chore: merge upstream/main and resolve docstring conflict

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

Author: mrowdy

.github/workflows/checks.yml

@@ -20,7 +20,7 @@ env:
     tests/test_asr_pipeline.py
     tests/test_threaded_pipeline.py
   PYTEST_TO_SKIP: |-
-  EXAMPLES_TO_SKIP: '^(batch_convert|compare_vlm_models|minimal|minimal_vlm_pipeline|minimal_asr_pipeline|export_multimodal|custom_convert|develop_picture_enrichment|rapidocr_with_custom_models|suryaocr_with_custom_models|offline_convert|pictures_description|pictures_description_api|vlm_pipeline_api_model|granitedocling_repetition_stopping|mlx_whisper_example|gpu_standard_pipeline|gpu_vlm_pipeline|demo_layout_vlm|post_process_ocr_with_vlm)\.py$'
+  EXAMPLES_TO_SKIP: '^(batch_convert|compare_vlm_models|minimal|minimal_vlm_pipeline|minimal_asr_pipeline|export_multimodal|custom_convert|develop_picture_enrichment|rapidocr_with_custom_models|suryaocr_with_custom_models|offline_convert|pictures_description|pictures_description_api|vlm_pipeline_api_model|granitedocling_repetition_stopping|mlx_whisper_example|gpu_standard_pipeline|gpu_vlm_pipeline|demo_layout_vlm|post_process_ocr_with_vlm)\.py$|xbrl_conversion\.ipynb$'
 
 jobs:
   lint:

