Update app.py

app.py

@@ -534,6 +534,20 @@ def align_and_grade_pipeline(qp_path, ms_path, ans_path, imprint=False):
         ms_graph_mapping = extract_graph_questions_from_ms(qpms_text)
         print("🖼️ Graph-expected questions in MS:", ms_graph_mapping)
 
+        # NEW: Separate out graph-expected pages as images
+        if ms_graph_mapping:
+            print("📤 Separating graph-expected pages as images...")
+            ms_pages = convert_from_path(merged_qpms_path, dpi=200)
+            for qnum, page_num in ms_graph_mapping.items():
+                # Page numbers in PDF are 1-indexed
+                if 1 <= page_num <= len(ms_pages):
+                    img = ms_pages[page_num-1]
+                    img_path = f"graph_q{qnum}_p{page_num}.png"
+                    img.save(img_path)
+                    print(f"✅ Saved graph image for Question {qnum} (Page {page_num}) as {img_path}")
+                else:
+                    print(f"⚠️ Page {page_num} for Question {qnum} is out of range (PDF has {len(ms_pages)} pages)")
+
         # Step 2: extract serial numbers (question IDs) using regex from qpms_text
         extracted_ids = extract_question_ids_from_qpms(qpms_text)
         if not extracted_ids:
@@ -562,32 +576,6 @@ def align_and_grade_pipeline(qp_path, ms_path, ans_path, imprint=False):
                 })
         print("🔗 Graph bundles for grading:", graph_bundles)
 
-        # --- NEW: Separate and save graph pages as images ---
-        ms_images = convert_from_path(ms_path, dpi=200)
-        ans_images = convert_from_path(ans_path, dpi=200)
-        graph_image_paths = []  # To collect for grading prompt if needed
-        for bundle in graph_bundles:
-            q = bundle['question']
-            ms_page = bundle['ms_page']
-            as_page = bundle['as_page']
-            # Markscheme graph page
-            ms_img_path = f"ms_graph_Q{q}.png"
-            try:
-                ms_images[ms_page-1].save(ms_img_path)
-                print(f"🖼️ Separated Markscheme graph page for Q{q}: source=MS, page={ms_page}, output={ms_img_path}")
-                graph_image_paths.append((f"ms_graph_Q{q}", ms_img_path))
-            except Exception as e:
-                print(f"❌ Failed to save Markscheme graph page for Q{q}: {e}")
-            # Answer Sheet graph page
-            as_img_path = f"as_graph_Q{q}.png"
-            try:
-                ans_images[as_page-1].save(as_img_path)
-                print(f"🖼️ Separated Answer Sheet graph page for Q{q}: source=AS, page={as_page}, output={as_img_path}")
-                graph_image_paths.append((f"as_graph_Q{q}", as_img_path))
-            except Exception as e:
-                print(f"❌ Failed to save Answer Sheet graph page for Q{q}: {e}")
-        # --- END NEW ---
-
         # Step 4: Grading - send both transcripts to grading model, inject graph bundle info
         print("2) Preparing grading input and sending to Gemini for grading...")
         grading_input = (
@@ -607,15 +595,7 @@ def align_and_grade_pipeline(qp_path, ms_path, ans_path, imprint=False):
             grading_input += graph_note
 
         grading_prompt_system = PROMPTS["GRADING_PROMPT"]["content"]
-        # --- NEW: Attach graph images to grading prompt if graph bundles exist ---
-        if graph_bundles and graph_image_paths:
-            print(f"📎 Attaching {len(graph_image_paths)} graph images to grading prompt for visual grading.")
-            # If gemini_generate_content supports image_obj list, pass it here. Otherwise, just print for now.
-            # grading_text = gemini_generate_content(model, grading_prompt_system + "\n\nPlease grade the following transcripts:\n" + grading_input, image_obj=[p[1] for p in graph_image_paths])
-            grading_text = gemini_generate_content(model, grading_prompt_system + "\n\nPlease grade the following transcripts:\n" + grading_input)
-        else:
-            grading_text = gemini_generate_content(model, grading_prompt_system + "\n\nPlease grade the following transcripts:\n" + grading_input)
-        # --- END NEW ---
+        grading_text = gemini_generate_content(model, grading_prompt_system + "\n\nPlease grade the following transcripts:\n" + grading_input)
         print("🧾 Grading output received. Saving debug file: debug_grading.md")
         with open("debug_grading.md", "w", encoding="utf-8") as f:
             f.write(grading_text)