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 /// Link destinations at this item (#849): the identifiers a document's
197 /// internal links (`hyperlink: "#id"`) point at — a DOCX bookmark in
198 /// a heading, say. docling's model has no slot for them, so the JSON
199 /// shows them only on the Pandoc path (`_anchors`), where a heading's
200 /// first one is its `Header` id and the rest empty `Span .anchor`s.
201 pub anchors: Vec<String>,
202 /// The generated numbering `text` opens with, when the backend put one
203 /// there (#850: a DOCX heading numbered through `w:numPr` is docling's
204 /// `1.2 Title`): `"1.2"` — the authored text follows it after one
205 /// space. docling's JSON keeps the numbered text; the Pandoc AST writes
206 /// the authored text alone (`_number` on that path), since Pandoc
207 /// numbers sections itself (`--number-sections`) and its docx reader
208 /// never writes the number into the `Header`.
209 pub number: Option<String>,
210 /// The tracked change this run carries under
211 /// [`TrackChanges::All`](crate::TrackChanges::All) (#851): the Pandoc
212 /// AST wraps the run in `Span .insertion` / `.deletion` (`_revision`
213 /// on that path); docling's JSON has no field for it.
214 pub revision: Option<crate::Revision>,
215}
216
217/// A note call inside a text item ([`TreeItem::notes`]).
218#[derive(Debug, Clone, PartialEq, Eq)]
219pub struct TreeNote {
220 /// The call's position, in chars into the item's `text`.
221 pub offset: usize,
222 /// The note's plain text.
223 pub text: String,
224}
225
226impl ItemTree {
227 /// Where each note call of one paragraph lands (#538): `full_text` is the
228 /// paragraph's text, `offsets` the calls' positions in it (chars), and
229 /// the paragraph's items are those created from `first_new` on. Each text
230 /// item is a trimmed slice of the paragraph (a formatting run, a link, or
231 /// the whole heading / list item), matched in order; a call inside an
232 /// item lands there, one between items ends the earlier (or starts the
233 /// first). When no item matches, the call ends the paragraph's last text
234 /// item; with no text item at all it is `None`.
235 pub fn place_note_calls(
236 &self,
237 first_new: usize,
238 full_text: &str,
239 offsets: &[usize],
240 ) -> Vec<Option<(usize, usize)>> {
241 let texts: Vec<(usize, &str)> = (first_new..self.items.len())
242 .filter(|&i| !self.items[i].deleted)
243 .filter_map(|i| match &self.items[i].kind {
244 TreeKind::Text { text, .. } | TreeKind::Code { text, .. } => {
245 Some((i, text.as_str()))
246 }
247 _ => None,
248 })
249 .collect();
250 // Each found item's (id, start, end) in chars of `full_text`.
251 let mut spans: Vec<(usize, usize, usize)> = Vec::new();
252 let mut cursor = 0usize; // bytes
253 for &(item, text) in &texts {
254 if text.is_empty() {
255 continue;
256 }
257 if let Some(pos) = full_text[cursor..].find(text) {
258 let start_b = cursor + pos;
259 let start = full_text[..start_b].chars().count();
260 spans.push((item, start, start + text.chars().count()));
261 cursor = start_b + text.len();
262 }
263 }
264 offsets
265 .iter()
266 .map(|&offset| {
267 spans
268 .iter()
269 .find(|&&(_, start, end)| offset >= start && offset <= end)
270 .map(|&(item, start, _)| (item, offset - start))
271 .or_else(
272 || match spans.iter().rev().find(|&&(_, _, end)| end <= offset) {
273 Some(&(item, start, end)) => Some((item, end - start)),
274 None => spans.first().map(|&(item, _, _)| (item, 0)),
275 },
276 )
277 .or_else(|| {
278 texts
279 .last()
280 .map(|&(item, text)| (item, text.chars().count()))
281 })
282 })
283 .collect()
284 }
285}
286
287/// docling's item tree in creation order (see the [module docs](self)).
288#[derive(Debug, Clone, PartialEq, Default)]
289pub struct ItemTree {
290 /// Every item, indexed by creation order — which is how docling numbers
291 /// `#/texts/N`, `#/groups/N`, … within each bucket.
