docling-core 2.38.0__py3-none-any.whl → 2.38.1__py3-none-any.whl

docling_core/transforms/serializer/common.py

@@ -349,6 +349,7 @@ class DocSerializer(BaseModel, BaseDocSerializer):
                         doc_serializer=self,
                         doc=self.doc,
                         is_inline_scope=is_inline_scope,
+                        visited=my_visited,
                         **my_kwargs,
                     )
                     if item.self_ref not in self.get_excluded_refs(**kwargs)

docling_core/transforms/serializer/markdown.py

@@ -106,26 +106,49 @@ class MarkdownTextSerializer(BaseModel, BaseTextSerializer):
         doc_serializer: BaseDocSerializer,
         doc: DoclingDocument,
         is_inline_scope: bool = False,
+        visited: Optional[set[str]] = None,  # refs of visited items
         **kwargs: Any,
     ) -> SerializationResult:
         """Serializes the passed item."""
+        my_visited = visited if visited is not None else set()
         params = MarkdownParams(**kwargs)
         res_parts: list[SerializationResult] = []
+        text = item.text
         escape_html = True
         escape_underscores = True
-        if isinstance(item, TitleItem):
-            text_part = f"# {item.text}"
-        elif isinstance(item, SectionHeaderItem):
-            text_part = f"{(item.level + 1) * '#'} {item.text}"
+        processing_pending = True
+        if isinstance(item, (TitleItem, SectionHeaderItem)):
+            # case where processing/formatting should be applied first (in inner scope)
+            processing_pending = False
+            if (
+                text == ""
+                and len(item.children) == 1
+                and isinstance(
+                    (child_group := item.children[0].resolve(doc)), InlineGroup
+                )
+            ):
+                # case of heading with inline
+                ser_res = doc_serializer.serialize(item=child_group)
+                text = ser_res.text
+                for span in ser_res.spans:
+                    my_visited.add(span.item.self_ref)
+            else:
+                text = doc_serializer.post_process(
+                    text=text,
+                    escape_html=escape_html,
+                    escape_underscores=escape_underscores,
+                    formatting=item.formatting,
+                    hyperlink=item.hyperlink,
+                )
+            num_hashes = 1 if isinstance(item, TitleItem) else item.level + 1
+            text_part = f"{num_hashes * '#'} {text}"
         elif isinstance(item, CodeItem):
-            text_part = (
-                f"`{item.text}`" if is_inline_scope else f"```\n{item.text}\n```"
-            )
+            text_part = f"`{text}`" if is_inline_scope else f"```\n{text}\n```"
             escape_html = False
             escape_underscores = False
         elif isinstance(item, FormulaItem):
-            if item.text:
-                text_part = f"${item.text}$" if is_inline_scope else f"$${item.text}$$"
+            if text:
+                text_part = f"${text}$" if is_inline_scope else f"$${text}$$"
             elif item.orig:
                 text_part = "<!-- formula-not-decoded -->"
             else:
@@ -133,9 +156,10 @@ class MarkdownTextSerializer(BaseModel, BaseTextSerializer):
             escape_html = False
             escape_underscores = False
         elif params.wrap_width:
-            text_part = textwrap.fill(item.text, width=params.wrap_width)
+            # although wrapping is not guaranteed if post-processing makes changes
+            text_part = textwrap.fill(text, width=params.wrap_width)
         else:
-            text_part = item.text
+            text_part = text
 
         if text_part:
             text_res = create_ser_result(text=text_part, span_source=item)
@@ -147,13 +171,14 @@ class MarkdownTextSerializer(BaseModel, BaseTextSerializer):
                 res_parts.append(cap_res)
 
         text = (" " if is_inline_scope else "\n\n").join([r.text for r in res_parts])
-        text = doc_serializer.post_process(
-            text=text,
-            escape_html=escape_html,
-            escape_underscores=escape_underscores,
-            formatting=item.formatting,
-            hyperlink=item.hyperlink,
-        )
+        if processing_pending:
+            text = doc_serializer.post_process(
+                text=text,
+                escape_html=escape_html,
+                escape_underscores=escape_underscores,
+                formatting=item.formatting,
+                hyperlink=item.hyperlink,
+            )
         return create_ser_result(text=text, span_source=res_parts)
 
