docling-core 2.28.1__tar.gz → 2.29.0__tar.gz

{docling_core-2.28.1 → docling_core-2.29.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: docling-core
-Version: 2.28.1
+Version: 2.29.0
 Summary: A python library to define and validate data types in Docling.
 Home-page: https://github.com/docling-project
 License: MIT

{docling_core-2.28.1 → docling_core-2.29.0}/docling_core/transforms/chunker/hierarchical_chunker.py

@@ -14,19 +14,19 @@ from typing import Any, ClassVar, Final, Iterator, Literal, Optional
 from pydantic import ConfigDict, Field, StringConstraints, field_validator
 from typing_extensions import Annotated, override
 
-from docling_core.experimental.serializer.base import (
+from docling_core.search.package import VERSION_PATTERN
+from docling_core.transforms.chunker import BaseChunk, BaseChunker, BaseMeta
+from docling_core.transforms.serializer.base import (
     BaseDocSerializer,
     BaseSerializerProvider,
     BaseTableSerializer,
     SerializationResult,
 )
-from docling_core.experimental.serializer.common import create_ser_result
-from docling_core.experimental.serializer.markdown import (
+from docling_core.transforms.serializer.common import create_ser_result
+from docling_core.transforms.serializer.markdown import (
     MarkdownDocSerializer,
     MarkdownParams,
 )
-from docling_core.search.package import VERSION_PATTERN
-from docling_core.transforms.chunker import BaseChunk, BaseChunker, BaseMeta
 from docling_core.types import DoclingDocument as DLDocument
 from docling_core.types.doc.base import ImageRefMode
 from docling_core.types.doc.document import (

{docling_core-2.28.1 → docling_core-2.29.0}/docling_core/transforms/chunker/hybrid_chunker.py

@@ -25,10 +25,6 @@ except ImportError:
         "`pip install 'docling-core[chunking-openai]'`"
     )
 
-from docling_core.experimental.serializer.base import (
-    BaseDocSerializer,
-    BaseSerializerProvider,
-)
 from docling_core.transforms.chunker import (
     BaseChunk,
     BaseChunker,
@@ -36,6 +32,10 @@ from docling_core.transforms.chunker import (
     DocMeta,
     HierarchicalChunker,
 )
+from docling_core.transforms.serializer.base import (
+    BaseDocSerializer,
+    BaseSerializerProvider,
+)
 from docling_core.types import DoclingDocument
 
 

{docling_core-2.28.1/docling_core/experimental → docling_core-2.29.0/docling_core/transforms}/serializer/common.py

@@ -14,7 +14,7 @@ from typing import Any, Iterable, Optional, Tuple, Union
 from pydantic import AnyUrl, BaseModel, NonNegativeInt, computed_field
 from typing_extensions import Self, override
 
-from docling_core.experimental.serializer.base import (
+from docling_core.transforms.serializer.base import (
     BaseDocSerializer,
     BaseFallbackSerializer,
     BaseFormSerializer,

{docling_core-2.28.1/docling_core/experimental → docling_core-2.29.0/docling_core/transforms}/serializer/doctags.py

@@ -6,7 +6,7 @@ from typing import Any, Dict, List, Optional, Union
 from pydantic import BaseModel
 from typing_extensions import override
 
-from docling_core.experimental.serializer.base import (
+from docling_core.transforms.serializer.base import (
     BaseDocSerializer,
     BaseFallbackSerializer,
     BaseFormSerializer,
@@ -18,7 +18,7 @@ from docling_core.experimental.serializer.base import (
     BaseTextSerializer,
     SerializationResult,
 )
-from docling_core.experimental.serializer.common import (
+from docling_core.transforms.serializer.common import (
     CommonParams,
     DocSerializer,
     create_ser_result,

{docling_core-2.28.1/docling_core/experimental → docling_core-2.29.0/docling_core/transforms}/serializer/html.py

@@ -19,7 +19,7 @@ import latex2mathml.converter
 from pydantic import AnyUrl, BaseModel
 from typing_extensions import override
 
-from docling_core.experimental.serializer.base import (
+from docling_core.transforms.serializer.base import (
     BaseDocSerializer,
     BaseFallbackSerializer,
     BaseFormSerializer,
@@ -31,12 +31,12 @@ from docling_core.experimental.serializer.base import (
     BaseTextSerializer,
     SerializationResult,
 )
-from docling_core.experimental.serializer.common import (
+from docling_core.transforms.serializer.common import (
     CommonParams,
     DocSerializer,
     create_ser_result,
 )
-from docling_core.experimental.serializer.html_styles import (
+from docling_core.transforms.serializer.html_styles import (
     _get_css_for_single_column,
     _get_css_for_split_page,
 )

{docling_core-2.28.1/docling_core/experimental → docling_core-2.29.0/docling_core/transforms}/serializer/markdown.py

@@ -14,7 +14,7 @@ from pydantic import AnyUrl, BaseModel, PositiveInt
 from tabulate import tabulate
 from typing_extensions import override
 
-from docling_core.experimental.serializer.base import (
+from docling_core.transforms.serializer.base import (
     BaseDocSerializer,
     BaseFallbackSerializer,
     BaseFormSerializer,
@@ -26,7 +26,7 @@ from docling_core.experimental.serializer.base import (
     BaseTextSerializer,
     SerializationResult,
 )
-from docling_core.experimental.serializer.common import (
+from docling_core.transforms.serializer.common import (
     CommonParams,
     DocSerializer,
     _PageBreakSerResult,

{docling_core-2.28.1 → docling_core-2.29.0}/docling_core/transforms/visualizer/layout_visualizer.py

@@ -149,38 +149,41 @@ class LayoutVisualizer(BaseVisualizer):
                 continue
             if len(elem.prov) == 0:
                 continue  # Skip elements without provenances
-            prov = elem.prov[0]
-            page_nr = prov.page_no
-
-            if page_nr in my_images:
-                image = my_images[page_nr]
-            else:
-                raise RuntimeError(f"Cannot visualize page-image for {page_nr}")
-
-            if prev_page_nr is None or page_nr > prev_page_nr:  # new page begins
-                # complete previous drawing
-                if prev_page_nr is not None and prev_image and clusters:
-                    self._draw_clusters(
-                        image=prev_image,
-                        clusters=clusters,
-                        scale_x=prev_image.width / doc.pages[prev_page_nr].size.width,
-                        scale_y=prev_image.height / doc.pages[prev_page_nr].size.height,
-                    )
-                    clusters = []
 
-            tlo_bbox = prov.bbox.to_top_left_origin(
-                page_height=doc.pages[prov.page_no].size.height
-            )
-            cluster = _TLCluster(
-                id=idx,
-                label=elem.label,
-                brec=_TLBoundingRectangle.from_bounding_box(bbox=tlo_bbox),
-                cells=[],
-            )
-            clusters.append(cluster)
+            for prov in elem.prov:
+                page_nr = prov.page_no
+
+                if page_nr in my_images:
+                    image = my_images[page_nr]
+                else:
+                    raise RuntimeError(f"Cannot visualize page-image for {page_nr}")
+
+                if prev_page_nr is None or page_nr > prev_page_nr:  # new page begins
+                    # complete previous drawing
+                    if prev_page_nr is not None and prev_image and clusters:
+                        self._draw_clusters(
+                            image=prev_image,
+                            clusters=clusters,
+                            scale_x=prev_image.width
+                            / doc.pages[prev_page_nr].size.width,
+                            scale_y=prev_image.height
+                            / doc.pages[prev_page_nr].size.height,
+                        )
+                        clusters = []
+
+                tlo_bbox = prov.bbox.to_top_left_origin(
+                    page_height=doc.pages[prov.page_no].size.height
+                )
+                cluster = _TLCluster(
+                    id=idx,
+                    label=elem.label,
+                    brec=_TLBoundingRectangle.from_bounding_box(bbox=tlo_bbox),
+                    cells=[],
+                )
+                clusters.append(cluster)
 
-            prev_page_nr = page_nr
-            prev_image = image
+                prev_page_nr = page_nr
+                prev_image = image
 
