Tweak pgf escapes.

anntzer · anntzer · commit ecc25ca32378 · 2022-07-15T18:25:19.000+02:00
- We don't need to escape underscores manually, but can rely on the
  underscore package like we already do for usetex.
- We don't actually escape dollars (we parse them as math delimiters
  first).
- Slightly tweak error message generation.
- Move escaping tests before the big `create_figure` definition, which
  is used for further tests below.
diff --git a/lib/matplotlib/backends/backend_pgf.py b/lib/matplotlib/backends/backend_pgf.py
@@ -58,7 +58,9 @@ def get_preamble():
 
 def _get_preamble():
     """Prepare a LaTeX preamble based on the rcParams configuration."""
-    preamble = [mpl.rcParams["pgf.preamble"]]
+    preamble = [
+        mpl.rcParams["pgf.preamble"],
+    ]
     if mpl.rcParams["pgf.texsystem"] != "pdflatex":
         preamble.append("\\usepackage{fontspec}")
         if mpl.rcParams["pgf.rcfonts"]:
@@ -70,6 +72,8 @@ def _get_preamble():
                 path = pathlib.Path(fm.findfont(family))
                 preamble.append(r"\%s{%s}[Path=\detokenize{%s/}]" % (
                     command, path.name, path.parent.as_posix()))
+    preamble.append(mpl.texmanager._usepackage_if_not_loaded(
+        "underscore", option="strings"))  # Documented as "must come last".
     return "\n".join(preamble)
 
 
@@ -84,9 +88,8 @@ def _get_preamble():
 _NO_ESCAPE = r"(?<!\\)(?:\\\\)*"
 _split_math = re.compile(_NO_ESCAPE + r"\$").split
 _replace_escapetext = functools.partial(
-    # When the next character is _, ^, $, or % (not preceded by an escape),
-    # insert a backslash.
-    re.compile(_NO_ESCAPE + "(?=[_^$%])").sub, "\\\\")
+    # When the next character is an unescaped % or ^, insert a backslash.
+    re.compile(_NO_ESCAPE + "(?=[%^])").sub, "\\\\")
 _replace_mathdefault = functools.partial(
     # Replace \mathdefault (when not preceded by an escape) by empty string.
     re.compile(_NO_ESCAPE + r"(\\mathdefault)").sub, "")
@@ -106,7 +109,7 @@ def _tex_escape(text):
     ``$`` with ``\(\displaystyle %s\)``. Escaped math separators (``\$``)
     are ignored.
 
-    The following characters are escaped in text segments: ``_^$%``
+    The following characters are escaped in text segments: ``^%``
     """
     # Sometimes, matplotlib adds the unknown command \mathdefault.
     # Not using \mathnormal instead since this looks odd for the latex cm font.
@@ -358,14 +361,16 @@ def _get_box_metrics(self, tex):
         try:
             answer = self._expect_prompt()
         except LatexError as err:
-            raise ValueError("Error measuring {!r}\nLaTeX Output:\n{}"
+            # Here and below, use '{}' instead of {!r} to avoid doubling all
+            # backslashes.
+            raise ValueError("Error measuring {}\nLaTeX Output:\n{}"
                              .format(tex, err.latex_output)) from err
         try:
             # Parse metrics from the answer string.  Last line is prompt, and
             # next-to-last-line is blank line from \typeout.
             width, height, offset = answer.splitlines()[-3].split(",")
         except Exception as err:
-            raise ValueError("Error measuring {!r}\nLaTeX Output:\n{}"
+            raise ValueError("Error measuring {}\nLaTeX Output:\n{}"
                              .format(tex, answer)) from err
         w, h, o = float(width[:-2]), float(height[:-2]), float(offset[:-2])
         # The height returned from LaTeX goes from base to top;
@@ -864,8 +869,8 @@ def print_pdf(self, fname_or_fh, *, metadata=None, **kwargs):
                     r"\documentclass[12pt]{minimal}",
                     r"\usepackage[papersize={%fin,%fin}, margin=0in]{geometry}"
                     % (w, h),
-                    _get_preamble(),
                     r"\usepackage{pgf}",
+                    _get_preamble(),
                     r"\begin{document}",
                     r"\centering",
                     r"\input{figure.pgf}",
@@ -975,8 +980,8 @@ def _write_header(self, width_inches, height_inches):
             r"\documentclass[12pt]{minimal}",
             r"\usepackage[papersize={%fin,%fin}, margin=0in]{geometry}"
             % (width_inches, height_inches),
-            _get_preamble(),
             r"\usepackage{pgf}",
+            _get_preamble(),
             r"\setlength{\parindent}{0pt}",
             r"\begin{document}%",
         ])
diff --git a/lib/matplotlib/tests/test_backend_pgf.py b/lib/matplotlib/tests/test_backend_pgf.py
@@ -33,6 +33,23 @@ def compare_figure(fname, savefig_kwargs={}, tol=0):
         raise ImageComparisonFailure(err)
 
 
+@pytest.mark.parametrize('plain_text, escaped_text', [
+    (r'quad_sum: $\sum x_i^2$', r'quad_sum: \(\displaystyle \sum x_i^2\)'),
+    ('% not a comment', r'\% not a comment'),
+    ('^not', r'\^not'),
+])
+def test_tex_escape(plain_text, escaped_text):
+    assert _tex_escape(plain_text) == escaped_text
+
+
+@needs_pgf_xelatex
+@pytest.mark.backend('pgf')
+def test_tex_special_chars(tmp_path):
+    fig = plt.figure()
+    fig.text(.5, .5, "_^ $a_b^c$")
+    fig.savefig(tmp_path / "test.pdf")  # Should not error.
+
+
 def create_figure():
     plt.figure()
     x = np.linspace(0, 1, 15)
@@ -59,17 +76,6 @@ def create_figure():
     plt.ylim(0, 1)
 
 
-@pytest.mark.parametrize('plain_text, escaped_text', [
-    (r'quad_sum: $\sum x_i^2$', r'quad\_sum: \(\displaystyle \sum x_i^2\)'),
-    (r'no \$splits \$ here', r'no \$splits \$ here'),
-    ('with_underscores', r'with\_underscores'),
-    ('% not a comment', r'\% not a comment'),
-    ('^not', r'\^not'),
-])
-def test_tex_escape(plain_text, escaped_text):
-    assert _tex_escape(plain_text) == escaped_text
-
-
 # test compiling a figure to pdf with xelatex
 @needs_pgf_xelatex
 @pytest.mark.backend('pgf')
