BUG: Correct Escaping for LaTeX, #20859 (#20860)

Author: retkowski

doc/source/whatsnew/v0.23.0.txt

@@ -1164,6 +1164,7 @@ I/O
 - Bug in :func:`DataFrame.to_latex()` where a non-string index-level name would result in an ``AttributeError`` (:issue:`19981`)
 - Bug in :func:`DataFrame.to_latex()` where the combination of an index name and the `index_names=False` option would result in incorrect output (:issue:`18326`)
 - Bug in :func:`DataFrame.to_latex()` where a ``MultiIndex`` with an empty string as its name would result in incorrect output (:issue:`18669`)
+- Bug in :func:`DataFrame.to_latex()` where missing space characters caused wrong escaping and produced non-valid latex in some cases (:issue:`20859`)
 - Bug in :func:`read_json` where large numeric values were causing an ``OverflowError`` (:issue:`18842`)
 - Bug in :func:`DataFrame.to_parquet` where an exception was raised if the write destination is S3 (:issue:`19134`)
 - :class:`Interval` now supported in :func:`DataFrame.to_excel` for all Excel file types (:issue:`19242`)

pandas/io/formats/latex.py

@@ -134,11 +134,13 @@ def pad_empties(x):
                     buf.write('\\endlastfoot\n')
             if self.fmt.kwds.get('escape', True):
                 # escape backslashes first
-                crow = [(x.replace('\\', '\\textbackslash').replace('_', '\\_')
+                crow = [(x.replace('\\', '\\textbackslash ')
+                         .replace('_', '\\_')
                          .replace('%', '\\%').replace('$', '\\$')
                          .replace('#', '\\#').replace('{', '\\{')
-                         .replace('}', '\\}').replace('~', '\\textasciitilde')
-                         .replace('^', '\\textasciicircum').replace('&', '\\&')
+                         .replace('}', '\\}').replace('~', '\\textasciitilde ')
+                         .replace('^', '\\textasciicircum ')
+                         .replace('&', '\\&')
                          if (x and x != '{}') else '{}') for x in row]
             else:
                 crow = [x if x else '{}' for x in row]

pandas/tests/io/formats/test_to_latex.py

@@ -369,7 +369,7 @@ def test_to_latex_escape(self):
 
         escaped_expected = r'''\begin{tabular}{lll}
 \toprule
-{} & co\$e\textasciicircumx\$ & co\textasciicircuml1 \\
+{} & co\$e\textasciicircum x\$ & co\textasciicircum l1 \\
 \midrule
 a &       a &     a \\
 b &       b &     b \\
@@ -380,6 +380,22 @@ def test_to_latex_escape(self):
         assert unescaped_result == unescaped_expected
         assert escaped_result == escaped_expected
 
+    def test_to_latex_special_escape(self):
+        df = DataFrame([r"a\b\c", r"^a^b^c", r"~a~b~c"])
+
+        escaped_result = df.to_latex()
+        escaped_expected = r"""\begin{tabular}{ll}
+\toprule
+{} &       0 \\
+\midrule
+0 &   a\textbackslash b\textbackslash c \\
+1 &  \textasciicircum a\textasciicircum b\textasciicircum c \\
+2 &  \textasciitilde a\textasciitilde b\textasciitilde c \\
+\bottomrule
+\end{tabular}
+"""
+        assert escaped_result == escaped_expected
+
     def test_to_latex_longtable(self, frame):
         frame.to_latex(longtable=True)
 
@@ -447,9 +463,9 @@ def test_to_latex_escape_special_chars(self):
 4 &  \_ \\
 5 &  \{ \\
 6 &  \} \\
-7 &  \textasciitilde \\
-8 &  \textasciicircum \\
-9 &  \textbackslash \\
+7 &  \textasciitilde  \\
+8 &  \textasciicircum  \\
+9 &  \textbackslash  \\
 \bottomrule
 \end{tabular}
 """