feat: add chart serialization support for pptx files and related classes

docling_core/transforms/serializer/base.py

@@ -11,6 +11,7 @@
 from pydantic import AnyUrl, BaseModel
 
 from docling_core.types.doc.document import (
+    ChartItem,
     DocItem,
     DoclingDocument,
     FloatingItem,
@@ -82,6 +83,22 @@ def serialize(
         ...
 
 
+class BaseChartSerializer(ABC):
+    """Basr class for chart item serializers."""
+
+    @abstractmethod
+    def serialize(
+        self,
+        *,
+        item: ChartItem,
+        doc_serializer: "BaseDocSerializer",
+        doc: DoclingDocument,
+        **kwargs: Any,
+    ) -> SerializationResult:
+        """Serializes the passed item."""
+        ...
+
+
 class BasePictureSerializer(ABC):
     """Base class for picture item serializers."""
 

docling_core/transforms/serializer/common.py

@@ -16,6 +16,7 @@
 
 from docling_core.transforms.serializer.base import (
     BaseAnnotationSerializer,
+    BaseChartSerializer,
     BaseDocSerializer,
     BaseFallbackSerializer,
     BaseFormSerializer,
@@ -30,6 +31,7 @@
 )
 from docling_core.types.doc.document import (
     DOCUMENT_TOKENS_EXPORT_LABELS,
+    ChartItem,
     ContentLayer,
     DescriptionAnnotation,
     DocItem,
@@ -207,6 +209,7 @@ class DocSerializer(BaseModel, BaseDocSerializer):
 
     text_serializer: BaseTextSerializer
     table_serializer: BaseTableSerializer
+    chart_serializer: BaseChartSerializer
     picture_serializer: BasePictureSerializer
     key_value_serializer: BaseKeyValueSerializer
     form_serializer: BaseFormSerializer
@@ -362,6 +365,14 @@ def serialize(
                 visited=my_visited,
                 **my_kwargs,
             )
+        elif isinstance(item, ChartItem):
+            part = self.chart_serializer.serialize(
+                item=item,
+                doc_serializer=self,
+                doc=self.doc,
+                visited=my_visited,
+                **my_kwargs,
+            )
         elif isinstance(item, PictureItem):
             part = self.picture_serializer.serialize(
                 item=item,

docling_core/transforms/serializer/doctags.py

@@ -8,6 +8,7 @@
 
 from docling_core.transforms.serializer.base import (
     BaseAnnotationSerializer,
+    BaseChartSerializer,
     BaseDocSerializer,
     BaseFallbackSerializer,
     BaseFormSerializer,
@@ -27,6 +28,7 @@
 )
 from docling_core.types.doc.base import BoundingBox
 from docling_core.types.doc.document import (
+    ChartItem,
     CodeItem,
     DocItem,
     DoclingDocument,
@@ -207,6 +209,23 @@ def serialize(
         return create_ser_result(text=text_res, span_source=res_parts)
 
 
+class DocTagsChartSerializer(BaseChartSerializer):
+    """DocTags-specific chart item serializer."""
+
+    @override
+    def serialize(
+        self,
+        *,
+        item: ChartItem,
+        doc_serializer: BaseDocSerializer,
+        doc: DoclingDocument,
+        **kwargs: Any,
+    ) -> SerializationResult:
+        """Serializes the passed item."""
+        # TODO add actual implementation
+        return create_ser_result()
+
+
 class DocTagsPictureSerializer(BasePictureSerializer):
     """DocTags-specific picture item serializer."""
 
@@ -539,6 +558,7 @@ class DocTagsDocSerializer(DocSerializer):
 
     text_serializer: BaseTextSerializer = DocTagsTextSerializer()
     table_serializer: BaseTableSerializer = DocTagsTableSerializer()
+    chart_serializer: BaseChartSerializer = DocTagsChartSerializer()
     picture_serializer: BasePictureSerializer = DocTagsPictureSerializer()
     key_value_serializer: BaseKeyValueSerializer = DocTagsKeyValueSerializer()
     form_serializer: BaseFormSerializer = DocTagsFormSerializer()

docling_core/transforms/serializer/html.py

@@ -22,6 +22,7 @@
 
 from docling_core.transforms.serializer.base import (
     BaseAnnotationSerializer,
+    BaseChartSerializer,
     BaseDocSerializer,
     BaseFallbackSerializer,
     BaseFormSerializer,
@@ -46,6 +47,7 @@
 from docling_core.transforms.visualizer.base import BaseVisualizer
 from docling_core.types.doc.base import ImageRefMode
 from docling_core.types.doc.document import (
+    ChartItem,
     CodeItem,
     ContentLayer,
     DescriptionAnnotation,
@@ -406,6 +408,23 @@ def serialize(
         return create_ser_result(text=text_res, span_source=res_parts)
 
 
+class HTMLChartSerializer(BaseChartSerializer):
+    """HTML-specific chart item serializer."""
+
+    @override
+    def serialize(
+        self,
+        *,
+        item: ChartItem,
+        doc_serializer: BaseDocSerializer,
+        doc: DoclingDocument,
+        **kwargs: Any,
+    ) -> SerializationResult:
+        """Export chart to HTML format."""
+        # TODO add actual implementation
+        return create_ser_result()
+
+
 class HTMLPictureSerializer(BasePictureSerializer):
     """HTML-specific picture item serializer."""
 
@@ -850,6 +869,7 @@ class HTMLDocSerializer(DocSerializer):
 
     text_serializer: BaseTextSerializer = HTMLTextSerializer()
     table_serializer: BaseTableSerializer = HTMLTableSerializer()
+    chart_serializer: BaseChartSerializer = HTMLChartSerializer()
     picture_serializer: BasePictureSerializer = HTMLPictureSerializer()
     key_value_serializer: BaseKeyValueSerializer = HTMLKeyValueSerializer()
     form_serializer: BaseFormSerializer = HTMLFormSerializer()

docling_core/transforms/serializer/markdown.py

@@ -17,6 +17,7 @@
 
 from docling_core.transforms.serializer.base import (
     BaseAnnotationSerializer,
+    BaseChartSerializer,
     BaseDocSerializer,
     BaseFallbackSerializer,
     BaseFormSerializer,
@@ -36,6 +37,7 @@
 )
 from docling_core.types.doc.base import ImageRefMode
 from docling_core.types.doc.document import (
+    ChartItem,
     CodeItem,
     ContentLayer,
     DescriptionAnnotation,
@@ -357,6 +359,62 @@ def serialize(
         return create_ser_result(text=text_res, span_source=res_parts)
 
 
+class MarkdownChartSerializer(BaseChartSerializer):
+    """Mardown-specific chart item serializer."""
+
+    @override
+    def serialize(
+        self,
+        *,
+        item: ChartItem,
+        doc_serializer: BaseDocSerializer,
+        doc: DoclingDocument,
+        **kwargs: Any,
+    ) -> SerializationResult:
+        """Serializes the passed item."""
+        params = MarkdownParams(**kwargs)
+        res_parts: list[SerializationResult] = []
+        cap_res = doc_serializer.serialize_captions(
+            item=item,
+            **kwargs,
+        )
+
+        if cap_res.text:
+            res_parts.append(cap_res)
+
+        if item.self_ref not in doc_serializer.get_excluded_refs():
+            if params.include_annotations:
+                ann_res = doc_serializer.serialize_annotations(
+                    item=item,
+                    **kwargs,
+                )
+                if ann_res.text:
+                    res_parts.append(ann_res)
+
+            res_parts.append(create_ser_result(text=item.data.title, span_source=item))
+            res_parts.append(create_ser_result(text=item.data.kind, span_source=item))
+            if item.data.is_categorical and item.data.categories:
+                categories = ", ".join(item.data.categories)
+                if categories:
+                    res_parts.append(
+                        create_ser_result(
+                            text=f"Categories: {categories}", span_source=item
+                        )
+                    )
+            if item.data.series:
+                series_text = ""
+                for series_name, series_data in item.data.series:
+                    series_text += f"- {series_name}: {series_data}\n"
+                if series_text:
+                    res_parts.append(
+                        create_ser_result(text=series_text.strip(), span_source=item)
+                    )
+
+        text_res = "\n\n".join([r.text for r in res_parts])
+
+        return create_ser_result(text=text_res, span_source=res_parts)
+
+
 class MarkdownPictureSerializer(BasePictureSerializer):
     """Markdown-specific picture item serializer."""
 
@@ -621,6 +679,7 @@ class MarkdownDocSerializer(DocSerializer):
 
     text_serializer: BaseTextSerializer = MarkdownTextSerializer()
     table_serializer: BaseTableSerializer = MarkdownTableSerializer()
+    chart_serializer: BaseChartSerializer = MarkdownChartSerializer()
     picture_serializer: BasePictureSerializer = MarkdownPictureSerializer()
     key_value_serializer: BaseKeyValueSerializer = MarkdownKeyValueSerializer()
     form_serializer: BaseFormSerializer = MarkdownFormSerializer()

docling_core/types/doc/__init__.py

@@ -10,6 +10,8 @@
     AnyTableCell,
     BaseAnnotation,
     ChartBar,
+    ChartData,
+    ChartItem,
     ChartLine,
     ChartPoint,
     ChartSlice,