Update app.py

app.py

@@ -1808,6 +1808,84 @@ def _normalize_daily_balance_table(grid):
     return [new_header] + new_rows
 
 
+def _normalize_checks_paid_table(grid):
+    """
+    Fix OCR artifact in CHECKS PAID SUMMARY tables where:
+    - Header col0 = "DATE CHECK # DATE CHECK # DATE CHECK #" (fused repeating groups)
+    - Remaining headers = "AMOUNT AMOUNT AMOUNT"
+    - Data col0 = "01/24 4701 01/18 4704 * 01/18 916831 *" (all groups fused)
+    - Remaining data cells = amount values
+
+    Expands to proper: DATE | CHECK # | AMOUNT | DATE | CHECK # | AMOUNT | ...
+
+    Detection: first header cell matches pattern (DATE CHECK #)+ 
+    and remaining headers are all AMOUNT.
+    """
+    if not grid or len(grid) < 2:
+        return grid
+
+    ncols = len(grid[0])
+    if ncols < 2:
+        return grid
+
+    first_h = str(grid[0][0] or "").strip()
+    _CHECKS_HDR = re.compile(r"^(DATE\s+CHECK\s+#\s*)+$", re.IGNORECASE)
+    if not _CHECKS_HDR.match(first_h):
+        return grid
+
+    # Remaining headers must be AMOUNT (or empty)
+    rest_h = [str(c or "").strip().upper() for c in grid[0][1:]]
+    if not all(h in ("AMOUNT", "") for h in rest_h if h):
+        return grid
+
+    # Count groups from how many times DATE appears in the fused header
+    n_groups = len(re.findall(r"\bDATE\b", first_h, re.IGNORECASE))
+    if n_groups < 1:
+        return grid
+
+    # Build new header: DATE | CHECK # | AMOUNT repeated n_groups times
+    new_header = []
+    for _ in range(n_groups):
+        new_header.extend(["DATE", "CHECK #", "AMOUNT"])
+
+    _DATE_RE = re.compile(r"^\d{1,2}/\d{2}$")
+
+    new_rows = [new_header]
+    for row in grid[1:]:
+        cells = [str(c or "").strip() for c in row]
+        while len(cells) < ncols:
+            cells.append("")
+
+        fused_data = cells[0]
+        amounts    = cells[1:]
+
+        # Split fused data into groups by date boundary
+        tokens = fused_data.split()
+        groups = []
+        current = []
+        for tok in tokens:
+            if _DATE_RE.match(tok) and current:
+                groups.append(current)
+                current = [tok]
+            else:
+                current.append(tok)
+        if current and current[0]:
+            groups.append(current)
+
+        # Build one output row
+        new_row = [""] * len(new_header)
+        for i, grp in enumerate(groups):
+            if i >= n_groups:
+                break
+            new_row[i * 3]     = grp[0]                       # DATE
+            new_row[i * 3 + 1] = " ".join(grp[1:]).strip()    # CHECK #
+            new_row[i * 3 + 2] = amounts[i] if i < len(amounts) else ""  # AMOUNT
+
+        new_rows.append(new_row)
+
+    return new_rows
+
+
 def normalize_html_tables(text: str) -> str:
     """
     For every <table>...</table>:
@@ -1862,6 +1940,9 @@ def normalize_html_tables(text: str) -> str:
             # Fix 9: normalize DAILY BALANCE SUMMARY tables with stacked dates
             grid = _normalize_daily_balance_table(grid)
 
+            # Fix 10: normalize CHECKS PAID SUMMARY repeating-group tables
+            grid = _normalize_checks_paid_table(grid)
+
             # Fix 6: merge split rows where description wraps to next line
             # (Hardin County Bank and similar monospace statement formats)
             grid = _merge_split_rows(grid)