docling-core 2.25.0__py3-none-any.whl → 2.26.1__py3-none-any.whl

docling_core/experimental/serializer/html_styles.py

@@ -0,0 +1,212 @@
+"""HTML styles for different export modes."""
+
+
+def _get_css_with_no_styling() -> str:
+    """Return default CSS styles for the HTML document."""
+    return "<style></style>"
+
+
+def _get_css_for_split_page() -> str:
+    """Return default CSS styles for the HTML document."""
+    return """<style>
+    html {
+        background-color: #e1e1e1;
+        font-family: Arial, sans-serif;
+        line-height: 1.6;
+    }
+    img {
+        min-width: 500px;
+        max-width: 100%;
+    }
+    table {
+        border-collapse: collapse;
+        border: 0px solid #fff;
+        width: 100%;
+    }
+    td {
+        vertical-align: top;
+    }
+    .page {
+        background-color: white;
+        margin-top:15px;
+        padding: 30px;
+        border: 1px solid black;
+        width:100%;
+        max-width:1000px;
+        box-shadow: 0 0 10px rgba(0,0,0,0.5);
+    }
+    .page figure {
+        text-align: center;
+    }
+    .page img {
+        max-width: 900px;
+        min-width: auto;
+    }
+    .page table {
+        border-collapse: collapse;
+        margin: 1em 0;
+        width: 100%;
+    }
+    .page table td {
+        border: 1px solid #ddd;
+        padding: 8px;
+        text-align: left;
+    }
+    .page table th {
+        border: 1px solid #ddd;
+        padding: 8px;
+        text-align: left;
+        background-color: #f2f2f2;
+        font-weight: bold;
+    }
+    .page table caption {
+        color: #666;
+        font-style: italic;
+        margin-top: 0.5em;
+        padding: 8px;
+        margin-top: 5px;
+        margin-bottom: 5px;
+    }
+    .page figcaption {
+        color: #666;
+        font-style: italic;
+        margin-top: 0.5em;
+        padding: 8px;
+        margin-top: 5px;
+        margin-bottom: 5px;
+    }
+    code {
+        background-color: rgb(228, 228, 228);
+        border: 1px solid darkgray;
+        padding: 10px;
+        display: inline-block;
+        font-family: monospace;
+        max-width:980px;
+        word-wrap: normal;
+        white-space: pre-wrap;
+        word-wrap: break-word;
+        /*overflow-wrap: break-word;*/
+    }
+</style>
+"""
+
+
+def _get_css_for_single_column() -> str:
+    """Return CSS styles for the single-column HTML document."""
+    return """<style>
+    html {
+        background-color: #f5f5f5;
+        font-family: Arial, sans-serif;
+        line-height: 1.6;
+    }
+    body {
+        max-width: 800px;
+        margin: 0 auto;
+        padding: 2rem;
+        background-color: white;
+        box-shadow: 0 0 10px rgba(0,0,0,0.1);
+    }
+    h1, h2, h3, h4, h5, h6 {
+        color: #333;
+        margin-top: 1.5em;
+        margin-bottom: 0.5em;
+    }
+    h1 {
+        font-size: 2em;
+        border-bottom: 1px solid #eee;
+        padding-bottom: 0.3em;
+    }
+    table {
+        border-collapse: collapse;
+        margin: 1em 0;
+        width: 100%;
+    }
+    th, td {
+        border: 1px solid #ddd;
+        padding: 8px;
+        text-align: left;
+    }
+    th {
+        background-color: #f2f2f2;
+        font-weight: bold;
+    }
+    figure {
+        margin: 1.5em 0;
+        text-align: center;
+    }
+    figcaption {
+        color: #666;
+        font-style: italic;
+        margin-top: 0.5em;
+    }
+    img {
+        max-width: 100%;
+        height: auto;
+    }
+    pre {
+        background-color: #f6f8fa;
+        border-radius: 3px;
+        padding: 1em;
+        overflow: auto;
+    }
+    code {
+        font-family: monospace;
+        background-color: #f6f8fa;
+        padding: 0.2em 0.4em;
+        border-radius: 3px;
+    }
+    pre code {
+        background-color: transparent;
+        padding: 0;
+    }
+    .formula {
+        text-align: center;
+        padding: 0.5em;
+        margin: 1em 0;
+        background-color: #f9f9f9;
+    }
+    .formula-not-decoded {
+        text-align: center;
+        padding: 0.5em;
+        margin: 1em 0;
+        background: repeating-linear-gradient(
+            45deg,
+            #f0f0f0,
+            #f0f0f0 10px,
+            #f9f9f9 10px,
+            #f9f9f9 20px
+        );
+    }
+    .page-break {
+        page-break-after: always;
+        border-top: 1px dashed #ccc;
+        margin: 2em 0;
+    }
+    .key-value-region {
+        background-color: #f9f9f9;
+        padding: 1em;
+        border-radius: 4px;
+        margin: 1em 0;
+    }
+    .key-value-region dt {
+        font-weight: bold;
+    }
+    .key-value-region dd {
+        margin-left: 1em;
+        margin-bottom: 0.5em;
+    }
+    .form-container {
+        border: 1px solid #ddd;
+        padding: 1em;
+        border-radius: 4px;
+        margin: 1em 0;
+    }
+    .form-item {
+        margin-bottom: 0.5em;
+    }
+    .image-classification {
+        font-size: 0.9em;
+        color: #666;
+        margin-top: 0.5em;
+    }
+</style>"""

