rehan953 commited on
Commit
f1dbd49
·
verified ·
1 Parent(s): 309a064

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +2 -1
app.py CHANGED
@@ -733,7 +733,8 @@ def strip_degenerate_html_tables(md: str) -> str:
733
  return re.sub(r"\n{3,}", "\n\n", out)
734
 
735
 
736
- _LONG_NUM_TOKEN = re.compile(r"\b\d{7,}\b")
 
737
 
738
 
739
  def mask_non_monetary_long_numbers(md: str) -> str:
 
733
  return re.sub(r"\n{3,}", "\n\n", out)
734
 
735
 
736
+ # Match long digit runs even when OCR glues trailing letters (e.g. 9257403599CCD).
737
+ _LONG_NUM_TOKEN = re.compile(r"(?<!\d)\d{7,}(?!\d)")
738
 
739
 
740
  def mask_non_monetary_long_numbers(md: str) -> str: