Update app.py

app.py

@@ -47,6 +47,26 @@ print("✅ result_formatter.py fixed")
 
 ABANDON = set(config["pipeline"]["result_formatter"]["abandon"])
 
+def _get_regions(page_item):
+    """Get list of regions from one page. Page can be list of regions or dict with 'regions' key."""
+    if isinstance(page_item, list):
+        return page_item
+    if isinstance(page_item, dict):
+        return page_item.get("regions", page_item.get("content", [page_item]))
+    return []
+
+def _get_content(region):
+    """Get text from a region. Supports 'content', 'text', 'raw'."""
+    if not isinstance(region, dict):
+        return str(region)
+    return str(region.get("content") or region.get("text") or region.get("raw") or "")
+
+def _get_label(region):
+    """Get label from a region."""
+    if not isinstance(region, dict):
+        return "text"
+    return region.get("label") or region.get("type") or "text"
+
 # ── STEP 3: OCR via Zhipu MaaS (no local model) ─────────────────────────────
 def run_ocr(uploaded_file):
     if uploaded_file is None:
@@ -55,7 +75,7 @@ def run_ocr(uploaded_file):
     api_key = "4570c28bdea5493c9efae9dae68edc66.sGbA9DLlcX1GlvqV"
     if not api_key:
         return (
-            "Error: ZHIPU_API_KEY is not set. Add it in Space Settings → Variables and secrets.",
+            "Error: ZHIPU_API_KEY is not set.",
             "Failed."
         )
 
@@ -64,7 +84,6 @@ def run_ocr(uploaded_file):
 
         path = uploaded_file.name if hasattr(uploaded_file, "name") else str(uploaded_file)
 
-        # PDF → list of image paths (SDK expects images per page)
         if path.lower().endswith(".pdf"):
             import fitz
             doc = fitz.open(path)
@@ -79,8 +98,6 @@ def run_ocr(uploaded_file):
         else:
             result = parse(path)
 
-        # parse() returns a list directly (list of pages, each page = list of regions)
-        # Region: {"label": "header"|"footer"|"text"|..., "content": "...", ...}
         pages_data = result if isinstance(result, list) else []
 
         total_headers = 0
@@ -88,17 +105,16 @@ def run_ocr(uploaded_file):
         all_markdown = []
         all_summary = []
 
-        for page_num, page_regions in enumerate(pages_data):
+        for page_num, page_item in enumerate(pages_data):
+            page_regions = _get_regions(page_item)
             if not isinstance(page_regions, list):
                 page_regions = [page_regions] if page_regions else []
             page_md = []
             page_summary = [f"── PAGE {page_num + 1} of {len(pages_data)} ──"]
 
             for region in page_regions:
-                if not isinstance(region, dict):
-                    continue
-                label = region.get("label", "text")
-                content = str(region.get("content", ""))
+                label = _get_label(region)
+                content = _get_content(region)
 
                 if label in ABANDON:
                     continue