docling-core 2.26.4__tar.gz → 2.28.0__tar.gz

{docling_core-2.26.4 → docling_core-2.28.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: docling-core
-Version: 2.26.4
+Version: 2.28.0
 Summary: A python library to define and validate data types in Docling.
 Home-page: https://github.com/docling-project
 License: MIT
@@ -26,6 +26,7 @@ Classifier: Topic :: Scientific/Engineering :: Information Analysis
 Classifier: Topic :: Software Development :: Libraries :: Python Modules
 Classifier: Typing :: Typed
 Provides-Extra: chunking
+Provides-Extra: chunking-openai
 Requires-Dist: jsonref (>=1.1.0,<2.0.0)
 Requires-Dist: jsonschema (>=4.16.0,<5.0.0)
 Requires-Dist: latex2mathml (>=3.77.0,<4.0.0)
@@ -33,8 +34,9 @@ Requires-Dist: pandas (>=2.1.4,<3.0.0)
 Requires-Dist: pillow (>=10.0.0,<12.0.0)
 Requires-Dist: pydantic (>=2.6.0,<3.0.0,!=2.10.0,!=2.10.1,!=2.10.2)
 Requires-Dist: pyyaml (>=5.1,<7.0.0)
-Requires-Dist: semchunk (>=2.2.0,<3.0.0) ; extra == "chunking"
+Requires-Dist: semchunk (>=2.2.0,<3.0.0) ; extra == "chunking" or extra == "chunking-openai"
 Requires-Dist: tabulate (>=0.9.0,<0.10.0)
+Requires-Dist: tiktoken (>=0.9.0,<0.10.0) ; extra == "chunking-openai"
 Requires-Dist: transformers (>=4.34.0,<5.0.0) ; extra == "chunking"
 Requires-Dist: typer (>=0.12.5,<0.16.0)
 Requires-Dist: typing-extensions (>=4.12.2,<5.0.0)

{docling_core-2.26.4 → docling_core-2.28.0}/docling_core/experimental/serializer/base.py

@@ -6,7 +6,7 @@
 """Define base classes for serialization."""
 from abc import ABC, abstractmethod
 from pathlib import Path
-from typing import Optional, Union
+from typing import Any, Optional, Union
 
 from pydantic import AnyUrl, BaseModel
 
@@ -51,7 +51,7 @@ class BaseTextSerializer(ABC):
         item: TextItem,
         doc_serializer: "BaseDocSerializer",
         doc: DoclingDocument,
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serializes the passed item."""
         ...
@@ -67,7 +67,7 @@ class BaseTableSerializer(ABC):
         item: TableItem,
         doc_serializer: "BaseDocSerializer",
         doc: DoclingDocument,
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serializes the passed item."""
         ...
@@ -83,7 +83,7 @@ class BasePictureSerializer(ABC):
         item: PictureItem,
         doc_serializer: "BaseDocSerializer",
         doc: DoclingDocument,
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serializes the passed item."""
         ...
@@ -99,7 +99,7 @@ class BaseKeyValueSerializer(ABC):
         item: KeyValueItem,
         doc_serializer: "BaseDocSerializer",
         doc: DoclingDocument,
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serializes the passed item."""
         ...
@@ -115,7 +115,7 @@ class BaseFormSerializer(ABC):
         item: FormItem,
         doc_serializer: "BaseDocSerializer",
         doc: DoclingDocument,
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serializes the passed item."""
         ...
@@ -131,7 +131,7 @@ class BaseListSerializer(ABC):
         item: Union[UnorderedList, OrderedList],
         doc_serializer: "BaseDocSerializer",
         doc: DoclingDocument,
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serializes the passed item."""
         ...
@@ -147,7 +147,7 @@ class BaseInlineSerializer(ABC):
         item: InlineGroup,
         doc_serializer: "BaseDocSerializer",
         doc: DoclingDocument,
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serializes the passed item."""
         ...
@@ -163,7 +163,7 @@ class BaseFallbackSerializer(ABC):
         item: NodeItem,
         doc_serializer: "BaseDocSerializer",
         doc: DoclingDocument,
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serializes the passed item."""
         ...
@@ -174,34 +174,40 @@ class BaseDocSerializer(ABC):
 
     @abstractmethod
     def serialize(
-        self, *, item: Optional[NodeItem] = None, **kwargs
+        self,
+        *,
+        item: Optional[NodeItem] = None,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Run the serialization."""
         ...
 
