sidoutcome commited on
Commit
e24be4d
·
1 Parent(s): bb28e9c

fix: prevent LaTeX regex from stripping currency dollar signs

Browse files
Files changed (1) hide show
  1. postprocess.py +3 -2
postprocess.py CHANGED
@@ -54,8 +54,9 @@ _LATEX_SUPERSCRIPT = re.compile(r"\$\s*\^?\{([^}]*)\}\s*\$")
54
  _LATEX_SUBSCRIPT = re.compile(r"\$\s*_\{([^}]*)\}\s*\$")
55
  # $\pm value$ → ± value
56
  _LATEX_PLUSMINUS = re.compile(r"\$\s*\\pm\s*([^$]*?)\s*\$")
57
- # Generic remaining $...$ inline math — strip dollar signs
58
- _LATEX_INLINE = re.compile(r"\$\s*([^$]{1,60}?)\s*\$")
 
59
  # LaTeX command remnants: \%, \$, \&, etc.
60
  _LATEX_ESCAPED_CHARS = re.compile(r"\\([%$&_#])")
61
 
 
54
  _LATEX_SUBSCRIPT = re.compile(r"\$\s*_\{([^}]*)\}\s*\$")
55
  # $\pm value$ → ± value
56
  _LATEX_PLUSMINUS = re.compile(r"\$\s*\\pm\s*([^$]*?)\s*\$")
57
+ # Generic remaining $...$ inline math — only match when content has LaTeX chars
58
+ # (backslash, caret, underscore, braces). Avoids stripping currency like $8,125.00
59
+ _LATEX_INLINE = re.compile(r"\$\s*([^$]{1,60}?[\\^_{}][^$]{0,60}?)\s*\$")
60
  # LaTeX command remnants: \%, \$, \&, etc.
61
  _LATEX_ESCAPED_CHARS = re.compile(r"\\([%$&_#])")
62