Update app.py

app.py

@@ -1035,6 +1035,132 @@ def _extract_rows_plain_from_table(full_table: str) -> List[List[str]]:
     return rows
 
 
+def _fmt_money(v: float) -> str:
+    return f"{v:,.2f}"
+
+
+def _is_date_like(s: str) -> bool:
+    t = (s or "").strip()
+    return bool(re.match(r"^(?:\d{1,2}[/-]\d{1,2}(?:[/-]\d{2,4})?)$", t))
+
+
+def normalize_ambiguous_amount_balance_tables(md: str) -> str:
+    """
+    For transaction-like tables that expose only a single Amount column plus
+    Balance, convert to explicit Credit/Debit using generic running-balance
+    inference. This avoids sign ambiguity and improves reconcile stability.
+    """
+    if not md or "<table" not in md.lower():
+        return md
+
+    credit_hint_re = re.compile(
+        r"\b(deposit|deposits|credit|credits|incoming|received|interest earned|refund|return)\b",
+        re.IGNORECASE,
+    )
+    debit_hint_re = re.compile(
+        r"\b(debit|debits|withdrawal|withdrawals|check|checks|fee|fees|charge|charges|payment|transfer out|purchase)\b",
+        re.IGNORECASE,
+    )
+
+    def repl_table(m: re.Match) -> str:
+        full = m.group(0)
+        rows = _extract_rows_plain_from_table(full)
+        if len(rows) < 3:
+            return full
+
+        hdr = [c.strip().lower() for c in rows[0]]
+        amount_idxs = [i for i, c in enumerate(hdr) if "amount" in c]
+        balance_idxs = [i for i, c in enumerate(hdr) if "balance" in c]
+        has_credit = any("credit" in c for c in hdr)
+        has_debit = any("debit" in c for c in hdr)
+        date_idxs = [i for i, c in enumerate(hdr) if "date" in c]
+        desc_idxs = [i for i, c in enumerate(hdr) if "description" in c or "memo" in c or "details" in c]
+
+        if has_credit or has_debit or not amount_idxs or not balance_idxs:
+            return full
+
+        amount_idx = amount_idxs[0]
+        balance_idx = balance_idxs[0]
+        date_idx = date_idxs[0] if date_idxs else 0
+        desc_idx = desc_idxs[0] if desc_idxs else min(1, max(0, len(rows[0]) - 1))
+
+        new_rows: List[List[str]] = [["Date", "Description", "Credit", "Debit", "Balance"]]
+        prev_bal = None
+        section_hint = ""
+        converted = 0
+
+        for r in rows[1:]:
+            if not r:
+                continue
+            padded = r + [""] * (max(amount_idx, balance_idx, date_idx, desc_idx) + 1 - len(r))
+            date_txt = (padded[date_idx] or "").strip()
+            desc_txt = (padded[desc_idx] or "").strip()
+            amt_txt = (padded[amount_idx] or "").strip()
+            bal_txt = (padded[balance_idx] or "").strip()
+            amt = _parse_amount_or_none(amt_txt)
+            bal = _parse_amount_or_none(bal_txt)
+
+            row_blob = " ".join((c or "").strip() for c in padded).strip()
+            if row_blob and not _is_date_like(date_txt) and amt is None:
+                section_hint = row_blob
+                continue
+            if amt is None and bal is None:
+                continue
+
+            direction = ""
+            # Primary: infer from balance delta between consecutive rows.
+            if amt is not None and bal is not None and prev_bal is not None:
+                delta = bal - prev_bal
+                tol = max(0.55, 0.025 * abs(amt))
+                if abs(abs(delta) - abs(amt)) <= tol:
+                    direction = "credit" if delta > 0 else "debit"
+
+            # Secondary: section + description lexical hints.
+            ctx = f"{section_hint} {desc_txt}"
+            if not direction and credit_hint_re.search(ctx):
+                direction = "credit"
+            if not direction and debit_hint_re.search(ctx):
+                direction = "debit"
+
+            credit = ""
+            debit = ""
+            if amt is not None:
+                if direction == "credit":
+                    credit = _fmt_money(abs(amt))
+                elif direction == "debit":
+                    debit = _fmt_money(abs(amt))
+                else:
+                    # Conservative fallback for ambiguous rows: preserve as debit
+                    # to avoid overstating inflows.
+                    debit = _fmt_money(abs(amt))
+
+            balance_out = _fmt_money(bal) if bal is not None else ""
+            new_rows.append([date_txt, desc_txt, credit, debit, balance_out])
+            if bal is not None:
+                prev_bal = bal
+            converted += 1
+
+        if converted < 5:
+            return full
+
+        html_rows = []
+        html_rows.append(
+            "<tr>" + "".join(f"<th>{html.escape(c)}</th>" for c in new_rows[0]) + "</tr>"
+        )
+        for rr in new_rows[1:]:
+            html_rows.append(
+                "<tr>" + "".join(f"<td>{html.escape(c)}</td>" for c in rr) + "</tr>"
+            )
+        return "<table>\n" + "\n".join(html_rows) + "\n</table>"
+
+    return re.sub(
+        r"<table\b[^>]*>.*?</table>",
+        repl_table,
+        md,
+        flags=re.IGNORECASE | re.DOTALL,
+    )
+
+
 def _extract_daily_balance_by_date(md: str):
     """
     Build a date->balance map from daily-balance style tables.
@@ -1545,6 +1671,7 @@ def run_ocr(uploaded_file):
         if merged and merged != "(No content)" and not merged.lstrip().startswith("Error:"):
             merged = mask_non_monetary_long_numbers(merged)
             merged = directionalize_amount_headers(merged)
+            merged = normalize_ambiguous_amount_balance_tables(merged)
             merged = stabilize_table_markup(merged, rounds=4)
             merged = enrich_transaction_tables_with_daily_balances(merged)
             # Keep transaction amounts untouched; add explicit snapshot for metadata extraction.