Apply line ops: split/join splice `\n`; set ops touch metadata only.
810 pub fn apply_line_ops(&mut self, ops: &[LineOp]) -> Result<(), ApplyError> {
811 self.apply_line_ops_inner(ops)?;
812 self.normalize();
813 Ok(())
814 }
815
816 /// [`apply_line_ops`](Self::apply_line_ops) without the terminal normalize:
817 /// a bundle stage canonicalized once by
818 /// [`apply_field_change`](Self::apply_field_change).
819 fn apply_line_ops_inner(&mut self, ops: &[LineOp]) -> Result<(), ApplyError> {
820 for op in ops {
821 match op {
822 LineOp::Split { at } => self.split_line(*at)?,
823 LineOp::Join { line } => self.join_line(*line)?,
824 LineOp::SetKind { line, kind } => {
825 // The kind must agree with the text already on the line:
826 // export reads the kind and never the segment, so an
827 // `Island`/`Rule` tag over prose projects the text away.
828 // Checked before the write (line ops stage on a scratch copy,
829 // so an error leaves the content untouched).
830 let seg = self
831 .text
832 .split('\n')
833 .nth(*line)
834 .ok_or(ApplyError::LineOutOfRange {
835 line: *line,
836 lines: self.lines.len(),
837 })?;
838 if let Some(mismatch) = line_kind_mismatch(kind, seg) {
839 return Err(ApplyError::LineKindMismatch {
840 line: *line,
841 mismatch,
842 });
843 }
844 let line = self.line_mut(*line)?;
845 line.kind = kind.clone();
846 }
847 LineOp::SetContainers { line, containers } => {
848 // Both emitters recurse one frame per container, so an
849 // over-deep path is a stack overflow at render, not a render
850 // error. Same cap as import, refused before the write.
851 if containers.len() > crate::MAX_NESTING_DEPTH {
852 return Err(ApplyError::NestingTooDeep {
853 line: *line,
854 depth: containers.len(),
855 max: crate::MAX_NESTING_DEPTH,
856 });
857 }
858 let line = self.line_mut(*line)?;
859 line.containers = containers.clone();
860 }
861 LineOp::SetContinues { line, continues } => {
862 // Line 0 has nothing before it to continue: setting the flag
863 // there would forge the `FirstLineContinues` invariant that
864 // `normalize` does not repair. Reject before the write so the
865 // content stays valid (`apply_field_change` stages line ops on
866 // a scratch copy, so this leaves `self` untouched).
867 if *line == 0 && *continues {
868 return Err(ApplyError::FirstLineContinues);
869 }
870 let l = self.line_mut(*line)?;
871 l.continues = *continues;
872 }
873 }
874 }
875 Ok(())
876 }
877
878 /// One committed field edit bundle: text delta, then island ops, then line
879 /// ops, then marks, canonicalized by a single terminal
880 /// [`normalize`](Self::normalize).
881 ///
882 /// All-or-nothing: on any op's error `self` is left exactly as it was, so a
883 /// caller need not snapshot-and-restore around a failed bundle. A bundle
884 /// carrying ops has several fallible stages that would otherwise partially
885 /// commit, so it is staged on a scratch copy and swapped in only once every
886 /// stage succeeds. The pure-text-delta path (the per-keystroke hot path)
887 /// skips the clone: `apply_text_delta` validates the delta before mutating,
888 /// so it is already atomic on the errors a caller can provoke.
889 ///
890 /// **Stage order is a coordinate contract**, not a convenience: each stage
891 /// reads the text the earlier ones left. Island ops sit between the delta
892 /// and the line ops because both neighbors need them there. An island insert
893 /// splices a slot, so a `LineOp::SetKind { kind: Island }` in the same bundle
894 /// can only validate against a line that already carries it; `Split`/`Join`
895 /// and every mark range are then measured in a frame that includes the new
896 /// slots. The one-bundle block island ([`IslandOp::Insert`]) follows from
897 /// that.
898 ///
899 /// The stages run on their non-normalizing inner forms and `normalize` runs
900 /// once at the end. One terminal normalize suffices because split/join
901 /// rebase marks through their `\n` splice
902 /// ([`map_pos`](crate::delta::Delta::map_pos) semantics): the
903 /// formatting-edge `\n`-trim then commutes with the line ops (trim-per-stage
904 /// and trim-once converge), and `MarkOp::Remove` is coverage-set
905 /// subtraction, which commutes with `normalize`'s same-kind union
906 /// (`(A ∪ B) \ R = (A\R) ∪ (B\R)`). One canonicalization point, one pass.
907 pub fn apply_field_change(&mut self, bundle: &ChangeBundle) -> Result<(), ApplyError> {
908 if bundle.is_delta_only() {
909 return self.apply_text_delta(&bundle.delta);
910 }
911 let mut scratch = self.clone();
912 scratch.apply_text_delta_inner(&bundle.delta)?;
913 scratch.apply_island_ops_inner(&bundle.island_ops)?;
914 scratch.apply_line_ops_inner(&bundle.line_ops)?;
915 scratch.apply_mark_ops_inner(&bundle.mark_ops)?;
916 scratch.normalize();
917 *self = scratch;
918 Ok(())
919 }
920
921 fn line_mut(&mut self, line: usize) -> Result<&mut Line, ApplyError> {
922 let lines = self.lines.len();
923 self.lines
924 .get_mut(line)
925 .ok_or(ApplyError::LineOutOfRange { line, lines })
926 }
927
928 fn split_line(&mut self, at: Usv) -> Result<(), ApplyError> {
929 let char_indices: Vec<(usize, char)> = self.text.char_indices().collect();
930 let len = char_indices.len();
931 if at > len {
932 return Err(ApplyError::SplitPositionOutOfRange { at, len });
933 }
934 if at > 0 && char_indices[at - 1].1 == '\n' {
935 return Err(ApplyError::SplitAtNewline { at });
936 }
937 if at < len && char_indices[at].1 == '\n' {
938 return Err(ApplyError::SplitAtNewline { at });
939 }
940
941 // `at`'s newline-adjacency neighbors, `at`'s byte offset, and the
942 // newline count before `at` (== the post-insert line index, since the
943 // insertion lands at index `at`, not before it) all come from this
944 // one pass over `char_indices`, instead of four separate text scans.
945 let byte = char_indices.get(at).map_or(self.text.len(), |&(b, _)| b);
946 let line_idx = char_indices[..at].iter().filter(|&(_, c)| *c == '\n').count();
947 self.text.insert(byte, '\n');
948
949 // Rebase marks through the one-char `\n` insertion: the same map_pos
950 // rule the text-delta channel uses, so a split does not drift a mark's
951 // coordinates (a mark spanning `at` grows by the inserted char; the
952 // terminal normalize trims any `\n` edge it lands on).
953 self.rebase_marks(&Delta {
954 ops: vec![Op::Retain(at), Op::Insert("\n".to_string())],
955 });
956
957 let template = self
958 .lines
959 .get(line_idx)
960 .cloned()
961 .unwrap_or_else(default_para_line);
962 let mut new_line = template;
963 new_line.continues = false;
964 self.lines.insert(line_idx + 1, new_line);
965
966 if self.lines.len() != self.segment_count() {
967 return Err(ApplyError::LineCountMismatch {
968 lines: self.lines.len(),
969 segments: self.segment_count(),
970 });
971 }
972 Ok(())
973 }
974
975 fn join_line(&mut self, line: usize) -> Result<(), ApplyError> {
976 if line + 1 >= self.lines.len() {
977 return Err(ApplyError::LineOutOfRange {
978 line,
979 lines: self.lines.len(),
980 });
981 }
982 let nl = newline_at_line_boundary(&self.text, line)?;
983 let byte = char_to_byte(&self.text, nl);
984 self.text.remove(byte);
985
986 // Rebase marks through the one-char `\n` deletion, as the text-delta
987 // channel would: a mark spanning the boundary shrinks by one; one that
988 // covered only the `\n` collapses to zero-width and the terminal
989 // normalize drops it.
990 self.rebase_marks(&Delta {
991 ops: vec![Op::Retain(nl), Op::Delete(1)],
992 });
993
994 self.lines.remove(line + 1);
995
996 if self.lines.len() != self.segment_count() {
997 return Err(ApplyError::LineCountMismatch {
998 lines: self.lines.len(),
999 segments: self.segment_count(),
1000 });
1001 }
1002 Ok(())
1003 }
1004}
1005
1006fn default_para_line() -> Line {
1007 Line {
1008 kind: LineKind::Para,
1009 containers: Vec::new(),
1010 continues: false,
1011 }
1012}
1013
1014fn ranges_overlap(a0: Usv, a1: Usv, b0: Usv, b1: Usv) -> bool {
1015 a0 < b1 && b0 < a1
1016}
1017
1018/// A char the content text may not carry (`validate()` rejects it): a bare `\r`
1019/// or a Unicode bidi formatting control. `\n` is a real line boundary and a raw
1020/// [`ISLAND_SLOT`] is refused separately, so neither belongs here.
