Update app.py

app.py

@@ -133,6 +133,9 @@ Primary goals for reconcile rate:
             table when bbox/image markers + repeated check lines are present; skips chunks with YOUR CHECKS SEQUENCED.
 - Fix MD-4: YOUR CHECKS SEQUENCED wide grid (N×DATE|CHECK#|AMOUNT per row) is unfolded to one DATE|CHECK#|AMOUNT
             table so narrow viewers show every check, not only the first date column.
+- Fix FH-1: Drop a leading Date|Description|Amount (DA3) table when its (date, amount) multiset is a
+            strict subset of a later DEPOSIT/WITHDRAWAL register (e.g. First Horizon OCR duplicate before
+            ACCOUNT SUMMARY). Safe no-op when no such pair exists.
 """
 
 # Patch asyncio first (before Gradio imports it) to suppress Python 3.13 cleanup noise
@@ -4523,6 +4526,140 @@ def _dedupe_da3_by_date_amount_loose(text: str) -> str:
         return text
 
 
+def _fh_withdrawal_column_index(grid) -> Optional[int]:
+    for row in grid[: min(6, len(grid))]:
+        for ci, c in enumerate(row):
+            if str(c or "").strip().upper() == "WITHDRAWAL":
+                return ci
+    return None
+
+
+def _fh_is_deposit_withdrawal_register(grid) -> bool:
+    if not grid or len(grid) < 2:
+        return False
+    blob = " ".join(
+        " ".join(str(c or "") for c in row) for row in grid[: min(5, len(grid))]
+    ).upper()
+    return "DEPOSIT" in blob and "WITHDRAWAL" in blob and "DATE" in blob
+
+
+def _drop_da3_duplicate_of_later_deposit_withdrawal_register(text: str) -> str:
+    """
+    Fix FH-1: Vision OCR may emit a spurious DA3 table (subset of withdrawals) before the real
+    ACCOUNT HISTORY grid. Drop the DA3 when every (date, amount) row is covered by a later
+    register's WITHDRAWAL column and that register has strictly more withdrawal rows.
+    """
+    if not text or "<table" not in text.lower():
+        return text
+    try:
+        pattern = re.compile(r"<table[^>]*>.*?</table>", re.DOTALL | re.IGNORECASE)
+        matches = list(pattern.finditer(text))
+        if len(matches) < 2:
+            return text
+
+        def _parse_grid(html: str):
+            try:
+                p = TableGridParser()
+                p.feed(html)
+                return _build_grid(p.rows)
+            except Exception:
+                return None
+
+        def _norm_da_key_date_amt(date_s: str, amt_s: str) -> Tuple[str, str]:
+            d = re.sub(r"\s+", "", str(date_s or "").strip())
+            a = re.sub(r"[\s$,]", "", str(amt_s or "").strip()).lower()
+            a = a.replace("−", "-").replace("\u2013", "-")
+            return (d, a)
+
+        def _frag_counter(g) -> Optional[Counter]:
+            if not g or not _is_da3_header(g):
+                return None
+            c = Counter()
+            for row in g[1:]:
+                cells = [str(x or "").strip() for x in row]
+                while len(cells) < 3:
+                    cells.append("")
+                if not _DA3_LEADING_DATE.match(cells[0]):
+                    continue
+                if not _DA3_STRICT_AMOUNT.match(cells[2]):
+                    continue
+                c[_norm_da_key_date_amt(cells[0], cells[2])] += 1
+            if sum(c.values()) < 2:
+                return None
+            return c
+
+        def _register_wd_counter(g) -> Optional[Counter]:
+            if not _fh_is_deposit_withdrawal_register(g):
+                return None
+            wci = _fh_withdrawal_column_index(g)
+            if wci is None:
+                return None
+            c = Counter()
+            for row in g[1:]:
+                cells = [str(x or "").strip() for x in row]
+                while len(cells) <= wci:
+                    cells.append("")
+                if not _DA3_LEADING_DATE.match(cells[0]):
+                    continue
+                amt = cells[wci]
+                if not amt.strip():
+                    continue