         # complete last drawing
         if prev_page_nr is not None and prev_image and clusters:

{docling_core-2.28.1 → docling_core-2.29.0}/docling_core/transforms/visualizer/reading_order_visualizer.py

@@ -77,57 +77,59 @@ class ReadingOrderVisualizer(BaseVisualizer):
                 continue
             if len(elem.prov) == 0:
                 continue  # Skip elements without provenances
-            prov = elem.prov[0]
-            page_no = prov.page_no
-            image = my_images.get(page_no)
-
-            if image is None or prev_page is None or page_no > prev_page:
-                # new page begins
-                prev_page = page_no
-                x0 = y0 = None
-
-                if image is None:
-                    page_image = doc.pages[page_no].image
-                    if page_image is None or (pil_img := page_image.pil_image) is None:
-                        raise RuntimeError("Cannot visualize document without images")
-                    else:
-                        image = deepcopy(pil_img)
-                        my_images[page_no] = image
-            draw = ImageDraw.Draw(image)
-
-            # if prov.page_no not in true_doc.pages or prov.page_no != 1:
-            #     logging.error(f"{prov.page_no} not in true_doc.pages -> skipping! ")
-            #     continue
-
-            tlo_bbox = prov.bbox.to_top_left_origin(
-                page_height=doc.pages[prov.page_no].size.height
-            )
-            ro_bbox = tlo_bbox.normalized(doc.pages[prov.page_no].size)
-            ro_bbox.l = round(ro_bbox.l * image.width)  # noqa: E741
-            ro_bbox.r = round(ro_bbox.r * image.width)
-            ro_bbox.t = round(ro_bbox.t * image.height)
-            ro_bbox.b = round(ro_bbox.b * image.height)
-
-            if ro_bbox.b > ro_bbox.t:
-                ro_bbox.b, ro_bbox.t = ro_bbox.t, ro_bbox.b
-
-            if x0 is None and y0 is None:
-                x0 = (ro_bbox.l + ro_bbox.r) / 2.0
-                y0 = (ro_bbox.b + ro_bbox.t) / 2.0
-            else:
-                assert x0 is not None
-                assert y0 is not None
-
-                x1 = (ro_bbox.l + ro_bbox.r) / 2.0
-                y1 = (ro_bbox.b + ro_bbox.t) / 2.0
-
-                draw = self._draw_arrow(
-                    draw=draw,
-                    arrow_coords=(x0, y0, x1, y1),
-                    line_width=2,
-                    color="red",
+
+            for prov in elem.prov:
+                page_no = prov.page_no
+                image = my_images.get(page_no)
+
+                if image is None or prev_page is None or page_no > prev_page:
+                    # new page begins
+                    prev_page = page_no
+                    x0 = y0 = None
+
+                    if image is None:
+                        page_image = doc.pages[page_no].image
+                        if (
+                            page_image is None
+                            or (pil_img := page_image.pil_image) is None
+                        ):
+                            raise RuntimeError(
+                                "Cannot visualize document without images"
+                            )
+                        else:
+                            image = deepcopy(pil_img)
+                            my_images[page_no] = image
+                draw = ImageDraw.Draw(image)
+
+                tlo_bbox = prov.bbox.to_top_left_origin(
+                    page_height=doc.pages[prov.page_no].size.height
                 )
-                x0, y0 = x1, y1
+                ro_bbox = tlo_bbox.normalized(doc.pages[prov.page_no].size)
+                ro_bbox.l = round(ro_bbox.l * image.width)  # noqa: E741
+                ro_bbox.r = round(ro_bbox.r * image.width)
+                ro_bbox.t = round(ro_bbox.t * image.height)
+                ro_bbox.b = round(ro_bbox.b * image.height)
+
+                if ro_bbox.b > ro_bbox.t:
+                    ro_bbox.b, ro_bbox.t = ro_bbox.t, ro_bbox.b
+
+                if x0 is None and y0 is None:
+                    x0 = (ro_bbox.l + ro_bbox.r) / 2.0
+                    y0 = (ro_bbox.b + ro_bbox.t) / 2.0
+                else:
+                    assert x0 is not None
+                    assert y0 is not None
+
+                    x1 = (ro_bbox.l + ro_bbox.r) / 2.0
+                    y1 = (ro_bbox.b + ro_bbox.t) / 2.0
+
+                    draw = self._draw_arrow(
+                        draw=draw,
+                        arrow_coords=(x0, y0, x1, y1),
+                        line_width=2,
+                        color="red",
+                    )
+                    x0, y0 = x1, y1
         return my_images
 
