Update app.py

app.py

@@ -24,6 +24,7 @@ Primary goals for reconcile rate:
 - Fix 13: collapse Date|Description|Amount|Description|Amount tables to 3 columns; fix fused
             Beginning/Ending balance rows (UCB-style wide summary tables)
 - Fix 14: UCB 3-col tables where Beginning Balance amount is fused into description and Amount is $0.00
+- Fix 15: UCB (and similar) 3-col rows where Date+Description are fused in col1 and Amount sits in col2 with col3 empty
 - Post-pass: drop later duplicate subset Date|Description|Amount tables (repeated deposit blocks)
 """
 
@@ -2863,6 +2864,60 @@ def _normalize_ucb_three_col_beginning_balance_fusion(grid):
     return out if changed else grid
 
 
+def _fix_three_col_date_desc_amount_left_shift(grid):
+    """
+    Fix 15: OCR shifts transaction rows one column left:
+      col0 = "02/24/2025 ITM DEPOSIT", col1 = "$1,440.00", col2 = empty
+    Expected: Date | Description | Amount
+    Activates only on Date|Description|Amount headers and strict money in col1 with empty col2.
+    """
+    if not grid or len(grid) < 2:
+        return grid
+
+    def _tight(s):
+        return re.sub(r"\s+", "", str(s or "").strip().lower())
+
+    h0 = [str(c or "").strip() for c in grid[0]]
+    if len(h0) < 3:
+        return grid
+    if _tight(h0[0]) != "date" or _tight(h0[1]) != "description" or _tight(h0[2]) != "amount":
+        return grid
+
+    date_then_desc = re.compile(
+        r"^(\d{1,2}/\d{1,2}/\d{2,4})\s+(.+)$"
+    )
+    money_re = re.compile(r"^-?\$?\d{1,3}(?:,\d{3})*(?:\.\d{2})-?$")
+
+    out = [list(grid[0])]
+    changed = False
+    for row in grid[1:]:
+        cells = [str(c or "").strip() for c in row]
+        while len(cells) < 3:
+            cells.append("")
+        c0, c1, c2 = cells[0], cells[1], cells[2]
+        if not c0:
+            out.append([c0, c1, c2])
+            continue
+        m = date_then_desc.match(c0)
+        if not m:
+            out.append([c0, c1, c2])
+            continue
+        desc_part = (m.group(2) or "").strip()
+        if not desc_part:
+            out.append([c0, c1, c2])
+            continue
+        if not (c1 and money_re.match(c1.strip())):
+            out.append([c0, c1, c2])
+            continue
+        if c2 and str(c2).strip():
+            out.append([c0, c1, c2])
+            continue
+        out.append([m.group(1).strip(), desc_part, c1.strip()])
+        changed = True
+
+    return out if changed else grid
+
+
 def _normalize_fused_date_posted_amount_table(grid):
     """
     Normalize 2-column OCR tables with fused header like:
@@ -2995,6 +3050,8 @@ def normalize_html_tables(text: str) -> str:
             grid = _normalize_double_desc_amount_header_table(grid)
             # Fix 14: same bank, 3-col OCR fused Beginning Balance + $0.00 amount column.
             grid = _normalize_ucb_three_col_beginning_balance_fusion(grid)
+            # Fix 15: "Deposits (continued)" etc. — date+desc in col1, amount in col2, empty col3.
+            grid = _fix_three_col_date_desc_amount_left_shift(grid)
 
             grid = _drop_truly_empty_columns(grid)
             grid = _merge_blank_header_text_columns(grid)