docling_core/experimental/serializer/markdown.py

@@ -26,7 +26,12 @@ from docling_core.experimental.serializer.base import (
     BaseTextSerializer,
     SerializationResult,
 )
-from docling_core.experimental.serializer.common import CommonParams, DocSerializer
+from docling_core.experimental.serializer.common import (
+    CommonParams,
+    DocSerializer,
+    _PageBreakSerResult,
+    create_ser_result,
+)
 from docling_core.types.doc.base import ImageRefMode
 from docling_core.types.doc.document import (
     CodeItem,
@@ -43,6 +48,7 @@ from docling_core.types.doc.document import (
     NodeItem,
     OrderedList,
     PictureItem,
+    PictureTabularChartData,
     SectionHeaderItem,
     TableItem,
     TextItem,
@@ -57,10 +63,12 @@ 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 -->"
     escape_underscores: bool = True
+    escape_html: bool = True
 
 
 class MarkdownTextSerializer(BaseModel, BaseTextSerializer):
@@ -78,46 +86,51 @@ class MarkdownTextSerializer(BaseModel, BaseTextSerializer):
     ) -> SerializationResult:
         """Serializes the passed item."""
         params = MarkdownParams(**kwargs)
-        parts: list[str] = []
+        res_parts: list[SerializationResult] = []
         escape_html = True
         escape_underscores = True
         if isinstance(item, TitleItem):
-            text = f"# {item.text}"
+            text_part = f"# {item.text}"
         elif isinstance(item, SectionHeaderItem):
-            text = f"{(item.level + 1) * '#'} {item.text}"
+            text_part = f"{(item.level + 1) * '#'} {item.text}"
         elif isinstance(item, CodeItem):
-            text = f"`{item.text}`" if is_inline_scope else f"```\n{item.text}\n```"
+            text_part = (
+                f"`{item.text}`" if is_inline_scope else f"```\n{item.text}\n```"
+            )
             escape_html = False
             escape_underscores = False
         elif isinstance(item, FormulaItem):
             if item.text:
-                text = f"${item.text}$" if is_inline_scope else f"$${item.text}$$"
+                text_part = f"${item.text}$" if is_inline_scope else f"$${item.text}$$"
             elif item.orig:
-                text = "<!-- formula-not-decoded -->"
+                text_part = "<!-- formula-not-decoded -->"
             else:
-                text = ""
+                text_part = ""
             escape_html = False
             escape_underscores = False
         elif params.wrap_width:
-            text = textwrap.fill(item.text, width=params.wrap_width)
+            text_part = textwrap.fill(item.text, width=params.wrap_width)
         else:
-            text = item.text
-        parts.append(text)
+            text_part = item.text
+
+        if text_part:
+            text_res = create_ser_result(text=text_part, span_source=item)
+            res_parts.append(text_res)
 
         if isinstance(item, FloatingItem):
-            cap_text = doc_serializer.serialize_captions(item=item, **kwargs).text
-            if cap_text:
-                parts.append(cap_text)
+            cap_res = doc_serializer.serialize_captions(item=item, **kwargs)
+            if cap_res.text:
+                res_parts.append(cap_res)
 
-        text_res = (" " if is_inline_scope else "\n\n").join(parts)
-        text_res = doc_serializer.post_process(
-            text=text_res,
+        text = (" " if is_inline_scope else "\n\n").join([r.text for r in res_parts])
+        text = doc_serializer.post_process(
+            text=text,
             escape_html=escape_html,
             escape_underscores=escape_underscores,
             formatting=item.formatting,
             hyperlink=item.hyperlink,
         )
-        return SerializationResult(text=text_res)
+        return create_ser_result(text=text, span_source=res_parts)
 
 
 class MarkdownTableSerializer(BaseTableSerializer):
@@ -133,14 +146,14 @@ class MarkdownTableSerializer(BaseTableSerializer):
         **kwargs,
     ) -> SerializationResult:
         """Serializes the passed item."""
-        text_parts: list[str] = []
+        res_parts: list[SerializationResult] = []
 
         cap_res = doc_serializer.serialize_captions(
             item=item,
             **kwargs,
         )
         if cap_res.text:
-            text_parts.append(cap_res.text)
+            res_parts.append(cap_res)
 
         if item.self_ref not in doc_serializer.get_excluded_refs(**kwargs):
             rows = [
@@ -165,11 +178,11 @@ class MarkdownTableSerializer(BaseTableSerializer):
             else:
                 table_text = ""
             if table_text:
-                text_parts.append(table_text)
+                res_parts.append(create_ser_result(text=table_text, span_source=item))
 
