mirrors/docling
1
0
Fork 0
mirror of https://github.com/DS4SD/docling.git synced 2025-07-26 20:14:47 +00:00
Code Issues Packages Projects Releases Wiki Activity

Merge 7955903e9b into b356b33059

Browse Source
This commit is contained in:
ka-weihe 2025-05-28 13:13:48 +02:00 committed by GitHub
commit c303265526
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 29 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

26
docling/backend/html_backend.py
View File

@ -181,6 +181,8 @@ class HTMLDocumentBackend(DeclarativeDocumentBackend):
self.handle_figure(tag, doc) self.handle_figure(tag, doc)
elif tag.name == "img": elif tag.name == "img":
self.handle_image(tag, doc) self.handle_image(tag, doc)
elif tag.name == "a":
self.handle_anchor(tag, doc)
elif tag.name == "details": elif tag.name == "details":
self.handle_details(tag, doc) self.handle_details(tag, doc)
else: else:
@ -293,7 +295,6 @@ class HTMLDocumentBackend(DeclarativeDocumentBackend):
def handle_list(self, element: Tag, doc: DoclingDocument) -> None: def handle_list(self, element: Tag, doc: DoclingDocument) -> None:
"""Handles list tags (ul, ol) and their list items.""" """Handles list tags (ul, ol) and their list items."""
if element.name == "ul": if element.name == "ul":
# create a list group # create a list group
self.parents[self.level + 1] = doc.add_group( self.parents[self.level + 1] = doc.add_group(
@ -508,9 +509,7 @@ class HTMLDocumentBackend(DeclarativeDocumentBackend):
def handle_table(self, element: Tag, doc: DoclingDocument) -> None: def handle_table(self, element: Tag, doc: DoclingDocument) -> None:
"""Handles table tags.""" """Handles table tags."""
table_data = HTMLDocumentBackend.parse_table_data(element) table_data = HTMLDocumentBackend.parse_table_data(element)
if table_data is not None: if table_data is not None:
doc.add_table( doc.add_table(
data=table_data, data=table_data,
@ -545,12 +544,10 @@ class HTMLDocumentBackend(DeclarativeDocumentBackend):
nested_list = li.find(["ul", "ol"]) nested_list = li.find(["ul", "ol"])
if isinstance(nested_list, Tag): if isinstance(nested_list, Tag):
result.extend(self.get_list_text(nested_list, level + 1)) result.extend(self.get_list_text(nested_list, level + 1))
return result return result
def handle_figure(self, element: Tag, doc: DoclingDocument) -> None: def handle_figure(self, element: Tag, doc: DoclingDocument) -> None:
"""Handles image tags (img).""" """Handles image tags (img)."""
# Extract the image URI from the <img> tag # Extract the image URI from the <img> tag
# image_uri = root.xpath('//figure//img/@src')[0] # image_uri = root.xpath('//figure//img/@src')[0]
@ -580,9 +577,26 @@ class HTMLDocumentBackend(DeclarativeDocumentBackend):
def handle_image(self, element: Tag, doc: DoclingDocument) -> None: def handle_image(self, element: Tag, doc: DoclingDocument) -> None:
"""Handles image tags (img).""" """Handles image tags (img)."""
_log.debug(f"ignoring <img> tags at the moment: {element}") _log.debug(f"ignoring <img> tags at the moment: {element}")
doc.add_picture( doc.add_picture(
parent=self.parents[self.level], parent=self.parents[self.level],
caption=None, caption=None,
content_layer=self.content_layer, content_layer=self.content_layer,
) )
def handle_anchor(self, element: Tag, doc: DoclingDocument) -> None:
"""Handles anchor tags (<a>) by extracting the visible text and setting the hyperlink property."""
# Extract the visible text and href URL
text = element.get_text().strip()
href = element.get("href", "").strip()
# If no text is present, use the hyperlink itself as the text.
display_text = text if text else href
if display_text or href:
doc.add_text(
parent=self.parents[self.level],
label=DocItemLabel.TEXT,
text=display_text,
hyperlink=href if href else None,
content_layer=self.content_layer,
)

9
tests/data/html/example_08.html Normal file
View File

@ -0,0 +1,9 @@
<html>
<body>
<a href="https://example.com">Example Website</a><br>
<a href="https://github.com"></a><br>
<a href="mailto:contact@example.com">Contact Us</a><br>
<a href="#top">Go to Top</a><br>
<a href="https://wikipedia.org">Wikipedia</a>
</body>
</html>