     @abstractmethod
-    def serialize_bold(self, text: str, **kwargs) -> str:
+    def serialize_bold(self, text: str, **kwargs: Any) -> str:
         """Hook for bold formatting serialization."""
         ...
 
     @abstractmethod
-    def serialize_italic(self, text: str, **kwargs) -> str:
+    def serialize_italic(self, text: str, **kwargs: Any) -> str:
         """Hook for italic formatting serialization."""
         ...
 
     @abstractmethod
-    def serialize_underline(self, text: str, **kwargs) -> str:
+    def serialize_underline(self, text: str, **kwargs: Any) -> str:
         """Hook for underline formatting serialization."""
         ...
 
     @abstractmethod
-    def serialize_strikethrough(self, text: str, **kwargs) -> str:
+    def serialize_strikethrough(self, text: str, **kwargs: Any) -> str:
         """Hook for strikethrough formatting serialization."""
         ...
 
     @abstractmethod
     def serialize_hyperlink(
-        self, text: str, hyperlink: Union[AnyUrl, Path], **kwargs
+        self,
+        text: str,
+        hyperlink: Union[AnyUrl, Path],
+        **kwargs: Any,
     ) -> str:
         """Hook for hyperlink serialization."""
         ...
@@ -210,7 +216,7 @@ class BaseDocSerializer(ABC):
     def get_parts(
         self,
         item: Optional[NodeItem] = None,
-        **kwargs,
+        **kwargs: Any,
     ) -> list[SerializationResult]:
         """Get the components to be combined for serializing this node."""
         ...
@@ -219,7 +225,7 @@ class BaseDocSerializer(ABC):
     def post_process(
         self,
         text: str,
-        **kwargs,
+        **kwargs: Any,
     ) -> str:
         """Apply some text post-processing steps."""
         ...
@@ -228,13 +234,13 @@ class BaseDocSerializer(ABC):
     def serialize_captions(
         self,
         item: FloatingItem,
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serialize the item's captions."""
         ...
 
     @abstractmethod
-    def get_excluded_refs(self, **kwargs) -> set[str]:
+    def get_excluded_refs(self, **kwargs: Any) -> set[str]:
         """Get references to excluded items."""
         ...
 

{docling_core-2.26.4 → docling_core-2.28.0}/docling_core/experimental/serializer/common.py

@@ -214,7 +214,7 @@ class DocSerializer(BaseModel, BaseDocSerializer):
         return refs
 
     @override
-    def get_excluded_refs(self, **kwargs) -> set[str]:
+    def get_excluded_refs(self, **kwargs: Any) -> set[str]:
         """References to excluded items."""
         params = self.params.merge_with_patch(patch=kwargs)
         params_json = params.model_dump_json()
@@ -252,7 +252,10 @@ class DocSerializer(BaseModel, BaseDocSerializer):
 
     @abstractmethod
     def serialize_doc(
-        self, *, parts: list[SerializationResult], **kwargs
+        self,
+        *,
+        parts: list[SerializationResult],
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serialize a document out of its pages."""
         ...
@@ -271,7 +274,7 @@ class DocSerializer(BaseModel, BaseDocSerializer):
         list_level: int = 0,
         is_inline_scope: bool = False,
         visited: Optional[set[str]] = None,  # refs of visited items
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serialize a given node."""
         my_visited: set[str] = visited if visited is not None else set()
@@ -380,7 +383,7 @@ class DocSerializer(BaseModel, BaseDocSerializer):
         list_level: int = 0,
         is_inline_scope: bool = False,
         visited: Optional[set[str]] = None,  # refs of visited items
-        **kwargs,
+        **kwargs: Any,
     ) -> list[SerializationResult]:
         """Get the components to be combined for serializing this node."""
         parts: list[SerializationResult] = []
@@ -415,7 +418,7 @@ class DocSerializer(BaseModel, BaseDocSerializer):
         *,
         formatting: Optional[Formatting] = None,
         hyperlink: Optional[Union[AnyUrl, Path]] = None,
-        **kwargs,
+        **kwargs: Any,
     ) -> str:
         """Apply some text post-processing steps."""
         params = self.params.merge_with_patch(patch=kwargs)
@@ -434,28 +437,31 @@ class DocSerializer(BaseModel, BaseDocSerializer):
         return res
 