CHANGELOG.md

@@ -1,3 +1,64 @@
+## [v2.78.0](https://github.com/docling-project/docling/releases/tag/v2.78.0) - 2026-03-10
+
+### Feature
+
+* Add support for TableFormer v2 ([#3013](https://github.com/docling-project/docling/issues/3013)) ([`4ccd1d4`](https://github.com/docling-project/docling/commit/4ccd1d465deb8d521c09e2da61b537a9236d6560))
+* Add gRPC transport for KServe v2 API engine ([#3074](https://github.com/docling-project/docling/issues/3074)) ([`3d90778`](https://github.com/docling-project/docling/commit/3d90778e3e5762b16758e1c121f42890e32f0560))
+
+### Fix
+
+* **html:** Fix broken document tree and quadratic complexity in rich table cells ([#3025](https://github.com/docling-project/docling/issues/3025)) ([`80f75b8`](https://github.com/docling-project/docling/commit/80f75b8896a6b15c5422c56e9a423e4d2e6673cd))
+* Loosen dependency for pandas3 ([#3095](https://github.com/docling-project/docling/issues/3095)) ([`5188180`](https://github.com/docling-project/docling/commit/5188180ea31dd90567140affc564ce2729b6e4a1))
+* Add parse timeout to legacy LaTeX documents ([#3019](https://github.com/docling-project/docling/issues/3019)) ([`1192714`](https://github.com/docling-project/docling/commit/1192714b536ebb8117785b06ed85e7d203e0996d))
+* **msword:** Skip GroupItem targets without comments attribute ([#3080](https://github.com/docling-project/docling/issues/3080)) ([`ee16285`](https://github.com/docling-project/docling/commit/ee16285651e5c2f963e051b1ee32b50a043191e2))
+
+### Documentation
+
+* Fix code in rag langchain chunker tokenizer ([#2993](https://github.com/docling-project/docling/issues/2993)) ([`d113e61`](https://github.com/docling-project/docling/commit/d113e611c445db6793fd94b3fee9c4109513d04a))
+* Update code snippet to use modern pipeline options syntax ([#3087](https://github.com/docling-project/docling/issues/3087)) ([`95b759e`](https://github.com/docling-project/docling/commit/95b759e5199f1142fb66dc2088c0c36177c5c284))
+* Set HuggingFaceEndpoint task for Mixtral examples ([#2945](https://github.com/docling-project/docling/issues/2945)) ([`5d3ac38`](https://github.com/docling-project/docling/commit/5d3ac38a65000cd39766f87557c685668224ad7f))
+
+## [v2.77.0](https://github.com/docling-project/docling/releases/tag/v2.77.0) - 2026-03-06
+
+### Feature
+
+* Track vlm_inference time for mlx_model pipeline ([#3060](https://github.com/docling-project/docling/issues/3060)) ([`38c4bb2`](https://github.com/docling-project/docling/commit/38c4bb26e8e3a7797d1caec3f690a7c8d5d9a735))
+* Add configurable graph_optimization_level for ONNX Runtime engines ([#3071](https://github.com/docling-project/docling/issues/3071)) ([`cfc6636`](https://github.com/docling-project/docling/commit/cfc6636a2a0e6b149dd51714d20e9b93f3f6463b))
+
+### Fix
+
+* **docx:** Preserve URL fragments and query params in hyperlinks ([#3050](https://github.com/docling-project/docling/issues/3050)) ([`cd9dd10`](https://github.com/docling-project/docling/commit/cd9dd10ccfe2a112af10ad135f8293d3bf845e1a))
+* Detect Office Open XML formats from ZIP contents when filename has no extension ([#3073](https://github.com/docling-project/docling/issues/3073)) ([`56f06fe`](https://github.com/docling-project/docling/commit/56f06fe372e3bfda29c14d66de0a066afb4c79c0))
+* **readingorder:** Assign FURNITURE content_layer to footer/header in container groups ([#3044](https://github.com/docling-project/docling/issues/3044)) ([`f7cb304`](https://github.com/docling-project/docling/commit/f7cb304daa7b7bfe49ba23b81d53fb16da4024af))
+* **docx:** Handle list items immediately after numbered headings ([#3070](https://github.com/docling-project/docling/issues/3070)) ([`56eb127`](https://github.com/docling-project/docling/commit/56eb12782c804b7ec36145bf52c1e005839c816b))
+* **rapidocr:** ORT thread configuration for RapidOCR backend ([#3062](https://github.com/docling-project/docling/issues/3062)) ([`68336c2`](https://github.com/docling-project/docling/commit/68336c2bda2b79f10759ad1587626c47500f4fb4))
+
+### Documentation
+
+* Add examples and fix docstring bug in DocumentConverter ([#3064](https://github.com/docling-project/docling/issues/3064)) ([`653940e`](https://github.com/docling-project/docling/commit/653940e0251e1bc5f311aded31690c64f42d9819))
+* Add docstrings to PipelineOptions classes ([#3065](https://github.com/docling-project/docling/issues/3065)) ([`8b99085`](https://github.com/docling-project/docling/commit/8b990856cd48fec12c68d940e665d8187d349753))
+
+## [v2.76.0](https://github.com/docling-project/docling/releases/tag/v2.76.0) - 2026-03-02
+
+### Feature
+
+* Export to WebVTT format ([#3036](https://github.com/docling-project/docling/issues/3036)) ([`d276e60`](https://github.com/docling-project/docling/commit/d276e6056106b6aa04fee65def96d3e10557d632))
+
+### Fix
+
+* **xlsx:** Handle OneCellAnchor images in Excel backend ([#3045](https://github.com/docling-project/docling/issues/3045)) ([`859c302`](https://github.com/docling-project/docling/commit/859c302310289c5bab45a6e160e7cc3b9c538343))
+* Normalize Unicode ligatures in PDF text extraction ([#3057](https://github.com/docling-project/docling/issues/3057)) ([`6198e69`](https://github.com/docling-project/docling/commit/6198e69dec33d9c14b3be279b19924d73e5eb3fb))
+* **ocr:** Update RapidOCR torch GPU config key ([#3049](https://github.com/docling-project/docling/issues/3049)) ([`477359b`](https://github.com/docling-project/docling/commit/477359b772039b9c9c0d31c9dabcd755abdeb560))
+* Convert PIL images to RGB before picture description ([#3014](https://github.com/docling-project/docling/issues/3014)) ([`90ce93d`](https://github.com/docling-project/docling/commit/90ce93d8a095ea17040bd6a91ded0b463998bea9))
+* **msword:** Use outlineLvl for heading levels and clamp to minimum 1 ([#2916](https://github.com/docling-project/docling/issues/2916)) ([`a3d2b4b`](https://github.com/docling-project/docling/commit/a3d2b4bcc07fc00fff3039ae2046ee69b7587ab2))
+
+### Documentation
+
+* Add metaxy integration ([#3058](https://github.com/docling-project/docling/issues/3058)) ([`7aacc6c`](https://github.com/docling-project/docling/commit/7aacc6c18da3e856babb0f06afd7c985774f118e))
+* Removes merge conflict artifacts ([#3055](https://github.com/docling-project/docling/issues/3055)) ([`672125c`](https://github.com/docling-project/docling/commit/672125cd1bb5e22bb7a677f48157a55ca93f9ff6))
+* Add audio & video processing guide ([#3038](https://github.com/docling-project/docling/issues/3038)) ([`1321b39`](https://github.com/docling-project/docling/commit/1321b39cd8203d5e1cd60191cc9e979c5b939f98))
+* Add XBRL conversion example notebook and update feature listings ([#3039](https://github.com/docling-project/docling/issues/3039)) ([`1eb5c21`](https://github.com/docling-project/docling/commit/1eb5c21dabfed02bfe71cb7fc502d124562f1ba8))
+
 ## [v2.75.0](https://github.com/docling-project/docling/releases/tag/v2.75.0) - 2026-02-24
 
