docling-core

Feature

• experimental: Add new DocTags serializer (#412) (c9e5fb4)
• Convert regions into TableData (#430) (c80b583)

Fix

• Improve meta migration (#422) (bc0e96b)
• DoclingDocument model validator should deal with any raw input (#419) (56b3c42)

Feature

• Add code chunking functionality (#398) (3097645)

Fix

• Improve meta migration and warning handling (#417) (3d13b02)
• Fix import handling of extra dependencies for chunking (#418) (567d3ad)

Fix

• Add JSON to CodeLanguageLabel (#413) (09ef91c)

Feature

• Add metadata model hierarchy (#408) (2ee3cac)
• Add split view & YAML support to CLI viewer (#407) (a3feae0)
• New picture classes for doctags (#404) (ada4068)

Feature

• Python 3.14 compatibility (#403) (47b70d3)
• Rendering of checkboxes in markdown (#381) (abfa81f)

Fix

• Switch to safe YAML loader (#396) (3e8d628)

Fix

• Pin larger range of typer (#394) (26d21e3)

Fix

• Expose escape_html param to DoclingDocument md serialization (#388) (dd6ebc3)

Fix

• markdown: Fix single-row table serialization (#385) (9df7208)

Feature

• Introduction of fillable TableCell (#384) (b13267f)
• Add support for heading with inline in HTML & DocTags (#379) (b60ac19)

Fix

• Add doc param to all export_to_dataframe() calls (#380) (0512f44)
• Fix handling of generic groups in rich table cells (#383) (2dc57c1)

Feature

• Add page filtering to DoclingDocument (#378) (9dc526d)

Feature

• Add rich table cells (#368) (1d04154)

Fix

• Fix text dir attribute (#281) (3f25267)
• Use in-OTSL DocTags for rich table cells (#375) (b2095b3)

Performance

• Cache grid property in HTMLTableSerializer (#373) (339bbd4)

Feature

• Add document concatenation (#365) (99eabb3)

Fix

• Add forward slashes to singleton tags (#369) (23badf2)

Fix

• HTML: Fix nested list serialization edge cases (#367) (807d972)

Fix

• Referenced artifacts relative to the document location (#361) (5afa99e)

Feature

• Key-value visualizer (#360) (0f7ca77)

Fix

• LayoutVisualizer should traverse pictures (#358) (f9b3b49)
• HTML serialization of nested lists (#359) (5a7883c)

Feature

• Add page chunking (#337) (3a0b747)

Fix

• Only save applicable page images (#226) (ebd9147)

Feature

• Extend and expose float serialization control (#353) (c339171)
• Additional DoclingDocument methods for use in MCP document manipulation (#344) (cb59fd3)

Feature

• Enable precision control in float serialization (#352) (baa2cc3)

Feature

• Added different content-layers (#345) (eb2538e)

Fix

• BoundingRectangle angle computation when in CoordOrigin.TOPLEFT (#347) (9fa0c9f)

Feature

• Remodel lists, add MD & HTML ser. params, enable unset marker (#339) (14a4fde)
• Download Google docs and drive files via export url (#335) (3eeb259)

Fix

• Add missing mimetypes for asr inputs (#341) (c2fd20f)
• Add text direction to export_to_textlines (#338) (425b191)

Fix

• markdown: Add heading formatting, fix code & formula formatting (#336) (c9374e8)

Feature

• viz: Add reading order branch numbering, fix cross-page lists (#334) (78b7962)
• Add parameter to choose of which pages export the doctags (#290) (0fd3c1c)

Fix

• Expose base types consistently (#332) (2e14a74)
• HybridChunker: Improve long heading handling (#333) (5c99722)

Feature

• Add improved table serializer and visualizer (#328) (3b99879)

Feature

• New labels for CVAT annotation (#314) (aa430d3)

Feature

• Visualizer for table cells (#325) (45667c7)

Fix

• Fix doc traversal for item deletion (#324) (076ad2b)

Fix

• Warn when adding misplaced ListItem via API (#321) (01b27b5)

Feature

• doctags: Add enclosing bbox to inline (#302) (dcc198f)
• Add subscript & superscript formatting (#319) (ae96129)
• Add table annotations (#304) (d8a5256)

Fix

• Fix misplaced list items (#317) (c383f64)

Fix

• New typer version with new click (#315) (e17eabf)
• Support section_header levels in doctags deserialization (#313) (defd49e)

Feature

• Add BoundingBox methods for overlap and union calculations (#311) (c521766)

Feature

• Add annotations in MD & HTML serialization (#295) (f067c51)

Fix

• HybridChunker: Refine max_tokens auto-detection (#306) (87b72d6)

Fix

• Fix hybrid chunker legacy patching (#300) (ebc356a)

