PyPI - docling-core - Versions diffs - 2.30.0__py3-none-any.whl → 2.31.0__py3-none-any.whl - Mend

docling-core 2.30.0py3-none-any.whl → 2.31.0py3-none-any.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.

Potentially problematic release.

This version of docling-core might be problematic. Click here for more details.

Files changed (8) hide show

docling_core/transforms/serializer/common.py CHANGED Viewed

@@ -169,7 +169,7 @@ class CommonParams(BaseModel):
     def merge_with_patch(self, patch: dict[str, Any]) -> Self:
         """Create an instance by merging the provided patch dict on top of self."""
-        res = self.model_validate({**self.model_dump(), **patch})
+        res = self.model_copy(update=patch)
         return res
@@ -260,10 +260,10 @@ class DocSerializer(BaseModel, BaseDocSerializer):
         """Serialize a document out of its pages."""
         ...
-    def _serialize_body(self) -> SerializationResult:
+    def _serialize_body(self, **kwargs) -> SerializationResult:
         """Serialize the document body."""
         subparts = self.get_parts()
-        res = self.serialize_doc(parts=subparts)
+        res = self.serialize_doc(parts=subparts, **kwargs)
         return res
     @override
@@ -278,12 +278,12 @@ class DocSerializer(BaseModel, BaseDocSerializer):
     ) -> SerializationResult:
         """Serialize a given node."""
         my_visited: set[str] = visited if visited is not None else set()
-        my_kwargs = self.params.merge_with_patch(patch=kwargs).model_dump()
+        my_kwargs = {**self.params.model_dump(), **kwargs}
         empty_res = create_ser_result()
         if item is None or item == self.doc.body:
             if self.doc.body.self_ref not in my_visited:
                 my_visited.add(self.doc.body.self_ref)
-                return self._serialize_body()
+                return self._serialize_body(**my_kwargs)
             else:
                 return empty_res

docling_core/transforms/serializer/html.py CHANGED Viewed

@@ -16,6 +16,7 @@ from xml.etree.cElementTree import SubElement, tostring
 from xml.sax.saxutils import unescape
 import latex2mathml.converter
+from PIL.Image import Image
 from pydantic import AnyUrl, BaseModel
 from typing_extensions import override
@@ -40,6 +41,7 @@ from docling_core.transforms.serializer.html_styles import (
     _get_css_for_single_column,
     _get_css_for_split_page,
 )
+from docling_core.transforms.visualizer.base import BaseVisualizer
 from docling_core.types.doc.base import ImageRefMode
 from docling_core.types.doc.document import (
     CodeItem,
@@ -821,9 +823,22 @@ class HTMLDocSerializer(DocSerializer):
     def serialize_doc(
         self,
         parts: list[SerializationResult],
+        visualizer: Optional[BaseVisualizer] = None,
         **kwargs: Any,
     ) -> SerializationResult:
         """Serialize a document out of its pages."""
+        def _serialize_page_img(page_img: Image):
+            buffered = BytesIO()
+            page_img.save(buffered, format="PNG")  # Save the image to the byte stream
+            img_bytes = buffered.getvalue()  # Get the byte data
+            # Encode to Base64 and decode to string
+            img_base64 = base64.b64encode(img_bytes).decode("utf-8")
+            img_text = f'<img src="data:image/png;base64,{img_base64}">'
+            return f"<figure>{img_text}</figure>"
         # Create HTML structure
         html_parts = [
             "<!DOCTYPE html>",
@@ -853,19 +868,26 @@ class HTMLDocSerializer(DocSerializer):
             html_parts.append("<table>")
             html_parts.append("<tbody>")
+            vized_pages_dict: dict[Optional[int], Image] = {}
+            if visualizer:
+                vized_pages_dict = visualizer.get_visualization(doc=self.doc)
             for page_no, page in pages.items():
                 if isinstance(page_no, int):
                     if applicable_pages is not None and page_no not in applicable_pages:
                         continue
                     page_img = self.doc.pages[page_no].image
+                    vized_page = vized_pages_dict.get(page_no)
                     html_parts.append("<tr>")
                     html_parts.append("<td>")
+                    if vized_page:
+                        html_parts.append(_serialize_page_img(page_img=vized_page))
                     # short-cut: we already have the image in base64
-                    if (
+                    elif (
                         (page_img is not None)
                         and isinstance(page_img, ImageRef)
                         and isinstance(page_img.uri, AnyUrl)
@@ -875,18 +897,7 @@ class HTMLDocSerializer(DocSerializer):
                         html_parts.append(f"<figure>{img_text}</figure>")
                     elif (page_img is not None) and (page_img._pil is not None):
-                        buffered = BytesIO()
-                        page_img._pil.save(
-                            buffered, format="PNG"
-                        )  # Save the image to the byte stream
-                        img_bytes = buffered.getvalue()  # Get the byte data
-                        # Encode to Base64 and decode to string
-                        img_base64 = base64.b64encode(img_bytes).decode("utf-8")
-                        img_text = f'<img src="data:image/png;base64,{img_base64}">'
-                        html_parts.append(f"<figure>{img_text}</figure>")
+                        html_parts.append(_serialize_page_img(page_img=page_img._pil))
                     else:
                         html_parts.append("<figure>no page-image found</figure>")