 ### Feature

README.md

@@ -33,7 +33,8 @@ Docling simplifies document processing, parsing diverse formats — including ad
 * 🗂️ Parsing of [multiple document formats][supported_formats] incl. PDF, DOCX, PPTX, XLSX, HTML, WAV, MP3, WebVTT, images (PNG, TIFF, JPEG, ...), LaTeX, and more
 * 📑 Advanced PDF understanding incl. page layout, reading order, table structure, code, formulas, image classification, and more
 * 🧬 Unified, expressive [DoclingDocument][docling_document] representation format
-* ↪️ Various [export formats][supported_formats] and options, including Markdown, HTML, [DocTags](https://arxiv.org/abs/2503.11576) and lossless JSON
+* ↪️ Various [export formats][supported_formats] and options, including Markdown, HTML, WebVTT, [DocTags](https://arxiv.org/abs/2503.11576) and lossless JSON
+* 📜 Support of several application-specifc XML schemas incl. [USPTO](https://www.uspto.gov/patents) patents, [JATS](https://jats.nlm.nih.gov/) articles, and [XBRL](https://www.xbrl.org/) financial reports.
 * 🔒 Local execution capabilities for sensitive data and air-gapped environments
 * 🤖 Plug-and-play [integrations][integrations] incl. LangChain, LlamaIndex, Crew AI & Haystack for agentic AI
 * 🔍 Extensive OCR support for scanned PDFs and images
@@ -46,7 +47,8 @@ Docling simplifies document processing, parsing diverse formats — including ad
 * 📤 Structured [information extraction][extraction] \[🧪 beta\]
 * 📑 New layout model (**Heron**) by default, for faster PDF parsing
 * 🔌 [MCP server](https://docling-project.github.io/docling/usage/mcp/) for agentic applications
-* 💬 Parsing of Web Video Text Tracks (WebVTT) files
+* 💼 Parsing of XBRL (eXtensible Business Reporting Language) documents for financial reports
+* 💬 Parsing of WebVTT (Web Video Text Tracks) files and export to WebVTT format
 * 💬 Parsing of LaTeX files
 
 ### Coming soon

docling/backend/html_backend.py

@@ -7,7 +7,7 @@
 from copy import deepcopy
 from io import BytesIO
 from pathlib import Path
-from typing import Final, Optional, Union, cast
+from typing import Final, Iterator, Optional, Union, cast
 from urllib.parse import urljoin, urlparse
 
 import requests
@@ -656,7 +656,7 @@ def _flush_buffer() -> None:
                 return
 
             for annotated_text_list in parts:
-                with self._use_inline_group(annotated_text_list, doc):
+                with self._use_inline_group(annotated_text_list, doc) as inline_ref:
                     for annotated_text in annotated_text_list:
                         if annotated_text.text.strip():
                             seg_clean = HTMLDocumentBackend._clean_unicode(
@@ -670,7 +670,8 @@ def _flush_buffer() -> None:
                                     formatting=annotated_text.formatting,
                                     hyperlink=annotated_text.hyperlink,
                                 )
-                                added_refs.append(docling_code2.get_ref())
+                                if inline_ref is None:
+                                    added_refs.append(docling_code2.get_ref())
                             else:
                                 docling_text2 = doc.add_text(
                                     parent=self.parents[self.level],
@@ -680,7 +681,10 @@ def _flush_buffer() -> None:
                                     formatting=annotated_text.formatting,
                                     hyperlink=annotated_text.hyperlink,
                                 )
-                                added_refs.append(docling_text2.get_ref())
+                                if inline_ref is None:
+                                    added_refs.append(docling_text2.get_ref())
+                    if inline_ref is not None:
+                        added_refs.append(inline_ref)
 
         for node in element.contents:
             if isinstance(node, Tag):
@@ -866,7 +870,7 @@ def _use_format(self, tags: list[str]):
     @contextmanager
     def _use_inline_group(
         self, annotated_text_list: AnnotatedTextList, doc: DoclingDocument
-    ):
+    ) -> Iterator[RefItem | None]:
         """Create an inline group for annotated texts.
 
