rehan953 commited on
Commit
678d4a1
·
verified ·
1 Parent(s): 1c191a0

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +9 -10
app.py CHANGED
@@ -207,18 +207,17 @@ def ocr_zone(image_path, y_start_frac, y_end_frac):
207
  return ""
208
 
209
 
210
- def fix_account_number(page_md, hdr):
211
- """If page_md has 'Account Number:' with no value after it,
212
- find the account number from header text and inject it."""
213
  import re
214
- if not page_md or not hdr:
215
- return page_md
216
- if not re.search(r'Account Number:\s*\S', page_md):
217
- match = re.search(r'(\d{10,})', hdr)
218
  if match:
219
- acct = match.group(1)
220
- page_md = page_md.replace('Account Number:', 'Account Number: ' + acct)
221
- return page_md
222
 
223
 
224
  def close_unclosed_html(md):
 
207
  return ""
208
 
209
 
210
+ def fix_account_number(hdr):
211
+ """If hdr contains Account Number: with no value, inject the number from hdr."""
 
212
  import re
213
+ if not hdr:
214
+ return hdr
215
+ if "Account Number:" in hdr and not re.search("Account Number: [0-9]", hdr):
216
+ match = re.search("[0-9]{5,}", hdr)
217
  if match:
218
+ acct = match.group(0)
219
+ hdr = hdr.replace("Account Number:", "Account Number: " + acct)
220
+ return hdr
221
 
222
 
223
  def close_unclosed_html(md):