fix: preserve malformed docx math content

packages/markitdown/src/markitdown/converter_utils/docx/math/omml.py

@@ -272,7 +272,9 @@ def do_fname(self, elm):
                 if FUNC.get(t):
                     latex_chars.append(FUNC[t])
                 else:
-                    raise NotImplementedError("Not support func %s" % t)
+                    latex_chars.append(
+                        "\\operatorname{%s}(%s)" % (escape_latex(t), FUNC_PLACE)
+                    )
             else:
                 latex_chars.append(t)
         t = BLANK.join(latex_chars)

packages/markitdown/src/markitdown/converter_utils/docx/pre_process.py

@@ -44,6 +44,8 @@ def _convert_omath_to_latex(tag: Tag) -> str:
     math_root = ET.fromstring(MATH_ROOT_TEMPLATE.format(str(tag)))
     # Find the 'oMath' element within the XML document
     math_element = math_root.find(OMML_NS + "oMath")
+    if math_element is None:
+        return tag.get_text("", strip=True)
     # Convert the 'oMath' element to LaTeX using the oMath2Latex function
     latex = oMath2Latex(math_element).latex
     return latex

packages/markitdown/tests/test_docx_math.py

@@ -0,0 +1,26 @@
+from bs4 import BeautifulSoup
+from defusedxml import ElementTree as ET
+
+from markitdown.converter_utils.docx.math.omml import oMath2Latex
+from markitdown.converter_utils.docx.pre_process import _convert_omath_to_latex
+
+
+def test_convert_omath_without_namespaced_child_returns_text() -> None:
+    soup = BeautifulSoup(b"<oMath><r><t>x</t></r></oMath>", "xml")
+
+    assert _convert_omath_to_latex(soup.find("oMath")) == "x"
+
+
+def test_unknown_omml_function_uses_operatorname() -> None:
+    root = ET.fromstring(
+        """
+        <m:oMath xmlns:m="http://schemas.openxmlformats.org/officeDocument/2006/math">
+          <m:func>
+            <m:fName><m:r><m:t>log</m:t></m:r></m:fName>
+            <m:e><m:r><m:t>x</m:t></m:r></m:e>
+          </m:func>
+        </m:oMath>
+        """
+    )
+
+    assert oMath2Latex(root).latex == r"\operatorname{log}(x)"