feat: update DoclingDocument model with correct wire formats#558
Merged
edeandrea merged 1 commit intoJun 22, 2026
Merged
Conversation
krickert
force-pushed
the
chore/update-docling-document-model
branch
from
b212f26 to
88e20e4
Compare
krickert
changed the title
Contributor
There was a problem hiding this comment.
Pull request overview
Updates docling-core’s DoclingDocument model to align Java serialization/deserialization with the Python docling-core v2.83.x wire format, adding new metadata fields and new document node types while tightening JSON shapes used on the wire.
Changes:
- Added new wire-format model types (e.g.,
FineRef,TrackSource, field region/item nodes, and new meta fields) and extended existing nodes withsourceandcomments. - Extended metadata (
language,entities,keywords,topics) and addedPictureMeta.code; addedOrientationand typedTableData.table_cellsasList<TableCell>. - Added/expanded Jackson round-trip tests for the new shapes (tuple-like arrays for
charspan/range, flatTrackSource, field text polymorphism, null-as-empty lists, typedtable_cells, etc.).
Reviewed changes
Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.
| File | Description |
|---|---|
| docling-core/src/main/java/ai/docling/core/DoclingDocument.java | Adds/updates model types and Jackson annotations to match the updated Python wire format (new meta fields, field nodes, source/comments, typed table cells, orientation, picture code meta). |
| docling-core/src/test/java/ai/docling/core/DoclingDocumentTests.java | Adds coverage for the new JSON wire shapes and null/empty collection handling. |
krickert
force-pushed
the
chore/update-docling-document-model
branch
from
88e20e4 to
af2fb07
Compare
krickert
force-pushed
the
chore/update-docling-document-model
branch
from
af2fb07 to
43a3eb1
Compare
Add new types from the Python docling-core model: - FineRef: $ref + range as [int,int] JSON array (not object) - TrackSource: flat discriminated object with kind/start_time/end_time - EntityMention: charspan as [int,int] JSON array matching CharSpan tuple - EntitiesMetaField, KeywordsMetaField, TopicsMetaField, LanguageMetaField - FieldHeadingItem, FieldValueItem (as sealed BaseTextItem subtypes) - FieldRegionItem, FieldItem (new top-level document item types) - CodeMetaField (PictureMeta.code) for code-backed picture nodes Update existing types: - BaseMeta/FloatingMeta/PictureMeta: add language, entities, keywords, topics; PictureMeta also gains the code (CodeMetaField) field - All BaseTextItem implementations: add source (List<TrackSource>) and comments (List<FineRef>) fields matching DocItem in the Python model - DocItemLabel: add FIELD_REGION, FIELD_HEADING, FIELD_ITEM, FIELD_KEY, FIELD_VALUE, FIELD_HINT, MARKER - DoclingDocument: add field_regions and field_items lists; add missing @JsonSetter(nulls = AS_EMPTY) to tables, key_value_items, form_items - TableData.table_cells: type as List<TableCell> (was List<Object>) so cells deserialize into typed TableCell instances, matching list[AnyTableCell] Add tests verifying correct JSON array wire format for charspan/range, flat TrackSource deserialization, polymorphic FieldHeadingItem/FieldValueItem dispatch, null-safe field_regions/field_items handling, PictureMeta sub-fields (description/classification/molecule/tabular_chart/code), and typed table_cells. Signed-off-by: Kristian Rickert <krickert@gmail.com>
edeandrea
force-pushed
the
chore/update-docling-document-model
branch
from
43a3eb1 to
4e3961e
Compare
:java_duke: JaCoCo coverage report
|
|
||||||||||||||
|
HTML test reports are available as workflow artifacts (zipped HTML). • Download: Artifacts for this run |
|
HTML test reports are available as workflow artifacts (zipped HTML). • Download: Artifacts for this run |
Contributor
Author
No problem. I have to maintain the model on my branch anyway. Now .. If there was only a technology out there so you just do model updates once ... Like a blockchain integration? Langblock4j?! |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Updates the
docling-coreJavaDoclingDocumentmodel to match the Python docling-core wire format (v2.83.x).New types
FineRef,TrackSource,EntityMention,EntitiesMetaField,KeywordsMetaField,TopicsMetaField,LanguageMetaField,CodeMetaField,FieldHeadingItem,FieldValueItem,FieldRegionItem,FieldItem.Updated types
BaseMeta/FloatingMeta/PictureMetagainlanguage,entities,keywords,topics;PictureMetaalso gainscode(CodeMetaField).BaseTextItemimplementations gainsource(List<TrackSource>) andcomments(List<FineRef>) matchingDocItem.DocItemLabeladdsFIELD_REGION,FIELD_HEADING,FIELD_ITEM,FIELD_KEY,FIELD_VALUE,FIELD_HINT,MARKER.DoclingDocumentaddsfield_regionsandfield_itemslists.TableData.table_cellstyped asList<TableCell>(wasList<Object>).Wire-format details
charspan/rangeserialize as 2-element[int, int]JSON arrays (Pythontuple[int, int]parity).TrackSourceis a flat object with akind: "track"discriminator.labelfor field heading/value text items.Tests
Adds round-trip tests for the new wire shapes (charspan/range arrays, flat
TrackSource, polymorphic field text items, null-safe field collections,PictureMetasub-fields incl.code, typedtable_cells)../gradlew :docling-core:testis green and spotless is clean.