docling-core 2.29.0__py3-none-any.whl → 2.30.0__py3-none-any.whl

docling_core/transforms/serializer/html.py

@@ -370,6 +370,13 @@ class HTMLPictureSerializer(BasePictureSerializer):
         **kwargs: Any,
     ) -> SerializationResult:
         """Export picture to HTML format."""
+
+        def get_img_row(imgb64: str, ind: int) -> str:
+            row = '<tr><td style="border: 2px solid black; padding: 8px;">'
+            row += f'<img src="data:image/png;base64,{imgb64}" alt="image {ind}">'
+            row += "</td></tr>\n"
+            return row
+
         params = HTMLParams(**kwargs)
 
         res_parts: list[SerializationResult] = []
@@ -393,6 +400,22 @@ class HTMLPictureSerializer(BasePictureSerializer):
                     and item.image.uri.scheme == "data"
                 ):
                     img_text = f'<img src="{item.image.uri}">'
+                elif len(item.prov) > 1:  # more than 1 provenance
+
+                    img_text = (
+                        '<table style="border-collapse: collapse; width: 100%;">\n'
+                    )
+                    for ind, prov in enumerate(item.prov):
+                        img = item.get_image(doc, prov_index=ind)
+
+                        if img is not None:
+                            imgb64 = item._image_to_base64(img)
+                            img_text += get_img_row(imgb64=imgb64, ind=ind)
+                        else:
+                            _logger.warning("Could not get image")
+
+                    img_text += "</table>\n"
+
                 else:
                     # get the item.image._pil or crop it out of the page-image
                     img = item.get_image(doc)
@@ -400,6 +423,9 @@ class HTMLPictureSerializer(BasePictureSerializer):
                     if img is not None:
                         imgb64 = item._image_to_base64(img)
                         img_text = f'<img src="data:image/png;base64,{imgb64}">'
+                    else:
+                        _logger.warning("Could not get image")
+
             elif params.image_mode == ImageRefMode.REFERENCED:
                 if isinstance(item.image, ImageRef) and not (
                     isinstance(item.image.uri, AnyUrl)

docling_core/types/doc/document.py

@@ -790,7 +790,9 @@ class DocItem(
 
         return location
 
-    def get_image(self, doc: "DoclingDocument") -> Optional[PILImage.Image]:
+    def get_image(
+        self, doc: "DoclingDocument", prov_index: int = 0
+    ) -> Optional[PILImage.Image]:
         """Returns the image of this DocItem.
 
         The function returns None if this DocItem has no valid provenance or
@@ -800,7 +802,7 @@ class DocItem(
         if not len(self.prov):
             return None
 
-        page = doc.pages.get(self.prov[0].page_no)
+        page = doc.pages.get(self.prov[prov_index].page_no)
         if page is None or page.size is None or page.image is None:
             return None
 
@@ -808,7 +810,7 @@ class DocItem(
         if not page_image:
             return None
         crop_bbox = (
-            self.prov[0]
+            self.prov[prov_index]
             .bbox.to_top_left_origin(page_height=page.size.height)
             .scale_to_size(old_size=page.size, new_size=page.image.size)
             # .scaled(scale=page_image.height / page.size.height)
@@ -973,7 +975,9 @@ class FloatingItem(DocItem):
             text += cap.resolve(doc).text
         return text
 
-    def get_image(self, doc: "DoclingDocument") -> Optional[PILImage.Image]:
+    def get_image(
+        self, doc: "DoclingDocument", prov_index: int = 0
+    ) -> Optional[PILImage.Image]:
         """Returns the image corresponding to this FloatingItem.
 
