rehan953 commited on
Commit
8a7e53d
·
verified ·
1 Parent(s): 499765d

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +4 -3
app.py CHANGED
@@ -216,7 +216,7 @@ def fix_account_number(hdr):
216
  if not hdr:
217
  return hdr
218
  # Step 1: inject account number if missing after label
219
- if "Account Number:" in hdr and not re.search("Account Number: [0-9]", hdr):
220
  match = re.search("[0-9]{5,}", hdr)
221
  if match:
222
  acct = match.group(0)
@@ -224,11 +224,12 @@ def fix_account_number(hdr):
224
  # Step 2: remove leading number that is duplicated after Account Number:
225
  # e.g. "000000731823008 JPMorgan ... Account Number: 000000731823008"
226
  # -> "JPMorgan ... Account Number: 000000731823008"
227
- acct_match = re.search("Account Number:\s*([0-9]{5,})", hdr)
228
  if acct_match:
229
  acct = acct_match.group(1)
230
  # If hdr starts with that same number, remove it from the start
231
- hdr = re.sub("^" + re.escape(acct) + "\s*", "", hdr)
 
232
  return hdr
233
 
234
 
 
216
  if not hdr:
217
  return hdr
218
  # Step 1: inject account number if missing after label
219
+ if "Account Number:" in hdr and "Account Number: " not in hdr:
220
  match = re.search("[0-9]{5,}", hdr)
221
  if match:
222
  acct = match.group(0)
 
224
  # Step 2: remove leading number that is duplicated after Account Number:
225
  # e.g. "000000731823008 JPMorgan ... Account Number: 000000731823008"
226
  # -> "JPMorgan ... Account Number: 000000731823008"
227
+ acct_match = re.search(r"Account Number: ([0-9]{5,})", hdr)
228
  if acct_match:
229
  acct = acct_match.group(1)
230
  # If hdr starts with that same number, remove it from the start
231
+ if hdr.startswith(acct):
232
+ hdr = hdr[len(acct):].lstrip()
233
  return hdr
234
 
235