Update app.py

app.py

@@ -126,6 +126,9 @@ 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.
+- 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.
 """
 
 # Patch asyncio first (before Gradio imports it) to suppress Python 3.13 cleanup noise
@@ -4918,6 +4921,74 @@ def _fix29_move_debit_heading_before_orphan_table(text: str) -> str:
         return text
 
 
+_CHECK_IMAGE_GALLERY_LINE_RE = re.compile(
+    r"(?<![\d])(\d{3,5})\s*\$\s*([\d,]+\.\d{2})\s*(\d{1,2}/\d{1,2}/\d{4})",
+    re.IGNORECASE,
+)
+
+
+def _normalize_check_image_gallery_pages(text: str) -> str:
+    """
+    Fix MD-3: Replace GLM check-scan page noise (markdown images with bbox + scattered div/line text)
+    with a single HTML table when the page looks like a check image gallery, not a structured table.
+    """
+    if not text or ("bbox=" not in text and "![](" not in text):
+        return text
+    try:
+        sep = "---page-separator---"
+        parts = text.split(sep)
+        out: List[str] = []
+        for chunk in parts:
+            out.append(_maybe_convert_check_image_gallery_chunk(chunk))
+        return sep.join(out)
+    except Exception as e:
+        log.warning("_normalize_check_image_gallery_pages failed: %s", e)
+        return text
+
+
+def _maybe_convert_check_image_gallery_chunk(chunk: str) -> str:
+    if not chunk or not chunk.strip():
+        return chunk
+    if "YOUR CHECKS SEQUENCED" in chunk.upper():
+        return chunk
+
+    matches = list(_CHECK_IMAGE_GALLERY_LINE_RE.finditer(chunk))
+    if len(matches) < 2:
+        return chunk
+
+    bbox_hits = chunk.count("bbox=")
+    img_hits = len(re.findall(r"!\[\]\s*\(", chunk))
+    if bbox_hits < 2 and img_hits < 2:
+        return chunk
+
+    rows: List[Tuple[str, str, str]] = []
+    seen = set()
+    for m in matches:
+        key = (m.group(1), m.group(2), m.group(3))
+        if key in seen:
+            continue
+        seen.add(key)
+        rows.append(key)
+
+    pos_img = chunk.find("![](")
+    pos_first = matches[0].start()
+    if pos_img == -1:
+        start = pos_first
+    else:
+        start = min(pos_img, pos_first)
+
+    prefix = chunk[:start].rstrip()
+    lines = [
+        "<table>",
+        "<tr><th>Check #</th><th>Amount</th><th>Date</th></tr>",
+    ]
+    for num, amt, dt in rows:
+        lines.append(f"<tr><td>{num}</td><td>{amt}</td><td>{dt}</td></tr>")
+    lines.append("</table>")
+    table_html = "\n".join(lines)
+    return (prefix + "\n\n" + table_html) if prefix else table_html
+
+
 def _strip_check_image_ocr_junk_html(text: str) -> str:
     """
     Fix 31: Remove hallucinated rows on scanned check/receipt regions and normalize amount cells.
@@ -5219,6 +5290,7 @@ def stabilize_tables_and_text(page_md: str) -> str:
     stabilized = _strip_orphan_continued_da3_flatline(stabilized)
     stabilized = _strip_standalone_continued_banner_line(stabilized)
     stabilized = _strip_check_image_ocr_junk_html(stabilized)  # Fix 31
+    stabilized = _normalize_check_image_gallery_pages(stabilized)  # Fix MD-3
     stabilized = _fix_markdown_fidelity_typos_and_escapes(stabilized)
     return close_unclosed_html(stabilized)
 
@@ -5354,6 +5426,7 @@ def run_ocr(uploaded_file):
                     stabilized = _fix29_move_debit_heading_before_orphan_table(stabilized)  # Fix 29
                     stabilized = _strip_standalone_continued_banner_line(stabilized)
                     stabilized = _strip_check_image_ocr_junk_html(stabilized)  # Fix 31
+                    stabilized = _normalize_check_image_gallery_pages(stabilized)  # Fix MD-3
                 elif is_pdf and is_nfcu_doc:
                     nfcu_summary = _extract_nfcu_summary_table(path, page_num)
                     if nfcu_summary:
@@ -5436,6 +5509,7 @@ def run_ocr(uploaded_file):
             merged = _strip_orphan_continued_da3_flatline(merged)
             merged = _strip_standalone_continued_banner_line(merged)
             merged = _strip_check_image_ocr_junk_html(merged)  # Fix 31
+            merged = _normalize_check_image_gallery_pages(merged)  # Fix MD-3
         return merged
 
     except Exception as e: