docling-core 2.23.1__tar.gz → 2.23.3__tar.gz

{docling_core-2.23.1 → docling_core-2.23.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: docling-core
-Version: 2.23.1
+Version: 2.23.3
 Summary: A python library to define and validate data types in Docling.
 Home-page: https://github.com/docling-project
 License: MIT

{docling_core-2.23.1 → docling_core-2.23.3}/docling_core/experimental/serializer/markdown.py

@@ -310,9 +310,17 @@ class MarkdownListSerializer(BaseModel, BaseListSerializer):
             is_inline_scope=is_inline_scope,
             visited=my_visited,
         )
+        sep = "\n"
+        my_parts: list[SerializationResult] = []
+        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
+            else:
+                my_parts.append(p)
+
         indent_str = list_level * self.indent * " "
         is_ol = isinstance(item, OrderedList)
-        text_res = "\n".join(
+        text_res = sep.join(
             [
                 # avoid additional marker on already evaled sublists
                 (
@@ -320,7 +328,7 @@ class MarkdownListSerializer(BaseModel, BaseListSerializer):
                     if c.text and c.text[0] == " "
                     else f"{indent_str}{f'{i + 1}.' if is_ol else '-'} {c.text}"
                 )
-                for i, c in enumerate(parts)
+                for i, c in enumerate(my_parts)
             ]
         )
         return SerializationResult(text=text_res)

{docling_core-2.23.1 → docling_core-2.23.3}/docling_core/types/doc/document.py

@@ -3051,6 +3051,25 @@ class DoclingDocument(BaseModel):
             """Strip all <...> tags inside the chunk to get the raw text content."""
             return re.sub(r"<.*?>", "", text_chunk, flags=re.DOTALL).strip()
 
+        def extract_caption(
+            text_chunk: str,
+        ) -> tuple[Optional[TextItem], Optional[BoundingBox]]:
+            """Extract caption text from the chunk."""
+            caption = re.search(r"<caption>(.*?)</caption>", text_chunk)
+            if caption is not None:
+                caption_content = caption.group(1)
+                bbox = extract_bounding_box(caption_content)
+                caption_text = extract_inner_text(caption_content)
+                caption_item = self.add_text(
+                    label=DocItemLabel.CAPTION,
+                    text=caption_text,
+                    parent=None,
+                )
+            else:
+                caption_item = None
+                bbox = None
+            return caption_item, bbox
+
         def otsl_parse_texts(texts, tokens):
             split_word = TableToken.OTSL_NL.value
             split_row_tokens = [
@@ -3261,16 +3280,24 @@ class DoclingDocument(BaseModel):
                 if tag_name == DocumentToken.OTSL.value:
                     table_data = parse_table_content(full_chunk)
                     bbox = extract_bounding_box(full_chunk) if image else None
-
+                    caption, caption_bbox = extract_caption(full_chunk)
+                    if caption is not None and caption_bbox is not None:
+                        caption.prov.append(
+                            ProvenanceItem(
+                                bbox=caption_bbox.resize_by_scale(pg_width, pg_height),
+                                charspan=(0, 0),
+                                page_no=page_no,
+                            )
+                        )
                     if bbox:
                         prov = ProvenanceItem(
                             bbox=bbox.resize_by_scale(pg_width, pg_height),
                             charspan=(0, 0),
                             page_no=page_no,
                         )
-                        self.add_table(data=table_data, prov=prov)
+                        self.add_table(data=table_data, prov=prov, caption=caption)
                     else:
-                        self.add_table(data=table_data)
+                        self.add_table(data=table_data, caption=caption)
 
                 elif tag_name == DocItemLabel.PICTURE:
                     text_caption_content = extract_inner_text(full_chunk)

{docling_core-2.23.1 → docling_core-2.23.3}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "docling-core"
-version = "2.23.1"
+version = "2.23.3"
 description = "A python library to define and validate data types in Docling."
 license = "MIT"
 authors = [