Update app.py

app.py

@@ -40,13 +40,14 @@ else:
 # Prompts are now imported from prompts.py
 
 # ---------------- SUPABASE HELPERS ---------------- 
-def upload_file_to_supabase(local_path, file_type="unknown"):
+def upload_file_to_supabase(local_path, file_type="unknown", timestamp=None):
     """
     Upload a file to Supabase Storage.
     
     Args:
         local_path (str): Local file path
         file_type (str): Type of file (qp, ms, ans, graded, imprinted)
+        timestamp (str): Unix timestamp for folder organization (optional)
     
     Returns:
         str: Public URL of uploaded file or None if upload failed
@@ -56,9 +57,12 @@ def upload_file_to_supabase(local_path, file_type="unknown"):
         return None
     
     try:
-        timestamp = str(int(time.time()))
+        if timestamp is None:
+            timestamp = str(int(time.time()))
+        
         original_name = os.path.basename(local_path)
-        remote_path = f"{timestamp}/{file_type}_{original_name}"
+        # Use original filename without prefix for cleaner storage
+        remote_path = f"{timestamp}/{original_name}"
         
         print(f"📤 Uploading {file_type} to Supabase: {remote_path}")
         
@@ -79,7 +83,7 @@ def upload_file_to_supabase(local_path, file_type="unknown"):
 
 def process_and_upload_input_files(qp_file_obj, ms_file_obj, ans_file_obj):
     """
-    Process uploaded files and upload them to Supabase.
+    Process uploaded files and upload them to Supabase using a shared timestamp.
     
     Args:
         qp_file_obj: Gradio file object for Question Paper
@@ -87,12 +91,16 @@ def process_and_upload_input_files(qp_file_obj, ms_file_obj, ans_file_obj):
         ans_file_obj: Gradio file object for Answer Sheet
     
     Returns:
-        tuple: (qp_path, ms_path, ans_path, upload_urls_dict)
+        tuple: (qp_path, ms_path, ans_path, upload_urls_dict, timestamp)
     """
     print("\n" + "="*60)
     print("📁 PROCESSING INPUT FILES")
     print("="*60)
     
+    # Generate single timestamp for this entire run
+    run_timestamp = str(int(time.time()))
+    print(f"🕐 Run timestamp: {run_timestamp}")
+    
     upload_urls = {
         "qp_url": None,
         "ms_url": None,
@@ -104,18 +112,18 @@ def process_and_upload_input_files(qp_file_obj, ms_file_obj, ans_file_obj):
     ms_path = ms_file_obj.name if ms_file_obj else None
     ans_path = ans_file_obj.name if ans_file_obj else None
     
-    # Upload to Supabase if configured
+    # Upload to Supabase if configured (all files use same timestamp)
     if supabase_client:
         if qp_path:
-            upload_urls["qp_url"] = upload_file_to_supabase(qp_path, "qp")
+            upload_urls["qp_url"] = upload_file_to_supabase(qp_path, "qp", run_timestamp)
         if ms_path:
-            upload_urls["ms_url"] = upload_file_to_supabase(ms_path, "ms")
+            upload_urls["ms_url"] = upload_file_to_supabase(ms_path, "ms", run_timestamp)
         if ans_path:
-            upload_urls["ans_url"] = upload_file_to_supabase(ans_path, "ans")
+            upload_urls["ans_url"] = upload_file_to_supabase(ans_path, "ans", run_timestamp)
     
     print("="*60 + "\n")
     
-    return qp_path, ms_path, ans_path, upload_urls 
+    return qp_path, ms_path, ans_path, upload_urls, run_timestamp 
 
 
 
@@ -1051,9 +1059,17 @@ def extract_pdf_pages_as_images(pdf_path, page_numbers, prefix):
     return out_paths
 
 # ---------------- PIPELINE ----------------
-def align_and_grade_pipeline(qp_path, ms_path, ans_path, subject="Maths", imprint=False):
+def align_and_grade_pipeline(qp_path, ms_path, ans_path, subject="Maths", imprint=False, run_timestamp=None):
     """
     Final pipeline with graph-aware grading logic using NEW SDK.
+    
+    Args:
+        qp_path: Path to Question Paper PDF
+        ms_path: Path to Markscheme PDF
+        ans_path: Path to Answer Sheet PDF
+        subject: Subject name (Maths or Science)
+        imprint: Whether to generate imprinted PDF
+        run_timestamp: Unix timestamp for organizing files in Supabase
     """
     try:
         print("🔁 Starting pipeline...")
@@ -1146,7 +1162,7 @@ def align_and_grade_pipeline(qp_path, ms_path, ans_path, subject="Maths", imprin
             imprinted_pdf_path = imprint_marks_using_mapping(ans_path, grading_json, imprinted_pdf_path, extracted_ids)
             print("✅ Imprinting finished. Imprinted PDF at:", imprinted_pdf_path)
 
-        # Upload output files to Supabase
+        # Upload output files to Supabase (using same timestamp as input files)
         output_urls = {
             "graded_pdf_url": None,
             "imprinted_pdf_url": None
@@ -1155,9 +1171,9 @@ def align_and_grade_pipeline(qp_path, ms_path, ans_path, subject="Maths", imprin
         if supabase_client:
             print("\n📤 Uploading output files to Supabase...")
             if grading_pdf_path:
-                output_urls["graded_pdf_url"] = upload_file_to_supabase(grading_pdf_path, "graded")
+                output_urls["graded_pdf_url"] = upload_file_to_supabase(grading_pdf_path, "graded", run_timestamp)
             if imprinted_pdf_path:
-                output_urls["imprinted_pdf_url"] = upload_file_to_supabase(imprinted_pdf_path, "imprinted")
+                output_urls["imprinted_pdf_url"] = upload_file_to_supabase(imprinted_pdf_path, "imprinted", run_timestamp)
 
         print("🏁 Pipeline finished successfully.")
         return qpms_text, as_text, grading_text, grading_pdf_path, imprinted_pdf_path, output_urls
@@ -1215,20 +1231,20 @@ with gr.Blocks(title="AI Grading (Pandoc + pdflatex)") as demo:
             else:
                 return error_msg, "", "", None, None
         
-        # Process and upload input files
-        qp_path, ms_path, ans_path, input_urls = process_and_upload_input_files(
+        # Process and upload input files (generates shared timestamp)
+        qp_path, ms_path, ans_path, input_urls, run_timestamp = process_and_upload_input_files(
             qp_file_obj, ms_file_obj, ans_file_obj
         )
 
-        # Run the grading pipeline
+        # Run the grading pipeline (pass timestamp to keep all files together)
         qpms_text, as_text, grading_text, grading_pdf_path, imprinted_pdf_path, output_urls = align_and_grade_pipeline( 
-            qp_path, ms_path, ans_path, subject=subject_choice, imprint=imprint_flag 
+            qp_path, ms_path, ans_path, subject=subject_choice, imprint=imprint_flag, run_timestamp=run_timestamp
         )
         
         # Build URLs summary
         urls_summary = ""
         if supabase_client:
-            urls_summary = "📤 UPLOADED FILES:\n\n"
+            urls_summary = f"📤 UPLOADED FILES (Timestamp: {run_timestamp}):\n\n"
             urls_summary += "INPUT FILES:\n"
             if input_urls.get("qp_url"):
                 urls_summary += f"• Question Paper: {input_urls['qp_url']}\n"
@@ -1243,6 +1259,8 @@ with gr.Blocks(title="AI Grading (Pandoc + pdflatex)") as demo:
             if output_urls.get("imprinted_pdf_url"):
                 urls_summary += f"• Imprinted PDF: {output_urls['imprinted_pdf_url']}\n"
             
+            urls_summary += f"\n📁 All files stored in: examfiles/{run_timestamp}/\n"
+            
             if not any(input_urls.values()) and not any(output_urls.values()):
                 urls_summary += "\n⚠️ No files were uploaded to Supabase"