292 pub items: Vec<TreeItem>,
293 /// The body's `children`, as item indices.
294 pub body: Vec<usize>,
295}
296
297impl ItemTree {
298 /// Append an item under `parent` (`None` = body) on `layer`, registering
299 /// it as its parent's last child — docling's `add_*` calls do exactly that.
300 pub fn add(
301 &mut self,
302 parent: Option<usize>,
303 layer: Option<ContentLayer>,
304 kind: TreeKind,
305 ) -> usize {
306 let id = self.items.len();
307 self.items.push(TreeItem {
308 parent,
309 children: Vec::new(),
310 layer,
311 kind,
312 prov: None,
313 comments: Vec::new(),
314 source: None,
315 deleted: false,
316 notes: Vec::new(),
317 note_body: false,
318 anchors: Vec::new(),
319 number: None,
320 revision: None,
321 });
322 match parent {
323 Some(p) => self.items[p].children.push(id),
324 None => self.body.push(id),
325 }
326 id
327 }
328
329 /// [`add`](Self::add) with the item's provenance — docling's
330 /// `add_text(…, prov=prov)`.
331 pub fn add_with_prov(
332 &mut self,
333 parent: Option<usize>,
334 layer: Option<ContentLayer>,
335 kind: TreeKind,
336 prov: TreeProv,
337 ) -> usize {
338 let id = self.add(parent, layer, kind);
339 self.items[id].prov = Some(prov);
340 id
341 }
342
343 /// Append every item of `other` after this tree's, renumbering its
344 /// indices (parents, children, comments, table/picture caption and
345 /// rich-cell refs) and adding its body children to this body — so a
346 /// backend can build independent fragments in parallel (one per PPTX
347 /// slide) and still hand the export one tree in creation order, exactly
348 /// as if it had been built sequentially.
349 pub fn append(&mut self, other: ItemTree) {
350 let off = self.items.len();
351 let shift = |i: usize| i + off;
352 for mut item in other.items {
353 item.parent = item.parent.map(shift);
354 for c in item.children.iter_mut().chain(item.comments.iter_mut()) {
355 *c = shift(*c);
356 }
357 match &mut item.kind {
358 TreeKind::Table {
359 rich_cells,
360 captions,
361 ..
362 } => {
363 for (_, _, g) in rich_cells.iter_mut() {
364 *g = shift(*g);
365 }
366 for c in captions.iter_mut() {
367 *c = shift(*c);
368 }
369 }
370 TreeKind::Picture { captions, .. } | TreeKind::Code { captions, .. } => {
371 for c in captions.iter_mut() {
372 *c = shift(*c);
373 }
374 }
375 _ => {}
376 }
377 self.items.push(item);
378 }
379 self.body.extend(other.body.into_iter().map(shift));
380 }
381
382 /// Move `id` under `new_parent`, dropping it from its current parent's
383 /// children and appending it to the new one's — docling's
384 /// `group_cell_elements` re-parenting of a rich cell's items.
385 pub fn reparent(&mut self, id: usize, new_parent: Option<usize>) {
386 let old = self.items[id].parent;
387 let siblings = match old {
388 Some(p) => &mut self.items[p].children,
389 None => &mut self.body,
390 };
391 siblings.retain(|&c| c != id);
392 self.items[id].parent = new_parent;
393 match new_parent {
394 Some(p) => self.items[p].children.push(id),
395 None => self.body.push(id),
396 }
397 }
398
399 /// Re-number the items in traversal order — a pre-order walk of the body
400 /// through every layer, which is how docling's
401 /// `DoclingDocument.concatenate` (and `_normalize_references`) re-creates
402 /// a document's items: a group created after the content it was later
403 /// wrapped around comes before that content afterwards. Items the walk
404 /// does not reach (deleted ones) are dropped. Returns each old id's new
405 /// id.
406 pub fn renumber_in_traversal_order(&mut self) -> Vec<Option<usize>> {
407 let mut order = Vec::with_capacity(self.items.len());
408 let mut stack: Vec<usize> = self.body.iter().rev().copied().collect();
409 while let Some(id) = stack.pop() {
410 if self.items[id].deleted {
411 continue;
412 }
413 order.push(id);
414 stack.extend(self.items[id].children.iter().rev());
415 }
416 let mut new_of: Vec<Option<usize>> = vec![None; self.items.len()];
417 for (new, &old) in order.iter().enumerate() {
418 new_of[old] = Some(new);
419 }
420 let remap = |ids: &mut Vec<usize>| {
421 *ids = ids.iter().filter_map(|&i| new_of[i]).collect();
422 };
423 let mut old_items: Vec<Option<TreeItem>> = std::mem::take(&mut self.items)
424 .into_iter()
425 .map(Some)
426 .collect();
427 for &old in &order {
428 let mut item = old_items[old].take().expect("each item visited once");
429 item.parent = item.parent.and_then(|p| new_of[p]);
430 remap(&mut item.children);
431 remap(&mut item.comments);
432 match &mut item.kind {
433 TreeKind::Table {
434 rich_cells,
435 captions,
436 ..
437 } => {
438 rich_cells.retain_mut(|(_, _, g)| match new_of[*g] {
439 Some(n) => {
440 *g = n;
441 true
442 }
443 None => false,
444 });
445 remap(captions);
446 }
447 TreeKind::Picture { captions, .. } | TreeKind::Code { captions, .. } => {
448 remap(captions)
449 }
450 _ => {}
451 }
452 self.items.push(item);
453 }
454 remap(&mut self.body);
455 new_of
456 }
457
458 /// Remove `id` from the tree — docling's `delete_items`, which the DOCX
459 /// backend uses to drop the empty text item a blank spacer paragraph left
460 /// between two items of a resumed list. The item leaves its parent's
461 /// children and is neither numbered nor written; its slot stays so the
462 /// indices held elsewhere stay valid.
463 pub fn delete(&mut self, id: usize) {
464 match self.items[id].parent {
465 Some(p) => self.items[p].children.retain(|&c| c != id),
466 None => self.body.retain(|&c| c != id),
467 }
468 self.items[id].deleted = true;
469 }
470
471 /// docling-core's `DoclingDocument.validate_misplaced_list_items` (a
472 /// model validator, so it runs whenever docling-core serializes or loads
473 /// a document): every `list_item` whose parent is not a `list` group is
474 /// re-homed into a new one. A pre-order walk of the body (groups
475 /// included) collects them; consecutive misplaced items directly on the
476 /// body share one group, any other misplaced item gets its own. Working
477 /// from the last run back, each run gets a `ListGroup` (name `group`)
478 /// inserted where its first item stood, the items are deleted and
479 /// re-added under the group as fresh items, so they move to the end of
480 /// the text numbering (#527: the DOCX backend leaves such items in rich
481 /// table cells). A no-op for a well-formed tree.
482 ///
483 /// One deliberate divergence (#586): docling-core deletes each item *with
484 /// its children* and re-adds it from its text alone, so a mixed-format
485 /// item — an empty `list_item` over an `inline` group of text runs —
486 /// comes back empty and its text is gone from the JSON (and from the
487 /// Markdown/HTML docling serializes after that validator ran; the JSON
488 /// docling saves *before* exporting still has it). Here the children
489 /// follow the item into the group: the structure docling-core requires,
490 /// the content the document had.
491 pub fn wrap_misplaced_list_items(&mut self) {
492 let is_list_item = |t: &Self, id: usize| matches!(&t.items[id].kind, TreeKind::Text { label, .. } if label == "list_item");
493 let in_list_group = |t: &Self, id: usize| {
494 t.items[id].parent.is_some_and(
495 |p| matches!(&t.items[p].kind, TreeKind::Group { label, .. } if label == "list"),
496 )
497 };
498 let mut runs: Vec<Vec<usize>> = Vec::new();
499 // `None` = the body itself, which the walk yields first.