docling_core/types/doc/labels.py CHANGED Viewed

@@ -25,6 +25,8 @@ class DocItemLabel(str, Enum):
     CHECKBOX_UNSELECTED = "checkbox_unselected"
     FORM = "form"
     KEY_VALUE_REGION = "key_value_region"
+    GRADING_SCALE = "grading_scale"  # for elements in forms, questionaires representing a grading scale
+    # e.g. [strongly disagree | ... | ... | strongly agree]
     # Additional labels for markup-based formats (e.g. HTML, Word)
     PARAGRAPH = "paragraph"
@@ -144,17 +146,43 @@ class TableCellLabel(str, Enum):
         """Get string value."""
         return str(self.value)
+    @staticmethod
+    def get_color(label: "TableCellLabel") -> Tuple[int, int, int]:
+        """Return the RGB color associated with a given label."""
+        color_map = {
+            TableCellLabel.COLUMN_HEADER: (255, 0, 0),
+            TableCellLabel.ROW_HEADER: (0, 255, 0),
+            TableCellLabel.ROW_SECTION: (0, 0, 255),
+            TableCellLabel.BODY: (0, 255, 255),
+        }
+        return color_map.get(label, (0, 0, 0))
 class GraphCellLabel(str, Enum):
     """GraphCellLabel."""
     UNSPECIFIED = "unspecified"
-    KEY = "key"
-    VALUE = "value"
+    KEY = "key"  # used to designate a key (label) of a key-value element
+    VALUE = "value"  # Data value with or without explicit Key, but filled in,
+    # e.g. telephone number, address, quantity, name, date
+    EMPTY_VALUE = "empty_value"  # used for empty value fields in fillable forms
     CHECKBOX = "checkbox"
+    def __str__(self):
+        """Get string value."""
+        return str(self.value)
+    @staticmethod
+    def get_color(label: "GraphCellLabel") -> Tuple[int, int, int]:
+        """Return the RGB color associated with a given label."""
+        color_map = {
+            GraphCellLabel.KEY: (255, 0, 0),
+            GraphCellLabel.VALUE: (0, 255, 0),
+            GraphCellLabel.EMPTY_VALUE: (0, 0, 255),
+        }
+        return color_map.get(label, (0, 0, 0))
 class GraphLinkLabel(str, Enum):
     """GraphLinkLabel."""

{docling_core-2.30.0.dist-info → docling_core-2.31.0.dist-info}/METADATA RENAMED Viewed

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: docling-core
-Version: 2.30.0
+Version: 2.31.0
 Summary: A python library to define and validate data types in Docling.
 Home-page: https://github.com/docling-project
 License: MIT

{docling_core-2.30.0.dist-info → docling_core-2.31.0.dist-info}/RECORD RENAMED Viewed

