Update app.py

app.py

@@ -26,6 +26,8 @@ Primary goals for reconcile rate:
 - 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
 - Fix 16: remove duplicate rows within a single Date|Description|Amount table; multi-pass table dedupe + final doc-wide dedupe
+- Fix 17: UCB page 3 — split one merged table under Deposits (continued) into Deposits + Electronic Credits +
+            Electronic Debits (PDF text order); drop orphan empty EC/ED headers that followed the merged table
 - Post-pass: dedupe 3-col Date|Description|Amount tables when one is a duplicate fragment
             (later subset of earlier, earlier subset of later, or identical row sets)
 """
@@ -3154,6 +3156,116 @@ def _dedupe_duplicate_rows_in_da3_table(grid):
     return out if len(out) < len(grid) else grid
 
 
+def _row_text_full_ucb(grid, r: int) -> str:
+    """Lowercased join of all cells in row (for UCB section split heuristics)."""
+    if not grid or r < 0 or r >= len(grid):
+        return ""
+    parts = [str(c or "").strip().lower() for c in grid[r]]
+    return " ".join(parts)
+
+
+def _is_itm_deposit_row_ucb(grid, r: int) -> bool:
+    if not grid or r >= len(grid) or len(grid[r]) < 2:
+        return False
+    desc = str(grid[r][1] or "").strip().lower()
+    return "itm deposit" in desc
+
+
+def _ucb_verify_electronic_credits_rows(grid, ec_start: int) -> bool:
+    """POS Return, Acct Fund (Apple instant), DDA Transfer IN — PDF Electronic Credits block."""
+    if ec_start + 2 >= len(grid):
+        return False
+    t0 = _row_text_full_ucb(grid, ec_start)
+    t1 = _row_text_full_ucb(grid, ec_start + 1)
+    t2 = _row_text_full_ucb(grid, ec_start + 2)
+    ok0 = "pos return" in t0 or ("pos" in t0 and "return" in t0)
+    ok1 = "acct fund" in t1 or ("apple" in t1 and "inst" in t1)
+    ok2 = "dda transfer" in t2 and "in" in t2
+    return bool(ok0 and ok1 and ok2)
+
+
+def _ucb_split_grid_deposits_ec_ed(grid):
+    """
+    Return (grid_dep, grid_ec, grid_ed) each with same Date|Description|Amount header,
+    or None if this does not look like the merged UCB page-3 pattern.
+    """
+    if not grid or len(grid) < 2:
+        return None
+    if not _is_da3_header(grid):
+        return None
+    data_start = 1
+    i = data_start
+    while i < len(grid) and _is_itm_deposit_row_ucb(grid, i):
+        i += 1
+    if i == data_start:
+        return None
+    if i + 3 > len(grid):
+        return None
+    if not _ucb_verify_electronic_credits_rows(grid, i):
+        return None
+    ec_end = i + 3
+    if ec_end >= len(grid):
+        return None
+    hdr = [list(grid[0])]
+    g_dep = hdr + grid[data_start:i]
+    g_ec = hdr + grid[i:ec_end]
+    g_ed = hdr + grid[ec_end:]
+    if len(g_ed) <= 1:
+        return None
+    return g_dep, g_ec, g_ed
+
+
+def _try_split_ucb_deposits_table_to_three(prefix: str, table_html: str):
+    """
+    Build HTML: Deposits (continued) + table, Electronic Credits + table, Electronic Debits + table.
+    Returns None if split heuristics do not match.
+    """
+    try:
+        p = TableGridParser()
+        p.feed(table_html)
+        grid = _build_grid(p.rows)
+    except Exception:
+        return None
+    triple = _ucb_split_grid_deposits_ec_ed(grid)
+    if not triple:
+        return None
+    g_dep, g_ec, g_ed = triple
+    sect = '<div align="center">\n\n{title}\n\n</div>\n\n'
+    parts = [prefix.rstrip(), _grid_to_html(g_dep), sect.format(title="Electronic Credits"), _grid_to_html(g_ec)]
+    parts.append(sect.format(title="Electronic Debits"))
+    parts.append(_grid_to_html(g_ed))
+    return "\n\n".join(parts)
+
+
+_UCB_DEP_CONT_BLOCK = re.compile(
+    r'(<div\s+align=["\']center["\']\s*>\s*Deposits\s*\(continued\)\s*</div>\s*)'
+    r'(<table[^>]*>.*?</table>)'
+    r'(?:\s*<div\s+align=["\']center["\']\s*>\s*Electronic\s+Credits\s*</div>\s*'
+    r'<div\s+align=["\']center["\']\s*>\s*Electronic\s+Debits\s*</div>)?',
+    re.DOTALL | re.IGNORECASE,
+)
+
+
+def _split_ucb_deposits_electronic_sections_html(text: str) -> str:
+    """
+    Fix 17: Split merged UCB Deposits (continued) table into Deposits + Electronic Credits + Electronic Debits.
+    Removes orphan empty Electronic Credits / Electronic Debits div pairs when replaced by real tables.
+    Safe no-op when heuristics do not match.
+    """
+    if not text or "deposits (continued)" not in text.lower():
+        return text
+
+    def _repl(m):
+        prefix = m.group(1)
+        table_html = m.group(2)
+        new = _try_split_ucb_deposits_table_to_three(prefix, table_html)
+        if new is None:
+            return m.group(0)
+        return new
+
+    return _UCB_DEP_CONT_BLOCK.sub(_repl, text)
+
+
 def _dedupe_subset_transaction_tables_html(text: str) -> str:
     """
     Remove duplicate 3-column Date|Description|Amount tables:
@@ -3285,6 +3397,7 @@ def stabilize_tables_and_text(page_md: str) -> str:
     stabilized = "\n\n".join(out_blocks)
     stabilized = convert_plaintext_bank_sections(stabilized)
     stabilized = normalize_html_tables(stabilized)
+    stabilized = _split_ucb_deposits_electronic_sections_html(stabilized)
     stabilized = _dedupe_subset_transaction_tables_html(stabilized)
     return close_unclosed_html(stabilized)
 
@@ -3412,6 +3525,7 @@ def run_ocr(uploaded_file):
                     # _patch_ocr_with_textlayer may append Case B rows from the PDF text layer
                     # with fused UCB-style cells; run the same HTML table pass again (Fix 13/14).
                     stabilized = normalize_html_tables(stabilized)
+                    stabilized = _split_ucb_deposits_electronic_sections_html(stabilized)
                     stabilized = _dedupe_subset_transaction_tables_html(stabilized)
                 elif is_pdf and is_nfcu_doc:
                     # Navy-only additive fix: rebuild malformed summary table from
@@ -3491,8 +3605,9 @@ def run_ocr(uploaded_file):
                 all_pages.append("\n\n".join(parts))
 
         merged = "\n\n---page-separator---\n\n".join(all_pages) if all_pages else "(No content)"
-        # Cross-page: same transaction tables split across page boundaries must dedupe once.
+        # Cross-page: UCB section split + same transaction tables dedupe once on full doc.
         if all_pages and merged and not merged.startswith("Error") and "<table" in merged.lower():
+            merged = _split_ucb_deposits_electronic_sections_html(merged)
             merged = _dedupe_subset_transaction_tables_html(merged)
         return merged