feat: Markdown chart serializer (picture+table) (#235)

* Markdown serializer for charts

Signed-off-by: Maksym Lysak <[email protected]>

* Added serializer parameter disable_chart_tables, that prevents printing of chart table into MarkDown

Signed-off-by: Maksym Lysak <[email protected]>

* cleaning

Signed-off-by: Maksym Lysak <[email protected]>

* propagating paramerer disable_chart_tables

Signed-off-by: Maksym Lysak <[email protected]>

* Updated parameter name from negative to positive notation enable_chart_tables

Signed-off-by: Maksym Lysak <[email protected]>

---------

Signed-off-by: Maksym Lysak <[email protected]>
Co-authored-by: Maksym Lysak <[email protected]>

Author: maxmnemonic

docling_core/experimental/serializer/markdown.py

@@ -43,6 +43,7 @@
     NodeItem,
     OrderedList,
     PictureItem,
+    PictureTabularChartData,
     SectionHeaderItem,
     TableItem,
     TextItem,
@@ -57,6 +58,7 @@ class MarkdownParams(CommonParams):
     layers: set[ContentLayer] = {ContentLayer.BODY}
     image_mode: ImageRefMode = ImageRefMode.PLACEHOLDER
     image_placeholder: str = "<!-- image -->"
+    enable_chart_tables: bool = True
     indent: int = 4
     wrap_width: Optional[PositiveInt] = None
     page_break_placeholder: Optional[str] = None  # e.g. "<!-- page break -->"
@@ -206,6 +208,21 @@ def serialize(
             if img_res.text:
                 texts.append(img_res.text)
 
+        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
+                )
+                md_table_content = temp_table.export_to_markdown(temp_doc)
+                if len(md_table_content) > 0:
+                    texts.append(md_table_content)
         text_res = "\n\n".join(texts)
 
         return SerializationResult(text=text_res)

docling_core/types/doc/document.py

@@ -3130,6 +3130,7 @@ def export_to_markdown(  # noqa: C901
         strict_text: bool = False,
         escape_underscores: bool = True,
         image_placeholder: str = "<!-- image -->",
+        enable_chart_tables: bool = True,
         image_mode: ImageRefMode = ImageRefMode.PLACEHOLDER,
         indent: int = 4,
         text_width: int = -1,
@@ -3197,6 +3198,7 @@ def export_to_markdown(  # noqa: C901
                 stop_idx=to_element,
                 escape_underscores=escape_underscores,
                 image_placeholder=image_placeholder,
+                enable_chart_tables=enable_chart_tables,
                 image_mode=image_mode,
                 indent=indent,
                 wrap_width=text_width if text_width > 0 else None,