@@ -27,9 +27,9 @@ docling_core/transforms/chunker/tokenizer/huggingface.py,sha256=JQ-D3b5vTPQbvu4H
 docling_core/transforms/chunker/tokenizer/openai.py,sha256=zt2kwcC-r8MafeEG0CESab8E4RIC9aaFXxxnxOGyTMA,918
 docling_core/transforms/serializer/__init__.py,sha256=CECQlMoCDUxkg4RAUdC3itA3I3qFhKhe2HcYghN6_xw,105
 docling_core/transforms/serializer/base.py,sha256=9bgpWA0oMmZNRc3yIuZVnu5bJ1glClBsswtVF1vYwMI,6046
-docling_core/transforms/serializer/common.py,sha256=xBwhsgDZbNWMpp6ExpyUWO8_NvHPfbPF1ak2z9API5M,17435
+docling_core/transforms/serializer/common.py,sha256=TC1EwHIp9PYcI8jeTKeavUAPtounYmS0V1bfS_wDKm0,17427
 docling_core/transforms/serializer/doctags.py,sha256=mEmRWVuebcG5pZcR1_HX146cyUk0_FjaLQtMXSgh9hs,17870
-docling_core/transforms/serializer/html.py,sha256=2zlV8B-xsXHfTM13sb2pRwtKaxLcNUlW0p0NCgbiatk,34088
+docling_core/transforms/serializer/html.py,sha256=Xq9CU5qZTDdwstizYqWNL_TFNDs9NHK_6JvvZk0TP98,34571
 docling_core/transforms/serializer/html_styles.py,sha256=-jBwS4EU7yfKoz0GSoxhwx90OmIKieO6TwPw57IuxcA,4692
 docling_core/transforms/serializer/markdown.py,sha256=YqThAYMsOWSg6nZnnmrUHZohn0QvfZzRqpLrB-Keev8,17873
 docling_core/transforms/visualizer/__init__.py,sha256=gUfF25yiJ_KO46ZIUNqZQOZGy2PLx6gnnr6AZYxKHXI,35
@@ -41,7 +41,7 @@ docling_core/types/base.py,sha256=PusJskRVL19y-hq0BgXr5e8--QEqSqLnFNJ8UbOqW88,83
 docling_core/types/doc/__init__.py,sha256=bysJn2iwjAHwThSWDPXEdVUUij7p_ax12_nx2_0CMdg,653
 docling_core/types/doc/base.py,sha256=sM3IyFXzVh2WT8IGh5nejXYh8sf39yBh8TBSlHeJ9CI,12611
 docling_core/types/doc/document.py,sha256=eboNYL-QVnDNnw3vL7PPVdPosfs5oNfsrVofxmdBDHw,140884
-docling_core/types/doc/labels.py,sha256=xyrQC89OfedBaBvulXtgIGgN3Sv0D4J5k0qSXnGw2hk,5846
+docling_core/types/doc/labels.py,sha256=vp4h3e7AmBvezRmgrfuPehjAHTZOufphErLB4ENhdME,7171
 docling_core/types/doc/page.py,sha256=1JMPwglaTITBvg959L_pcWPb-fXoDYGh-e_tGZMzVMQ,41060
 docling_core/types/doc/tokens.py,sha256=z22l9J81_sg9CYMvOuLmPuLsNT7h_s7wao2UT89DvI8,9278
 docling_core/types/doc/utils.py,sha256=SaiQD-WMMooFm1bMqwatU-IGhtG048iKJb-ppnJit_k,2250
@@ -73,8 +73,8 @@ docling_core/utils/generate_jsonschema.py,sha256=uNX1O5XnjyB5nA66XqZXTt3YbGuR2ty
 docling_core/utils/legacy.py,sha256=DrI3QGoL755ZCIoKHF74-pTWm8R0zfFo2C2vB5dT2aY,24463
 docling_core/utils/validate.py,sha256=aQ11UbFyl8iD_N7yTTZmm_VVeXz8KcCyn3GLXgkfYRM,2049
 docling_core/utils/validators.py,sha256=azcrndLzhNkTWnbFSu9shJ5D3j_znnLrIFA5R8hzmGU,2798
-docling_core-2.30.0.dist-info/LICENSE,sha256=2M9-6EoQ1sxFztTOkXGAtwUDJvnWaAHdB9BYWVwGkIw,1087
-docling_core-2.30.0.dist-info/METADATA,sha256=mzObLyfGYJ-ZLvh5X4plERiaEI_58l7fXs7zp0uqn_M,5976
-docling_core-2.30.0.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
-docling_core-2.30.0.dist-info/entry_points.txt,sha256=oClcdb2L2RKx4jdqUykY16Kum_f0_whwWhGzIodyidc,216
-docling_core-2.30.0.dist-info/RECORD,,
+docling_core-2.31.0.dist-info/LICENSE,sha256=2M9-6EoQ1sxFztTOkXGAtwUDJvnWaAHdB9BYWVwGkIw,1087
+docling_core-2.31.0.dist-info/METADATA,sha256=OohNxPwKcbRVVKm_kpa3HRFhYh9ZMVyBIlf3apF9hm4,5976
+docling_core-2.31.0.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
+docling_core-2.31.0.dist-info/entry_points.txt,sha256=oClcdb2L2RKx4jdqUykY16Kum_f0_whwWhGzIodyidc,216
+docling_core-2.31.0.dist-info/RECORD,,

{docling_core-2.30.0.dist-info → docling_core-2.31.0.dist-info}/LICENSE RENAMED Viewed

File without changes

{docling_core-2.30.0.dist-info → docling_core-2.31.0.dist-info}/WHEEL RENAMED Viewed

File without changes

{docling_core-2.30.0.dist-info → docling_core-2.31.0.dist-info}/entry_points.txt RENAMED Viewed

File without changes

docling-core 2.30.0__py3-none-any.whl → 2.31.0__py3-none-any.whl

Potentially problematic release.

docling-core 2.30.0py3-none-any.whl → 2.31.0py3-none-any.whl