Update app.py

app.py

@@ -2682,6 +2682,20 @@ def run_ocr(uploaded_file):
 
             parts = []
 
+            # Fix NF-1: Navy Federal full text-layer parse.
+            # Must run FIRST — before header/OCR/footer — so that when NFCU is
+            # detected, the entire page output is replaced with clean text-layer
+            # data and none of the garbled OCR pipeline runs at all.
+            # Safe no-op for all other banks (_parse_nfcu_page returns "").
+            if is_pdf:
+                try:
+                    nfcu_html = _parse_nfcu_page(path, page_num)
+                    if nfcu_html:
+                        all_pages.append(nfcu_html)
+                        continue  # skip ALL OCR processing for this page
+                except Exception:
+                    pass  # fall through to normal OCR pipeline on any error
+
             # Header inclusion: PDF text -> band words -> OCR band
             hdr = ""
             if is_pdf:
@@ -2732,22 +2746,6 @@ def run_ocr(uploaded_file):
 
                 parts.append(stabilized)
 
-            # Fix NF-1: full text-layer parse for Navy Federal statements.
-            # GLM-OCR fails on NFCU pages (summary table, transaction columns,
-            # multi-line descriptions). When NFCU is detected, replace the OCR
-            # body with a clean text-layer extraction and clear the parts list
-            # to avoid mixing garbled OCR output with correct text-layer data.
-            # Safe no-op for all non-NFCU PDFs (_parse_nfcu_page returns "").
-            if is_pdf:
-                try:
-                    nfcu_html = _parse_nfcu_page(path, page_num)
-                    if nfcu_html:
-                        # Replace ALL parts with the text-layer result —
-                        # discard garbled OCR header/body/footer for this page
-                        parts = [nfcu_html]
-                except Exception:
-                    pass  # safe no-op — falls back to OCR output on any error
-
             # Footer extraction: PDF text layer -> band words -> OCR crop
             # Deduplication guard prevents double-printing when body OCR already captured it.
             if ENABLE_FOOTER_OCR and page_num < len(page_images):