Update app.py

app.py

@@ -208,15 +208,27 @@ def ocr_zone(image_path, y_start_frac, y_end_frac):
 
 
 def fix_account_number(hdr):
-    """If hdr contains Account Number: with no value, inject the number from hdr."""
+    """Fix header text:
+    1. If Account Number: has no value, find the number and inject it.
+    2. Remove any leading standalone number that is duplicated elsewhere in hdr.
+    """
     import re
     if not hdr:
         return hdr
+    # Step 1: inject account number if missing after label
     if "Account Number:" in hdr and not re.search("Account Number: [0-9]", hdr):
         match = re.search("[0-9]{5,}", hdr)
         if match:
             acct = match.group(0)
             hdr = hdr.replace("Account Number:", "Account Number: " + acct)
+    # Step 2: remove leading number that is duplicated after Account Number:
+    # e.g. "000000731823008 JPMorgan ... Account Number: 000000731823008"
+    # -> "JPMorgan ... Account Number: 000000731823008"
+    acct_match = re.search("Account Number:\s*([0-9]{5,})", hdr)
+    if acct_match:
+        acct = acct_match.group(1)
+        # If hdr starts with that same number, remove it from the start
+        hdr = re.sub("^" + re.escape(acct) + "\s*", "", hdr)
     return hdr