rehan953 commited on
Commit
208b105
·
verified ·
1 Parent(s): 2b7d582

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +12 -1
app.py CHANGED
@@ -2760,8 +2760,19 @@ def _normalize_fused_date_posted_amount_table(grid):
2760
  date = parts[0]
2761
  desc = " ".join(parts[1:]).strip()
2762
 
2763
- # If amount is empty in column 2, try extracting trailing amount from description.
 
2764
  amount = c2
 
 
 
 
 
 
 
 
 
 
2765
  if not amount and desc:
2766
  m = re.search(r"(-?\$?\d{1,3}(?:,\d{3})*(?:\.\d{2})-?)\s*$", desc)
2767
  if m:
 
2760
  date = parts[0]
2761
  desc = " ".join(parts[1:]).strip()
2762
 
2763
+ # Amount may live in any shifted column for malformed OCR tables.
2764
+ # Prefer explicit second-column value, then rightmost strict money token in row.
2765
  amount = c2
2766
+ if not amount:
2767
+ money_cells = []
2768
+ for x in cells[1:]:
2769
+ t = str(x or "").strip()
2770
+ if re.match(r"^-?\$?\d{1,3}(?:,\d{3})*(?:\.\d{2})-?$", t):
2771
+ money_cells.append(t)
2772
+ if money_cells:
2773
+ amount = money_cells[-1]
2774
+
2775
+ # If still empty, try extracting trailing amount from description.
2776
  if not amount and desc:
2777
  m = re.search(r"(-?\$?\d{1,3}(?:,\d{3})*(?:\.\d{2})-?)\s*$", desc)
2778
  if m: