docling_core/tree.rs
1//! docling's **item tree**, for a backend that knows the exact shape upstream
2//! gives a document and wants the JSON export to reproduce it.
3//!
4//! [`DoclingDocument::nodes`](crate::DoclingDocument::nodes) is a flat,
5//! reading-order stream tuned for Markdown / DocLang / LaTeX; the JSON export
6//! rebuilds docling's parent/child structure from it with generic rules (runs
7//! of list items become list groups, a heading is a flat sibling of the text
8//! that follows it). Upstream's backends do not all agree on that structure:
9//! the HTML backend nests everything after a heading *under* the heading,
10//! splits a paragraph of mixed formatting into an `inline` group of one text
11//! item per formatting run, parents a rich table cell's content to a group
12//! under the table, keeps site chrome on the `furniture` layer… and numbers
13//! every item in the order it *creates* them. A backend that ports those
14//! rules call-for-call (HTML's `html_tree.rs`, DOCX's `docx_tree.rs`) records
15//! the result here — an arena of items in
16//! creation order, each with its parent and children — and the JSON export
17//! ([`DoclingDocument::export_to_json`](crate::DoclingDocument::export_to_json))
18//! serializes this tree instead of deriving one from the nodes. Every other
19//! serializer keeps reading the flat nodes, so their output is unaffected.
20
21use crate::{ContentLayer, FieldItem, PictureImage, Script, Table};
22
23/// docling-core's `Formatting`: the inline styles an item carries in JSON.
24#[derive(Debug, Clone, Copy, PartialEq, Eq, Default)]
25pub struct Formatting {
26 pub bold: bool,
27 pub italic: bool,
28 pub underline: bool,
29 pub strikethrough: bool,
30 pub script: Script,
31}
32
33/// A `list_item`'s docling fields.
34#[derive(Debug, Clone, PartialEq, Eq, Default)]
35pub struct ListMeta {
36 pub enumerated: bool,
37 /// docling's `marker` — the HTML backend writes `""` unless an ordered
38 /// list carries an explicit `start`, then `"{n}."`.
39 pub marker: String,
40}
41
42/// What an item in the tree is. Mirrors the docling-core item classes the
43/// JSON `texts` / `groups` / `tables` / `pictures` / `field_regions` buckets
44/// hold.
45#[derive(Debug, Clone, PartialEq)]
46pub enum TreeKind {
47 /// A `TextItem` / `TitleItem` / `SectionHeaderItem` / `ListItem`, told
48 /// apart by `label` (`text`, `title`, `section_header`, `list_item`,
49 /// `caption`, `checkbox_selected`, `checkbox_unselected`, …).
50 Text {
51 label: String,
52 text: String,
53 /// docling's `orig` when it differs from `text` (the heading text
54 /// before unicode cleanup, say); `None` = same as `text`.
55 orig: Option<String>,
56 formatting: Option<Formatting>,
57 hyperlink: Option<String>,
58 /// `section_header` only: docling's heading level.
59 level: Option<u8>,
60 /// `list_item` only.
61 list: Option<ListMeta>,
62 },
63 /// A `CodeItem`.
64 Code {
65 text: String,
66 orig: Option<String>,
67 /// The language hint (a highlighter class token such as `python`),
68 /// mapped onto docling's `CodeLanguageLabel` at export; `None` →
69 /// `unknown`.
70 language: Option<String>,
71 formatting: Option<Formatting>,
72 hyperlink: Option<String>,
73 /// Caption text items in the tree — docling's `add_code(caption=…)`
74 /// (an AsciiDoc block title in front of a literal block).
75 captions: Vec<usize>,
76 },
77 /// A `GroupItem`: `label` is docling's `GroupLabel` value (`inline`,
78 /// `list`, `section`, `unspecified`, …), `name` its name (`group`, `list`,
79 /// `ordered list`, `header-2`, `rich_cell_group_1_0_3`, …).
80 Group { label: String, name: String },
81 /// A `TableItem`. `rich_cells` marks the cells docling serialized as a
82 /// `RichTableCell`: `(row, col)` grid anchor → the group item (a child of
83 /// the table) that holds the cell's content. `captions` are caption text
84 /// items in the tree.
85 Table {
86 table: Table,
87 rich_cells: Vec<(usize, usize, usize)>,
88 captions: Vec<usize>,
89 },
90 /// A `PictureItem`, its caption text items and optional payload.
91 /// `classification` is a `PictureClassificationLabel` value written as
92 /// the picture's `meta.classification` (an HTML `<stamp>` / `<signature>`).
93 Picture {
94 captions: Vec<usize>,
95 image: Option<PictureImage>,
96 classification: Option<String>,
97 /// Further predictions after `classification`, in order: the other
98 /// plot types of a PPTX combination chart (docling#4773).
99 extra_classes: Vec<String>,
100 /// The picture-OCR enrichment's text (#645), serialized like the flat
101 /// node's: `meta.description` + the `description` annotation.
102 description: Option<crate::PictureDescription>,
103 /// The prediction's `confidence`, when the backend writes one: the
104 /// DocLang deserializer stamps `1.0`; the office and HTML backends
105 /// leave it out (`None`).
106 confidence: Option<f64>,
107 /// A native chart's data grid (docling's `meta.tabular_chart.chart_data`,
108 /// the series reconstructed as a `TableData`), for a DOCX chart drawing.
109 chart: Option<Table>,
110 /// The `ImageRef.dpi` docling writes for `image` when the backend
111 /// read one from the file (python-pptx's `Image.dpi`: PIL's `dpi`
112 /// info, rounded, 72 when absent or out of 1–2048); `None` → 72,
113 /// which is what upstream's other office backends pass.
114 dpi: Option<u32>,
115 },
116 /// A form key-value region (`field_regions` / `field_items`).
117 FieldRegion { items: Vec<FieldItem> },
118 /// A `KeyValueItem` (`key_value_items`): docling's `GraphData` of key and
119 /// value cells and their links, written verbatim.
120 KeyValueGraph {
121 cells: Vec<crate::GraphCell>,
122 links: Vec<crate::GraphLink>,
123 },
124}
125
126/// docling's `ProvenanceItem` for a tree item, written verbatim: the
127/// backend's own geometry in the page's units — a PPTX shape's EMU box,
128/// whose `pages` entry is the slide size in EMU — rather than the 0–511
129/// DocLang grid the flat [`Node::Located`](crate::Node::Located) carries
130/// (which cannot round-trip those integers).
131#[derive(Debug, Clone, PartialEq)]
132pub struct TreeProv {
133 /// 1-based page (slide) number.
134 pub page_no: usize,
135 /// `[l, t, r, b]`, exactly as docling computed them.
136 pub bbox: [f64; 4],
137 /// docling's `coord_origin` tag. The office backends tag their
138 /// top-left-based boxes `TOPLEFT` (the PPTX backend since docling#4294 —
139 /// it used to tag them `BOTTOMLEFT`, which read the tuple as `(l, b, r,
140 /// t)` and swapped the vertical edges); a backend that really works in a
141 /// bottom-left space sets this.
142 pub bottom_left: bool,
143 /// `[0, len(text)]` in characters for a text item, `[0, 0]` for a table
144 /// or picture.
145 pub charspan: [usize; 2],
146}
147
148/// docling's `TrackSource` — where in a time-based track (a WebVTT cue) a
149/// text item came from. Written as the item's `source: [{"kind": "track", …}]`.
150#[derive(Debug, Clone, PartialEq)]
151pub struct TreeTrack {
152 /// The cue's start offset in seconds (docling's `WebVTTTimestamp.seconds`:
153 /// `h*3600 + m*60 + s + millis/1000.0`, so the float is bit-identical).
154 pub start_time: f64,
155 /// The cue's end offset in seconds.
156 pub end_time: f64,
157 /// The cue identifier line, when the cue has one.
158 pub identifier: Option<String>,
159 /// The `<v …>` voice annotation the text sits in, when any.
160 pub voice: Option<String>,
161}
162
163/// One item of an [`ItemTree`].
164#[derive(Debug, Clone, PartialEq)]
165pub struct TreeItem {
166 /// The parent item's index; `None` = the document body.
167 pub parent: Option<usize>,
168 /// Child item indices, in docling's `children` order.
169 pub children: Vec<usize>,
170 /// The content layer; `None` = `body`.
171 pub layer: Option<ContentLayer>,
172 pub kind: TreeKind,
173 /// The item's `prov` entry, when the backend has page geometry for it
174 /// (`None` → `prov: []`, what the HTML and DOCX backends write).
175 pub prov: Option<TreeProv>,
176 /// docling's `DocItem.comments`: the `comment_section` groups (or note
177 /// text items) annotating this item, as item indices — written after
178 /// `prov` when non-empty.
179 pub comments: Vec<usize>,
180 /// docling's `DocItem.source`: the track segment a text item was taken
181 /// from (WebVTT cues) — written after `prov` when set.
182 pub source: Option<TreeTrack>,
183 /// Removed by [`ItemTree::delete`] (docling's `delete_items`): the slot
184 /// stays so every other index keeps its meaning, but the item is not
185 /// numbered or written.
186 pub deleted: bool,
187 /// Footnotes / endnotes referenced from inside this text item (#538): the
188 /// note call's position (chars into the item's text) and the note's text.
189 /// docling keeps notes as unlinked furniture `footnote` items, so the
190 /// JSON never shows this; the Pandoc AST writes each as a `Note` there.
191 pub notes: Vec<TreeNote>,
192 /// This furniture `footnote` item is the body of a note some text item
193 /// calls (it travels in that item's [`Self::notes`]); the Pandoc AST then
194 /// leaves it out as a standalone block.
195 pub note_body: bool,
196}
197
198/// A note call inside a text item ([`TreeItem::notes`]).
199#[derive(Debug, Clone, PartialEq, Eq)]
200pub struct TreeNote {
201 /// The call's position, in chars into the item's `text`.
202 pub offset: usize,
203 /// The note's plain text.
204 pub text: String,
205}
206
207impl ItemTree {
208 /// Where each note call of one paragraph lands (#538): `full_text` is the
209 /// paragraph's text, `offsets` the calls' positions in it (chars), and
210 /// the paragraph's items are those created from `first_new` on. Each text
211 /// item is a trimmed slice of the paragraph (a formatting run, a link, or
212 /// the whole heading / list item), matched in order; a call inside an
213 /// item lands there, one between items ends the earlier (or starts the
214 /// first). When no item matches, the call ends the paragraph's last text
215 /// item; with no text item at all it is `None`.
216 pub fn place_note_calls(
217 &self,
218 first_new: usize,
219 full_text: &str,
220 offsets: &[usize],
221 ) -> Vec<Option<(usize, usize)>> {
222 let texts: Vec<(usize, &str)> = (first_new..self.items.len())
223 .filter(|&i| !self.items[i].deleted)
224 .filter_map(|i| match &self.items[i].kind {
225 TreeKind::Text { text, .. } | TreeKind::Code { text, .. } => {
226 Some((i, text.as_str()))
227 }
228 _ => None,
229 })
230 .collect();
231 // Each found item's (id, start, end) in chars of `full_text`.
232 let mut spans: Vec<(usize, usize, usize)> = Vec::new();
233 let mut cursor = 0usize; // bytes
234 for &(item, text) in &texts {
235 if text.is_empty() {
236 continue;
237 }
238 if let Some(pos) = full_text[cursor..].find(text) {
239 let start_b = cursor + pos;
240 let start = full_text[..start_b].chars().count();
241 spans.push((item, start, start + text.chars().count()));
242 cursor = start_b + text.len();
243 }
244 }
245 offsets
246 .iter()
247 .map(|&offset| {
248 spans
249 .iter()
250 .find(|&&(_, start, end)| offset >= start && offset <= end)
251 .map(|&(item, start, _)| (item, offset - start))
252 .or_else(
253 || match spans.iter().rev().find(|&&(_, _, end)| end <= offset) {
254 Some(&(item, start, end)) => Some((item, end - start)),
255 None => spans.first().map(|&(item, _, _)| (item, 0)),
256 },
257 )
258 .or_else(|| {
259 texts
260 .last()
261 .map(|&(item, text)| (item, text.chars().count()))
262 })
263 })
264 .collect()
265 }
266}
267
268/// docling's item tree in creation order (see the [module docs](self)).
269#[derive(Debug, Clone, PartialEq, Default)]
270pub struct ItemTree {
271 /// Every item, indexed by creation order — which is how docling numbers
272 /// `#/texts/N`, `#/groups/N`, … within each bucket.
273 pub items: Vec<TreeItem>,
274 /// The body's `children`, as item indices.
275 pub body: Vec<usize>,
276}
277
278impl ItemTree {
279 /// Append an item under `parent` (`None` = body) on `layer`, registering
280 /// it as its parent's last child — docling's `add_*` calls do exactly that.
281 pub fn add(
282 &mut self,
283 parent: Option<usize>,
284 layer: Option<ContentLayer>,
285 kind: TreeKind,
286 ) -> usize {
287 let id = self.items.len();
288 self.items.push(TreeItem {
289 parent,
290 children: Vec::new(),
291 layer,
292 kind,
293 prov: None,
294 comments: Vec::new(),
295 source: None,
296 deleted: false,
297 notes: Vec::new(),
298 note_body: false,
299 });
300 match parent {
301 Some(p) => self.items[p].children.push(id),
302 None => self.body.push(id),
303 }
304 id
305 }
306
307 /// [`add`](Self::add) with the item's provenance — docling's
308 /// `add_text(…, prov=prov)`.
309 pub fn add_with_prov(
310 &mut self,
311 parent: Option<usize>,
312 layer: Option<ContentLayer>,
313 kind: TreeKind,
314 prov: TreeProv,
315 ) -> usize {
316 let id = self.add(parent, layer, kind);
317 self.items[id].prov = Some(prov);
318 id
319 }
320
321 /// Append every item of `other` after this tree's, renumbering its
322 /// indices (parents, children, comments, table/picture caption and
323 /// rich-cell refs) and adding its body children to this body — so a
324 /// backend can build independent fragments in parallel (one per PPTX
325 /// slide) and still hand the export one tree in creation order, exactly
326 /// as if it had been built sequentially.
327 pub fn append(&mut self, other: ItemTree) {
328 let off = self.items.len();
329 let shift = |i: usize| i + off;
330 for mut item in other.items {
331 item.parent = item.parent.map(shift);
332 for c in item.children.iter_mut().chain(item.comments.iter_mut()) {
333 *c = shift(*c);
334 }
335 match &mut item.kind {
336 TreeKind::Table {
337 rich_cells,
338 captions,
339 ..
340 } => {
341 for (_, _, g) in rich_cells.iter_mut() {
342 *g = shift(*g);
343 }
344 for c in captions.iter_mut() {
345 *c = shift(*c);
346 }
347 }
348 TreeKind::Picture { captions, .. } | TreeKind::Code { captions, .. } => {
349 for c in captions.iter_mut() {
350 *c = shift(*c);
351 }
352 }
353 _ => {}
354 }
355 self.items.push(item);
356 }
357 self.body.extend(other.body.into_iter().map(shift));
358 }
359
360 /// Move `id` under `new_parent`, dropping it from its current parent's
361 /// children and appending it to the new one's — docling's
362 /// `group_cell_elements` re-parenting of a rich cell's items.
363 pub fn reparent(&mut self, id: usize, new_parent: Option<usize>) {
364 let old = self.items[id].parent;
365 let siblings = match old {
366 Some(p) => &mut self.items[p].children,
367 None => &mut self.body,
368 };
369 siblings.retain(|&c| c != id);
370 self.items[id].parent = new_parent;
371 match new_parent {
372 Some(p) => self.items[p].children.push(id),
373 None => self.body.push(id),
374 }
375 }
376
377 /// Re-number the items in traversal order — a pre-order walk of the body
378 /// through every layer, which is how docling's
379 /// `DoclingDocument.concatenate` (and `_normalize_references`) re-creates
380 /// a document's items: a group created after the content it was later
381 /// wrapped around comes before that content afterwards. Items the walk
382 /// does not reach (deleted ones) are dropped. Returns each old id's new
383 /// id.
384 pub fn renumber_in_traversal_order(&mut self) -> Vec<Option<usize>> {
385 let mut order = Vec::with_capacity(self.items.len());
386 let mut stack: Vec<usize> = self.body.iter().rev().copied().collect();
387 while let Some(id) = stack.pop() {
388 if self.items[id].deleted {
389 continue;
390 }
391 order.push(id);
392 stack.extend(self.items[id].children.iter().rev());
393 }
394 let mut new_of: Vec<Option<usize>> = vec![None; self.items.len()];
395 for (new, &old) in order.iter().enumerate() {
396 new_of[old] = Some(new);
397 }
398 let remap = |ids: &mut Vec<usize>| {
399 *ids = ids.iter().filter_map(|&i| new_of[i]).collect();
400 };
401 let mut old_items: Vec<Option<TreeItem>> = std::mem::take(&mut self.items)
402 .into_iter()
403 .map(Some)
404 .collect();
405 for &old in &order {
406 let mut item = old_items[old].take().expect("each item visited once");
407 item.parent = item.parent.and_then(|p| new_of[p]);
408 remap(&mut item.children);
409 remap(&mut item.comments);
410 match &mut item.kind {
411 TreeKind::Table {
412 rich_cells,
413 captions,
414 ..
415 } => {
416 rich_cells.retain_mut(|(_, _, g)| match new_of[*g] {
417 Some(n) => {
418 *g = n;
419 true
420 }
421 None => false,
422 });
423 remap(captions);
424 }
425 TreeKind::Picture { captions, .. } | TreeKind::Code { captions, .. } => {
426 remap(captions)
427 }
428 _ => {}
429 }
430 self.items.push(item);
431 }
432 remap(&mut self.body);
433 new_of
434 }
435
436 /// Remove `id` from the tree — docling's `delete_items`, which the DOCX
437 /// backend uses to drop the empty text item a blank spacer paragraph left
438 /// between two items of a resumed list. The item leaves its parent's
439 /// children and is neither numbered nor written; its slot stays so the
440 /// indices held elsewhere stay valid.
441 pub fn delete(&mut self, id: usize) {
442 match self.items[id].parent {
443 Some(p) => self.items[p].children.retain(|&c| c != id),
444 None => self.body.retain(|&c| c != id),
445 }
446 self.items[id].deleted = true;
447 }
448
449 /// docling-core's `DoclingDocument.validate_misplaced_list_items` (a
450 /// model validator, so it runs whenever docling-core serializes or loads
451 /// a document): every `list_item` whose parent is not a `list` group is
452 /// re-homed into a new one. A pre-order walk of the body (groups
453 /// included) collects them; consecutive misplaced items directly on the
454 /// body share one group, any other misplaced item gets its own. Working
455 /// from the last run back, each run gets a `ListGroup` (name `group`)
456 /// inserted where its first item stood, the items are deleted and
457 /// re-added under the group as fresh items, so they move to the end of
458 /// the text numbering (#527: the DOCX backend leaves such items in rich
459 /// table cells). A no-op for a well-formed tree.
460 ///
461 /// One deliberate divergence (#586): docling-core deletes each item *with
462 /// its children* and re-adds it from its text alone, so a mixed-format
463 /// item — an empty `list_item` over an `inline` group of text runs —
464 /// comes back empty and its text is gone from the JSON (and from the
465 /// Markdown/HTML docling serializes after that validator ran; the JSON
466 /// docling saves *before* exporting still has it). Here the children
467 /// follow the item into the group: the structure docling-core requires,
468 /// the content the document had.
469 pub fn wrap_misplaced_list_items(&mut self) {
470 let is_list_item = |t: &Self, id: usize| matches!(&t.items[id].kind, TreeKind::Text { label, .. } if label == "list_item");
471 let in_list_group = |t: &Self, id: usize| {
472 t.items[id].parent.is_some_and(
473 |p| matches!(&t.items[p].kind, TreeKind::Group { label, .. } if label == "list"),
474 )
475 };
476 let mut runs: Vec<Vec<usize>> = Vec::new();
477 // `None` = the body itself, which the walk yields first.
478 let mut prev: Option<usize> = None;
479 let mut stack: Vec<usize> = self.body.iter().rev().copied().collect();
480 while let Some(id) = stack.pop() {
481 if self.items[id].deleted {
482 continue;
483 }
484 if is_list_item(self, id) && !in_list_group(self, id) {
485 let continues =
486 prev.is_some_and(|p| is_list_item(self, p) && self.items[p].parent.is_none());
487 match runs.last_mut() {
488 Some(run) if continues => run.push(id),
489 _ => runs.push(vec![id]),
490 }
491 }
492 prev = Some(id);
493 stack.extend(self.items[id].children.iter().rev().copied());
494 }
495 for run in runs.into_iter().rev() {
496 let parent = self.items[run[0]].parent;
497 let group = self.add(
498 parent,
499 None,
500 TreeKind::Group {
501 label: "list".into(),
502 name: "group".into(),
503 },
504 );
505 let siblings = match parent {
506 Some(p) => &mut self.items[p].children,
507 None => &mut self.body,
508 };
509 siblings.pop();
510 let at = siblings
511 .iter()
512 .position(|&c| c == run[0])
513 .unwrap_or(siblings.len());
514 siblings.insert(at, group);
515 for &li in &run {
516 // Not `delete_subtree`: the children move to the copy (#586).
517 self.delete(li);
518 }
519 // `add_list_item` keeps the text, marker, formatting, hyperlink
520 // and first provenance — not comments or a source. The children
521 // (the inline group of a mixed-format item) are carried over.
522 for &li in &run {
523 let children = std::mem::take(&mut self.items[li].children);
524 let copy = TreeItem {
525 parent: Some(group),
526 children: children.clone(),
527 comments: Vec::new(),
528 source: None,
529 deleted: false,
530 ..self.items[li].clone()
531 };
532 let id = self.items.len();
533 self.items.push(copy);
534 for c in children {
535 self.items[c].parent = Some(id);
536 }
537 self.items[group].children.push(id);
538 }
539 }
540 }
541
542 /// The last live text-bucket item (docling's `doc.texts[-1]`).
543 pub fn last_text(&self) -> Option<usize> {
544 self.items.iter().rposition(|it| {
545 !it.deleted && matches!(it.kind, TreeKind::Text { .. } | TreeKind::Code { .. })
546 })
547 }
548
549 /// How many items of a bucket precede `id` — its `#/{bucket}/N` index.
550 pub fn bucket_index(&self, id: usize) -> usize {
551 let same = |k: &TreeKind| {
552 std::mem::discriminant(k) == std::mem::discriminant(&self.items[id].kind)
553 || matches!(
554 (k, &self.items[id].kind),
555 (TreeKind::Text { .. }, TreeKind::Code { .. })
556 | (TreeKind::Code { .. }, TreeKind::Text { .. })
557 )
558 };
559 self.items[..id]
560 .iter()
561 .filter(|it| !it.deleted && same(&it.kind))
562 .count()
563 }
564
565 /// The number of tables created so far (docling's `len(doc.tables)`).
566 pub fn table_count(&self) -> usize {
567 self.items
568 .iter()
569 .filter(|it| !it.deleted && matches!(it.kind, TreeKind::Table { .. }))
570 .count()
571 }
572}
573
574#[cfg(test)]
575mod tests {
576 use super::*;
577
578 fn text(t: &str) -> TreeKind {
579 TreeKind::Text {
580 label: "text".into(),
581 text: t.into(),
582 orig: None,
583 formatting: None,
584 hyperlink: None,
585 level: None,
586 list: None,
587 }
588 }
589
590 /// `add` registers the item as its parent's (or the body's) last child;
591 /// `reparent` moves it — a rich cell's items leave the heading they were
592 /// created under for the table's group.
593 #[test]
594 fn add_and_reparent_keep_docling_children_order() {
595 let mut t = ItemTree::default();
596 let title = t.add(None, None, text("Title"));
597 let a = t.add(Some(title), None, text("a"));
598 let b = t.add(Some(title), None, text("b"));
599 let table = t.add(
600 Some(title),
601 None,
602 TreeKind::Table {
603 table: Table::default(),
604 rich_cells: Vec::new(),
605 captions: Vec::new(),
606 },
607 );
608 let group = t.add(
609 Some(table),
610 None,
611 TreeKind::Group {
612 label: "unspecified".into(),
613 name: "rich_cell_group_1_0_0".into(),
614 },
615 );
616 assert_eq!(t.body, vec![title]);
617 assert_eq!(t.items[title].children, vec![a, b, table]);
618 t.reparent(a, Some(group));
619 assert_eq!(t.items[title].children, vec![b, table]);
620 assert_eq!(t.items[group].children, vec![a]);
621 assert_eq!(t.items[a].parent, Some(group));
622 assert_eq!(t.table_count(), 1);
623 // Text and code share the `texts` bucket.
624 let code = t.add(
625 None,
626 None,
627 TreeKind::Code {
628 text: "x".into(),
629 orig: None,
630 language: None,
631 formatting: None,
632 hyperlink: None,
633 captions: Vec::new(),
634 },
635 );
636 assert_eq!(t.bucket_index(code), 3, "title, a, b precede it in `texts`");
637 assert_eq!(t.bucket_index(group), 0);
638 assert_eq!(t.body, vec![title, code]);
639 }
640
641 /// `append` renumbers a fragment built on its own (a slide converted in
642 /// parallel) so the merged tree reads as if built in one pass: parents,
643 /// children, comment back-refs and caption refs all shift together.
644 #[test]
645 fn append_renumbers_a_fragment_into_creation_order() {
646 let mut whole = ItemTree::default();
647 let slide0 = whole.add(
648 None,
649 None,
650 TreeKind::Group {
651 label: "chapter".into(),
652 name: "slide-0".into(),
653 },
654 );
655 whole.add(Some(slide0), None, text("first"));
656
657 let mut frag = ItemTree::default();
658 let slide1 = frag.add(
659 None,
660 None,
661 TreeKind::Group {
662 label: "chapter".into(),
663 name: "slide-1".into(),
664 },
665 );
666 let cap = frag.add_with_prov(
667 Some(slide1),
668 None,
669 TreeKind::Text {
670 label: "caption".into(),
671 text: "Title".into(),
672 orig: None,
673 formatting: None,
674 hyperlink: None,
675 level: None,
676 list: None,
677 },
678 TreeProv {
679 page_no: 2,
680 bbox: [1.0, 2.0, 3.0, 4.0],
681 bottom_left: true,
682 charspan: [0, 5],
683 },
684 );
685 let pic = frag.add(
686 Some(slide1),
687 None,
688 TreeKind::Picture {
689 captions: vec![cap],
690 image: None,
691 extra_classes: Vec::new(),
692 classification: Some("bar_chart".into()),
693 description: None,
694 confidence: None,
695 chart: None,
696 dpi: None,
697 },
698 );
699 let note = frag.add(
700 None,
701 Some(ContentLayer::Notes),
702 TreeKind::Group {
703 label: "comment_section".into(),
704 name: "comment-slide2-1".into(),
705 },
706 );
707 frag.items[pic].comments.push(note);
708
709 whole.append(frag);
710 assert_eq!(whole.body, vec![slide0, 2, 5]);
711 assert_eq!(whole.items[2].children, vec![3, 4]);
712 assert_eq!(whole.items[3].parent, Some(2));
713 assert_eq!(whole.items[3].prov.as_ref().map(|p| p.page_no), Some(2));
714 assert!(
715 matches!(&whole.items[4].kind, TreeKind::Picture { captions, .. } if captions == &[3])
716 );
717 assert_eq!(whole.items[4].comments, vec![5]);
718 assert_eq!(whole.items[5].parent, None);
719 assert_eq!(
720 whole.bucket_index(4),
721 0,
722 "the fragment's picture is #/pictures/0"
723 );
724 assert_eq!(
725 whole.bucket_index(5),
726 2,
727 "slide-0, slide-1 precede it in `groups`"
728 );
729 }
730
731 /// `renumber_in_traversal_order`: docling's `concatenate` re-creates the
732 /// items as a pre-order walk meets them, so a group created after the
733 /// content it was wrapped around (a rich cell's group) comes first, and a
734 /// deleted item disappears; every cross-reference follows.
735 #[test]
736 fn renumbering_follows_the_traversal() {
737 let mut t = ItemTree::default();
738 let a = t.add(None, None, text("a"));
739 let table = t.add(
740 None,
741 None,
742 TreeKind::Table {
743 table: Table::default(),
744 rich_cells: Vec::new(),
745 captions: Vec::new(),
746 },
747 );
748 let cell_text = t.add(None, None, text("cell"));
749 let group = t.add(
750 Some(table),
751 None,
752 TreeKind::Group {
753 label: "unspecified".into(),
754 name: "rich_cell_group_1_0_0".into(),
755 },
756 );
757 t.reparent(cell_text, Some(group));
758 if let TreeKind::Table { rich_cells, .. } = &mut t.items[table].kind {
759 rich_cells.push((0, 0, group));
760 }
761 let gone = t.add(None, None, text("gone"));
762 t.delete(gone);
763 let z = t.add(None, None, text("z"));
764
765 let new_of = t.renumber_in_traversal_order();
766 assert_eq!(
767 new_of,
768 vec![Some(0), Some(1), Some(3), Some(2), None, Some(4)]
769 );
770 assert_eq!(t.items.len(), 5);
771 assert_eq!(t.body, vec![0, 1, 4]);
772 assert_eq!(t.items[1].children, vec![2], "the group follows its table");
773 assert_eq!(t.items[2].parent, Some(1));
774 assert_eq!(
775 t.items[3].parent,
776 Some(2),
777 "the cell text follows its group"
778 );
779 assert!(matches!(&t.items[3].kind, TreeKind::Text { text, .. } if text == "cell"));
780 assert!(
781 matches!(&t.items[1].kind, TreeKind::Table { rich_cells, .. } if rich_cells == &[(0, 0, 2)])
782 );
783 let _ = (a, z);
784 }
785
786 /// #586: a misplaced `list_item` keeps its children when it is re-homed.
787 /// docling-core's `validate_misplaced_list_items` re-adds the item from
788 /// its text alone, so a mixed-format item (empty `list_item` over an
789 /// `inline` group of runs) lost every run — a DOCX table cell whose list
790 /// paragraph followed a `numId 0` spacer came out as an empty bullet in
791 /// the JSON and the HTML. The group goes where the item stood, the copy
792 /// is numbered last, the inline group and its texts move under it.
793 #[test]
794 fn wrapping_a_misplaced_list_item_keeps_its_runs() {
795 let mut t = ItemTree::default();
796 let cell = t.add(
797 None,
798 None,
799 TreeKind::Group {
800 label: "unspecified".into(),
801 name: "rich_cell_group_1_0_1".into(),
802 },
803 );
804 let before = t.add(Some(cell), None, text("before"));
805 let item = t.add(
806 Some(cell),
807 None,
808 TreeKind::Text {
809 label: "list_item".into(),
810 text: String::new(),
811 orig: None,
812 formatting: None,
813 hyperlink: None,
814 level: None,
815 list: Some(ListMeta {
816 enumerated: false,
817 marker: String::new(),
818 }),
819 },
820 );
821 let inline = t.add(
822 Some(item),
823 None,
824 TreeKind::Group {
825 label: "inline".into(),
826 name: "group".into(),
827 },
828 );
829 let run_a = t.add(Some(inline), None, text("Second item text"));
830 let run_b = t.add(Some(inline), None, text("[Optional]"));
831 let after = t.add(Some(cell), None, text("after"));
832
833 t.wrap_misplaced_list_items();
834
835 let group = t.items[cell].children[1];
836 assert_eq!(t.items[cell].children, vec![before, group, after]);
837 assert!(
838 matches!(&t.items[group].kind, TreeKind::Group { label, name } if label == "list" && name == "group")
839 );
840 assert!(t.items[item].deleted, "the original item is deleted");
841 let copy = t.items[group].children[0];
842 assert_ne!(copy, item);
843 assert!(
844 copy > after,
845 "the copy is numbered after every existing item"
846 );
847 assert_eq!(t.items[copy].children, vec![inline]);
848 assert_eq!(t.items[inline].parent, Some(copy));
849 assert_eq!(t.items[inline].children, vec![run_a, run_b]);
850 for id in [inline, run_a, run_b] {
851 assert!(!t.items[id].deleted, "item {id} must survive the re-homing");
852 }
853 }
854}