fix: drop python 3.9 and pin built tree-sitter versions

.github/workflows/checks.yml

@@ -18,7 +18,7 @@ jobs:
     runs-on: ubuntu-latest
     strategy:
       matrix:
-        python-version: ['3.9', '3.10', '3.11', '3.12', '3.13', '3.14']
+        python-version: ['3.10', '3.11', '3.12', '3.13', '3.14']
     steps:
       - uses: actions/checkout@v4
       - name: Cache Hugging Face models

docling_core/experimental/idoctags.py

@@ -1316,7 +1316,7 @@ def _serialize_single_item(
         if (
             (isinstance(item, CodeItem) and ContentType.TEXT_CODE in params.content_types)
             or (isinstance(item, FormulaItem) and ContentType.TEXT_FORMULA in params.content_types)
-            or (not isinstance(item, (CodeItem, FormulaItem)) and ContentType.TEXT_OTHER in params.content_types)
+            or (not isinstance(item, CodeItem | FormulaItem) and ContentType.TEXT_OTHER in params.content_types)
         ):
             # Check if we should serialize a single inline group child instead of text
             if len(item.children) > 0 and isinstance((first_child := item.children[0].resolve(doc)), InlineGroup):
@@ -1754,7 +1754,7 @@ def serialize(
         **kwargs: Any,
     ) -> SerializationResult:
         """Serialize unsupported nodes by concatenating their textual parts."""
-        if isinstance(item, (ListGroup, InlineGroup)):
+        if isinstance(item, ListGroup | InlineGroup):
             parts = doc_serializer.get_parts(item=item, **kwargs)
             text_res = "\n".join([p.text for p in parts if p.text])
             return create_ser_result(text=text_res, span_source=parts)

docling_core/transforms/chunker/code_chunking/_utils.py

@@ -38,16 +38,15 @@ def _get_tree_sitter_language(language: CodeLanguageLabel):
         CodeLanguageLabel.JAVASCRIPT: lambda: Lang(ts_js.language()),
         CodeLanguageLabel.C: lambda: Lang(ts_c.language()),
     }
-    if sys.version_info >= (3, 10):
-        try:
-            import tree_sitter_java_orchard as ts_java
-
-            language_map[CodeLanguageLabel.JAVA] = lambda: Lang(ts_java.language())
-        except ImportError:
-            _logger.warning(
-                "Code chunking for Java cannot be enabled because tree-sitter-java-orchard is missing. "
-                "Please installed it via `pip install tree-sitter-java-orchard`."
-            )
+    try:
+        import tree_sitter_java_orchard as ts_java
+
+        language_map[CodeLanguageLabel.JAVA] = lambda: Lang(ts_java.language())
+    except ImportError:
+        _logger.warning(
+            "Code chunking for Java cannot be enabled because tree-sitter-java-orchard is missing. "
+            "Please installed it via `pip install tree-sitter-java-orchard`."
+        )
 
     factory = language_map.get(language)
     return factory() if factory else None
@@ -157,6 +156,21 @@ def _to_str(node: Node) -> str:
 def _query_tree(language, tree: Tree, query: str):
     """Query a tree-sitter tree with the given query string."""
     if not language:
-        return []
-    q = language.query(query)
-    return q.captures(tree.root_node)
+        return {}
+    from tree_sitter import Query, QueryCursor
+
+    q = Query(language, query)
+    cursor = QueryCursor(q)
+    matches = list(cursor.matches(tree.root_node))
+
+    # Combine all captures from all matches into a single dict
+    # Old API returned: {"capture_name": [node1, node2, ...]}
+    # New API returns: [(pattern_idx, {"capture_name": [node1, ...]}), ...]
+    combined_captures: dict[str, list[Node]] = {}
+    for _, captures_dict in matches:
+        for capture_name, nodes in captures_dict.items():
+            if capture_name not in combined_captures:
+                combined_captures[capture_name] = []
+            combined_captures[capture_name].extend(nodes)
+
+    return combined_captures

docling_core/transforms/chunker/hierarchical_chunker.py

@@ -155,7 +155,7 @@ def chunk(
         for item, level in dl_doc.iterate_items(with_groups=True):
             if item.self_ref in excluded_refs:
                 continue
-            if isinstance(item, (TitleItem, SectionHeaderItem)):
+            if isinstance(item, TitleItem | SectionHeaderItem):
                 level = item.level if isinstance(item, SectionHeaderItem) else 0
 