500 let mut prev: Option<usize> = None;
501 let mut stack: Vec<usize> = self.body.iter().rev().copied().collect();
502 while let Some(id) = stack.pop() {
503 if self.items[id].deleted {
504 continue;
505 }
506 if is_list_item(self, id) && !in_list_group(self, id) {
507 let continues =
508 prev.is_some_and(|p| is_list_item(self, p) && self.items[p].parent.is_none());
509 match runs.last_mut() {
510 Some(run) if continues => run.push(id),
511 _ => runs.push(vec![id]),
512 }
513 }
514 prev = Some(id);
515 stack.extend(self.items[id].children.iter().rev().copied());
516 }
517 for run in runs.into_iter().rev() {
518 let parent = self.items[run[0]].parent;
519 let group = self.add(
520 parent,
521 None,
522 TreeKind::Group {
523 label: "list".into(),
524 name: "group".into(),
525 },
526 );
527 let siblings = match parent {
528 Some(p) => &mut self.items[p].children,
529 None => &mut self.body,
530 };
531 siblings.pop();
532 let at = siblings
533 .iter()
534 .position(|&c| c == run[0])
535 .unwrap_or(siblings.len());
536 siblings.insert(at, group);
537 for &li in &run {
538 // Not `delete_subtree`: the children move to the copy (#586).
539 self.delete(li);
540 }
541 // `add_list_item` keeps the text, marker, formatting, hyperlink
542 // and first provenance — not comments or a source. The children
543 // (the inline group of a mixed-format item) are carried over.
544 for &li in &run {
545 let children = std::mem::take(&mut self.items[li].children);
546 let copy = TreeItem {
547 parent: Some(group),
548 children: children.clone(),
549 comments: Vec::new(),
550 source: None,
551 deleted: false,
552 ..self.items[li].clone()
553 };
554 let id = self.items.len();
555 self.items.push(copy);
556 for c in children {
557 self.items[c].parent = Some(id);
558 }
559 self.items[group].children.push(id);
560 }
561 }
562 }
563
564 /// The last live text-bucket item (docling's `doc.texts[-1]`).
565 pub fn last_text(&self) -> Option<usize> {
566 self.items.iter().rposition(|it| {
567 !it.deleted && matches!(it.kind, TreeKind::Text { .. } | TreeKind::Code { .. })
568 })
569 }
570
571 /// How many items of a bucket precede `id` — its `#/{bucket}/N` index.
572 pub fn bucket_index(&self, id: usize) -> usize {
573 let same = |k: &TreeKind| {
574 std::mem::discriminant(k) == std::mem::discriminant(&self.items[id].kind)
575 || matches!(
576 (k, &self.items[id].kind),
577 (TreeKind::Text { .. }, TreeKind::Code { .. })
578 | (TreeKind::Code { .. }, TreeKind::Text { .. })
579 )
580 };
581 self.items[..id]
582 .iter()
583 .filter(|it| !it.deleted && same(&it.kind))
584 .count()
585 }
586
587 /// The number of tables created so far (docling's `len(doc.tables)`).
588 pub fn table_count(&self) -> usize {
589 self.items
590 .iter()
591 .filter(|it| !it.deleted && matches!(it.kind, TreeKind::Table { .. }))
592 .count()
593 }
594}
595
596#[cfg(test)]
597mod tests {
598 use super::*;
599
600 fn text(t: &str) -> TreeKind {
601 TreeKind::Text {
602 label: "text".into(),
603 text: t.into(),
604 orig: None,
605 formatting: None,
606 hyperlink: None,
607 level: None,
608 list: None,
609 }
610 }
611
612 /// `add` registers the item as its parent's (or the body's) last child;
613 /// `reparent` moves it — a rich cell's items leave the heading they were
614 /// created under for the table's group.
615 #[test]
616 fn add_and_reparent_keep_docling_children_order() {
617 let mut t = ItemTree::default();
618 let title = t.add(None, None, text("Title"));
619 let a = t.add(Some(title), None, text("a"));
620 let b = t.add(Some(title), None, text("b"));
621 let table = t.add(
622 Some(title),
623 None,
624 TreeKind::Table {
625 table: Table::default(),
626 rich_cells: Vec::new(),
627 captions: Vec::new(),
628 },
629 );
630 let group = t.add(
631 Some(table),
632 None,
633 TreeKind::Group {
634 label: "unspecified".into(),
635 name: "rich_cell_group_1_0_0".into(),
636 },
637 );
638 assert_eq!(t.body, vec![title]);
639 assert_eq!(t.items[title].children, vec![a, b, table]);
640 t.reparent(a, Some(group));
641 assert_eq!(t.items[title].children, vec![b, table]);
642 assert_eq!(t.items[group].children, vec![a]);
643 assert_eq!(t.items[a].parent, Some(group));
644 assert_eq!(t.table_count(), 1);
645 // Text and code share the `texts` bucket.
646 let code = t.add(
647 None,
648 None,
649 TreeKind::Code {
650 text: "x".into(),
651 orig: None,
652 language: None,
653 formatting: None,
654 hyperlink: None,
655 captions: Vec::new(),
656 },
657 );
658 assert_eq!(t.bucket_index(code), 3, "title, a, b precede it in `texts`");
659 assert_eq!(t.bucket_index(group), 0);
660 assert_eq!(t.body, vec![title, code]);
661 }
662
663 /// `append` renumbers a fragment built on its own (a slide converted in
664 /// parallel) so the merged tree reads as if built in one pass: parents,
665 /// children, comment back-refs and caption refs all shift together.
666 #[test]
667 fn append_renumbers_a_fragment_into_creation_order() {
668 let mut whole = ItemTree::default();
669 let slide0 = whole.add(
670 None,
671 None,
672 TreeKind::Group {
673 label: "chapter".into(),
674 name: "slide-0".into(),
675 },
676 );
677 whole.add(Some(slide0), None, text("first"));
678
679 let mut frag = ItemTree::default();
680 let slide1 = frag.add(
681 None,
682 None,
683 TreeKind::Group {
684 label: "chapter".into(),
685 name: "slide-1".into(),
686 },
687 );
688 let cap = frag.add_with_prov(
689 Some(slide1),
690 None,
691 TreeKind::Text {
692 label: "caption".into(),
693 text: "Title".into(),
694 orig: None,
695 formatting: None,
696 hyperlink: None,
697 level: None,
698 list: None,
699 },
700 TreeProv {
701 page_no: 2,
702 bbox: [1.0, 2.0, 3.0, 4.0],
703 bottom_left: true,
704 charspan: [0, 5],
705 },
706 );
707 let pic = frag.add(
708 Some(slide1),
709 None,
710 TreeKind::Picture {
711 captions: vec![cap],
712 image: None,
713 extra_classes: Vec::new(),
714 classification: Some("bar_chart".into()),
715 description: None,
716 confidence: None,
717 chart: None,
718 dpi: None,
719 },
720 );
721 let note = frag.add(
722 None,
723 Some(ContentLayer::Notes),
724 TreeKind::Group {
725 label: "comment_section".into(),
726 name: "comment-slide2-1".into(),
727 },
728 );
729 frag.items[pic].comments.push(note);
730
731 whole.append(frag);
732 assert_eq!(whole.body, vec![slide0, 2, 5]);
733 assert_eq!(whole.items[2].children, vec![3, 4]);
734 assert_eq!(whole.items[3].parent, Some(2));
735 assert_eq!(whole.items[3].prov.as_ref().map(|p| p.page_no), Some(2));
736 assert!(
737 matches!(&whole.items[4].kind, TreeKind::Picture { captions, .. } if captions == &[3])
738 );
739 assert_eq!(whole.items[4].comments, vec![5]);
740 assert_eq!(whole.items[5].parent, None);
741 assert_eq!(
742 whole.bucket_index(4),
743 0,
744 "the fragment's picture is #/pictures/0"
745 );
746 assert_eq!(
747 whole.bucket_index(5),
748 2,
749 "slide-0, slide-1 precede it in `groups`"
750 );
751 }
752
753 /// `renumber_in_traversal_order`: docling's `concatenate` re-creates the
754 /// items as a pre-order walk meets them, so a group created after the
755 /// content it was wrapped around (a rich cell's group) comes first, and a
756 /// deleted item disappears; every cross-reference follows.
757 #[test]
758 fn renumbering_follows_the_traversal() {
759 let mut t = ItemTree::default();
760 let a = t.add(None, None, text("a"));
761 let table = t.add(
762 None,
763 None,
764 TreeKind::Table {
765 table: Table::default(),
766 rich_cells: Vec::new(),
767 captions: Vec::new(),
768 },
769 );
770 let cell_text = t.add(None, None, text("cell"));
771 let group = t.add(
772 Some(table),
773 None,
774 TreeKind::Group {
775 label: "unspecified".into(),
776 name: "rich_cell_group_1_0_0".into(),
777 },
778 );
779 t.reparent(cell_text, Some(group));
780 if let TreeKind::Table { rich_cells, .. } = &mut t.items[table].kind {
781 rich_cells.push((0, 0, group));
782 }
783 let gone = t.add(None, None, text("gone"));
784 t.delete(gone);
785 let z = t.add(None, None, text("z"));
786
787 let new_of = t.renumber_in_traversal_order();
788 assert_eq!(
789 new_of,
790 vec![Some(0), Some(1), Some(3), Some(2), None, Some(4)]
791 );
792 assert_eq!(t.items.len(), 5);
793 assert_eq!(t.body, vec![0, 1, 4]);
794 assert_eq!(t.items[1].children, vec![2], "the group follows its table");
795 assert_eq!(t.items[2].parent, Some(1));
796 assert_eq!(
797 t.items[3].parent,
798 Some(2),
799 "the cell text follows its group"
800 );
801 assert!(matches!(&t.items[3].kind, TreeKind::Text { text, .. } if text == "cell"));
802 assert!(
803 matches!(&t.items[1].kind, TreeKind::Table { rich_cells, .. } if rich_cells == &[(0, 0, 2)])
804 );
805 let _ = (a, z);
806 }
807
808 /// #586: a misplaced `list_item` keeps its children when it is re-homed.
809 /// docling-core's `validate_misplaced_list_items` re-adds the item from
810 /// its text alone, so a mixed-format item (empty `list_item` over an
811 /// `inline` group of runs) lost every run — a DOCX table cell whose list
812 /// paragraph followed a `numId 0` spacer came out as an empty bullet in
813 /// the JSON and the HTML. The group goes where the item stood, the copy
814 /// is numbered last, the inline group and its texts move under it.
815 #[test]
816 fn wrapping_a_misplaced_list_item_keeps_its_runs() {
817 let mut t = ItemTree::default();
818 let cell = t.add(
819 None,
820 None,
821 TreeKind::Group {
822 label: "unspecified".into(),
823 name: "rich_cell_group_1_0_1".into(),
824 },
825 );
826 let before = t.add(Some(cell), None, text("before"));
827 let item = t.add(
828 Some(cell),
829 None,
830 TreeKind::Text {
831 label: "list_item".into(),
832 text: String::new(),
833 orig: None,
834 formatting: None,
835 hyperlink: None,
836 level: None,
837 list: Some(ListMeta {
838 enumerated: false,
839 marker: String::new(),
840 }),
841 },
842 );
843 let inline = t.add(
844 Some(item),
845 None,
846 TreeKind::Group {
847 label: "inline".into(),
848 name: "group".into(),
849 },
850 );
851 let run_a = t.add(Some(inline), None, text("Second item text"));
852 let run_b = t.add(Some(inline), None, text("[Optional]"));
853 let after = t.add(Some(cell), None, text("after"));
854
855 t.wrap_misplaced_list_items();
856
857 let group = t.items[cell].children[1];
858 assert_eq!(t.items[cell].children, vec![before, group, after]);
859 assert!(
860 matches!(&t.items[group].kind, TreeKind::Group { label, name } if label == "list" && name == "group")
861 );
862 assert!(t.items[item].deleted, "the original item is deleted");
863 let copy = t.items[group].children[0];
864 assert_ne!(copy, item);
865 assert!(
866 copy > after,
867 "the copy is numbered after every existing item"
868 );
869 assert_eq!(t.items[copy].children, vec![inline]);
870 assert_eq!(t.items[inline].parent, Some(copy));
871 assert_eq!(t.items[inline].children, vec![run_a, run_b]);
872 for id in [inline, run_a, run_b] {
873 assert!(!t.items[id].deleted, "item {id} must survive the re-homing");
874 }
875 }
876}