fix(docx): read styles with the invalid type "list"

docling/backend/msword_backend.py

@@ -68,6 +68,7 @@
     from docx.enum.style import WD_STYLE_TYPE
     from docx.opc.constants import RELATIONSHIP_TYPE as DOCX_RT
     from docx.oxml import parse_xml as docx_parse_xml
+    from docx.oxml.ns import qn
     from docx.oxml.simpletypes import ST_Merge
     from docx.oxml.table import CT_Tc
     from docx.oxml.xmlchemy import BaseOxmlElement
@@ -649,6 +650,19 @@ def _normalize_strict_ooxml(archive: zipfile.ZipFile) -> BytesIO:
     return normalized
 
 
+def _normalize_list_style_types(docx_obj: DocxDocument) -> None:
+    """Change the style type ``list`` to ``numbering`` in the styles part.
+
+    ``ST_StyleType`` has no ``list`` value; the numbering style type is
+    ``WD_STYLE_TYPE.LIST`` in python-docx. python-docx raises ``ValueError``
+    when it reads the type ``list``, and it reads the type of every style to
+    find a default style, so one such style fails the whole conversion.
+    """
+    for style in docx_obj.styles.element.findall(qn("w:style")):
+        if style.get(qn("w:type")) == "list":
+            style.set(qn("w:type"), "numbering")
+
+
 class _NestedList(NamedTuple):
     """A list opened under an open list's item for its larger left indentation.
 
@@ -879,6 +893,7 @@ def __init__(
         if self.docx_obj:
             self.valid = True
             self.current_part = self.docx_obj.part
+            _normalize_list_style_types(self.docx_obj)
             self._default_paragraph_style = self.docx_obj.styles.default(
                 WD_STYLE_TYPE.PARAGRAPH
             )

tests/test_backend_msword.py

@@ -2353,6 +2353,30 @@ def test_fragment_only_rel_does_not_crash_backend():
     assert text_items[0].hyperlink is None
 
 
+@pytest.mark.parametrize("body_style", ["Body Text", None])
+def test_list_style_type_does_not_fail_conversion(tmp_path, body_style):
+    """A style with the invalid type ``list`` must not fail the conversion.
+
+    python-docx raises ``ValueError`` when it reads that type. Both the default
+    style lookup on load and the style of a paragraph without ``w:pStyle``
+    read the type of every style.
+    """
+    document = Document()
+    document.add_paragraph("Quarterly results", style="Heading 1")
+    document.add_paragraph("Revenue grew 3%.", style=body_style)
+    list_style = OxmlElement("w:style")
+    list_style.set(qn("w:type"), "list")
+    list_style.set(qn("w:styleId"), "OutlineList")
+    document.styles.element.append(list_style)
+
+    doc = _convert_built(document, tmp_path)
+
+    assert [item.text for item in doc.texts] == [
+        "Quarterly results",
+        "Revenue grew 3%.",
+    ]
+
+
 _PLAIN_FOOTNOTE_BODY = (
     "<w:p><w:r><w:t>Smith, J. (2020). Example Study.</w:t></w:r></w:p>"
 )