         This function returns the PIL image from self.image if one is available.
@@ -985,7 +989,7 @@ class FloatingItem(DocItem):
         """
         if self.image is not None:
             return self.image.pil_image
-        return super().get_image(doc=doc)
+        return super().get_image(doc=doc, prov_index=prov_index)
 
 
 class CodeItem(FloatingItem, TextItem):
@@ -1073,7 +1077,7 @@ class PictureItem(FloatingItem):
             image_bytes = self.image._pil.tobytes()
 
             # Create a hash object (e.g., SHA-256)
-            hasher = hashlib.sha256()
+            hasher = hashlib.sha256(usedforsecurity=False)
 
             # Feed the image bytes into the hash object
             hasher.update(image_bytes)
@@ -2657,16 +2661,25 @@ class DoclingDocument(BaseModel):
         if should_yield:
             yield root, my_stack
 
-        # Handle picture traversal - only traverse children if requested
-        if isinstance(root, PictureItem) and not traverse_pictures:
-            return
-
         my_stack.append(-1)
 
+        allowed_pic_refs: set[str] = (
+            {r.cref for r in root.captions}
+            if (root_is_picture := isinstance(root, PictureItem))
+            else set()
+        )
+
         # Traverse children
         for child_ind, child_ref in enumerate(root.children):
-            my_stack[-1] = child_ind
             child = child_ref.resolve(self)
+            if (
+                root_is_picture
+                and not traverse_pictures
+                and isinstance(child, DocItem)
+                and child.self_ref not in allowed_pic_refs
+            ):
+                continue
+            my_stack[-1] = child_ind
 
             if isinstance(child, NodeItem):
                 yield from self._iterate_items_with_stack(
@@ -3597,7 +3610,9 @@ class DoclingDocument(BaseModel):
                 rf"{DocumentToken.UNORDERED_LIST.value}|"
                 rf"{DocItemLabel.KEY_VALUE_REGION}|"
                 rf"{DocumentToken.CHART.value}|"
-                rf"{DocumentToken.OTSL.value})>.*?</(?P=tag)>"
+                rf"{DocumentToken.OTSL.value})>"
+                rf"(?P<content>.*?)"
+                rf"(?:(?P<closed></(?P=tag)>)|(?P<eof>$))"
             )
             pattern = re.compile(tag_pattern, re.DOTALL)
 
@@ -3607,6 +3622,10 @@ class DoclingDocument(BaseModel):
                 tag_name = match.group("tag")
 
                 bbox = extract_bounding_box(full_chunk)  # Extracts first bbox
+                if not match.group("closed"):
+                    # no closing tag; only the existence of the item is recovered
+                    full_chunk = f"<{tag_name}></{tag_name}>"
+
                 doc_label = tag_to_doclabel.get(tag_name, DocItemLabel.PARAGRAPH)
 
                 if tag_name == DocumentToken.OTSL.value:

docling_core/types/doc/labels.py

@@ -78,6 +78,7 @@ class GroupLabel(str, Enum):
     KEY_VALUE_AREA = "key_value_area"
     COMMENT_SECTION = "comment_section"
     INLINE = "inline"
+    PICTURE_AREA = "picture_area"
 
     def __str__(self):
         """Get string value."""

docling_core/types/doc/page.py

@@ -472,8 +472,27 @@ class SegmentedPage(BaseModel):
     word_cells: List[TextCell] = []
     textline_cells: List[TextCell] = []
 
+    # These flags are set to differentiate if above lists of this SegmentedPage
+    # are empty (page had no content) or if they have not been computed (i.e. textline_cells may be present
+    # but word_cells are not)
+    has_chars: bool = False
+    has_words: bool = False
+    has_lines: bool = False
+
     image: Optional[ImageRef] = None
 
+    @model_validator(mode="after")
+    def validate_page(self) -> "SegmentedPage":
+        """Validate page."""
+        if len(self.textline_cells) > 0:
+            self.has_lines = True
+        if len(self.word_cells) > 0:
+            self.has_words = True
+        if len(self.char_cells) > 0:
+            self.has_chars = True
+
+        return self
+
     def iterate_cells(self, unit_type: TextCellUnit) -> Iterator[TextCell]:
         """Iterate through text cells of the specified unit type.
 

docling_core/utils/legacy.py

@@ -47,7 +47,7 @@ from docling_core.types.legacy_doc.document import ExportedCCSDocument as DsDocu
 
 
 def _create_hash(string: str):
-    hasher = hashlib.sha256()
+    hasher = hashlib.sha256(usedforsecurity=False)
     hasher.update(string.encode("utf-8"))
 
     return hasher.hexdigest()

