fix(latex): skip minted's language and bracketed options before code

docling/backend/latex/utils/text.py

@@ -41,6 +41,13 @@
     pass  # guarded by LatexDocumentBackend.__init__
 
 
+# A {...} group, allowing one level of nested braces.
+_BRACED_GROUP = r"\{(?:[^{}]|\{[^{}]*\})*\}"
+# A [...] optional argument whose braced values may contain brackets,
+# as in [caption={A [b] c}].
+_OPTIONAL_ARGUMENT = rf"\[(?:[^\[\]{{}}]|{_BRACED_GROUP})*\]"
+
+
 def iter_child_nodelists(node: Any) -> Iterator[list]:
     """Yield the node lists of ``node`` and of its group arguments."""
     if isinstance(node, (LatexGroupNode, LatexEnvironmentNode, LatexMathNode)):
@@ -125,7 +132,13 @@ def _process_group_node(
             self._process_nodes(node.nodelist, doc, parent, formatting, text_label)
 
     def _extract_verbatim_content(self, latex_str: str, env_name: str) -> str:
-        pattern = rf"\\begin\{{{re.escape(env_name)}\}}(?:\[.*?\])?(.*?)\\end\{{{re.escape(env_name)}\}}"
+        name = re.escape(env_name)
+        # minted takes the code's language as a mandatory {argument}.
+        language = _BRACED_GROUP if env_name == "minted" else ""
+        pattern = (
+            rf"\\begin\{{{name}\}}(?:{_OPTIONAL_ARGUMENT})?{language}"
+            rf"(.*?)\\end\{{{name}\}}"
+        )
         match = re.search(pattern, latex_str, re.DOTALL)
         if match:
             return re.sub(r"\A(?:[ \t]*\r?\n)+", "", match.group(1)).rstrip()

tests/test_latex/test_environments.py

@@ -126,6 +126,38 @@ def test_latex_code_environment_keeps_first_line_indentation(env_name, lead):
     assert code_texts == ["  key: 1\n  sub:\n    x: 2"]
 
 
+@pytest.mark.parametrize(
+    "opening",
+    [
+        "\\begin{minted}{python}",
+        "\\begin{minted}[linenos, frame=lines]{python}",
+        "\\begin{lstlisting}[language=Python, caption={A [nested] cap}]",
+    ],
+)
+def test_latex_code_environment_skips_its_arguments(opening):
+    """Environment arguments are markup, not the first line of the code."""
+    env_name = opening[len("\\begin{") : opening.index("}")]
+    latex_content = (
+        "\\documentclass{article}\n"
+        "\\begin{document}\n"
+        f"{opening}\n"
+        "    x = 1\n"
+        f"\\end{{{env_name}}}\n"
+        "\\end{document}\n"
+    ).encode()
+    in_doc = InputDocument(
+        path_or_stream=BytesIO(latex_content),
+        format=InputFormat.LATEX,
+        backend=LatexDocumentBackend,
+        filename="test.tex",
+    )
+    backend = LatexDocumentBackend(in_doc=in_doc, path_or_stream=BytesIO(latex_content))
+    doc = backend.convert()
+
+    code_texts = [t.text for t in doc.texts if t.label == DocItemLabel.CODE]
+    assert code_texts == ["    x = 1"]
+
+
 def test_latex_subequations_environment():
     """Test subequations wrapper environment passes through inner equations."""
     latex_content = rb"""