1021fn insert_forbidden(c: char) -> bool {
1022 c == '\r' || is_bidi_char(c)
1023}
1024
1025/// Drop [`insert_forbidden`] chars from every `Op::Insert`, returning the delta
1026/// borrowed untouched when no insert carries one (the common keystroke). Mirrors
1027/// the forbidden-char stripping `import` applies (`push_text`, `strip_bidi_
1028/// formatting`); a raw `\r`/bidi arriving through the text-delta channel would
1029/// otherwise persist a content that fails `validate()`.
1030fn sanitize_inserts(delta: &Delta) -> Cow<'_, Delta> {
1031 let needs_cleaning = delta
1032 .ops
1033 .iter()
1034 .any(|op| matches!(op, Op::Insert(s) if s.chars().any(insert_forbidden)));
1035 if !needs_cleaning {
1036 return Cow::Borrowed(delta);
1037 }
1038 let ops = delta
1039 .ops
1040 .iter()
1041 .map(|op| match op {
1042 Op::Insert(s) => Op::Insert(s.chars().filter(|c| !insert_forbidden(*c)).collect()),
1043 other => other.clone(),
1044 })
1045 .collect();
1046 Cow::Owned(Delta { ops })
1047}
1048
1049/// Walk `delta` over `old_chars` and mirror `\n` insert/delete in `lines`,
1050/// building the result in one forward pass: O(old_chars walked + inserts),
1051/// no per-`\n` mid-`Vec` `remove`/`insert`.
1052///
1053/// The cursor sits *in* a line, `cur`; downstream of it is always the untouched
1054/// original suffix (`rest`), because a split lands its clone right at the cursor
1055/// and a delete drops the next original. So the three `\n` events reduce to:
1056/// a retained `\n` finalizes `cur` and pulls the next original into it; a
1057/// deleted `\n` drops the next original (merging it in), when one exists; an
1058/// inserted `\n` finalizes `cur` and makes a clone (its `continues` cleared) the
1059/// new `cur`. `cur == None` is the past-the-end state on a malformed content
1060/// (more `\n` than lines), where a split clones a default line.
1061fn sync_lines_for_delta(old_chars: &[char], old_lines: Vec<Line>, delta: &Delta) -> Vec<Line> {
1062 let cap = old_lines.len();
1063 let mut rest = old_lines.into_iter();
1064 let mut out: Vec<Line> = Vec::with_capacity(cap);
1065 let mut cur: Option<Line> = rest.next();
1066 let mut old = 0usize;
1067
1068 for op in &delta.ops {
1069 match op {
1070 Op::Retain(n) => {
1071 for _ in 0..*n {
1072 if old >= old_chars.len() {
1073 break;
1074 }
1075 if old_chars[old] == '\n' {
1076 out.extend(cur.take());
1077 cur = rest.next();
1078 }
1079 old += 1;
1080 }
1081 }
1082 Op::Delete(n) => {
1083 for _ in 0..*n {
1084 if old >= old_chars.len() {
1085 break;
1086 }
1087 // A deleted '\n' merges the next original into `cur`: drop
1088 // it. With no next original there is nothing to drop.
1089 if old_chars[old] == '\n' {
1090 rest.next();
1091 }
1092 old += 1;
1093 }
1094 }
1095 Op::Insert(s) => {
1096 for c in s.chars() {
1097 if c == '\n' {
1098 let mut new_line = match cur.take() {
1099 Some(line) => {
1100 let clone = line.clone();
1101 out.push(line);
1102 clone
1103 }
1104 None => default_para_line(),
1105 };
1106 new_line.continues = false;
1107 cur = Some(new_line);
1108 }
1109 }
1110 }
1111 }
1112 }
1113
1114 out.extend(cur);
1115 out.extend(rest);
1116 out
1117}
1118
1119/// Walk `delta` over `old_chars` and drop any island whose [`ISLAND_SLOT`] char
1120/// was deleted (cascade removal: the island's content goes away with its slot).
1121/// Islands are stored in slot order, so the Nth slot backs the Nth island; a
1122/// deleted slot drops its island and the survivors renumber implicitly. Raw
1123/// slot *inserts* are rejected upstream, so an insert never mints a new slot.
1124fn sync_islands_for_delta(
1125 old_chars: &[char],
1126 old_islands: Vec<Island>,
1127 delta: &Delta,
1128) -> Vec<Island> {
1129 let mut keep = vec![true; old_islands.len()];
1130 let mut old = 0usize;
1131 let mut slot_idx = 0usize;
1132
1133 for op in &delta.ops {
1134 match op {
1135 Op::Retain(n) => {
1136 for _ in 0..*n {
1137 if old >= old_chars.len() {
1138 break;
1139 }
1140 if old_chars[old] == ISLAND_SLOT {
1141 slot_idx += 1;
1142 }
1143 old += 1;
1144 }
1145 }
1146 Op::Delete(n) => {
1147 for _ in 0..*n {
1148 if old >= old_chars.len() {
1149 break;
1150 }
1151 if old_chars[old] == ISLAND_SLOT {
1152 if let Some(k) = keep.get_mut(slot_idx) {
1153 *k = false;
1154 }
1155 slot_idx += 1;
1156 }
1157 old += 1;
1158 }
1159 }
1160 // Inserts add no slots (a raw ISLAND_SLOT insert is rejected before
1161 // this walk), so they never touch the island list.
1162 Op::Insert(_) => {}
1163 }
1164 }
1165
1166 old_islands
1167 .into_iter()
1168 .zip(keep)
1169 .filter_map(|(island, keep)| keep.then_some(island))
1170 .collect()
1171}
1172
1173fn newline_at_line_boundary(text: &str, line: usize) -> Result<Usv, ApplyError> {
1174 let mut current = 0usize;
1175 for (i, c) in text.chars().enumerate() {
1176 if c == '\n' {
1177 if current == line {
1178 return Ok(i);
1179 }
1180 current += 1;
1181 }
1182 }
1183 Err(ApplyError::LineOutOfRange {
1184 line,
1185 lines: text.chars().filter(|&c| c == '\n').count() + 1,
1186 })
1187}
1188
1189#[cfg(test)]
1190mod tests {
1191 use super::*;
1192 use crate::delta::diff;
1193 use crate::import::from_markdown;
1194
1195 #[test]
1196 fn mark_op_wire_round_trips_each_variant() {
1197 let ops = vec![
1198 MarkOp::Add {
1199 start: 0,
1200 end: 3,
1201 kind: MarkKind::Strong,
1202 },
1203 MarkOp::Add {
1204 start: 1,
1205 end: 2,
1206 kind: MarkKind::Link {
1207 url: "https://x".into(),
1208 },
1209 },
1210 MarkOp::Remove {
1211 start: 4,
1212 end: 6,
1213 kind: MarkKind::Anchor { id: "c1".into() },
1214 },
1215 MarkOp::RemoveAnchor { id: "c2".into() },
1216 ];
1217 for op in ops {
1218 let v = mark_op_to_value(&op);
1219 assert_eq!(mark_op_from_value(&v).unwrap(), op, "round-trip: {v}");
1220 }
1221 }
1222
1223 #[test]
1224 fn line_op_wire_round_trips_each_variant() {
1225 let ops = vec![
1226 LineOp::Split { at: 5 },
1227 LineOp::Join { line: 1 },
1228 LineOp::SetKind {
1229 line: 0,
1230 kind: LineKind::Heading { level: 2 },
1231 },
1232 LineOp::SetContainers {
1233 line: 2,
1234 containers: vec![Container::Quote],
1235 },
1236 // The open block vocabulary rides the same op wire:
1237 // a host can set a role or a container this build does not know.
1238 LineOp::SetKind {
1239 line: 0,
1240 kind: LineKind::Unknown {
1241 tag: "callout".into(),
1242 attrs: serde_json::json!({"variant": "warn"}),
1243 },
1244 },
1245 LineOp::SetContainers {
1246 line: 2,
1247 containers: vec![Container::Unknown {
1248 tag: "indent".into(),
1249 attrs: serde_json::json!({"depth": 2}),
1250 }],
1251 },
1252 LineOp::SetContinues {
1253 line: 1,
1254 continues: true,
1255 },
1256 LineOp::SetContinues {
1257 line: 3,
1258 continues: false,
1259 },
1260 ];
1261 for op in ops {
1262 let v = line_op_to_value(&op);
1263 assert_eq!(line_op_from_value(&v).unwrap(), op, "round-trip: {v}");
1264 }
1265 }
1266
1267 /// On the op lane, `attrs` beside a built-in discriminator is a
1268 /// shape error. A host that emits one classified a built-in as unknown
1269 /// (stale copy of the built-in list) and the lenient reader would resolve the
1270 /// name and drop the payload unread, corrupting the line with no diagnostic.
