Update app.py

app.py

@@ -28,6 +28,10 @@ Primary goals for reconcile rate:
 - 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
+- Fix 17b: When markdown has Electronic Credits (+ optional Electronic Debits headings) but a single fused
+            Date|Description|Amount table (common when Deposits (continued) is absent from OCR), split after a
+            leading ITM DEPOSIT run when the next row matches generic electronic-debit patterns (POS purchase,
+            ATM withdrawal, etc.). Safe no-op when there is no ITM prefix or no such debit row after it.
 - 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)
 - Fix 18: adjacent DA3 tables — if the last K data rows of table N match the first K rows of table N+1
@@ -2898,6 +2902,108 @@ def _ucb_split_grid_deposits_ec_ed(grid):
     return g_dep, g_ec, g_ed
 
 
+def _is_strong_ucb_electronic_debit_row(grid, r: int) -> bool:
+    """True for typical card/ACH debit lines (generic; avoids splitting on POS return / transfer-in credits)."""
+    t = _row_text_full_ucb(grid, r)
+    if not t or not t.strip():
+        return False
+    if "pos return" in t:
+        return False
+    if "dda transfer" in t and "in" in t:
+        return False
+    if "pos purchase" in t:
+        return True
+    if "atm withdrawal" in t:
+        return True
+    if "atm" in t and "withdrawal" in t:
+        return True
+    if "debit card" in t:
+        return True
+    if "point of sale" in t:
+        return True
+    if "recur payment" in t:
+        return True
+    return False
+
+
+def _ucb_split_grid_electronic_credits_debits_after_itm_prefix(grid):
+    """
+    Fix 17b: one fused DA3 table under Electronic Credits (+ optional empty EC/ED headings) where OCR
+    merged ITM deposits with POS/ATM debits. Split after the leading ITM DEPOSIT run when the next
+    data row is a strong electronic-debit pattern.
+    """
+    if not grid or len(grid) < 3:
+        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 or i >= len(grid):
+        return None
+    if not _is_strong_ucb_electronic_debit_row(grid, i):
+        return None
+    hdr = [list(grid[0])]
+    g_ec = hdr + grid[data_start:i]
+    g_ed = hdr + grid[i:]
+    if len(g_ec) <= 1 or len(g_ed) <= 1:
+        return None
+    return g_ec, g_ed
+
+
+def _try_split_ucb_ec_ed_from_fused_table_html(lead_before_ec: str, table_html: str):
+    try:
+        p = TableGridParser()
+        p.feed(table_html)
+        grid = _build_grid(p.rows)
+    except Exception:
+        return None
+    pair = _ucb_split_grid_electronic_credits_debits_after_itm_prefix(grid)
+    if not pair:
+        return None
+    g_ec, g_ed = pair
+    sect = '<div align="center">\n\n{title}\n\n</div>\n\n'
+    parts = []
+    if lead_before_ec and lead_before_ec.strip():
+        parts.append(lead_before_ec.rstrip())
+    parts.extend(
+        [
+            sect.format(title="Electronic Credits"),
+            _grid_to_html(g_ec),
+            sect.format(title="Electronic Debits"),
+            _grid_to_html(g_ed),
+        ]
+    )
+    return "\n\n".join(parts)
+
+
+# Optional empty centered banner only; EC/ED title divs are re-emitted after a successful split
+# so we do not duplicate headings.
+_UCB_EC_ED_SINGLE_FUSED_TABLE = re.compile(
+    r'(?P<lead>(?:<div\s+align=["\']center["\']\s*>\s*</div>\s*)?)'
+    r'<div\s+align=["\']center["\']\s*>\s*Electronic\s+Credits\s*</div>\s*'
+    r'(?:<div\s+align=["\']center["\']\s*>\s*Electronic\s+Debits\s*</div>\s*)?'
+    r'(?P<table><table[^>]*>.*?</table>)',
+    re.DOTALL | re.IGNORECASE,
+)
+
+
+def _split_ucb_fused_ec_ed_headings_single_table_html(text: str) -> str:
+    if not text or "electronic credits" not in text.lower():
+        return text
+    if "<table" not in text.lower():
+        return text
+
+    def _repl(m):
+        lead = m.group("lead")
+        table_html = m.group("table")
+        new = _try_split_ucb_ec_ed_from_fused_table_html(lead, table_html)
+        return new if new is not None else m.group(0)
+
+    return _UCB_EC_ED_SINGLE_FUSED_TABLE.sub(_repl, text)
+
+
 def _try_split_ucb_deposits_table_to_three(prefix: str, table_html: str):
     try:
         p = TableGridParser()
@@ -2926,18 +3032,20 @@ _UCB_DEP_CONT_BLOCK = re.compile(
 
 
 def _split_ucb_deposits_electronic_sections_html(text: str) -> str:
-    if not text or "deposits (continued)" not in text.lower():
+    if not text:
         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)
+    if "deposits (continued)" in text.lower():
+
+        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
+
+        text = _UCB_DEP_CONT_BLOCK.sub(_repl, text)
+    return _split_ucb_fused_ec_ed_headings_single_table_html(text)
 
 
 def _parse_da3_grid_from_table_html(table_html: str):