     @override
-    def serialize_bold(self, text: str, **kwargs) -> str:
+    def serialize_bold(self, text: str, **kwargs: Any) -> str:
         """Hook for bold formatting serialization."""
         return text
 
     @override
-    def serialize_italic(self, text: str, **kwargs) -> str:
+    def serialize_italic(self, text: str, **kwargs: Any) -> str:
         """Hook for italic formatting serialization."""
         return text
 
     @override
-    def serialize_underline(self, text: str, **kwargs) -> str:
+    def serialize_underline(self, text: str, **kwargs: Any) -> str:
         """Hook for underline formatting serialization."""
         return text
 
     @override
-    def serialize_strikethrough(self, text: str, **kwargs) -> str:
+    def serialize_strikethrough(self, text: str, **kwargs: Any) -> str:
         """Hook for strikethrough formatting serialization."""
         return text
 
     @override
     def serialize_hyperlink(
-        self, text: str, hyperlink: Union[AnyUrl, Path], **kwargs
+        self,
+        text: str,
+        hyperlink: Union[AnyUrl, Path],
+        **kwargs: Any,
     ) -> str:
         """Hook for hyperlink serialization."""
         return text
@@ -464,7 +470,7 @@ class DocSerializer(BaseModel, BaseDocSerializer):
     def serialize_captions(
         self,
         item: FloatingItem,
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serialize the item's captions."""
         params = self.params.merge_with_patch(patch=kwargs)

{docling_core-2.26.4 → docling_core-2.28.0}/docling_core/experimental/serializer/doctags.py

@@ -1,7 +1,7 @@
 """Define classes for Doctags serialization."""
 
 from enum import Enum
-from typing import Dict, List, Optional, Union
+from typing import Any, Dict, List, Optional, Union
 
 from pydantic import BaseModel
 from typing_extensions import override
@@ -91,7 +91,7 @@ class DocTagsTextSerializer(BaseModel, BaseTextSerializer):
         item: TextItem,
         doc_serializer: BaseDocSerializer,
         doc: DoclingDocument,
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serializes the passed item."""
         from docling_core.types.doc.document import SectionHeaderItem
@@ -154,7 +154,7 @@ class DocTagsTableSerializer(BaseTableSerializer):
         item: TableItem,
         doc_serializer: BaseDocSerializer,
         doc: DoclingDocument,
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serializes the passed item."""
         params = DocTagsParams(**kwargs)
@@ -201,7 +201,7 @@ class DocTagsPictureSerializer(BasePictureSerializer):
         item: PictureItem,
         doc_serializer: BaseDocSerializer,
         doc: DoclingDocument,
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serializes the passed item."""
         params = DocTagsParams(**kwargs)
@@ -284,7 +284,7 @@ class DocTagsKeyValueSerializer(BaseKeyValueSerializer):
         item: KeyValueItem,
         doc_serializer: "BaseDocSerializer",
         doc: DoclingDocument,
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serializes the passed item."""
         params = DocTagsParams(**kwargs)
@@ -356,7 +356,7 @@ class DocTagsFormSerializer(BaseFormSerializer):
         item: FormItem,
         doc_serializer: "BaseDocSerializer",
         doc: DoclingDocument,
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serializes the passed item."""
         # TODO add actual implementation
@@ -378,7 +378,7 @@ class DocTagsListSerializer(BaseModel, BaseListSerializer):
         list_level: int = 0,
         is_inline_scope: bool = False,
         visited: Optional[set[str]] = None,  # refs of visited items
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serializes the passed item."""
         my_visited = visited if visited is not None else set()
@@ -423,7 +423,7 @@ class DocTagsInlineSerializer(BaseInlineSerializer):
         doc: DoclingDocument,
         list_level: int = 0,
         visited: Optional[set[str]] = None,  # refs of visited items
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serializes the passed item."""
         my_visited = visited if visited is not None else set()
@@ -454,7 +454,7 @@ class DocTagsFallbackSerializer(BaseFallbackSerializer):
         item: NodeItem,
         doc_serializer: "BaseDocSerializer",
         doc: DoclingDocument,
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serializes the passed item."""
         return create_ser_result()
@@ -477,7 +477,10 @@ class DocTagsDocSerializer(DocSerializer):
 