1271 #[test]
1272 fn op_wire_rejects_attrs_beside_a_built_in_name() {
1273 let bad = serde_json::json!({
1274 "op": "setKind", "line": 0, "kind": "para", "attrs": {"tone": "warn"},
1275 });
1276 assert!(matches!(line_op_from_value(&bad), Err(ParseError::Shape(_))));
1277 let bad = serde_json::json!({
1278 "op": "setContainers", "line": 0,
1279 "containers": [{"container": "quote", "attrs": {"k": 1}}],
1280 });
1281 assert!(matches!(line_op_from_value(&bad), Err(ParseError::Shape(_))));
1282 let bad = serde_json::json!({
1283 "op": "add", "start": 0, "end": 1, "type": "strong", "attrs": {"k": 1},
1284 });
1285 assert!(matches!(mark_op_from_value(&bad), Err(ParseError::Shape(_))));
1286
1287 // An unknown name keeps carrying `attrs` (the rule is reserved-name
1288 // reuse, not `attrs` itself) and a built-in without `attrs` is untouched.
1289 for ok in [
1290 serde_json::json!({"op": "setKind", "line": 0, "kind": "callout", "attrs": {"tone": "warn"}}),
1291 serde_json::json!({"op": "setKind", "line": 0, "kind": "heading", "level": 2}),
1292 ] {
1293 assert!(line_op_from_value(&ok).is_ok(), "rejected: {ok}");
1294 }
1295 }
1296
1297 #[test]
1298 fn delta_serde_shape() {
1299 let d = Delta {
1300 ops: vec![Op::Retain(2), Op::Insert("hi".into()), Op::Delete(1)],
1301 };
1302 let v = serde_json::to_value(&d).unwrap();
1303 assert_eq!(
1304 v,
1305 serde_json::json!({"ops": [{"retain": 2}, {"insert": "hi"}, {"delete": 1}]})
1306 );
1307 assert_eq!(serde_json::from_value::<Delta>(v).unwrap(), d);
1308 }
1309
1310 #[test]
1311 fn apply_text_delta_rebases_marks() {
1312 let mut rt = from_markdown("hello").unwrap();
1313 rt.marks.push(Mark {
1314 start: 1,
1315 end: 4,
1316 kind: MarkKind::Strong,
1317 });
1318 rt.normalize();
1319 let d = diff("hello", "hXello");
1320 rt.apply_text_delta(&d).unwrap();
1321 let strong = rt
1322 .marks
1323 .iter()
1324 .find(|m| matches!(m.kind, MarkKind::Strong))
1325 .unwrap();
1326 assert_eq!((strong.start, strong.end), (2, 5));
1327 assert_eq!(rt.text, "hXello");
1328 }
1329
1330 #[test]
1331 fn apply_text_delta_pads_short_prepend() {
1332 // A bare prepend names only its inserted text (no trailing retain); it
1333 // still splices against the whole content rather than failing the base
1334 // check (regression for the per-field delta path).
1335 let mut rt = from_markdown("hello").unwrap();
1336 rt.apply_text_delta(&Delta {
1337 ops: vec![Op::Insert("NEW ".into())],
1338 })
1339 .unwrap();
1340 assert_eq!(rt.text, "NEW hello");
1341 }
1342
1343 #[test]
1344 fn apply_text_delta_rejects_over_long_delta() {
1345 // Consuming more base than exists is a wrong-revision delta, not an
1346 // abbreviated one: it still fails closed.
1347 let mut rt = from_markdown("hi").unwrap();
1348 assert!(matches!(
1349 rt.apply_text_delta(&Delta {
1350 ops: vec![Op::Retain(99)],
1351 }),
1352 Err(ApplyError::DeltaBaseMismatch { .. })
1353 ));
1354 assert_eq!(rt.text, "hi");
1355 }
1356
1357 #[test]
1358 fn apply_mark_ops_add_and_remove() {
1359 let mut rt = from_markdown("abcd").unwrap();
1360 rt.apply_mark_ops(&[MarkOp::Add {
1361 start: 0,
1362 end: 2,
1363 kind: MarkKind::Emph,
1364 }])
1365 .unwrap();
1366 assert!(rt.marks.iter().any(|m| matches!(m.kind, MarkKind::Emph)));
1367 rt.apply_mark_ops(&[MarkOp::Remove {
1368 start: 0,
1369 end: 4,
1370 kind: MarkKind::Emph,
1371 }])
1372 .unwrap();
1373 assert!(!rt.marks.iter().any(|m| matches!(m.kind, MarkKind::Emph)));
1374 }
1375
1376 #[test]
1377 fn apply_mark_ops_remove_punches_hole() {
1378 // Un-formatting the middle of a run leaves the two non-overlapping
1379 // fragments, not an empty mark set. Strong[0,6) over
1380 // "abcdef", Remove[2,4) -> Strong[0,2) + Strong[4,6).
1381 let mut rt = from_markdown("abcdef").unwrap();
1382 rt.apply_mark_ops(&[MarkOp::Add {
1383 start: 0,
1384 end: 6,
1385 kind: MarkKind::Strong,
1386 }])
1387 .unwrap();
1388 rt.apply_mark_ops(&[MarkOp::Remove {
1389 start: 2,
1390 end: 4,
1391 kind: MarkKind::Strong,
1392 }])
1393 .unwrap();
1394 let strong: Vec<_> = rt
1395 .marks
1396 .iter()
1397 .filter(|m| matches!(m.kind, MarkKind::Strong))
1398 .map(|m| (m.start, m.end))
1399 .collect();
1400 assert_eq!(strong, vec![(0, 2), (4, 6)]);
1401 }
1402
1403 #[test]
1404 fn apply_mark_ops_remove_at_edge_leaves_no_zero_width() {
1405 // A removal flush against the mark's start yields a zero-width left
1406 // fragment [0,0); normalize drops it, leaving only the right fragment.
1407 let mut rt = from_markdown("abcdef").unwrap();
1408 rt.apply_mark_ops(&[MarkOp::Add {
1409 start: 0,
1410 end: 6,
1411 kind: MarkKind::Strong,
1412 }])
1413 .unwrap();
1414 rt.apply_mark_ops(&[MarkOp::Remove {
1415 start: 0,
1416 end: 2,
1417 kind: MarkKind::Strong,
1418 }])
1419 .unwrap();
1420 let strong: Vec<_> = rt
1421 .marks
1422 .iter()
1423 .filter(|m| matches!(m.kind, MarkKind::Strong))
1424 .map(|m| (m.start, m.end))
1425 .collect();
1426 assert_eq!(strong, vec![(2, 6)]);
1427 }
1428
1429 #[test]
1430 fn apply_mark_ops_remove_covering_range_drops_mark() {
1431 // A removal that fully covers the mark leaves nothing (both fragments
1432 // zero-width or inverted): the whole-drop case still holds.
1433 let mut rt = from_markdown("abcdef").unwrap();
1434 rt.apply_mark_ops(&[MarkOp::Add {
1435 start: 2,
1436 end: 4,
1437 kind: MarkKind::Emph,
1438 }])
1439 .unwrap();
1440 rt.apply_mark_ops(&[MarkOp::Remove {
1441 start: 0,
1442 end: 6,
1443 kind: MarkKind::Emph,
1444 }])
1445 .unwrap();
1446 assert!(!rt.marks.iter().any(|m| matches!(m.kind, MarkKind::Emph)));
1447 }
1448
1449 #[test]
1450 fn apply_mark_ops_remove_non_formatting_drops_whole() {
1451 // Identity/unknown handles can't be range-fragmented: an overlapping
1452 // one is dropped whole, never split into fragments.
1453 let mut rt = from_markdown("abcdef").unwrap();
1454 rt.marks.push(Mark {
1455 start: 0,
1456 end: 6,
1457 kind: MarkKind::Unknown {
1458 tag: "x".into(),
1459 attrs: serde_json::json!({}),
1460 },
1461 });
1462 rt.normalize();
1463 rt.apply_mark_ops(&[MarkOp::Remove {
1464 start: 2,
1465 end: 4,
1466 kind: MarkKind::Unknown {
1467 tag: "x".into(),
1468 attrs: serde_json::json!({}),
1469 },
1470 }])
1471 .unwrap();
1472 assert!(!rt
1473 .marks
1474 .iter()
1475 .any(|m| matches!(m.kind, MarkKind::Unknown { .. })));
1476 }
1477
1478 #[test]
1479 fn apply_text_delta_splits_lines_on_newline_insert() {
1480 let mut rt = from_markdown("one two").unwrap();
1481 let d = diff("one two", "one\ntwo");
1482 rt.apply_text_delta(&d).unwrap();
1483 assert_eq!(rt.lines.len(), 2);
1484 assert_eq!(rt.segment_count(), 2);
1485 assert_eq!(rt.validate(), Ok(()));
1486 }
1487
1488 #[test]
1489 fn line_op_split_and_join() {
1490 let mut rt = from_markdown("onetwo").unwrap();
1491 rt.apply_line_ops(&[LineOp::Split { at: 3 }]).unwrap();
1492 assert_eq!(rt.text, "one\ntwo");
1493 assert_eq!(rt.lines.len(), 2);
1494
1495 rt.apply_line_ops(&[LineOp::Join { line: 0 }]).unwrap();
1496 assert_eq!(rt.text, "onetwo");
1497 assert_eq!(rt.lines.len(), 1);
1498 assert_eq!(rt.validate(), Ok(()));
1499 }
1500
1501 #[test]
1502 fn line_op_set_kind() {
1503 let mut rt = from_markdown("title").unwrap();
1504 rt.apply_line_ops(&[LineOp::SetKind {
1505 line: 0,
1506 kind: LineKind::Heading { level: 2 },
1507 }])
1508 .unwrap();
1509 assert!(matches!(rt.lines[0].kind, LineKind::Heading { level: 2 }));
1510 }
1511
1512 /// `SetKind` may not tag a line with a kind its text
1513 /// contradicts: export reads the kind and not the segment, so the write
1514 /// would project the line's content away. Refused before the write, so the
1515 /// content is untouched.
1516 #[test]
1517 fn line_op_set_kind_refuses_a_kind_the_text_contradicts() {
1518 let mut rt = from_markdown("hello world").unwrap();
1519 assert_eq!(
1520 rt.apply_line_ops(&[LineOp::SetKind {
1521 line: 0,
1522 kind: LineKind::Island,
1523 }]),
1524 Err(ApplyError::LineKindMismatch {
1525 line: 0,
1526 mismatch: LineKindMismatch::IslandNotOneSlot,
1527 })
1528 );
1529 assert_eq!(
1530 rt.apply_line_ops(&[LineOp::SetKind {
1531 line: 0,
1532 kind: LineKind::Rule,
1533 }]),
1534 Err(ApplyError::LineKindMismatch {
1535 line: 0,
1536 mismatch: LineKindMismatch::RuleNotEmpty,
1537 })
1538 );
1539 assert_eq!(rt.text, "hello world");
1540 assert_eq!(rt.lines[0].kind, LineKind::Para);
1541 assert_eq!(rt.validate(), Ok(()));
1542
1543 // A table island's line tagged `Code` would fence the slot, which
1544 // re-imports as nothing.
1545 let mut tbl = from_markdown("| a | b |\n|---|---|\n| 1 | 2 |").unwrap();
1546 assert_eq!(
1547 tbl.apply_line_ops(&[LineOp::SetKind {
1548 line: 0,
1549 kind: LineKind::Code { lang: None },
1550 }]),
1551 Err(ApplyError::LineKindMismatch {
1552 line: 0,
1553 mismatch: LineKindMismatch::CodeHasSlot,
1554 })
1555 );
1556 assert_eq!(tbl.lines[0].kind, LineKind::Island);
1557 }
1558
1559 /// `SetContainers` is capped at the depth both emitters can
1560 /// recurse: the op-time twin of the `validate` invariant.
1561 #[test]
1562 fn line_op_set_containers_is_depth_capped() {
1563 let mut rt = from_markdown("hi").unwrap();
1564 let deep = vec![Container::Quote; crate::MAX_NESTING_DEPTH + 1];
1565 assert_eq!(
1566 rt.apply_line_ops(&[LineOp::SetContainers {
1567 line: 0,
1568 containers: deep,
1569 }]),
1570 Err(ApplyError::NestingTooDeep {
1571 line: 0,
1572 depth: crate::MAX_NESTING_DEPTH + 1,
1573 max: crate::MAX_NESTING_DEPTH,
1574 })
1575 );
1576 assert!(rt.lines[0].containers.is_empty());
1577 }
1578
1579 #[test]
1580 fn line_op_set_continues_sets_and_clears() {
1581 // Two paragraph lines (delta-split → both `continues: false`, i.e. two
1582 // blocks). `setContinues` on line 1 turns the boundary into a within-block
1583 // hard break, and export then emits one block, not two paragraphs.
1584 let mut rt = from_markdown("one two").unwrap();
1585 rt.apply_text_delta(&diff("one two", "one\ntwo")).unwrap();
1586 assert!(!rt.lines[1].continues, "delta-split newline is a new block");
1587
1588 rt.apply_line_ops(&[LineOp::SetContinues {
1589 line: 1,
1590 continues: true,
1591 }])
1592 .unwrap();
1593 assert!(rt.lines[1].continues);
1594 assert_eq!(rt.validate(), Ok(()));
1595 assert_eq!(
1596 crate::export::to_markdown(&rt).matches("\n\n").count(),
1597 0,
1598 "a within-block hard break is not a paragraph boundary"
1599 );
1600
1601 // Clearing restores the block boundary.
1602 rt.apply_line_ops(&[LineOp::SetContinues {
1603 line: 1,
1604 continues: false,
1605 }])
1606 .unwrap();
1607 assert!(!rt.lines[1].continues);
1608 assert_eq!(rt.validate(), Ok(()));
1609 }
1610
1611 #[test]
1612 fn line_op_set_continues_rejects_first_line() {
1613 let mut rt = from_markdown("one two").unwrap();
1614 rt.apply_text_delta(&diff("one two", "one\ntwo")).unwrap();
1615 let before = rt.clone();
1616 // `continues: true` on line 0 forges `FirstLineContinues`; refused, and
1617 // the content is left untouched.
1618 assert_eq!(
1619 rt.apply_line_ops(&[LineOp::SetContinues {
1620 line: 0,
1621 continues: true,
1622 }]),
1623 Err(ApplyError::FirstLineContinues)
1624 );
1625 assert_eq!(rt, before, "rejected op leaves the content untouched");
1626 // Clearing line 0 (already `false`) is a no-op, not an error.
1627 rt.apply_line_ops(&[LineOp::SetContinues {
1628 line: 0,
1629 continues: false,
1630 }])
1631 .unwrap();
1632 assert_eq!(rt.validate(), Ok(()));
1633 }
1634
1635 fn island(id: &str) -> Island {
1636 Island {
1637 id: id.into(),
1638 island_type: "image".into(),
1639 props: serde_json::json!({}),
1640 loss: crate::model::Loss::LOSSLESS,
1641 }
1642 }
1643
1644 /// A single-line content `ab` (one inline island slot, one backing island).
1645 fn content_with_island() -> Content {
1646 let mut rt = Content::empty();
1647 rt.text = format!("a{ISLAND_SLOT}b");
1648 rt.lines = vec![Line {
1649 kind: LineKind::Para,
1650 containers: vec![],
1651 continues: false,
1652 }];
1653 rt.islands = vec![island("i1")];
1654 assert_eq!(rt.validate(), Ok(()));
1655 rt
1656 }
1657
1658 #[test]
1659 fn delete_slot_cascades_island_removal() {
1660 let mut rt = content_with_island();
1661 // Delete the slot char at index 1 (`ab` -> `ab`).
1662 let d = Delta {
1663 ops: vec![Op::Retain(1), Op::Delete(1), Op::Retain(1)],
1664 };
1665 rt.apply_text_delta(&d).unwrap();
1666 assert_eq!(rt.text, "ab");
1667 assert!(rt.islands.is_empty(), "island cascaded away with its slot");
1668 // slot count now equals islands.len(): validate confirms the sync.
1669 assert_eq!(rt.validate(), Ok(()));
1670 }
1671
1672 #[test]
1673 fn delete_one_of_two_slots_removes_the_matching_island() {
1674 let mut rt = Content::empty();
1675 rt.text = format!("{ISLAND_SLOT}x{ISLAND_SLOT}");
1676 rt.lines = vec![Line {
1677 kind: LineKind::Para,
1678 containers: vec![],
1679 continues: false,
1680 }];
1681 rt.islands = vec![island("first"), island("second")];
1682 assert_eq!(rt.validate(), Ok(()));
1683
1684 // Delete the FIRST slot (index 0): `x` -> `x`.
1685 let d = Delta {
1686 ops: vec![Op::Delete(1), Op::Retain(2)],
1687 };
1688 rt.apply_text_delta(&d).unwrap();
1689 assert_eq!(rt.text, format!("x{ISLAND_SLOT}"));
1690 // The surviving island is the second one: the cascade removed the
1691 // island whose slot was deleted, not merely the last entry.
1692 assert_eq!(rt.islands.len(), 1);
1693 assert_eq!(rt.islands[0].id, "second");
1694 assert_eq!(rt.validate(), Ok(()));
1695 }
1696
1697 #[test]
1698 fn insert_raw_slot_is_rejected() {
1699 let mut rt = from_markdown("ab").unwrap();
1700 // An Op::Insert carrying a raw U+FFFC would orphan a slot: reject it.
1701 let d = Delta {
1702 ops: vec![
1703 Op::Retain(1),
1704 Op::Insert(ISLAND_SLOT.to_string()),
1705 Op::Retain(1),
1706 ],
1707 };
1708 assert_eq!(rt.apply_text_delta(&d), Err(ApplyError::IslandSlotInInsert));
1709 // Content untouched on the rejected insert (checked before any mutation).
1710 assert_eq!(rt.text, "ab");
1711 assert!(rt.islands.is_empty());
1712 assert_eq!(rt.validate(), Ok(()));
1713 }
1714
1715 #[test]
1716 fn insert_carriage_return_is_stripped() {
1717 // A `\r` in an insert is dropped, not persisted: the content stays
1718 // valid instead of the op returning Ok over a `CarriageReturn`
1719 // violation. `\r\n` still yields the line-boundary `\n`.
1720 let mut rt = from_markdown("ab").unwrap();
1721 let d = Delta {
1722 ops: vec![Op::Retain(1), Op::Insert("\r".into()), Op::Retain(1)],
1723 };
1724 rt.apply_text_delta(&d).unwrap();
1725 assert_eq!(rt.text, "ab");
1726 assert_eq!(rt.validate(), Ok(()));
1727 }
1728
1729 #[test]
1730 fn insert_bidi_control_is_stripped() {
1731 // A bidi override (U+202E) in an insert is dropped: the content stays
1732 // valid and import's Trojan-source defense is not bypassed.
1733 let mut rt = from_markdown("ab").unwrap();
1734 let d = Delta {
1735 ops: vec![
1736 Op::Retain(1),
1737 Op::Insert("\u{202E}".into()),
1738 Op::Retain(1),
1739 ],
1740 };
1741 rt.apply_text_delta(&d).unwrap();
1742 assert_eq!(rt.text, "ab");
1743 assert_eq!(rt.validate(), Ok(()));
1744 }
1745
1746 #[test]
1747 fn insert_crlf_keeps_the_newline_and_splits() {
1748 // Stripping only the `\r` of a `\r\n` leaves a real line boundary: the
1749 // insert still splits the line, and slot/line sync stays intact.
1750 let mut rt = from_markdown("ab").unwrap();
1751 let d = Delta {
1752 ops: vec![Op::Retain(1), Op::Insert("\r\n".into()), Op::Retain(1)],
1753 };
1754 rt.apply_text_delta(&d).unwrap();
1755 assert_eq!(rt.text, "a\nb");
1756 assert_eq!(rt.lines.len(), 2);
1757 assert_eq!(rt.validate(), Ok(()));
1758 }
1759
1760 #[test]
1761 fn insert_of_clean_text_is_not_reallocated() {
1762 // The hot path: a delta whose inserts carry no forbidden char borrows
1763 // through `sanitize_inserts` unchanged.
1764 let d = Delta {
1765 ops: vec![Op::Retain(1), Op::Insert("clean\n".into()), Op::Retain(1)],
1766 };
1767 assert!(matches!(sanitize_inserts(&d), Cow::Borrowed(_)));
1768 }
1769
1770 /// A bundle carrying a text delta and mark ops alone.
1771 fn mark_bundle(delta: Delta, mark_ops: Vec<MarkOp>) -> ChangeBundle {
1772 ChangeBundle {
1773 delta,
1774 mark_ops,
1775 ..Default::default()
1776 }
1777 }
1778
1779 /// A bundle carrying island ops alone.
1780 fn island_bundle(island_ops: Vec<IslandOp>) -> ChangeBundle {
1781 ChangeBundle {
1782 island_ops,
1783 ..Default::default()
1784 }
1785 }
1786
1787 /// A one-cell table island's props, so a `Set` lands a shape `normalize`
1788 /// leaves alone and `validate` accepts.
1789 fn table_props(header: &str, cell: &str) -> serde_json::Value {
1790 serde_json::json!({
1791 "header": [{ "text": header, "marks": [] }],
1792 "rows": [[{ "text": cell, "marks": [] }]],
1793 "aligns": ["none"],
1794 })
1795 }
1796
1797 /// A minimal image island: the cheapest well-formed `Insert` payload.
1798 fn image(id: &str) -> Island {
1799 Island::new(id.into(), "image".into())
1800 .with_props(serde_json::json!({ "url": "u", "alt": "a" }))
1801 }
1802
1803 #[test]
1804 fn island_op_wire_round_trips_each_variant() {
1805 let island = Island::new("isl-0".into(), "table".into())
1806 .with_props(table_props("H", "a"))
1807 .with_loss(crate::model::Loss::DEGRADED);
1808 let ops = vec![
1809 IslandOp::Set {
1810 island: island.clone(),
1811 },
1812 IslandOp::Insert { at: 7, island },
1813 ];
1814 for op in ops {
1815 let v = island_op_to_value(&op);
1816 assert_eq!(island_op_from_value(&v).unwrap(), op, "round-trip: {v}");
1817 }
1818 }
1819
1820 /// The motivating case: an island payload edit moves the island entry alone,
1821 /// so an anchor elsewhere in the field survives an edit that a whole-value
1822 /// `install` would have cleared.
1823 #[test]
1824 fn island_set_edits_props_and_keeps_the_field_anchors() {
1825 let mut rt = from_markdown("intro\n\n| H |\n| --- |\n| a |").unwrap();
1826 assert_eq!(rt.islands.len(), 1, "one table island");
1827 let id = rt.islands[0].id.clone();
1828 rt.apply_mark_ops(&[MarkOp::Add {
1829 start: 0,
1830 end: 5,
1831 kind: MarkKind::Anchor { id: "c1".into() },
1832 }])
1833 .unwrap();
1834
1835 rt.apply_field_change(&island_bundle(vec![IslandOp::Set {
1836 island: Island::new(id.clone(), "table".into()).with_props(table_props("H", "b")),
1837 }]))
1838 .unwrap();
1839
1840 assert_eq!(rt.islands.len(), 1);
1841 assert_eq!(rt.islands[0].id, id, "the id is target and stored value");
1842 assert_eq!(rt.islands[0].props, table_props("H", "b"));
1843 let anchor = rt
1844 .marks
1845 .iter()
1846 .find(|m| matches!(&m.kind, MarkKind::Anchor { id } if id == "c1"))
1847 .expect("the anchor above the table survives the island edit");
1848 assert_eq!((anchor.start, anchor.end), (0, 5));
1849 assert_eq!(rt.validate(), Ok(()));
1850 }
1851
1852 /// A `Set` whose id names no island is refused, never a silent no-op: the
1853 /// store must not keep the old island while the caller believes it committed.
1854 #[test]
1855 fn island_set_rejects_an_unknown_id() {
1856 let mut rt = from_markdown("| H |\n| --- |\n| a |").unwrap();
1857 let before = rt.clone();
1858 assert_eq!(
1859 rt.apply_field_change(&island_bundle(vec![IslandOp::Set {
1860 island: Island::new("isl-nope".into(), "table".into())
1861 .with_props(table_props("H", "b")),
1862 }])),
1863 Err(ApplyError::UnknownIslandId {
1864 id: "isl-nope".into()
1865 })
1866 );
1867 assert_eq!(rt, before);
1868 }
1869
1870 /// `Insert` mints the slot and its entry together, so the slot count and the
1871 /// island list stay in lockstep with no orphan window.
1872 #[test]
1873 fn island_insert_adds_the_slot_and_its_entry() {
1874 let mut rt = from_markdown("ab").unwrap();
1875 rt.apply_mark_ops(&[MarkOp::Add {
1876 start: 0,
1877 end: 1,
1878 kind: MarkKind::Anchor { id: "c1".into() },
1879 }])
1880 .unwrap();
1881
1882 rt.apply_field_change(&island_bundle(vec![IslandOp::Insert {
1883 at: 1,
1884 island: Island::new("isl-new".into(), "image".into())
1885 .with_props(serde_json::json!({ "url": "u", "alt": "a" })),
1886 }]))
1887 .unwrap();
1888
1889 assert_eq!(rt.text, format!("a{ISLAND_SLOT}b"));
1890 assert_eq!(rt.islands.len(), 1);
1891 assert_eq!(rt.islands[0].id, "isl-new");
1892 assert_eq!(rt.validate(), Ok(()), "slot count matches the island list");
1893 // The anchor before the slot is untouched; one after would have moved
1894 // with the splice.
1895 let anchor = rt
1896 .marks
1897 .iter()
1898 .find(|m| matches!(&m.kind, MarkKind::Anchor { id } if id == "c1"))
1899 .expect("anchor survives");
1900 assert_eq!((anchor.start, anchor.end), (0, 1));
1901 }
1902
1903 /// Island ops sequence: op *n*'s `at` counts the slots ops `0..n` already
1904 /// spliced, not the shared post-delta frame, and its entry files at the
1905 /// slot-order index that frame gives rather than at emission order. Slots
1906 /// after `a` and `b` of `abc` go in at 1 and 3, and an op at an earlier
1907 /// position emitted last still files first. Both assertions land
1908 /// differently under the post-delta-only reading, which errors neither way.
1909 #[test]
1910 fn island_inserts_apply_in_sequence() {
1911 let mut rt = from_markdown("xabc").unwrap();
1912 rt.apply_field_change(&ChangeBundle {
1913 // Post-delta: the deleted `x` is out of the frame the ops read.
1914 delta: diff("xabc", "abc"),
1915 island_ops: vec![
1916 IslandOp::Insert {
1917 at: 1,
1918 island: image("isl-b"),
1919 },
1920 // 3, not 2: op 0's slot is in the frame this op reads.
1921 IslandOp::Insert {
1922 at: 3,
1923 island: image("isl-c"),
1924 },
1925 // An earlier position emitted last, so its slot lands first.
1926 IslandOp::Insert {
1927 at: 1,
1928 island: image("isl-a"),
1929 },
1930 ],
1931 ..Default::default()
1932 })
1933 .unwrap();
1934
1935 assert_eq!(
1936 rt.text,
1937 format!("a{ISLAND_SLOT}{ISLAND_SLOT}b{ISLAND_SLOT}c")
1938 );
1939 let ids: Vec<&str> = rt.islands.iter().map(|i| i.id.as_str()).collect();
1940 assert_eq!(ids, ["isl-a", "isl-b", "isl-c"], "slot order, not emission");
1941 assert_eq!(rt.validate(), Ok(()));
1942 }
1943
1944 /// A computed splice carrying a slot is refused whole; the same edit lands
1945 /// as its split form, the slot-free delta plus one `Insert` per slot.
1946 #[test]
1947 fn slot_bearing_splice_splits_into_delta_and_insert() {
1948 let mut rt = from_markdown("ab").unwrap();
1949 let before = rt.clone();
1950
1951 let paste = format!("x{ISLAND_SLOT}y");
1952 assert_eq!(
1953 rt.apply_field_change(&ChangeBundle::from_delta(Delta {
1954 ops: vec![Op::Retain(1), Op::Insert(paste)],
1955 })),
1956 Err(ApplyError::IslandSlotInInsert)
1957 );
1958 assert_eq!(rt, before, "the refusal commits nothing");
1959
1960 rt.apply_field_change(&ChangeBundle {
1961 delta: Delta {
1962 ops: vec![Op::Retain(1), Op::Insert("xy".into())],
1963 },
1964 // The delta leaves `axyb`; the slot goes between `x` and `y`.
1965 island_ops: vec![IslandOp::Insert {
1966 at: 2,
1967 island: image("isl-p"),
1968 }],
1969 ..Default::default()
1970 })
1971 .unwrap();
1972 assert_eq!(rt.text, format!("ax{ISLAND_SLOT}yb"));
1973 assert_eq!(rt.islands[0].id, "isl-p");
1974 assert_eq!(rt.validate(), Ok(()));
1975 }
1976
1977 /// The delete cascade is whole: the payload leaves the store with its slot,
1978 /// so re-landing the island re-inserts the value the producer held, under
1979 /// the original id the drop freed.
1980 #[test]
1981 fn island_delete_then_restore_round_trips() {
1982 let mut rt = from_markdown("ab").unwrap();
1983 rt.apply_field_change(&island_bundle(vec![IslandOp::Insert {
1984 at: 1,
1985 island: image("isl-a"),
1986 }]))
1987 .unwrap();
1988 let before = rt.clone();
1989 let held = rt.islands[0].clone();
1990
1991 rt.apply_field_change(&ChangeBundle::from_delta(diff(&before.text, "ab")))
1992 .unwrap();
1993 assert!(rt.islands.is_empty(), "the payload goes with its slot");
1994
1995 rt.apply_field_change(&island_bundle(vec![IslandOp::Insert {
1996 at: 1,
1997 island: held,
1998 }]))
1999 .unwrap();
2000 assert_eq!(rt, before, "same content, original id included");
2001 }
2002
2003 /// A block island's line demotes to `Para` when its slot goes: the kind
2004 /// stops matching the text and `normalize` repairs rather than fails.
2005 /// Re-landing one therefore carries the re-tag, not the island op alone.
2006 #[test]
2007 fn block_island_restore_retags_its_line() {
2008 let mut rt = from_markdown("intro").unwrap();
2009 rt.apply_field_change(&ChangeBundle {
2010 delta: diff("intro", "intro\n"),
2011 island_ops: vec![IslandOp::Insert {
2012 at: 6,
2013 island: image("isl-a"),
2014 }],
2015 line_ops: vec![LineOp::SetKind {
2016 line: 1,
2017 kind: LineKind::Island,
2018 }],
2019 ..Default::default()
2020 })
2021 .unwrap();
2022 let before = rt.clone();
2023 let held = rt.islands[0].clone();
2024
2025 rt.apply_field_change(&ChangeBundle::from_delta(diff(&before.text, "intro\n")))
2026 .unwrap();
2027 assert!(rt.islands.is_empty());
2028 assert_eq!(rt.lines[1].kind, LineKind::Para, "demoted, not failed");
2029
2030 // The line stayed open, so the restore is the island op and the re-tag;
2031 // only a delete that took the `\n` too would need a delta.
2032 rt.apply_field_change(&ChangeBundle {
2033 island_ops: vec![IslandOp::Insert { at: 6, island: held }],
2034 line_ops: vec![LineOp::SetKind {
2035 line: 1,
2036 kind: LineKind::Island,
2037 }],
2038 ..Default::default()
2039 })
2040 .unwrap();
2041 assert_eq!(rt, before, "same content, original id and kind included");
2042 }
2043
2044 /// An inserted island's id is caller-supplied on an anchor id's terms:
2045 /// non-empty and unused, since `Set` addresses by it.
2046 #[test]
2047 fn island_insert_id_and_position_rules() {
2048 let mut rt = from_markdown("ab").unwrap();
2049 assert_eq!(
2050 rt.apply_field_change(&island_bundle(vec![IslandOp::Insert {
2051 at: 1,
2052 island: image(""),
2053 }])),
2054 Err(ApplyError::EmptyIslandId)
2055 );
2056 assert_eq!(
2057 rt.apply_field_change(&island_bundle(vec![IslandOp::Insert {
2058 at: 9,
2059 island: image("isl-a"),
2060 }])),
2061 Err(ApplyError::IslandInsertOutOfRange { at: 9, len: 2 })
2062 );
2063
2064 rt.apply_field_change(&island_bundle(vec![IslandOp::Insert {
2065 at: 1,
2066 island: image("isl-a"),
2067 }]))
2068 .unwrap();
2069 assert_eq!(
2070 rt.apply_field_change(&island_bundle(vec![IslandOp::Insert {
2071 at: 0,
2072 island: image("isl-a"),
2073 }])),
2074 Err(ApplyError::IslandIdCollision { id: "isl-a".into() })
2075 );
2076 }
2077
2078 /// A block island in one bundle, which is what the stage order buys: the
2079 /// delta opens the line, the island op fills it, `SetKind` tags it. Nothing
2080 /// here falls back to a whole-value install, so the field's anchors stay.
2081 #[test]
2082 fn block_island_lands_in_one_bundle() {
2083 let mut rt = from_markdown("intro").unwrap();
2084 rt.apply_mark_ops(&[MarkOp::Add {
2085 start: 0,
2086 end: 5,
2087 kind: MarkKind::Anchor { id: "c1".into() },
2088 }])
2089 .unwrap();
2090
2091 rt.apply_field_change(&ChangeBundle {
2092 delta: diff("intro", "intro\n"),
2093 island_ops: vec![IslandOp::Insert {
2094 at: 6,
2095 island: Island::new("isl-t".into(), "table".into())
2096 .with_props(table_props("H", "a")),
2097 }],
2098 line_ops: vec![LineOp::SetKind {
2099 line: 1,
2100 kind: LineKind::Island,
2101 }],
2102 ..Default::default()
2103 })
2104 .unwrap();
2105
2106 assert_eq!(rt.text, format!("intro\n{ISLAND_SLOT}"));
2107 assert_eq!(rt.lines[1].kind, LineKind::Island);
2108 assert_eq!(rt.validate(), Ok(()));
2109 assert!(rt
2110 .marks
2111 .iter()
2112 .any(|m| matches!(&m.kind, MarkKind::Anchor { id } if id == "c1")));
2113 assert!(
2114 crate::export::to_markdown(&rt).contains("| H |"),
2115 "the block island projects as a pipe table"
2116 );
2117 }
2118
2119 /// A bundle whose island op fails commits none of its earlier stages.
2120 #[test]
2121 fn island_op_failure_leaves_the_content_untouched() {
2122 let mut rt = from_markdown("ab").unwrap();
2123 let before = rt.clone();
2124 let err = rt.apply_field_change(&ChangeBundle {
2125 delta: diff("ab", "aXb"),
2126 island_ops: vec![IslandOp::Set {
2127 island: Island::new("isl-nope".into(), "image".into()),
2128 }],
2129 ..Default::default()
2130 });
2131 assert!(matches!(err, Err(ApplyError::UnknownIslandId { .. })));
2132 assert_eq!(rt, before, "failed bundle must not mutate the content");
2133 }
2134
2135 #[test]
2136 fn apply_field_change_bundle_order() {
2137 let mut rt = from_markdown("abc").unwrap();
2138 let d = diff("abc", "abXc");
2139 rt.apply_field_change(&mark_bundle(
2140 d,
2141 vec![MarkOp::Add {
2142 start: 3,
2143 end: 4,
2144 kind: MarkKind::Strong,
2145 }],
2146 ))
2147 .unwrap();
2148 let strong = rt
2149 .marks
2150 .iter()
2151 .find(|m| matches!(m.kind, MarkKind::Strong))
2152 .unwrap();
2153 assert_eq!((strong.start, strong.end), (3, 4));
2154 assert_eq!(rt.text, "abXc");
2155 }
2156
2157 #[test]
2158 fn apply_field_change_is_all_or_nothing() {
2159 // A bundle whose text delta and first mark op succeed but whose second
2160 // mark op is out of range must leave the content exactly as it was: the
2161 // successful earlier stages do not partially commit.
2162 let mut rt = from_markdown("abc").unwrap();
2163 let before = rt.clone();
2164 let d = diff("abc", "abXc");
2165 let err = rt.apply_field_change(&mark_bundle(
2166 d,
2167 vec![
2168 MarkOp::Add {
2169 start: 0,
2170 end: 2,
2171 kind: MarkKind::Strong,
2172 },
2173 MarkOp::Add {
2174 start: 99,
2175 end: 100,
2176 kind: MarkKind::Emph,
2177 },
2178 ],
2179 ));
2180 assert!(matches!(err, Err(ApplyError::MarkOutOfRange { .. })));
2181 assert_eq!(rt, before, "failed bundle must not mutate the content");
2182 }
2183
2184 /// `add` of an anchor rejects a live id collision and the empty id, but
2185 /// re-adds an id freed by an earlier `RemoveAnchor` in the same bundle.
2186 #[test]
2187 fn add_anchor_id_uniqueness() {
2188 let anchor = |id: &str| MarkKind::Anchor { id: id.into() };
2189 let add = |start, end, id: &str| MarkOp::Add {
2190 start,
2191 end,
2192 kind: anchor(id),
2193 };
2194
2195 let noop = || diff("abcd", "abcd");
2196
2197 // First anchor lands; a second `add` of the same id is a collision.
2198 let mut rt = from_markdown("abcd").unwrap();
2199 rt.apply_field_change(&mark_bundle(noop(), vec![add(0, 2, "x")]))
2200 .unwrap();
2201 assert_eq!(
2202 rt.apply_field_change(&mark_bundle(noop(), vec![add(2, 4, "x")])),
2203 Err(ApplyError::AnchorIdCollision { id: "x".into() })
2204 );
2205
2206 // The empty id is refused.
2207 let mut rt = from_markdown("abcd").unwrap();
2208 assert_eq!(
2209 rt.apply_field_change(&mark_bundle(noop(), vec![add(0, 2, "")])),
2210 Err(ApplyError::EmptyAnchorId)
2211 );
2212
2213 // Remove-then-add of the same id in one bundle is allowed: ops apply in
2214 // sequence, so the id is free by the time the `add` runs.
2215 let mut rt = from_markdown("abcd").unwrap();
2216 rt.apply_field_change(&mark_bundle(noop(), vec![add(0, 2, "x")]))
2217 .unwrap();
2218 rt.apply_field_change(&mark_bundle(
2219 noop(),
2220 vec![MarkOp::RemoveAnchor { id: "x".into() }, add(2, 4, "x")],
2221 ))
2222 .unwrap();
2223 let anchors: Vec<_> = rt
2224 .marks
2225 .iter()
2226 .filter(|m| matches!(m.kind, MarkKind::Anchor { .. }))
2227 .collect();
2228 assert_eq!(anchors.len(), 1);
2229 assert_eq!((anchors[0].start, anchors[0].end), (2, 4));
2230 }
2231
2232 // ── sync_lines_for_delta characterization ───────────────────────────────
2233 //
2234 // Pin the observable behavior of the line-sync walk: retain/insert/delete
2235 // interleavings, the split template-clone rule, and the malformed-content
2236 // guards: against a silent change to its internals.
2237
2238 /// A `Heading{level}` line, its level a visible tag so a test can trace
2239 /// which original line landed where; `continues` distinguishes a clone.
2240 fn tag_line(level: u8, continues: bool) -> Line {
2241 Line {
2242 kind: LineKind::Heading { level },
2243 containers: Vec::new(),
2244 continues,
2245 }
2246 }
2247
2248 /// `(tag, continues)` per line: `Heading{level}` reads its level, `Para` is
2249 /// tag 0 (the default line), any other kind is 255.
2250 fn tags(lines: &[Line]) -> Vec<(u8, bool)> {
2251 lines
2252 .iter()
2253 .map(|l| match l.kind {
2254 LineKind::Heading { level } => (level, l.continues),
2255 LineKind::Para => (0, l.continues),
2256 _ => (255, l.continues),
2257 })
2258 .collect()
2259 }
2260
2261 #[test]
2262 fn sync_lines_retain_only_is_identity() {
2263 let old_chars: Vec<char> = "a\nb\nc".chars().collect();
2264 let lines = vec![tag_line(1, false), tag_line(2, false), tag_line(3, false)];
2265 let d = Delta {
2266 ops: vec![Op::Retain(5)],
2267 };
2268 assert_eq!(sync_lines_for_delta(&old_chars, lines.clone(), &d), lines);
2269 }
2270
2271 #[test]
2272 fn sync_lines_insert_newline_clones_split_line_and_clears_continues() {
2273 // Split line 1 ("bc") mid-line: the first half stays the original line
2274 // (keeps kind, containers, and its `continues: true`); the second half
2275 // is a clone of it with `continues` forced false.
2276 let old_chars: Vec<char> = "a\nbc".chars().collect();
2277 let l1 = Line {
2278 kind: LineKind::Heading { level: 5 },
2279 containers: vec![Container::Quote],
2280 continues: true,
2281 };
2282 let lines = vec![tag_line(1, false), l1.clone()];
2283 // Retain(3)[a\nb] moves to line 1; Insert("\n") splits it; Retain(1)[c].
2284 let d = Delta {
2285 ops: vec![Op::Retain(3), Op::Insert("\n".into()), Op::Retain(1)],
2286 };
2287 let out = sync_lines_for_delta(&old_chars, lines, &d);
2288 assert_eq!(out.len(), 3);
2289 assert_eq!(out[1], l1, "first half is the untouched original line");
2290 assert_eq!(out[2].kind, LineKind::Heading { level: 5 });
2291 assert_eq!(out[2].containers, vec![Container::Quote]);
2292 assert!(!out[2].continues, "the split clone starts a new block");
2293 }
2294
2295 #[test]
2296 fn sync_lines_delete_newline_drops_following_line() {
2297 // Delete the first '\n' of "a\nb\nc": lines 0 and 1 merge, dropping line
2298 // 1; the current line (0) and line 2 survive.