{docling_core-2.29.0.dist-info → docling_core-2.30.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: docling-core
-Version: 2.29.0
+Version: 2.30.0
 Summary: A python library to define and validate data types in Docling.
 Home-page: https://github.com/docling-project
 License: MIT

{docling_core-2.29.0.dist-info → docling_core-2.30.0.dist-info}/RECORD

@@ -29,7 +29,7 @@ docling_core/transforms/serializer/__init__.py,sha256=CECQlMoCDUxkg4RAUdC3itA3I3
 docling_core/transforms/serializer/base.py,sha256=9bgpWA0oMmZNRc3yIuZVnu5bJ1glClBsswtVF1vYwMI,6046
 docling_core/transforms/serializer/common.py,sha256=xBwhsgDZbNWMpp6ExpyUWO8_NvHPfbPF1ak2z9API5M,17435
 docling_core/transforms/serializer/doctags.py,sha256=mEmRWVuebcG5pZcR1_HX146cyUk0_FjaLQtMXSgh9hs,17870
-docling_core/transforms/serializer/html.py,sha256=r37zA88ca0GAo1-I-il6ZMQA7OsT0gsjAYsqQFfH9o4,33064
+docling_core/transforms/serializer/html.py,sha256=2zlV8B-xsXHfTM13sb2pRwtKaxLcNUlW0p0NCgbiatk,34088
 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
@@ -40,9 +40,9 @@ docling_core/types/__init__.py,sha256=MVRSgsk5focwGyAplh_TRR3dEecIXpd98g_u3zZ5HX
 docling_core/types/base.py,sha256=PusJskRVL19y-hq0BgXr5e8--QEqSqLnFNJ8UbOqW88,8318
 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=wCUKOGngiqITddTinlU-qGys3QdfGn29eoRUEeKrHOQ,140220
-docling_core/types/doc/labels.py,sha256=3QgteZZ4jKi0fideTuTnuriviJBwew-5RKE4pse7Ppk,5812
-docling_core/types/doc/page.py,sha256=MvQnrCLDQDd9STawMFNXVCI4oYKYcQ3ECXlAUBbSz9A,40406
+docling_core/types/doc/document.py,sha256=eboNYL-QVnDNnw3vL7PPVdPosfs5oNfsrVofxmdBDHw,140884
+docling_core/types/doc/labels.py,sha256=xyrQC89OfedBaBvulXtgIGgN3Sv0D4J5k0qSXnGw2hk,5846
+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
 docling_core/types/gen/__init__.py,sha256=C6TuCfvpSnSL5XDOFMcYHUY2-i08vvfOGRcdu6Af0pI,124
@@ -70,11 +70,11 @@ docling_core/utils/alias.py,sha256=B6Lqvss8CbaNARHLR4qSmNh9OkB6LvqTpxfsFmkLAFo,8
 docling_core/utils/file.py,sha256=GzX0pclvewwPoqHJSaVUuULzSJwJgkCUwgKgJ7G5ohQ,5628
 docling_core/utils/generate_docs.py,sha256=BdKAoduWXOc7YMvcmlhjoJOFlUxij1ybxglj6LZDtC8,2290
 docling_core/utils/generate_jsonschema.py,sha256=uNX1O5XnjyB5nA66XqZXTt3YbGuR2tyi_OhHepHYtZg,1654
-docling_core/utils/legacy.py,sha256=SqNQAxl97aHfoJEsC9vZcMJg5FNkmqKPFi-wdSrnfI0,24442
+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-2.29.0.dist-info/LICENSE,sha256=2M9-6EoQ1sxFztTOkXGAtwUDJvnWaAHdB9BYWVwGkIw,1087
-docling_core-2.29.0.dist-info/METADATA,sha256=Lkbmi6367j_0fHabpIhLMgyv9qXBEI4fONRdMBvDEm4,5976
-docling_core-2.29.0.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
-docling_core-2.29.0.dist-info/entry_points.txt,sha256=oClcdb2L2RKx4jdqUykY16Kum_f0_whwWhGzIodyidc,216
-docling_core-2.29.0.dist-info/RECORD,,
+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,,