docling-core 2.26.4__tar.gz → 2.27.0__tar.gz

{docling_core-2.26.4 → docling_core-2.27.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: docling-core
-Version: 2.26.4
+Version: 2.27.0
 Summary: A python library to define and validate data types in Docling.
 Home-page: https://github.com/docling-project
 License: MIT

{docling_core-2.26.4 → docling_core-2.27.0}/docling_core/experimental/serializer/html.py

@@ -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."""
@@ -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>"
@@ -779,6 +802,8 @@ 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 +816,12 @@ 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):

{docling_core-2.26.4 → docling_core-2.27.0}/docling_core/types/doc/document.py

@@ -3143,6 +3143,7 @@ class DoclingDocument(BaseModel):
         from_element: int = 0,
         to_element: int = sys.maxsize,
         labels: Optional[set[DocItemLabel]] = None,
+        enable_chart_tables: bool = True,
         image_mode: ImageRefMode = ImageRefMode.PLACEHOLDER,
         formula_to_mathml: bool = True,
         page_no: Optional[int] = None,
@@ -3176,6 +3177,7 @@ class DoclingDocument(BaseModel):
             start_idx=from_element,
             stop_idx=to_element,
             image_mode=image_mode,
+            enable_chart_tables=enable_chart_tables,
             formula_to_mathml=formula_to_mathml,
             html_head=html_head,
             html_lang=html_lang,

{docling_core-2.26.4 → docling_core-2.27.0}/docling_core/types/doc/labels.py

@@ -45,6 +45,7 @@ class DocItemLabel(str, Enum):
             DocItemLabel.PAGE_FOOTER: (204, 255, 204),
             DocItemLabel.PAGE_HEADER: (204, 255, 204),
             DocItemLabel.PICTURE: (255, 204, 164),
+            DocItemLabel.CHART: (255, 204, 164),
             DocItemLabel.SECTION_HEADER: (255, 153, 153),
             DocItemLabel.TABLE: (255, 204, 204),
             DocItemLabel.TEXT: (255, 255, 153),
@@ -58,7 +59,7 @@ class DocItemLabel(str, Enum):
             DocItemLabel.PARAGRAPH: (255, 255, 153),
             DocItemLabel.REFERENCE: (176, 224, 230),
         }
-        return color_map[label]
+        return color_map.get(label, (0, 0, 0))
 
 
 class GroupLabel(str, Enum):

{docling_core-2.26.4 → docling_core-2.27.0}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "docling-core"
-version = "2.26.4"
+version = "2.27.0"
 description = "A python library to define and validate data types in Docling."
 license = "MIT"
 authors = [
@@ -80,6 +80,8 @@ types-setuptools = "^70.3.0"
 python-semantic-release = "^7.32.2"
 pandas-stubs = "^2.1.4.231227"
 ipykernel = "^6.29.5"
+coverage = "^7.6.2"
+pytest-cov = "^6.0.0"
 
 [tool.poetry.group.constraints]
 optional = true