Update app.py

app.py

@@ -123,8 +123,11 @@ Primary goals for reconcile rate:
             | Amount | Amount | (date+desc | amount | balance) or a four-column variant (date+desc
             fragment | desc tail | amount | balance). CONT may be omitted by OCR. Downstream mis-maps
             amount vs balance. Normalize to Date | Transaction Description | Transaction Amount | Balance.
-- Fix DA4-dedupe: In 4-column Date|…|Amount|Balance registers, consecutive rows identical on all four
-            cells (same running balance) are OCR stutters — collapse to one; honour text-layer counts when present.
+- Fix DA4-dedupe: In 4-column registers whose amount header is SEFCU-style (contains both \"transaction\"
+            and \"amount\"), consecutive rows identical on all four cells are OCR stutters — collapse to one;
+            honour text-layer counts when present. Skips tables with plain \"Amount\" / \"Balance($)\" headers.
+- Fix SE-CONT banner guard: reject cells containing product words (checking, savings, inspire, advantage, …)
+            so Amegy / BofA marketing rows and Navy \"Business Checking-…\" continuations are never rewritten.
 - Fix MD-1: Markdown fidelity — (a) collapse adjacent HTML tables that are the same register
             (identical multiset of date + debit + credit + balance per row, descriptions may differ);
             keep the table with richer description text so output matches the PDF once. (b) Common OCR
@@ -2985,9 +2988,22 @@ _BUSINESS_COOP_BANNER_FIRST_CELL_RE = re.compile(
 )
 
 
+_BUSINESS_COOP_BANNER_REJECT_RE = re.compile(
+    r"\b(checking|savings|inspire|advantage|mbr|premier|relationship)\b",
+    re.IGNORECASE,
+)
+
+
 def _first_cell_is_business_coop_banner(s: str) -> bool:
+    """
+    True only for CO-OP-style account banner cells (BUSINESS - <digits> [CONT]).
+    Reject product lines like 'BUSINESS INSPIRE CHECKING …', 'Business Checking-…',
+    or 'BUSINESS ADVANTAGE' so wide summary / Navy / Amegy tables are not rewritten.
+    """
     t = str(s or "").strip()
-    return bool(t and _BUSINESS_COOP_BANNER_FIRST_CELL_RE.match(t))
+    if not t or _BUSINESS_COOP_BANNER_REJECT_RE.search(t):
+        return False
+    return bool(_BUSINESS_COOP_BANNER_FIRST_CELL_RE.match(t))
 
 
 def _fix_business_coop_cont_three_col_register_grid(grid):
@@ -3007,9 +3023,11 @@ def _fix_business_coop_cont_three_col_register_grid(grid):
     scan = min(5, len(grid))
     for ri in range(scan):
         cells = [str(c or "").strip() for c in grid[ri]]
+        while len(cells) > 3 and cells[-1] == "":
+            cells.pop()
         while len(cells) < 3:
             cells.append("")
-        if len(cells) < 3:
+        if len(cells) != 3:
             continue
         if _first_cell_is_business_coop_banner(cells[0]) and _tight_cell(
             cells[1]
@@ -3094,13 +3112,13 @@ def _fix_business_coop_cont_four_col_register_grid(grid):
     hdr_idx = None
     for ri in range(min(6, len(grid))):
         cells = [str(c or "").strip() for c in grid[ri]]
-        while len(cells) < ncols:
-            cells.append("")
-        if len(cells) < 4:
+        while len(cells) > 4 and cells[-1] == "":
+            cells.pop()
+        if len(cells) != 4:
             continue
         if not _first_cell_is_business_coop_banner(cells[0]):
             continue
-        if _tight_cell(cells[-2]) != "amount" or _tight_cell(cells[-1]) != "amount":
+        if _tight_cell(cells[2]) != "amount" or _tight_cell(cells[3]) != "amount":
             continue
         hdr_idx = ri
         break
@@ -3856,7 +3874,9 @@ def _is_da4_date_desc_amount_balance_header(grid):
         return False
     if "description" not in h[1]:
         return False
-    if "amount" not in h[2]:
+    # Require "transaction" in the amount header (SEFCU-style) so fee / ledger /
+    # daily tables with plain "Amount" or "Balance($)" are not deduped.
+    if "transaction" not in h[2] or "amount" not in h[2]:
         return False
     if "balance" not in h[3]:
         return False