+                if not _DA3_STRICT_AMOUNT.match(amt):
+                    continue
+                c[_norm_da_key_date_amt(cells[0], amt)] += 1
+            if sum(c.values()) < 3:
+                return None
+            return c
+
+        grids = [(m, _parse_grid(m.group(0))) for m in matches]
+        drop_starts = set()
+
+        for i in range(len(grids) - 1):
+            mi, gi = grids[i]
+            if mi.start() in drop_starts:
+                continue
+            fk = _frag_counter(gi)
+            if fk is None:
+                continue
+            n_frag = sum(fk.values())
+            if n_frag > 48:
+                continue
+            for j in range(i + 1, len(grids)):
+                mj, gj = grids[j]
+                rk = _register_wd_counter(gj)
+                if rk is None:
+                    continue
+                n_reg = sum(rk.values())
+                if n_reg <= n_frag:
+                    continue
+                if not all(fk[k] <= rk.get(k, 0) for k in fk):
+                    continue
+                drop_starts.add(mi.start())
+                log.info(
+                    "Fix FH-1: dropped DA3 (%d rows) duplicate of later "
+                    "DEPOSIT/WITHDRAWAL register (%d wd rows)",
+                    n_frag,
+                    n_reg,
+                )
+                break
+
+        if not drop_starts:
+            return text
+
+        out: List[str] = []
+        last = 0
+        for m in matches:
+            out.append(text[last : m.start()])
+            if m.start() not in drop_starts:
+                out.append(m.group(0))
+            last = m.end()
+        out.append(text[last:])
+        return "".join(out)
+    except Exception as e:
+        log.warning("_drop_da3_duplicate_of_later_deposit_withdrawal_register failed: %s", e)
+        return text
+
+
 def _dedupe_subset_transaction_tables_html(text: str) -> str:
     if not text or "<table" not in text.lower():
         return text
@@ -5377,6 +5514,7 @@ def stabilize_tables_and_text(page_md: str) -> str:
     stabilized = _split_ucb_deposits_electronic_sections_html(stabilized)
     stabilized = _patch_da3_msft_g_ref_amounts_across_tables(stabilized)
     stabilized = _dedupe_subset_transaction_tables_html(stabilized)
+    stabilized = _drop_da3_duplicate_of_later_deposit_withdrawal_register(stabilized)  # Fix FH-1
     stabilized = _dedupe_da3_by_date_amount_loose(stabilized)  # Fix 30
     stabilized = _drop_da3_misplaced_debit_after_deposit_heading_tables(stabilized)
     stabilized = _strip_credits_prefix_from_debit_da3_table(stabilized)  # Fix 28
@@ -5514,6 +5652,7 @@ def run_ocr(uploaded_file):
                     stabilized = _split_ucb_deposits_electronic_sections_html(stabilized)
                     stabilized = _patch_da3_msft_g_ref_amounts_across_tables(stabilized)
                     stabilized = _dedupe_subset_transaction_tables_html(stabilized)
+                    stabilized = _drop_da3_duplicate_of_later_deposit_withdrawal_register(stabilized)  # Fix FH-1
                     stabilized = _dedupe_da3_by_date_amount_loose(stabilized)  # Fix 30
                     stabilized = _drop_da3_misplaced_debit_after_deposit_heading_tables(stabilized)
                     stabilized = _strip_credits_prefix_from_debit_da3_table(stabilized)  # Fix 28
@@ -5596,6 +5735,7 @@ def run_ocr(uploaded_file):
             merged = _split_ucb_deposits_electronic_sections_html(merged)
             merged = _patch_da3_msft_g_ref_amounts_across_tables(merged)
             merged = _dedupe_subset_transaction_tables_html(merged)
+            merged = _drop_da3_duplicate_of_later_deposit_withdrawal_register(merged)  # Fix FH-1
             merged = _dedupe_da3_by_date_amount_loose(merged)  # Fix 30
             merged = _drop_da3_misplaced_debit_after_deposit_heading_tables(merged)
             merged = _strip_credits_prefix_from_debit_da3_table(merged)  # Fix 28