     @override

{docling_core-2.28.1 → docling_core-2.29.0}/docling_core/types/doc/document.py

@@ -872,7 +872,7 @@ class TextItem(DocItem):
         :param add_content: bool:  (Default value = True)
 
         """
-        from docling_core.experimental.serializer.doctags import (
+        from docling_core.transforms.serializer.doctags import (
             DocTagsDocSerializer,
             DocTagsParams,
         )
@@ -930,7 +930,7 @@ class SectionHeaderItem(TextItem):
         :param add_content: bool:  (Default value = True)
 
         """
-        from docling_core.experimental.serializer.doctags import (
+        from docling_core.transforms.serializer.doctags import (
             DocTagsDocSerializer,
             DocTagsParams,
         )
@@ -1020,7 +1020,7 @@ class CodeItem(FloatingItem, TextItem):
         :param add_content: bool:  (Default value = True)
 
         """
-        from docling_core.experimental.serializer.doctags import (
+        from docling_core.transforms.serializer.doctags import (
             DocTagsDocSerializer,
             DocTagsParams,
         )
@@ -1091,7 +1091,7 @@ class PictureItem(FloatingItem):
         image_placeholder: str = "<!-- image -->",
     ) -> str:
         """Export picture to Markdown format."""
-        from docling_core.experimental.serializer.markdown import (
+        from docling_core.transforms.serializer.markdown import (
             MarkdownDocSerializer,
             MarkdownParams,
         )
@@ -1118,7 +1118,7 @@ class PictureItem(FloatingItem):
         image_mode: ImageRefMode = ImageRefMode.PLACEHOLDER,
     ) -> str:
         """Export picture to HTML format."""
-        from docling_core.experimental.serializer.html import (
+        from docling_core.transforms.serializer.html import (
             HTMLDocSerializer,
             HTMLParams,
         )
@@ -1159,7 +1159,7 @@ class PictureItem(FloatingItem):
         :param # not used at the moment
 
         """
-        from docling_core.experimental.serializer.doctags import (
+        from docling_core.transforms.serializer.doctags import (
             DocTagsDocSerializer,
             DocTagsParams,
         )
@@ -1235,7 +1235,7 @@ class TableItem(FloatingItem):
     def export_to_markdown(self, doc: Optional["DoclingDocument"] = None) -> str:
         """Export the table as markdown."""
         if doc is not None:
-            from docling_core.experimental.serializer.markdown import (
+            from docling_core.transforms.serializer.markdown import (
                 MarkdownDocSerializer,
             )
 
@@ -1282,7 +1282,7 @@ class TableItem(FloatingItem):
     ) -> str:
         """Export the table as html."""
         if doc is not None:
-            from docling_core.experimental.serializer.html import HTMLDocSerializer
+            from docling_core.transforms.serializer.html import HTMLDocSerializer
 
             serializer = HTMLDocSerializer(doc=doc)
             text = serializer.serialize(item=self).text
@@ -1414,7 +1414,7 @@ class TableItem(FloatingItem):
         :param add_caption: bool:  (Default value = True)
 
         """
-        from docling_core.experimental.serializer.doctags import (
+        from docling_core.transforms.serializer.doctags import (
             DocTagsDocSerializer,
             DocTagsParams,
         )
@@ -1512,7 +1512,7 @@ class KeyValueItem(FloatingItem):
         :param add_content: bool:  (Default value = True)
 