                 # prepare to remove shadowed headings as they just went out of scope
@@ -185,7 +185,7 @@ def chunk(
                 heading_by_level[level] = item
 
                 continue
-            elif isinstance(item, (ListGroup, InlineGroup, DocItem)) and item.self_ref not in visited:
+            elif isinstance(item, ListGroup | InlineGroup | DocItem) and item.self_ref not in visited:
                 if self.code_chunking_strategy is not None and isinstance(item, CodeItem):
                     yield from self.code_chunking_strategy.chunk_code_item(
                         item=item,

docling_core/transforms/chunker/hybrid_chunker.py

@@ -162,7 +162,7 @@ def _make_chunk_from_doc_items(
                     res_text
                     for doc_item in doc_items
                     if (res_text := doc_serializer.serialize(item=doc_item).text)
-                    and not isinstance(doc_item, (TitleItem, SectionHeaderItem))
+                    and not isinstance(doc_item, TitleItem | SectionHeaderItem)
                 ]
             )
         )

docling_core/transforms/serializer/common.py

@@ -97,7 +97,7 @@ def _iterate_items(
         traverse_pictures=traverse_pictures,
     ):
         if add_page_breaks:
-            if isinstance(item, (ListGroup, InlineGroup)) and item.self_ref not in my_visited:
+            if isinstance(item, ListGroup | InlineGroup) and item.self_ref not in my_visited:
                 # if group starts with new page, yield page break before group node
                 my_visited.add(item.self_ref)
                 for it, _ in _iterate_items(

docling_core/transforms/serializer/html.py

@@ -158,12 +158,12 @@ def serialize(
             post_processed = True
         else:
             text = item.text
-            if not isinstance(item, (CodeItem, FormulaItem)):
+            if not isinstance(item, CodeItem | FormulaItem):
                 text = html.escape(text, quote=False)
                 text = text.replace("\n", "<br>")
 
         # Prepare the HTML based on item type
-        if isinstance(item, (TitleItem, SectionHeaderItem)):
+        if isinstance(item, TitleItem | SectionHeaderItem):
             section_level = min(item.level + 1, 6) if isinstance(item, SectionHeaderItem) else 1
             text = get_html_tag_with_text_direction(html_tag=f"h{section_level}", text=text)
 
@@ -843,7 +843,7 @@ def serialize(
         for ann in item.get_annotations():
             if isinstance(
                 ann,
-                (PictureClassificationData, DescriptionAnnotation, PictureMoleculeData),
+                PictureClassificationData | DescriptionAnnotation | PictureMoleculeData,
             ):
                 if ann_text := _get_annotation_text(ann):
                     text_dir = get_text_direction(ann_text)
@@ -1055,7 +1055,7 @@ def serialize_captions(
 
         if (
             item.self_ref not in excluded_refs
-            and isinstance(item, (PictureItem, TableItem))
+            and isinstance(item, PictureItem | TableItem)
             and _should_use_legacy_annotations(params=params, item=item)
         ):
             ann_res = self.serialize_annotations(

docling_core/transforms/serializer/latex.py

@@ -155,7 +155,7 @@ def serialize(
             text = item.text
             post_process = True
 
-        if isinstance(item, (ListItem, TitleItem, SectionHeaderItem)):
+        if isinstance(item, ListItem | TitleItem | SectionHeaderItem):
             # For list items, defer environment wrapping to list serializer
             if isinstance(item, ListItem):
                 if post_process:
@@ -253,11 +253,7 @@ def serialize(
         for ann in item.get_annotations():
             if isinstance(
                 ann,
-                (
-                    PictureClassificationData,
-                    DescriptionAnnotation,
-                    PictureMoleculeData,
-                ),
+                PictureClassificationData | DescriptionAnnotation | PictureMoleculeData,
             ):
                 if ann_text := _get_annotation_text(ann):
                     # Ensure each line of the annotation is prefixed with '%'

docling_core/transforms/serializer/markdown.py

@@ -145,7 +145,7 @@ def serialize(
             text = f"- [x] {text}"
         if item.label == DocItemLabel.CHECKBOX_UNSELECTED:
             text = f"- [ ] {text}"
-        if isinstance(item, (ListItem, TitleItem, SectionHeaderItem)):
+        if isinstance(item, ListItem | TitleItem | SectionHeaderItem):
             if not has_inline_repr:
                 # case where processing/formatting should be applied first (in inner scope)
                 text = doc_serializer.post_process(
@@ -320,11 +320,7 @@ def serialize(
         for ann in item.get_annotations():
             if isinstance(
                 ann,
-                (
-                    PictureClassificationData,
-                    DescriptionAnnotation,
-                    PictureMoleculeData,
-                ),
+                PictureClassificationData | DescriptionAnnotation | PictureMoleculeData,
             ):
                 if ann_text := _get_annotation_text(ann):
                     ann_res = create_ser_result(

docling_core/types/doc/document.py

@@ -1855,12 +1855,10 @@ def _migrate_annotations_to_meta(self) -> Self:
                         )
                     elif not isinstance(
                         ann,
-                        (
-                            PictureClassificationData,
-                            DescriptionAnnotation,
-                            PictureMoleculeData,
-                            PictureTabularChartData,
-                        ),
+                        PictureClassificationData
+                        | DescriptionAnnotation
+                        | PictureMoleculeData
+                        | PictureTabularChartData,
                     ) and not hasattr(
                         self.meta,
                         MetaUtils.create_meta_field_name(
@@ -2639,7 +2637,7 @@ def _append_item(self, *, item: NodeItem, parent_ref: RefItem) -> RefItem:
 
             self.form_items.append(item)
 