Fix

• markdown: Fix case of empty page break string (#298) (c49a50e)

Feature

• Provide visualizer option in HTML split view (#294) (6a7eb53)

Fix

• Updates for labels and methods to support document GT annotation (#293) (aa957cf)

Feature

• Add image group serialization in html (#284) (7f83f1c)
• Adding the label picture_group (#283) (2f0f121)

Fix

• Add unit flags to SegmentedPage (#286) (ad88ecf)
• Update deserialization for better recovery (#282) (511fb98)
• Include captions regardless of traverse_pictures flag (#278) (7eb9fa9)
• Hashlib usage for FIPS (#280) (4b967ab)

Feature

• Promote serializers to stable API (#276) (d05fe08)

Fix

• Fix multi-provenance item visualization (#277) (8677d6e)
• Added return value for crop_text method in segmentedPdfPage Class (#275) (591fe59)
• Make load_from_doctags method static (#273) (8f85d05)

Fix

• Visualization of document pages without items (#271) (a947440)
• UnboundLocal variable (#269) (d9709d0)

Feature

• Add tiktoken tokenizers support to HybridChunker (#240) (763e136)
• Add visualizers (#263) (a258d52)

Feature

• Chart tabular data serialization for HTML serializer (#258) (caa8aee)

Fix

• HTML serialization for single image documents (#261) (d0a49da)
• codecov: Fix codecov argument and yaml file (#260) (1af0721)
• Safer label color API (#259) (159f61d)

Fix

• Fix page breaking in case page starts with group (#253) (928e5c5)

Fix

• deps: Widen typer upper bound (#252) (437aef0)

Fix

• Fix code handling in HTML serialization (#251) (15d2f2c)

Performance

• Fix serialization performance (#249) (98c60bb)

Feature

• Add HTML serializer (#232) (5d40600)
• Add serializer provider to chunkers (#239) (23036e1)
• Integrate serialization API into chunkers (#221) (5e4c0fd)
• Expose page number in Serialization API (#238) (73b9941)
• Markdown chart serializer (picture+table) (#235) (0482bac)
• Support of DocTags charts (serialization and deserialization) (#229) (e9259a5)
• Added initial delete and insert methods in DoclingDocument (#220) (f2fe1c1)

Fix

• Fix page filtering issue (#247) (ab78e0b)
• Propagate HTMLOutputStyle properly through (#246) (587e67f)
• Better BoundingRectangle.angle and BoundingRectangle.angle_360 computation (#237) (055742c)
• DocTags import location fix for tables, pictures, captions (#227) (a055e1a)

Performance

• Accelerate span deduplication (#248) (26f639d)

Feature

• Allow images in doctags deserializer to be optional and support multipage (#225) (e0943d2)

Fix

• DivisionByZero in intersection_over_self (#224) (2f380ab)
• Fix hyperlink deserialization (#223) (57d26ee)

Fix

• Automatic transformation of output cells bbox coord origin defined by input in get_cells_in_bbox (#219) (8e0e9b7)

[!IMPORTANT]

API notice

• export_to_document_tokens() methods have been deprecated in favor of newly introduced export_to_doctags() methods, to which they are now internally delegating.
• In export_to_doctags() methods, arguments new_line / delim have been deprecated — check minified instead.

Feature

• Expose MD page break & DocTags minification (#213) (ff13a93)
• Add document tokens from key value items (#170) (db119f4)
• Add DocTags serializers (#192) (1f4d57e)
• Add kv_item support for doctag to docling_document (#188) (2371c11)

Fix

• Enable caption serialization for all floating items (#216) (e1d0597)
• Allow captions without holding item (#215) (2efb71a)
• Add 'text/csv' mimetype to _extra_mimetypes type list (#210) (bc3f5d5)
• Add handling for str filenames in save/load methods (#205) (75d94ab)
• Markdown picture item export (#207) (510649e)
• DocTags support of furniture (#209) (337ff74)

Performance

• serialization: Cache excluded references (#214) (bcace5d)

Fix

• markdown: Fix ordered list numbering (#200) (7ed4d22)

Fix

• Add caption to the table in load_from_doctags (#197) (5cee486)

Fix

• New favicon url (#191) (11eb1dc)

Feature

• Add serializers, text formatting, update Markdown export (#182) (a7cdc87)
• Add data model types from docling-parse (#186) (a86a4a3)

Feature

• Add DoclingDocument.load_from_doctags method and DocTags data models (#187) (c065c4c)
• Add document tokens for SMILES (#176) (32398b8)

Fix

• Suppress warning for missing fallback case (#184) (ccde54a)
• doctags: Fix code export (#181) (53f6d09)
• markdown: Fix escaping in case of nesting (#180) (834db4b)
• HybridChunker: Remove max_length from tokenization (#178) (419252c)