-        text_res = "\n\n".join(text_parts)
+        text_res = "\n\n".join([r.text for r in res_parts])
 
-        return SerializationResult(text=text_res)
+        return create_ser_result(text=text_res, span_source=res_parts)
 
 
 class MarkdownPictureSerializer(BasePictureSerializer):
@@ -187,14 +200,14 @@ class MarkdownPictureSerializer(BasePictureSerializer):
         """Serializes the passed item."""
         params = MarkdownParams(**kwargs)
 
-        texts: list[str] = []
+        res_parts: list[SerializationResult] = []
 
         cap_res = doc_serializer.serialize_captions(
             item=item,
             **kwargs,
         )
         if cap_res.text:
-            texts.append(cap_res.text)
+            res_parts.append(cap_res)
 
         if item.self_ref not in doc_serializer.get_excluded_refs(**kwargs):
             img_res = self._serialize_image_part(
@@ -204,11 +217,28 @@ class MarkdownPictureSerializer(BasePictureSerializer):
                 image_placeholder=params.image_placeholder,
             )
             if img_res.text:
-                texts.append(img_res.text)
-
-        text_res = "\n\n".join(texts)
+                res_parts.append(img_res)
+
+        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:
+                    res_parts.append(
+                        create_ser_result(text=md_table_content, span_source=item)
+                    )
+        text_res = "\n\n".join([r.text for r in res_parts])
 
-        return SerializationResult(text=text_res)
+        return create_ser_result(text=text_res, span_source=res_parts)
 
     def _serialize_image_part(
         self,
@@ -255,7 +285,7 @@ class MarkdownPictureSerializer(BasePictureSerializer):
         else:
             text_res = image_placeholder
 
-        return SerializationResult(text=text_res)
+        return create_ser_result(text=text_res, span_source=item)
 
 
 class MarkdownKeyValueSerializer(BaseKeyValueSerializer):
@@ -272,12 +302,13 @@ class MarkdownKeyValueSerializer(BaseKeyValueSerializer):
     ) -> SerializationResult:
         """Serializes the passed item."""
         # TODO add actual implementation
-        text_res = (
-            "<!-- missing-key-value-item -->"
-            if item.self_ref not in doc_serializer.get_excluded_refs()
-            else ""
-        )
-        return SerializationResult(text=text_res)
+        if item.self_ref not in doc_serializer.get_excluded_refs():
+            return create_ser_result(
+                text="<!-- missing-key-value-item -->",
+                span_source=item,
+            )
+        else:
+            return create_ser_result()
 
 
 class MarkdownFormSerializer(BaseFormSerializer):
@@ -294,12 +325,13 @@ class MarkdownFormSerializer(BaseFormSerializer):
     ) -> SerializationResult:
         """Serializes the passed item."""
         # TODO add actual implementation
-        text_res = (
-            "<!-- missing-form-item -->"
-            if item.self_ref not in doc_serializer.get_excluded_refs()
-            else ""
-        )
-        return SerializationResult(text=text_res)
+        if item.self_ref not in doc_serializer.get_excluded_refs():
+            return create_ser_result(
+                text="<!-- missing-form-item -->",
+                span_source=item,
+            )
+        else:
+            return create_ser_result()
 
 
 class MarkdownListSerializer(BaseModel, BaseListSerializer):
@@ -319,7 +351,7 @@ class MarkdownListSerializer(BaseModel, BaseListSerializer):
     ) -> SerializationResult:
         """Serializes the passed item."""
         params = MarkdownParams(**kwargs)
-        my_visited = visited or set()
+        my_visited = visited if visited is not None else set()
         parts = doc_serializer.get_parts(
             item=item,
             list_level=list_level + 1,
@@ -332,6 +364,7 @@ class MarkdownListSerializer(BaseModel, BaseListSerializer):
         for p in parts:
             if p.text and p.text[0] == " " and my_parts:
                 my_parts[-1].text = sep.join([my_parts[-1].text, p.text])  # update last
+                my_parts[-1].spans.extend(p.spans)
             else:
                 my_parts.append(p)
 