         Checks if annotated_text_list has more than one item and if so creates an inline
@@ -876,6 +880,10 @@ def _use_inline_group(
         Args:
             annotated_text_list (AnnotatedTextList): Annotated text
             doc (DoclingDocument): Currently used document
+
+        Yields:
+            The RefItem of the created InlineGroup, or None when the list has only one
+                element and no group is created.
         """
         if len(annotated_text_list) > 1:
             inline_fmt = doc.add_group(
@@ -886,7 +894,7 @@ def _use_inline_group(
             self.parents[self.level + 1] = inline_fmt
             self.level += 1
             try:
-                yield None
+                yield inline_fmt.get_ref()
             finally:
                 self.parents[self.level] = None
                 self.level -= 1
@@ -1205,7 +1213,7 @@ def _handle_block(self, tag: Tag, doc: DoclingDocument) -> list[RefItem]:
             )
             annotated_texts: AnnotatedTextList = text_list.simplify_text_elements()
             for part in annotated_texts.split_by_newline():
-                with self._use_inline_group(part, doc):
+                with self._use_inline_group(part, doc) as inline_ref:
                     for annotated_text in part:
                         if seg := annotated_text.text.strip():
                             seg_clean = HTMLDocumentBackend._clean_unicode(seg)
@@ -1217,7 +1225,8 @@ def _handle_block(self, tag: Tag, doc: DoclingDocument) -> list[RefItem]:
                                     formatting=annotated_text.formatting,
                                     hyperlink=annotated_text.hyperlink,
                                 )
-                                added_refs.append(docling_code.get_ref())
+                                if inline_ref is None:
+                                    added_refs.append(docling_code.get_ref())
                             else:
                                 docling_text = doc.add_text(
                                     parent=self.parents[self.level],
@@ -1227,7 +1236,10 @@ def _handle_block(self, tag: Tag, doc: DoclingDocument) -> list[RefItem]:
                                     formatting=annotated_text.formatting,
                                     hyperlink=annotated_text.hyperlink,
                                 )
-                                added_refs.append(docling_text.get_ref())
+                                if inline_ref is None:
+                                    added_refs.append(docling_text.get_ref())
+                    if inline_ref is not None:
+                        added_refs.append(inline_ref)
 
             for img_tag in tag("img"):
                 if isinstance(img_tag, Tag):
@@ -1244,19 +1256,13 @@ def _handle_block(self, tag: Tag, doc: DoclingDocument) -> list[RefItem]:
             added_refs.append(docling_table.get_ref())
             self.parse_table_data(tag, doc, docling_table, num_rows, num_cols)
 
-            for img_tag in tag("img"):
-                if isinstance(img_tag, Tag):
-                    im_ref2 = self._emit_image(tag, doc)
-                    if im_ref2 is not None:
-                        added_refs.append(im_ref2)
-
         elif tag_name in {"pre"}:
             # handle monospace code snippets (pre).
             text_list = self._extract_text_and_hyperlink_recursively(
                 tag, find_parent_annotation=True, keep_newlines=True
             )
             annotated_texts = text_list.simplify_text_elements()
-            with self._use_inline_group(annotated_texts, doc):
+            with self._use_inline_group(annotated_texts, doc) as inline_ref:
                 for annotated_text in annotated_texts:
                     text_clean = HTMLDocumentBackend._clean_unicode(
                         annotated_text.text.strip()
@@ -1268,7 +1274,10 @@ def _handle_block(self, tag: Tag, doc: DoclingDocument) -> list[RefItem]:
                         formatting=annotated_text.formatting,
                         hyperlink=annotated_text.hyperlink,
                     )
-                    added_refs.append(docling_code2.get_ref())
+                    if inline_ref is None:
+                        added_refs.append(docling_code2.get_ref())
+            if inline_ref is not None:
+                added_refs.append(inline_ref)
 
         elif tag_name == "footer":
             with self._use_footer(tag, doc):
@@ -1416,7 +1425,9 @@ def _extract_text_recursively(item: PageElement) -> list[str]:
                 for child in tag:
                     parts.extend(_extract_text_recursively(child))
                 result.append(
-                    "".join(parts) + " " if tag.name in {"p", "li"} else "".join(parts)
+                    "".join(parts) + " "
+                    if tag.name in {"p", "li", "th", "td"}
+                    else "".join(parts)
                 )
 
             return result