2299 let old_chars: Vec<char> = "a\nb\nc".chars().collect();
2300 let lines = vec![tag_line(1, false), tag_line(2, false), tag_line(3, false)];
2301 let d = Delta {
2302 ops: vec![Op::Retain(1), Op::Delete(1), Op::Retain(3)],
2303 };
2304 let out = sync_lines_for_delta(&old_chars, lines, &d);
2305 assert_eq!(tags(&out), vec![(1, false), (3, false)]);
2306 }
2307
2308 #[test]
2309 fn sync_lines_delete_trailing_newline_without_following_line_is_guarded() {
2310 // Malformed content: text "a\n" is two segments but `lines` has one
2311 // entry. Deleting the '\n' when `line_idx + 1` is out of bounds removes
2312 // nothing (the guard), leaving the single line intact.
2313 let old_chars: Vec<char> = "a\n".chars().collect();
2314 let lines = vec![tag_line(1, false)];
2315 let d = Delta {
2316 ops: vec![Op::Retain(1), Op::Delete(1)],
2317 };
2318 let out = sync_lines_for_delta(&old_chars, lines, &d);
2319 assert_eq!(tags(&out), vec![(1, false)]);
2320 }
2321
2322 #[test]
2323 fn sync_lines_stops_at_end_of_old_chars() {
2324 // A retain running past the end of old_chars stops at the end rather
2325 // than indexing out of bounds (the `old >= old_chars.len()` guard).
2326 let old_chars: Vec<char> = "a\nb".chars().collect();
2327 let lines = vec![tag_line(1, false), tag_line(2, false)];
2328 let d = Delta {
2329 ops: vec![Op::Retain(99)],
2330 };
2331 assert_eq!(sync_lines_for_delta(&old_chars, lines.clone(), &d), lines);
2332 }
2333
2334 #[test]
2335 fn sync_lines_insert_two_newlines_adds_two_clones() {
2336 // Inserting "\n\n" mid-line adds two lines, each carrying the split
2337 // line's kind and containers with `continues: false`.
2338 let old_chars: Vec<char> = "abc".chars().collect();
2339 let src = Line {
2340 kind: LineKind::Heading { level: 7 },
2341 containers: vec![Container::Quote],
2342 continues: false,
2343 };
2344 let d = Delta {
2345 ops: vec![Op::Retain(1), Op::Insert("\n\n".into()), Op::Retain(2)],
2346 };
2347 let out = sync_lines_for_delta(&old_chars, vec![src], &d);
2348 assert_eq!(out.len(), 3);
2349 for l in &out {
2350 assert_eq!(l.kind, LineKind::Heading { level: 7 });
2351 assert_eq!(l.containers, vec![Container::Quote]);
2352 assert!(!l.continues);
2353 }
2354 }
2355
2356 // ── line-op mark remap + terminal-normalize collapse ────────────────────
2357
2358 #[test]
2359 fn split_line_rebases_mark_across_the_split_point() {
2360 // A strong mark spanning the split point grows by the inserted `\n`
2361 // rather than staying at its old coordinates. "abcd", strong[1..3)
2362 // ("bc"); split at 2 → "ab\ncd"; the mark must still cover "b"+"c",
2363 // i.e. [1..4) over "ab\ncd".
2364 let mut rt = from_markdown("abcd").unwrap();
2365 rt.apply_mark_ops(&[MarkOp::Add {
2366 start: 1,
2367 end: 3,
2368 kind: MarkKind::Strong,
2369 }])
2370 .unwrap();
2371 rt.apply_line_ops(&[LineOp::Split { at: 2 }]).unwrap();
2372 assert_eq!(rt.text, "ab\ncd");
2373 let strong: Vec<_> = rt
2374 .marks
2375 .iter()
2376 .filter(|m| matches!(m.kind, MarkKind::Strong))
2377 .map(|m| (m.start, m.end))
2378 .collect();
2379 // [1..4) spans "b\nc"; normalize keeps the interior `\n` (a mark may
2380 // legitimately span lines), trimming only leading/trailing boundaries.
2381 assert_eq!(strong, vec![(1, 4)]);
2382 assert_eq!(rt.validate(), Ok(()));
2383 }
2384
2385 #[test]
2386 fn join_line_rebases_marks_to_final_text_coordinates() {
2387 // The issue's concrete drift case: "ab\ncd", strong[2..4) (over "\nc").
2388 // Joining line 0 removes the `\n`; the remap + terminal normalize must
2389 // land strong on "c" (coordinate [2..3) over "abcd") not on "d"
2390 // (the un-remapped-mark bug) nor on "cd".
2391 let mut rt = from_markdown("ab").unwrap();
2392 rt.apply_text_delta(&diff("ab", "ab\ncd")).unwrap();
2393 rt.marks.push(Mark {
2394 start: 2,
2395 end: 4,
2396 kind: MarkKind::Strong,
2397 });
2398 rt.normalize();
2399 // Post-normalize the `\n` edge trims to [3..4) ("c"); either way the
2400 // join must converge to strong on "c".
2401 rt.apply_line_ops(&[LineOp::Join { line: 0 }]).unwrap();
2402 assert_eq!(rt.text, "abcd");
2403 let strong: Vec<_> = rt
2404 .marks
2405 .iter()
2406 .filter(|m| matches!(m.kind, MarkKind::Strong))
2407 .map(|m| (m.start, m.end))
2408 .collect();
2409 assert_eq!(strong, vec![(2, 3)], "strong lands on 'c', not 'd' or 'cd'");
2410 assert_eq!(rt.validate(), Ok(()));
2411 }
2412
2413 #[test]
2414 fn field_change_terminal_normalize_matches_per_stage_normalize() {
2415 // The collapse proof obligation: a bundle applied through
2416 // `apply_field_change` (one terminal normalize) must equal applying the
2417 // same stages each with its own normalize (the public wrappers). The
2418 // remap through split/join is what makes the two converge.
2419 let start = from_markdown("hello world").unwrap();
2420 let text_delta = diff("hello world", "hello brave world");
2421 let line_ops = vec![LineOp::Split { at: 5 }]; // after "hello"
2422 let mark_ops = vec![MarkOp::Add {
2423 start: 0,
2424 end: 5,
2425 kind: MarkKind::Strong,
2426 }];
2427
2428 let mut bundled = start.clone();
2429 bundled
2430 .apply_field_change(&ChangeBundle {
2431 delta: text_delta.clone(),
2432 line_ops: line_ops.clone(),
2433 mark_ops: mark_ops.clone(),
2434 ..Default::default()
2435 })
2436 .unwrap();
2437
2438 let mut staged = start;
2439 staged.apply_text_delta(&text_delta).unwrap();
2440 staged.apply_line_ops(&line_ops).unwrap();
2441 staged.apply_mark_ops(&mark_ops).unwrap();
2442
2443 assert_eq!(bundled, staged, "terminal normalize diverged from per-stage");
2444 assert_eq!(bundled.validate(), Ok(()));
2445 }
2446
2447 #[test]
2448 fn sync_lines_select_all_delete_collapses_to_first_line() {
2449 // The motivating case: deleting a whole
2450 // multi-line body drops every line but the first (each deleted '\n'
2451 // merges the next line away).
2452 let text: String = (0..50).map(|i| format!("line{i}\n")).collect();
2453 let old_chars: Vec<char> = text.chars().collect();
2454 let lines: Vec<Line> = (0..=50).map(|i| tag_line((i % 200) as u8, false)).collect();
2455 assert_eq!(lines.len(), old_chars.iter().filter(|&&c| c == '\n').count() + 1);
2456 let d = Delta {
2457 ops: vec![Op::Delete(old_chars.len())],
2458 };
2459 let out = sync_lines_for_delta(&old_chars, lines, &d);
2460 assert_eq!(tags(&out), vec![(0, false)], "only the first line survives");
2461 }
2462
2463 #[test]
2464 fn sync_lines_insert_newline_past_end_appends_default() {
2465 // Malformed content: after a retain walks past the sole line (line_idx ==
2466 // lines.len()), an inserted '\n' has no line to clone and appends a
2467 // default Para.
2468 let old_chars: Vec<char> = "a\n".chars().collect();
2469 let lines = vec![tag_line(1, false)];
2470 // Retain(2)[a\n] moves line_idx to 1 (== lines.len()); Insert("\n").
2471 let d = Delta {
2472 ops: vec![Op::Retain(2), Op::Insert("\n".into())],
2473 };
2474 let out = sync_lines_for_delta(&old_chars, lines, &d);
2475 assert_eq!(out.len(), 2);
2476 assert_eq!(tags(&out)[0], (1, false));
2477 assert_eq!(out[1].kind, LineKind::Para);
2478 assert!(out[1].containers.is_empty());
2479 assert!(!out[1].continues);
2480 }
2481}