-        elif isinstance(item, (ListGroup, InlineGroup)):
+        elif isinstance(item, ListGroup | InlineGroup):
             item_label = "groups"
             item_index = len(self.groups)
 

docling_core/types/legacy_doc/document.py

@@ -453,7 +453,7 @@ def export_to_markdown(
                 if item is None:
                     continue
 
-                if isinstance(item, (Table, Figure)) and item.text and item.obj_type in main_text_labels:
+                if isinstance(item, Table | Figure) and item.text and item.obj_type in main_text_labels:
                     embedded_captions.add(item.text)
 
             # serialize document to markdown

docling_core/utils/legacy.py

@@ -139,7 +139,7 @@ def docling_document_to_legacy(doc: DoclingDocument, fallback_filaname: str = "f
 
     embedded_captions = set()
     for ix, (item, level) in enumerate(doc.iterate_items(doc.body)):
-        if isinstance(item, (TableItem, PictureItem)) and len(item.captions) > 0:
+        if isinstance(item, TableItem | PictureItem) and len(item.captions) > 0:
             caption = item.caption_text(doc)
             if caption:
                 embedded_captions.add(caption)
@@ -148,7 +148,7 @@ def docling_document_to_legacy(doc: DoclingDocument, fallback_filaname: str = "f
         if isinstance(item, DocItem):
             item_type = item.label
 
-            if isinstance(item, (TextItem, ListItem, SectionHeaderItem)):
+            if isinstance(item, TextItem | ListItem | SectionHeaderItem):
                 if isinstance(item, ListItem) and item.marker:
                     text = f"{item.marker} {item.text}"
                 else:
@@ -422,7 +422,7 @@ def _transform_prov(item: BaseCell) -> Optional[ProvenanceItem]:
             if item is None:
                 continue
 
-            if isinstance(item, (DsSchemaTable, Figure)) and item.text:
+            if isinstance(item, DsSchemaTable | Figure) and item.text:
                 embedded_captions[item.text] = ix
 
         # build lookup from floating objects to their caption item

pyproject.toml

@@ -31,14 +31,13 @@ classifiers = [
     "Topic :: Software Development :: Libraries :: Python Modules",
     "Typing :: Typed",
     "Programming Language :: Python :: 3",
-    "Programming Language :: Python :: 3.9",
     "Programming Language :: Python :: 3.10",
     "Programming Language :: Python :: 3.11",
     "Programming Language :: Python :: 3.12",
     "Programming Language :: Python :: 3.13",
     "Programming Language :: Python :: 3.14",
 ]
-requires-python = '>=3.9,<4.0'
+requires-python = '>=3.10,<4.0'
 dependencies = [
     'jsonschema (>=4.16.0,<5.0.0)',
     'pydantic (>=2.6.0,<3.0.0,!=2.10.0,!=2.10.1,!=2.10.2)',
@@ -68,23 +67,23 @@ docling-view = "docling_core.cli.view:app"
 chunking = [
     # common:
     'semchunk (>=2.2.0,<3.0.0)',
-    'tree-sitter (>=0.23.2,<0.25.0)',
-    'tree-sitter-python (>=0.23.6,<0.25.0)',
-    'tree-sitter-c (>=0.23.4,<0.23.5)',
-    'tree-sitter-javascript (>=0.23.1,<0.25.0)',
-    'tree-sitter-typescript (>=0.23.2,<0.25.0)',
+    'tree-sitter (>=0.25.0,<0.27.0)',
+    'tree-sitter-python >=0.23.6',
+    'tree-sitter-c >=0.23.4',
+    'tree-sitter-javascript >=0.23.1',
+    'tree-sitter-typescript >=0.23.2',
 
     # specific:
     'transformers (>=4.34.0,<5.0.0)',
 ]
 chunking-openai = [
     # common:
     'semchunk (>=2.2.0,<3.0.0)',
-    'tree-sitter (>=0.23.2,<0.25.0)',
-    'tree-sitter-python (>=0.23.6,<0.25.0)',
-    'tree-sitter-c (>=0.23.4,<0.23.5)',
-    'tree-sitter-javascript (>=0.23.1,<0.25.0)',
-    'tree-sitter-typescript (>=0.23.2,<0.25.0)',
+    'tree-sitter (>=0.25.0,<0.27.0)',
+    'tree-sitter-python >=0.23.6',
+    'tree-sitter-c >=0.23.4',
+    'tree-sitter-javascript >=0.23.1',
+    'tree-sitter-typescript >=0.23.2',
 
     # specific:
     'tiktoken (>=0.9.0,<0.13.0)',
@@ -131,7 +130,7 @@ namespaces = true
 "*" = ["*.json"]
 
 [tool.ruff]
-target-version = "py39"
+target-version = "py310"
 line-length = 120
 respect-gitignore = true
 exclude = [
@@ -205,7 +204,7 @@ pretty = true
 no_implicit_optional = true
 namespace_packages = true
 show_error_codes = true
-python_version = "3.9"
+python_version = "3.10"
 plugins = ["pydantic.mypy"]
 exclude = "(^|/)test/data/.*"