Handle missing verbatim arguments and empty LaTeX matrices

great_ai/utilities/external/pylatexenc/latex2text/__init__.py

@@ -386,7 +386,7 @@ def fmt_matrix_environment_node(node, l2tobj):
     state.new_row()  # finish the last row
 
     # now format the contents as array --
-    max_char_width = max((len(x) for row in state.matrix_rows for x in row))
+    max_char_width = max((len(x) for row in state.matrix_rows for x in row), default=0)
     matrix_contents = "; ".join(
         (
             " ".join((x.rjust(max_char_width, " ") for x in row))

great_ai/utilities/external/pylatexenc/macrospec/_argparsers.py

@@ -460,12 +460,12 @@ class VerbatimArgsParser(MacroStandardArgsParser):
         if self.verbatim_arg_type == "verb-macro":
             # read the next nonwhitespace char. This is the delimiter of the
             # argument
-            while w.s[pos].isspace():
+            while pos < len(w.s) and w.s[pos].isspace():
                 pos += 1
-                if pos >= len(w.s):
-                    raise latexwalker.LatexWalkerParseError(
-                        s=w.s, pos=pos, msg=r"Missing argument to \verb command"
-                    )
+            if pos >= len(w.s):
+                raise latexwalker.LatexWalkerParseError(
+                    s=w.s, pos=pos, msg=r"Missing argument to \verb command"
+                )
             verbdelimchar = w.s[pos]
             beginpos = pos + 1
             endpos = w.s.find(verbdelimchar, beginpos)

tests/utilities/test_latex_regressions.py

@@ -1,6 +1,8 @@
 from pathlib import Path
 
+import pytest
 from great_ai.utilities.external.pylatexenc.latex2text import LatexNodes2Text
+from great_ai.utilities.external.pylatexenc.latexwalker import LatexWalkerParseError
 
 
 def test_strict_input_checks_path_ancestry(tmp_path: Path) -> None:
@@ -16,3 +18,15 @@ def test_strict_input_checks_path_ancestry(tmp_path: Path) -> None:
     assert converter.latex_to_text(r"\input{inside}") == "inside"
     assert converter.latex_to_text(r"\input{../allowed_neighbor/outside}") == ""
     assert converter.latex_to_text(r"\input{link}") == ""
+
+
+@pytest.mark.parametrize("text", [r"\verb", "\\verb   "])
+def test_missing_verbatim_arguments_use_normal_parse_errors(text: str) -> None:
+    converter = LatexNodes2Text()
+    with pytest.raises(LatexWalkerParseError):
+        converter.latex_to_text(text, tolerant_parsing=False)
+    assert converter.latex_to_text(text, tolerant_parsing=True) == ""
+
+
+def test_empty_matrix_has_a_text_representation() -> None:
+    assert LatexNodes2Text().latex_to_text(r"\begin{pmatrix}\end{pmatrix}") == "[  ]"