     @override
     def serialize_doc(
-        self, *, parts: list[SerializationResult], **kwargs
+        self,
+        *,
+        parts: list[SerializationResult],
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serialize a document out of its pages."""
         delim = _get_delim(params=self.params)
@@ -496,7 +499,7 @@ class DocTagsDocSerializer(DocSerializer):
     def serialize_captions(
         self,
         item: FloatingItem,
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serialize the item's captions."""
         params = DocTagsParams(**kwargs)

{docling_core-2.26.4 → docling_core-2.28.0}/docling_core/experimental/serializer/html.py

@@ -10,7 +10,7 @@ import logging
 from enum import Enum
 from io import BytesIO
 from pathlib import Path
-from typing import Optional, Union
+from typing import Any, Optional, Union
 from urllib.parse import quote
 from xml.etree.cElementTree import SubElement, tostring
 from xml.sax.saxutils import unescape
@@ -57,6 +57,7 @@ from docling_core.types.doc.document import (
     NodeItem,
     OrderedList,
     PictureItem,
+    PictureTabularChartData,
     SectionHeaderItem,
     TableCell,
     TableItem,
@@ -104,6 +105,9 @@ class HTMLParams(CommonParams):
     # Allow for different output styles
     output_style: HTMLOutputStyle = HTMLOutputStyle.SINGLE_COLUMN
 
+    # Enable charts to be printed into HTML as tables
+    enable_chart_tables: bool = True
+
 
 class HTMLTextSerializer(BaseModel, BaseTextSerializer):
     """HTML-specific text item serializer."""
@@ -116,7 +120,7 @@ class HTMLTextSerializer(BaseModel, BaseTextSerializer):
         doc_serializer: BaseDocSerializer,
         doc: DoclingDocument,
         is_inline_scope: bool = False,
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serializes the passed text item to HTML."""
         params = HTMLParams(**kwargs)
@@ -292,7 +296,7 @@ class HTMLTableSerializer(BaseTableSerializer):
         item: TableItem,
         doc_serializer: BaseDocSerializer,
         doc: DoclingDocument,
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serializes the passed table item to HTML."""
         nrows = item.data.num_rows
@@ -363,7 +367,7 @@ class HTMLPictureSerializer(BasePictureSerializer):
         item: PictureItem,
         doc_serializer: BaseDocSerializer,
         doc: DoclingDocument,
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Export picture to HTML format."""
         params = HTMLParams(**kwargs)
@@ -402,9 +406,28 @@ class HTMLPictureSerializer(BasePictureSerializer):
                     and item.image.uri.scheme == "data"
                 ):
                     img_text = f'<img src="{quote(str(item.image.uri))}">'
+
         if img_text:
             res_parts.append(create_ser_result(text=img_text, span_source=item))
 