 

{docling_core-2.38.0.dist-info → docling_core-2.38.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: docling-core
-Version: 2.38.0
+Version: 2.38.1
 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-2.38.0.dist-info → docling_core-2.38.1.dist-info}/RECORD

@@ -27,11 +27,11 @@ docling_core/transforms/chunker/tokenizer/huggingface.py,sha256=aZ_RNQIzcNkAHGHZ
 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=ZFIiZeplL-QbBs9EDUb1awqxapQ23PsApVetJtAs7Vs,6891
-docling_core/transforms/serializer/common.py,sha256=WP-qO-woidrKyvZ56m0vlKMysoLrMzzZtHSCIwsl3ek,19119
+docling_core/transforms/serializer/common.py,sha256=RO2KWl3sZq_PIvzWzuGJTWntKjLOAy3n17cgZi84AAs,19163
 docling_core/transforms/serializer/doctags.py,sha256=PuAExlP-2HxcDSP_R_phtYQU0yKBW94RrPgb85IUxck,19905
 docling_core/transforms/serializer/html.py,sha256=SZgQa0QnknEoRwMFLdgmVsLQqLF2rQl3D7XyEZzUHCE,37151
 docling_core/transforms/serializer/html_styles.py,sha256=-jBwS4EU7yfKoz0GSoxhwx90OmIKieO6TwPw57IuxcA,4692
-docling_core/transforms/serializer/markdown.py,sha256=wfMNrjA4wMehWLCejAhEN1eQPRixUO1SyL6ojkKkzZY,20614
+docling_core/transforms/serializer/markdown.py,sha256=2wV0ydqWKSm-HAW94gF0IRBpjWgoqUjL4JHRYS8DDgY,21803
 docling_core/transforms/visualizer/__init__.py,sha256=gUfF25yiJ_KO46ZIUNqZQOZGy2PLx6gnnr6AZYxKHXI,35
 docling_core/transforms/visualizer/base.py,sha256=aEF7b3rHq6DVdX8zDYEPoq55BHDYe4Hh_97lBdcW4lY,555
 docling_core/transforms/visualizer/layout_visualizer.py,sha256=zHzQTWcy-z1J2BcsjvakLkrp8pgStgnxhDl8YqIAotY,8035
@@ -74,9 +74,9 @@ 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-2.38.0.dist-info/licenses/LICENSE,sha256=2M9-6EoQ1sxFztTOkXGAtwUDJvnWaAHdB9BYWVwGkIw,1087
-docling_core-2.38.0.dist-info/METADATA,sha256=llcycAVzvc09CX0igt4VIGrGWT8UuMjnWN5rrQoEJ6s,6453
-docling_core-2.38.0.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-docling_core-2.38.0.dist-info/entry_points.txt,sha256=ER4zROQWkFMHIrY-oqY5E4HeCcCIg8dLkNztYGxdb7c,59
-docling_core-2.38.0.dist-info/top_level.txt,sha256=O-tcXpGiurlud-1ZxMq1b-OmrfAVA4sajcgWU32RtfA,13
-docling_core-2.38.0.dist-info/RECORD,,
+docling_core-2.38.1.dist-info/licenses/LICENSE,sha256=2M9-6EoQ1sxFztTOkXGAtwUDJvnWaAHdB9BYWVwGkIw,1087
+docling_core-2.38.1.dist-info/METADATA,sha256=MnIJIe_3840fVysMwWJFKu792iTKG2wzp_3dOjxi6Yg,6453
+docling_core-2.38.1.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+docling_core-2.38.1.dist-info/entry_points.txt,sha256=ER4zROQWkFMHIrY-oqY5E4HeCcCIg8dLkNztYGxdb7c,59
+docling_core-2.38.1.dist-info/top_level.txt,sha256=O-tcXpGiurlud-1ZxMq1b-OmrfAVA4sajcgWU32RtfA,13
+docling_core-2.38.1.dist-info/RECORD,,