docling-core 2.44.0__tar.gz → 2.44.2__tar.gz

{docling_core-2.44.0 → docling_core-2.44.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: docling-core
-Version: 2.44.0
+Version: 2.44.2
 Summary: A python library to define and validate data types in Docling.
 Author-email: Cesar Berrospi Ramis <ceb@zurich.ibm.com>, Panos Vagenas <pva@zurich.ibm.com>, Michele Dolfi <dol@zurich.ibm.com>, Christoph Auer <cau@zurich.ibm.com>, Peter Staar <taa@zurich.ibm.com>
 Maintainer-email: Panos Vagenas <pva@zurich.ibm.com>, Michele Dolfi <dol@zurich.ibm.com>, Christoph Auer <cau@zurich.ibm.com>, Peter Staar <taa@zurich.ibm.com>, Cesar Berrospi Ramis <ceb@zurich.ibm.com>

{docling_core-2.44.0 → docling_core-2.44.2}/docling_core/transforms/serializer/html.py

@@ -130,11 +130,14 @@ class HTMLTextSerializer(BaseModel, BaseTextSerializer):
         doc_serializer: BaseDocSerializer,
         doc: DoclingDocument,
         is_inline_scope: bool = False,
+        visited: Optional[set[str]] = None,
         **kwargs: Any,
     ) -> SerializationResult:
         """Serializes the passed text item to HTML."""
         params = HTMLParams(**kwargs)
+        my_visited: set[str] = visited if visited is not None else set()
         res_parts: list[SerializationResult] = []
+        post_processed = False
 
         # Prepare the HTML based on item type
         if isinstance(item, TitleItem):
@@ -162,7 +165,28 @@ class HTMLTextSerializer(BaseModel, BaseTextSerializer):
 
         elif isinstance(item, ListItem):
             # List items are handled by list serializer
-            text_inner = self._prepare_content(item.text)
+            text_parts: list[str] = []
+            if item_text := self._prepare_content(item.text):
+                item_text = doc_serializer.post_process(
+                    text=item_text,
+                    formatting=item.formatting,
+                    hyperlink=item.hyperlink,
+                )
+                post_processed = True
+                text_parts.append(item_text)
+            nested_parts = [
+                r.text
+                for r in doc_serializer.get_parts(
+                    item=item,
+                    is_inline_scope=is_inline_scope,
+                    visited=my_visited,
+                    **kwargs,
+                )
+            ]
+            text_parts.extend(nested_parts)
+            text_inner = "\n".join(text_parts)
+            if nested_parts:
+                text_inner = f"\n{text_inner}\n"
             text = (
                 get_html_tag_with_text_direction(
                     html_tag="li",
@@ -185,11 +209,12 @@ class HTMLTextSerializer(BaseModel, BaseTextSerializer):
             text = get_html_tag_with_text_direction(html_tag="p", text=text_inner)
 
         # Apply formatting and hyperlinks
-        text = doc_serializer.post_process(
-            text=text,
-            formatting=item.formatting,
-            hyperlink=item.hyperlink,
-        )
+        if not post_processed:
+            text = doc_serializer.post_process(
+                text=text,
+                formatting=item.formatting,
+                hyperlink=item.hyperlink,
+            )
 
         if text:
             text_res = create_ser_result(text=text, span_source=item)
@@ -703,7 +728,6 @@ class HTMLListSerializer(BaseModel, BaseListSerializer):
     ) -> SerializationResult:
         """Serializes a list to HTML."""
         my_visited: set[str] = visited if visited is not None else set()
-        params = HTMLParams(**kwargs)
         # Get all child parts
         parts = doc_serializer.get_parts(
             item=item,
@@ -713,72 +737,8 @@ class HTMLListSerializer(BaseModel, BaseListSerializer):
             **kwargs,
         )
 
-        # Append nested list to parent list item:
-        i = 0
-        while i < len(parts):
-            prt = parts[i]
-            if prt.text.startswith(("<ul>", "<ol>")):
-                for j in range(i - 1, -1, -1):
-                    if parts[j].text.startswith(("<li>", "<li ")) and parts[
-                        j
-                    ].text.endswith("</li>"):
-                        before, _, _ = parts[j].text.rpartition("</li>")
-                        parts[j].text = f"{before}\n{prt.text}\n</li>"
-                        break
-                if j > -1:
-                    parts.pop(i)
-            else:
-                i += 1
-
         # Add all child parts
-        text_res = "\n".join(
-            [
-                (
-                    p.text
-                    if (
-                        (
-                            p.text.startswith(("<li>", "<li "))
-                            and p.text.endswith("</li>")
-                        )
-                        or (
-                            p.text.startswith(("<ol>", "<ol "))
-                            and p.text.endswith("</ol>")
-                        )
-                        or (
-                            p.text.startswith(("<ul>", "<ul "))
-                            and p.text.endswith("</ul>")
-                        )
-                    )
-                    else (
-                        get_html_tag_with_text_direction(
-                            html_tag="li",
-                            text=p.text,
-                            attrs=(
-                                {
-                                    "style": f"list-style-type: '{grandparent_item.marker} ';"
-                                }
-                                if params.show_original_list_item_marker
-                                and grandparent_item.marker
-                                else {}
-                            ),
-                        )
-                        if p.spans
-                        and p.spans[0].item.parent
-                        and isinstance(
-                            (parent_item := p.spans[0].item.parent.resolve(doc)),
-                            InlineGroup,
-                        )
-                        and parent_item.parent
-                        and isinstance(
-                            (grandparent_item := parent_item.parent.resolve(doc)),
-                            ListItem,
-                        )
-                        else f"<li>{p.text}</li>"
-                    )
-                )
-                for p in parts
-            ]
-        )
+        text_res = "\n".join(p.text for p in parts if p.text)
         if text_res:
             tag = "ol" if item.first_item_is_enumerated(doc) else "ul"
             text_res = f"<{tag}>\n{text_res}\n</{tag}>"

{docling_core-2.44.0 → docling_core-2.44.2}/docling_core/types/doc/document.py

@@ -1373,11 +1373,12 @@ class PictureItem(FloatingItem):
         )  # Encode to Base64 and decode to string
         return img_base64
 
-    def _image_to_hexhash(self) -> Optional[str]:
+    @staticmethod
+    def _image_to_hexhash(img: Optional[PILImage.Image]) -> Optional[str]:
         """Hexash from the image."""
-        if self.image is not None and self.image._pil is not None:
+        if img is not None:
             # Convert the image to raw bytes
-            image_bytes = self.image._pil.tobytes()
+            image_bytes = img.tobytes()
 
             # Create a hash object (e.g., SHA-256)
             hasher = hashlib.sha256(usedforsecurity=False)
@@ -4116,16 +4117,10 @@ class DoclingDocument(BaseModel):
         if image_dir.is_dir():
             for item, level in result.iterate_items(page_no=page_no, with_groups=False):
                 if isinstance(item, PictureItem):
+                    img = item.get_image(doc=self)
+                    if img is not None:
 
-                    if (
-                        item.image is not None
-                        and isinstance(item.image.uri, AnyUrl)
-                        and item.image.uri.scheme == "data"
-                        and item.image.pil_image is not None
-                    ):
-                        img = item.image.pil_image
-
-                        hexhash = item._image_to_hexhash()
+                        hexhash = PictureItem._image_to_hexhash(img)
 
                         # loc_path = image_dir / f"image_{img_count:06}.png"
                         if hexhash is not None:
@@ -4140,6 +4135,11 @@ class DoclingDocument(BaseModel):
                             else:
                                 obj_path = loc_path
 
+                            if item.image is None:
+                                scale = img.size[0] / item.prov[0].bbox.width
+                                item.image = ImageRef.from_pil(
+                                    image=img, dpi=round(72 * scale)
+                                )
                             item.image.uri = Path(obj_path)
 
                         # if item.image._pil is not None:
@@ -4539,6 +4539,8 @@ class DoclingDocument(BaseModel):
             reference_path = None
         else:
             reference_path = filename.parent
+            artifacts_dir = reference_path / artifacts_dir
+
         return artifacts_dir, reference_path
 
     def _make_copy_with_refmode(

{docling_core-2.44.0 → docling_core-2.44.2}/docling_core.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: docling-core
-Version: 2.44.0
+Version: 2.44.2
 Summary: A python library to define and validate data types in Docling.
 Author-email: Cesar Berrospi Ramis <ceb@zurich.ibm.com>, Panos Vagenas <pva@zurich.ibm.com>, Michele Dolfi <dol@zurich.ibm.com>, Christoph Auer <cau@zurich.ibm.com>, Peter Staar <taa@zurich.ibm.com>
 Maintainer-email: Panos Vagenas <pva@zurich.ibm.com>, Michele Dolfi <dol@zurich.ibm.com>, Christoph Auer <cau@zurich.ibm.com>, Peter Staar <taa@zurich.ibm.com>, Cesar Berrospi Ramis <ceb@zurich.ibm.com>

{docling_core-2.44.0 → docling_core-2.44.2}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "docling-core"
-version = "2.44.0"  # DO NOT EDIT, updated automatically
+version = "2.44.2"  # DO NOT EDIT, updated automatically
 description = "A python library to define and validate data types in Docling."
 license = "MIT"
 license-files = ["LICENSE"]

{docling_core-2.44.0 → docling_core-2.44.2}/test/test_docling_doc.py

@@ -1442,10 +1442,11 @@ def test_save_to_disk():
 
     doc: DoclingDocument = _construct_doc()
 
-    image_dir = Path("./test/data/doc/constructed_images/")
+    test_dir = Path("./test/data/doc")
+    image_dir = Path("constructed_images/")  # will be relative to test_dir
 
     doc_with_references = doc._with_pictures_refs(
-        image_dir=image_dir,  # Path("./test/data/constructed_images/")
+        image_dir=(test_dir / image_dir),
         page_no=None,
     )
 
@@ -1455,19 +1456,19 @@ def test_save_to_disk():
 
     ### MarkDown
 
-    filename = Path("test/data/doc/constructed_doc.placeholder.md")
+    filename = test_dir / "constructed_doc.placeholder.md"
     doc.save_as_markdown(
         filename=filename, artifacts_dir=image_dir, image_mode=ImageRefMode.PLACEHOLDER
     )
     _verify_saved_output(filename=filename, paths=paths)
 
-    filename = Path("test/data/doc/constructed_doc.embedded.md")
+    filename = test_dir / "constructed_doc.embedded.md"
     doc.save_as_markdown(
         filename=filename, artifacts_dir=image_dir, image_mode=ImageRefMode.EMBEDDED
     )
     _verify_saved_output(filename=filename, paths=paths)
 
-    filename = Path("test/data/doc/constructed_doc.referenced.md")
+    filename = test_dir / "constructed_doc.referenced.md"
     doc.save_as_markdown(
         filename=filename, artifacts_dir=image_dir, image_mode=ImageRefMode.REFERENCED
     )
@@ -1475,19 +1476,19 @@ def test_save_to_disk():
 
     ### HTML
 
-    filename = Path("test/data/doc/constructed_doc.placeholder.html")
+    filename = test_dir / "constructed_doc.placeholder.html"
     doc.save_as_html(
         filename=filename, artifacts_dir=image_dir, image_mode=ImageRefMode.PLACEHOLDER
     )
     _verify_saved_output(filename=filename, paths=paths)
 
-    filename = Path("test/data/doc/constructed_doc.embedded.html")
+    filename = test_dir / "constructed_doc.embedded.html"
     doc.save_as_html(
         filename=filename, artifacts_dir=image_dir, image_mode=ImageRefMode.EMBEDDED
     )
     _verify_saved_output(filename=filename, paths=paths)
 
-    filename = Path("test/data/doc/constructed_doc.referenced.html")
+    filename = test_dir / "constructed_doc.referenced.html"
     doc.save_as_html(
         filename=filename, artifacts_dir=image_dir, image_mode=ImageRefMode.REFERENCED
     )
@@ -1495,13 +1496,13 @@ def test_save_to_disk():
 
     ### Document Tokens
 
-    filename = Path("test/data/doc/constructed_doc.dt")
+    filename = test_dir / "constructed_doc.dt"
     doc.save_as_doctags(filename=filename)
     _verify_saved_output(filename=filename, paths=paths)
 
     ### JSON
 
-    filename = Path("test/data/doc/constructed_doc.embedded.json")
+    filename = test_dir / "constructed_doc.embedded.json"
     doc.save_as_json(
         filename=filename,
         artifacts_dir=image_dir,
@@ -1512,7 +1513,7 @@ def test_save_to_disk():
     doc_emb_loaded = DoclingDocument.load_from_json(filename)
     _verify_loaded_output(filename=filename, pred=doc_emb_loaded)
 
-    filename = Path("test/data/doc/constructed_doc.referenced.json")
+    filename = test_dir / "constructed_doc.referenced.json"
     doc.save_as_json(
         filename=filename,
         artifacts_dir=image_dir,
@@ -1525,7 +1526,7 @@ def test_save_to_disk():
 
     ### YAML
 
-    filename = Path("test/data/doc/constructed_doc.embedded.yaml")
+    filename = test_dir / "constructed_doc.embedded.yaml"
     doc.save_as_yaml(
         filename=filename,
         artifacts_dir=image_dir,
@@ -1533,7 +1534,7 @@ def test_save_to_disk():
     )
     _verify_saved_output(filename=filename, paths=paths)
 
-    filename = Path("test/data/doc/constructed_doc.referenced.yaml")
+    filename = test_dir / "constructed_doc.referenced.yaml"
     doc.save_as_yaml(
         filename=filename,
         artifacts_dir=image_dir,

{docling_core-2.44.0 → docling_core-2.44.2}/test/test_serialization.py

@@ -352,6 +352,15 @@ def test_md_mark_annotations_true():
     )
 
 
+def test_md_nested_lists():
+    src = Path("./test/data/doc/polymers.json")
+    doc = DoclingDocument.load_from_json(src)
+
+    ser = MarkdownDocSerializer(doc=doc)
+    actual = ser.serialize().text
+    verify(exp_file=src.parent / f"{src.stem}.gt.md", actual=actual)
+
+
 def test_html_split_page():
     src = Path("./test/data/doc/2408.09869v3_enriched.json")
     doc = DoclingDocument.load_from_json(src)
@@ -482,6 +491,15 @@ def test_html_list_item_markers():
         )
 
 
+def test_html_nested_lists():
+    src = Path("./test/data/doc/polymers.json")
+    doc = DoclingDocument.load_from_json(src)
+
+    ser = HTMLDocSerializer(doc=doc)
+    actual = ser.serialize().text
+    verify(exp_file=src.parent / f"{src.stem}.gt.html", actual=actual)
+
+
 def test_doctags_inline_loc_tags():
     src = Path("./test/data/doc/2408.09869v3_enriched.json")
     doc = DoclingDocument.load_from_json(src)