Update app.py

app.py

@@ -1074,6 +1074,80 @@ def _extract_textlayer_rows(pdf_path: str, page_num: int):
         return []
 
 
+def _extract_jb_summary_sections(text_layer: str, needs_checks: bool = True, needs_dab: bool = True) -> str:
+    """
+    Extract Johnson Bank Checks and Daily Account Balance sections from the PDF
+    text layer and return them as HTML tables.
+
+    Only fires when MM-DD date format is detected (Johnson Bank unique identifier).
+    needs_checks / needs_dab flags prevent re-extraction when already present.
+    """
+    if not text_layer:
+        return ""
+
+    _CHK_SEC = re.compile(r"Checks\s*\(", re.IGNORECASE)
+    _DAB_SEC = re.compile(r"^Daily\s+Account\s+Balance\s*$", re.IGNORECASE | re.MULTILINE)
+    _CHK_ROW = re.compile(r"^(\d{2}-\d{2})\s+(\*?\d+)\s+(\d{1,3}(?:,\d{3})*\.\d{2})\s*$")
+    _BAL_ROW = re.compile(r"^(\d{2}-\d{2})\s+(\d{1,3}(?:,\d{3})*\.\d{2})\s*$")
+
+    lines  = text_layer.splitlines()
+    output = []
+
+    # ── Checks section ─────────────────────────────────────────────────────
+    if needs_checks:
+        checks = []
+        in_checks = False
+        for line in lines:
+            stripped = line.strip()
+            if _CHK_SEC.search(stripped):
+                in_checks = True
+                continue
+            if in_checks:
+                if _DAB_SEC.match(stripped):
+                    break
+                m = _CHK_ROW.match(stripped)
+                if m:
+                    checks.append((m.group(1), m.group(2), m.group(3)))
+
+        if checks:
+            rows_html = "\n".join(
+                "<tr><td>" + d + "</td><td>" + n + "</td><td>" + a + "</td></tr>"
+                for d, n, a in checks
+            )
+            output.append("Checks")
+            output.append(
+                "<table>\n<tr><th>Date</th><th>Number</th><th>Amount</th></tr>\n"
+                + rows_html + "\n</table>"
+            )
+
+    # ── Daily Account Balance section ───────────────────────────────────────
+    if needs_dab:
+        balances = []
+        in_bal = False
+        for line in lines:
+            stripped = line.strip()
+            if _DAB_SEC.match(stripped):
+                in_bal = True
+                continue
+            if in_bal:
+                m = _BAL_ROW.match(stripped)
+                if m:
+                    balances.append((m.group(1), m.group(2)))
+
+        if balances:
+            rows_html = "\n".join(
+                "<tr><td>" + d + "</td><td>" + b + "</td></tr>"
+                for d, b in balances
+            )
+            output.append("Daily Account Balance")
+            output.append(
+                "<table>\n<tr><th>Date</th><th>Balance</th></tr>\n"
+                + rows_html + "\n</table>"
+            )
+
+    return "\n\n".join(output)
+
+
 def _patch_ocr_with_textlayer(page_md: str, pdf_path: str, page_num: int) -> str:
     """
     Compare OCR table rows against PDF text-layer rows and inject any rows
@@ -1301,6 +1375,16 @@ def _patch_ocr_with_textlayer(page_md: str, pdf_path: str, page_num: int) -> str
                 log.info("page %d: Case C — new table (%d rows) from text layer",
                          page_num, len(_cnew_rows))
 
+        # Fix 4D: append Johnson Bank Checks + Daily Account Balance
+        # if missing from OCR but present in PDF text layer.
+        needs_checks = "checks" not in result.lower() or "<th>Number</th>" not in result
+        needs_dab    = "daily account balance" not in result.lower()
+        if needs_checks or needs_dab:
+            raw_tl = extract_zone_text_pdf(pdf_path, page_num, 0.0, 1.0) or ""
+            jb_extra = _extract_jb_summary_sections(raw_tl, needs_checks, needs_dab)
+            if jb_extra:
+                result = result.rstrip() + "\n\n" + jb_extra
+
         return result
 
     except Exception as e: