From 1c04508ed739d22ec674e664d048787fd6138d31 Mon Sep 17 00:00:00 2001
From: Chris Hoeppler <christian.hoeppler@de.bosch.com>
Date: Tue, 1 Oct 2024 16:31:28 +0200
Subject: [PATCH] fix(latex-renderer): special chars in raw text

Escape some additional characters in raw text which have special meaning
in LaTeX and which previously hadn't been dealt with by
`latex-renderer`'s `render_raw_text`-method.

Refactor the corresponding test case along the lines of the other
parameterized test cases to improve the code's consistency.
---
 mistletoe/latex_renderer.py | 20 +++++++++++++++-----
 test/test_latex_renderer.py | 28 ++++++++++++++++++++++------
 2 files changed, 37 insertions(+), 11 deletions(-)

diff --git a/mistletoe/latex_renderer.py b/mistletoe/latex_renderer.py
index ec76d76..3230392 100644
--- a/mistletoe/latex_renderer.py
+++ b/mistletoe/latex_renderer.py
@@ -2,6 +2,7 @@
 LaTeX renderer for mistletoe.
 """
 
+import re
 import string
 from itertools import chain
 from urllib.parse import quote
@@ -78,11 +79,20 @@ def render_escape_sequence(self, token):
         return self.render_inner(token)
 
     def render_raw_text(self, token, escape=True):
-        return (token.content.replace('$', '\\$').replace('#', '\\#')
-                             .replace('{', '\\{').replace('}', '\\}')
-                             .replace('&', '\\&').replace('_', '\\_')
-                             .replace('%', '\\%')
-               ) if escape else token.content
+        """Escape all latex special characters $#&%_{}^~\\ within `token.content`.
+        """
+        if not escape:
+            return token.content
+
+        if not hasattr(self, 'raw_escape_chars'):
+            self.raw_escape_chars = re.compile('([$#&%_{}])')
+
+        content = token.content.replace('\\', '\\textbackslash')
+        content = self.raw_escape_chars.sub(r'\\\1', content)
+        # The \text* commands gobble up whitespace behind them -> {} to prevent that.
+        return content.replace('~', '\\textasciitilde{}') \
+                      .replace('^', '\\textasciicircum{}') \
+                      .replace('\\textbackslash', '\\textbackslash{}')
 
     def render_heading(self, token):
         inner = self.render_inner(token)
diff --git a/test/test_latex_renderer.py b/test/test_latex_renderer.py
index c9b3760..e0c6407 100644
--- a/test/test_latex_renderer.py
+++ b/test/test_latex_renderer.py
@@ -3,6 +3,7 @@
 import mistletoe.latex_renderer
 from mistletoe.latex_renderer import LaTeXRenderer
 from mistletoe import markdown
+import markdown
 
 
 class TestLaTeXRenderer(TestCase):
@@ -13,14 +14,14 @@ def setUp(self):
         self.addCleanup(self.renderer.__exit__, None, None, None)
 
     def _test_token(self, token_name, expected_output, children=True,
-                    without_attrs=None, **kwargs):
+                    without_attrs=None, render_func_kwargs={}, **kwargs):
         render_func = self.renderer.render_map[token_name]
         children = mock.MagicMock(spec=list) if children else None
         mock_token = mock.Mock(children=children, **kwargs)
         without_attrs = without_attrs or []
         for attr in without_attrs:
             delattr(mock_token, attr)
-        self.assertEqual(render_func(mock_token), expected_output)
+        self.assertEqual(render_func(mock_token, **render_func_kwargs), expected_output)
 
     def test_strong(self):
         self._test_token('Strong', '\\textbf{inner}')
@@ -72,10 +73,25 @@ def test_math(self):
         self._test_token('Math', expected,
                          children=False, content='$ 1 + 2 = 3 $')
 
-    def test_raw_text(self):
-        expected = '\\$\\&\\#\\{\\}'
-        self._test_token('RawText', expected,
-                         children=False, content='$&#{}')
+    @parameterized.expand([
+        ('$', '\\$'),
+        ('&', '\\&'),
+        ('#', '\\#'),
+        ('%', '\\%'),
+        ('_', '\\_'),
+        ('{', '\\{'),
+        ('}', '\\}'),
+        ('~', '\\textasciitilde{}'),
+        ('^', '\\textasciicircum{}'),
+        ('\\', '\\textbackslash{}'),
+    ])
+    def test_raw_text(self, target, expected):
+        self._test_token('RawText', expected, children=False, content=target)
+
+    def test_raw_text_no_escape(self):
+        expected = '$&#%_{}~^\\'
+        self._test_token('RawText', expected, children=False, content=expected,
+                         render_func_kwargs={'escape': False})
 
     def test_heading(self):
         expected = '\n\\section{inner}\n'