@@ -343,12 +376,16 @@ class MarkdownListSerializer(BaseModel, BaseListSerializer):
                 (
                     c.text
                     if c.text and c.text[0] == " "
-                    else f"{indent_str}{f'{i + 1}.' if is_ol else '-'} {c.text}"
+                    else (
+                        f"{indent_str}"
+                        f"{'' if isinstance(c, _PageBreakSerResult) else (f'{i + 1}. ' if is_ol else '- ')}"  # noqa: E501
+                        f"{c.text}"
+                    )
                 )
                 for i, c in enumerate(my_parts)
             ]
         )
-        return SerializationResult(text=text_res)
+        return create_ser_result(text=text_res, span_source=my_parts)
 
 
 class MarkdownInlineSerializer(BaseInlineSerializer):
@@ -366,15 +403,16 @@ class MarkdownInlineSerializer(BaseInlineSerializer):
         **kwargs,
     ) -> SerializationResult:
         """Serializes the passed item."""
-        my_visited = visited or set()
+        my_visited = visited if visited is not None else set()
         parts = doc_serializer.get_parts(
             item=item,
             list_level=list_level,
             is_inline_scope=True,
             visited=my_visited,
+            **kwargs,
         )
         text_res = " ".join([p.text for p in parts if p.text])
-        return SerializationResult(text=text_res)
+        return create_ser_result(text=text_res, span_source=parts)
 
 
 class MarkdownFallbackSerializer(BaseFallbackSerializer):
@@ -391,10 +429,12 @@ class MarkdownFallbackSerializer(BaseFallbackSerializer):
     ) -> SerializationResult:
         """Serializes the passed item."""
         if isinstance(item, DocItem):
-            text_res = "<!-- missing-text -->"
+            return create_ser_result(
+                text="<!-- missing-text -->",
+                span_source=item,
+            )
         else:
-            text_res = ""  # TODO go with explicit None return type?
-        return SerializationResult(text=text_res)
+            return create_ser_result()
 
 
 class MarkdownDocSerializer(DocSerializer):
@@ -472,7 +512,7 @@ class MarkdownDocSerializer(DocSerializer):
         params = self.params.merge_with_patch(patch=kwargs)
         if escape_underscores and params.escape_underscores:
             res = self._escape_underscores(text)
-        if escape_html:
+        if escape_html and params.escape_html:
             res = html.escape(res, quote=False)
         res = super().post_process(
             text=res,
@@ -482,17 +522,19 @@ class MarkdownDocSerializer(DocSerializer):
         return res
 
     @override
-    def serialize_page(self, parts: list[SerializationResult]) -> SerializationResult:
-        """Serialize a page out of its parts."""
-        text_res = "\n\n".join([p.text for p in parts])
-        return SerializationResult(text=text_res)
+    def serialize_doc(
+        self, *, parts: list[SerializationResult], **kwargs
+    ) -> SerializationResult:
+        """Serialize a document out of its parts."""
+        text_res = "\n\n".join([p.text for p in parts if p.text])
+        if self.params.page_break_placeholder:
+            page_sep = self.params.page_break_placeholder or ""
+            for full_match, _, _ in self._get_page_breaks(text=text_res):
+                text_res = text_res.replace(full_match, page_sep)
+
+        return create_ser_result(text=text_res, span_source=parts)
 
     @override
-    def serialize_doc(self, pages: list[SerializationResult]) -> SerializationResult:
-        """Serialize a document out of its pages."""
-        if self.params.page_break_placeholder is not None:
-            sep = f"\n\n{self.params.page_break_placeholder}\n\n"
-            text_res = sep.join([p.text for p in pages if p.text])
-            return SerializationResult(text=text_res)
-        else:
-            return self.serialize_page(parts=pages)
+    def requires_page_break(self):
+        """Whether to add page breaks."""
+        return self.params.page_break_placeholder is not None

docling_core/transforms/chunker/base.py

@@ -9,6 +9,7 @@ from abc import ABC, abstractmethod
 from typing import Any, ClassVar, Iterator
 
 from pydantic import BaseModel
+from typing_extensions import deprecated
 
 from docling_core.types.doc import DoclingDocument as DLDocument
 
@@ -65,8 +66,8 @@ class BaseChunker(BaseModel, ABC):
         """
         raise NotImplementedError()
 
-    def serialize(self, chunk: BaseChunk) -> str:
-        """Serialize the given chunk. This base implementation is embedding-targeted.
+    def contextualize(self, chunk: BaseChunk) -> str:
+        """Contextualize the given chunk. This implementation is embedding-targeted.
 
         Args:
             chunk: chunk to serialize
@@ -93,3 +94,8 @@ class BaseChunker(BaseModel, ABC):
         items.append(chunk.text)
 
         return self.delim.join(items)
+
+    @deprecated("Use contextualize() instead.")
+    def serialize(self, chunk: BaseChunk) -> str:
+        """Contextualize the given chunk. This implementation is embedding-targeted."""
+        return self.contextualize(chunk=chunk)