+        if params.enable_chart_tables:
+            # Check if picture has attached PictureTabularChartData
+            tabular_chart_annotations = [
+                ann
+                for ann in item.annotations
+                if isinstance(ann, PictureTabularChartData)
+            ]
+            if len(tabular_chart_annotations) > 0:
+                temp_doc = DoclingDocument(name="temp")
+                temp_table = temp_doc.add_table(
+                    data=tabular_chart_annotations[0].chart_data
+                )
+                html_table_content = temp_table.export_to_html(temp_doc)
+                if len(html_table_content) > 0:
+                    res_parts.append(
+                        create_ser_result(text=html_table_content, span_source=item)
+                    )
+
         text_res = "".join([r.text for r in res_parts])
         if text_res:
             text_res = f"<figure>{text_res}</figure>"
@@ -551,7 +574,7 @@ class HTMLKeyValueSerializer(BaseKeyValueSerializer):
         item: KeyValueItem,
         doc_serializer: "BaseDocSerializer",
         doc: DoclingDocument,
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serializes the passed key-value item to HTML."""
         res_parts: list[SerializationResult] = []
@@ -588,7 +611,7 @@ class HTMLFormSerializer(BaseFormSerializer):
         item: FormItem,
         doc_serializer: "BaseDocSerializer",
         doc: DoclingDocument,
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serializes the passed form item to HTML."""
         res_parts: list[SerializationResult] = []
@@ -628,7 +651,7 @@ class HTMLListSerializer(BaseModel, BaseListSerializer):
         list_level: int = 0,
         is_inline_scope: bool = False,
         visited: Optional[set[str]] = None,  # refs of visited items
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serializes a list to HTML."""
         my_visited: set[str] = visited if visited is not None else set()
@@ -676,7 +699,7 @@ class HTMLInlineSerializer(BaseInlineSerializer):
         doc: DoclingDocument,
         list_level: int = 0,
         visited: Optional[set[str]] = None,  # refs of visited items
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serializes an inline group to HTML."""
         my_visited: set[str] = visited if visited is not None else set()
@@ -710,7 +733,7 @@ class HTMLFallbackSerializer(BaseFallbackSerializer):
         item: NodeItem,
         doc_serializer: "BaseDocSerializer",
         doc: DoclingDocument,
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Fallback serializer for items not handled by other serializers."""
         if isinstance(item, DocItem):
@@ -739,35 +762,40 @@ class HTMLDocSerializer(DocSerializer):
     params: HTMLParams = HTMLParams()
 
     @override
-    def serialize_bold(self, text: str, **kwargs) -> str:
+    def serialize_bold(self, text: str, **kwargs: Any) -> str:
         """Apply HTML-specific bold serialization."""
         return f"<strong>{text}</strong>"
 
     @override
-    def serialize_italic(self, text: str, **kwargs) -> str:
+    def serialize_italic(self, text: str, **kwargs: Any) -> str:
         """Apply HTML-specific italic serialization."""
         return f"<em>{text}</em>"
 
     @override
-    def serialize_underline(self, text: str, **kwargs) -> str:
+    def serialize_underline(self, text: str, **kwargs: Any) -> str:
         """Apply HTML-specific underline serialization."""
         return f"<u>{text}</u>"
 
     @override
-    def serialize_strikethrough(self, text: str, **kwargs) -> str:
+    def serialize_strikethrough(self, text: str, **kwargs: Any) -> str:
         """Apply HTML-specific strikethrough serialization."""
         return f"<del>{text}</del>"
 
     @override
     def serialize_hyperlink(
-        self, text: str, hyperlink: Union[AnyUrl, Path], **kwargs
+        self,
+        text: str,
+        hyperlink: Union[AnyUrl, Path],
+        **kwargs: Any,
     ) -> str:
         """Apply HTML-specific hyperlink serialization."""
         return f'<a href="{str(hyperlink)}">{text}</a>'
 
     @override
     def serialize_doc(
-        self, parts: list[SerializationResult], **kwargs
+        self,
+        parts: list[SerializationResult],
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serialize a document out of its pages."""
         # Create HTML structure
@@ -779,6 +807,8 @@ class HTMLDocSerializer(DocSerializer):
         ]
 
         if self.params.output_style == HTMLOutputStyle.SPLIT_PAGE:
+            applicable_pages = self._get_applicable_pages()
+
             html_content = "\n".join([p.text for p in parts if p.text])
             next_page: Optional[int] = None
             prev_full_match_end = 0
@@ -791,11 +821,12 @@ class HTMLDocSerializer(DocSerializer):
             # capture last page
             if next_page is not None:
                 pages[next_page] = html_content[prev_full_match_end:]
+            elif applicable_pages is not None and len(applicable_pages) == 1:
+                pages[applicable_pages[0]] = html_content
 
             html_parts.append("<table>")
             html_parts.append("<tbody>")
 
-            applicable_pages = self._get_applicable_pages()
             for page_no, page in pages.items():
 
                 if isinstance(page_no, int):
@@ -869,7 +900,7 @@ class HTMLDocSerializer(DocSerializer):
         self,
         item: FloatingItem,
         tag: str = "figcaption",
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serialize the item's captions."""
         params = self.params.merge_with_patch(patch=kwargs)