         """
-        from docling_core.experimental.serializer.doctags import (
+        from docling_core.transforms.serializer.doctags import (
             DocTagsDocSerializer,
             DocTagsParams,
         )
@@ -2999,7 +2999,7 @@ class DoclingDocument(BaseModel):
         :returns: The exported Markdown representation.
         :rtype: str
         """
-        from docling_core.experimental.serializer.markdown import (
+        from docling_core.transforms.serializer.markdown import (
             MarkdownDocSerializer,
             MarkdownParams,
         )
@@ -3153,7 +3153,7 @@ class DoclingDocument(BaseModel):
         split_page_view: bool = False,
     ) -> str:
         r"""Serialize to HTML."""
-        from docling_core.experimental.serializer.html import (
+        from docling_core.transforms.serializer.html import (
             HTMLDocSerializer,
             HTMLOutputStyle,
             HTMLParams,
@@ -3195,9 +3195,9 @@ class DoclingDocument(BaseModel):
 
         return ser_res.text
 
+    @staticmethod
     def load_from_doctags(  # noqa: C901
-        self,
-        doctag_document: DocTagsDocument,
+        doctag_document: DocTagsDocument, document_name: str = "Document"
     ) -> "DoclingDocument":
         r"""Load Docling document from lists of DocTags and Images."""
         # Maps the recognized tag to a Docling label.
@@ -3221,6 +3221,8 @@ class DoclingDocument(BaseModel):
             "key_value_region": DocItemLabel.KEY_VALUE_REGION,
         }
 
+        doc = DoclingDocument(name=document_name)
+
         def extract_bounding_box(text_chunk: str) -> Optional[BoundingBox]:
             """Extract <loc_...> coords from the chunk, normalized by / 500."""
             coords = re.findall(r"<loc_(\d+)>", text_chunk)
@@ -3244,7 +3246,7 @@ class DoclingDocument(BaseModel):
                 caption_content = caption.group(1)
                 bbox = extract_bounding_box(caption_content)
                 caption_text = extract_inner_text(caption_content)
-                caption_item = self.add_text(
+                caption_item = doc.add_text(
                     label=DocItemLabel.CAPTION,
                     text=caption_text,
                     parent=None,
@@ -3567,7 +3569,7 @@ class DoclingDocument(BaseModel):
                 pg_width = 1
                 pg_height = 1
 
-            self.add_page(
+            doc.add_page(
                 page_no=page_no,
                 size=Size(width=pg_width, height=pg_height),
                 image=ImageRef.from_pil(image=image, dpi=72) if image else None,
@@ -3624,9 +3626,9 @@ class DoclingDocument(BaseModel):
                             charspan=(0, 0),
                             page_no=page_no,
                         )
-                        self.add_table(data=table_data, prov=prov, caption=caption)
+                        doc.add_table(data=table_data, prov=prov, caption=caption)
                     else:
-                        self.add_table(data=table_data, caption=caption)
+                        doc.add_table(data=table_data, caption=caption)
 
                 elif tag_name in [DocItemLabel.PICTURE, DocItemLabel.CHART]:
                     caption, caption_bbox = extract_caption(full_chunk)
@@ -3646,7 +3648,7 @@ class DoclingDocument(BaseModel):
                                 int(bbox.b * im_height),
                             )
                             cropped_image = image.crop(crop_box)
-                            pic = self.add_picture(
+                            pic = doc.add_picture(
                                 parent=None,
                                 image=ImageRef.from_pil(image=cropped_image, dpi=72),
                                 prov=(
@@ -3692,7 +3694,7 @@ class DoclingDocument(BaseModel):
                     else:
                         if bbox:
                             # In case we don't have access to an binary of an image
-                            pic = self.add_picture(
+                            pic = doc.add_picture(
                                 parent=None,
                                 prov=ProvenanceItem(
                                     bbox=bbox, charspan=(0, 0), page_no=page_no
@@ -3733,7 +3735,7 @@ class DoclingDocument(BaseModel):
                     key_value_data, kv_item_prov = parse_key_value_item(
                         full_chunk, image
                     )
-                    self.add_key_values(graph=key_value_data, prov=kv_item_prov)
+                    doc.add_key_values(graph=key_value_data, prov=kv_item_prov)
                 elif tag_name in [
                     DocumentToken.ORDERED_LIST.value,
                     DocumentToken.UNORDERED_LIST.value,
@@ -3749,7 +3751,7 @@ class DoclingDocument(BaseModel):
                     )
                     li_pattern = re.compile(list_item_pattern, re.DOTALL)
                     # Add list group:
-                    new_list = self.add_group(label=list_label, name="list")
+                    new_list = doc.add_group(label=list_label, name="list")
                     # Pricess list items
                     for li_match in li_pattern.finditer(full_chunk):
                         enum_value += 1
@@ -3760,7 +3762,7 @@ class DoclingDocument(BaseModel):
                         li_bbox = extract_bounding_box(li_full_chunk) if image else None
                         text_content = extract_inner_text(li_full_chunk)
                         # Add list item
-                        self.add_list_item(
+                        doc.add_list_item(
                             marker=enum_marker,
                             enumerated=(tag_name == DocumentToken.ORDERED_LIST.value),
                             parent=new_list,
@@ -3792,13 +3794,13 @@ class DoclingDocument(BaseModel):
                     if tag_name in [DocItemLabel.PAGE_HEADER, DocItemLabel.PAGE_FOOTER]:
                         content_layer = ContentLayer.FURNITURE
 
-                    self.add_text(
+                    doc.add_text(
                         label=doc_label,
                         text=text_content,
                         prov=element_prov,
                         content_layer=content_layer,
                     )
-        return self
+        return doc
 
     @deprecated("Use save_as_doctags instead.")
     def save_as_document_tokens(self, *args, **kwargs):
@@ -3885,7 +3887,7 @@ class DoclingDocument(BaseModel):
         :returns: The content of the document formatted as a DocTags string.
         :rtype: str
         """
-        from docling_core.experimental.serializer.doctags import (
+        from docling_core.transforms.serializer.doctags import (
             DocTagsDocSerializer,
             DocTagsParams,
         )

{docling_core-2.28.1 → docling_core-2.29.0}/docling_core/types/doc/page.py

@@ -579,13 +579,17 @@ class SegmentedPdfPage(SegmentedPage):
         with open(filename, "r", encoding="utf-8") as f:
             return cls.model_validate_json(f.read())
 
-    def crop_text(self, cell_unit: TextCellUnit, bbox: BoundingBox, eps: float = 1.0):
+    def crop_text(
+        self, cell_unit: TextCellUnit, bbox: BoundingBox, eps: float = 1.0
+    ) -> str:
         """Extract text from cells within the specified bounding box.
 
         Args:
             cell_unit: Type of text unit to extract
             bbox: Bounding box to extract from
             eps: Epsilon value for position comparison
+        Returns:
+            Extracted text from the cells
         """
         selection = []
         for page_cell in self.iterate_cells(cell_unit):
@@ -605,7 +609,6 @@ class SegmentedPdfPage(SegmentedPage):
 
         text = ""
         for i, cell in enumerate(selection):
-
             if i == 0:
                 text += cell.text
             else:
@@ -619,6 +622,7 @@ class SegmentedPdfPage(SegmentedPage):
                 else:
                     text += " "
                     text += cell.text
+        return text
 
     def export_to_textlines(
         self,
@@ -640,7 +644,6 @@ class SegmentedPdfPage(SegmentedPage):
         """
         lines: List[str] = []
         for cell in self.iterate_cells(cell_unit):
-
             line = ""
             if add_location:
                 line += f"({cell.rect.r_x0:06.02f}, {cell.rect.r_y0:06.02f}) "
@@ -1104,7 +1107,6 @@ class SegmentedPdfPage(SegmentedPage):
 
         # Draw each rectangle by connecting its four points
         for line in self.lines:
-
             line.to_top_left_origin(page_height=page_height)
             for segment in line.iterate_segments():
                 draw.line(

{docling_core-2.28.1 → docling_core-2.29.0}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "docling-core"
-version = "2.28.1"
+version = "2.29.0"
 description = "A python library to define and validate data types in Docling."
 license = "MIT"
 authors = [