Update app.py

app.py

@@ -126,6 +126,8 @@ Primary goals for reconcile rate:
             Gap between tables must not contain another <table> or page-separator.
 - Fix MD-2: Skip blank-header column merge for YOUR CHECKS SEQUENCED–style tables (title + repeated
             Amount columns + blank th between triplets); merge was collapsing multi-column check rows.
+            Also skip when the table is ACCOUNT HISTORY / DEPOSIT+WITHDRAWAL (First Horizon–style): blank
+            <th> align subheaders on the next row; merge was fusing DEPOSIT/WITHDRAWAL/CARD into wrong columns.
 - Fix MD-3: Check image pages — GLM often emits ![](bbox…) plus <div align="center"> fragments
             (e.g. 2293 $6,721.00 4/29/2025). Collapse detected gallery pages into one Check #|Amount|Date
             table when bbox/image markers + repeated check lines are present; skips chunks with YOUR CHECKS SEQUENCED.
@@ -550,6 +552,28 @@ def _is_multigroup_check_list_header(header: List[str]) -> bool:
     return False
 
 
+def _should_skip_blank_header_merge_multiline_register(grid) -> bool:
+    """
+    Some banks (e.g. First Horizon) use a title row with blank <th> cells, then a second header
+    row with DATE DESCRIPTION | DEPOSIT | WITHDRAWAL | CARD #. _merge_blank_header_text_columns
+    would treat those blanks as OCR splits and merge columns — corrupting the grid.
+    """
+    if not grid or len(grid) < 2:
+        return False
+    h0_join = " ".join(str(c or "").strip() for c in grid[0]).upper()
+    if "ACCOUNT HISTORY" in h0_join:
+        return True
+    for ri in range(1, min(5, len(grid))):
+        cells = [str(c or "").strip().upper() for c in grid[ri]]
+        if not any(cells):
+            continue
+        row_join = " ".join(cells)
+        if "DEPOSIT" in row_join and "WITHDRAWAL" in row_join:
+            if "DATE" in row_join or "DESCRIPTION" in row_join:
+                return True
+    return False
+
+
 def _merge_blank_header_text_columns(grid):
     if not grid or len(grid) < 2:
         return grid
@@ -557,6 +581,8 @@ def _merge_blank_header_text_columns(grid):
     header = [str(c or "").strip() for c in grid[0]]
     if _is_multigroup_check_list_header(header):
         return grid
+    if _should_skip_blank_header_merge_multiline_register(grid):
+        return grid
 
     ncols = len(header)
     body = grid[1:]
@@ -5009,6 +5035,16 @@ def _maybe_convert_check_image_gallery_chunk(chunk: str) -> str:
         return chunk
     if "YOUR CHECKS SEQUENCED" in chunk.upper():
         return chunk
+    cu = chunk.upper()
+    # Statement summary / register pages (e.g. First Horizon) may include bbox images; do not
+    # replace the whole chunk with a synthetic check table.
+    if "DAILY BALANCE SUMMARY" in cu or "CHECKING ACCOUNT MONTHLY SUMMARY" in cu:
+        return chunk
+    if "ACCOUNT SUMMARY" in cu and "NEW BALANCE" in cu and "<table" in chunk.lower():
+        return chunk
+    n_tr = len(re.findall(r"<tr\b", chunk, re.IGNORECASE))
+    if n_tr > 24:
+        return chunk
 
     matches = list(_CHECK_IMAGE_GALLERY_LINE_RE.finditer(chunk))
     if len(matches) < 2: