Update app.py

app.py

@@ -43,24 +43,23 @@ def process_contract(uploaded_files, object_name, manual_mappings):
     if not uploaded_files:
         return "❌ No files uploaded.", None, failed_records, "0/0"
 
-    # Debug: Log uploaded files and their raw data (first few bytes for debugging)
-    print(f"Received files (bytes): {len(uploaded_files)} files")
+    # Debug: Log uploaded files and their raw data
+    print(f"Received files (bytes): {len(uploaded_files)} files at {len(uploaded_files)}")
     for i, file_bytes in enumerate(uploaded_files):
-        print(f"File {i} header: {file_bytes[:5]}")
+        print(f"File {i} header: {file_bytes[:5]} - Starting processing")
 
     # Check for poppler-utils
     if not check_poppler():
         return "❌ Error: poppler-utils is not installed or not in PATH. Please install it (e.g., 'sudo apt-get install poppler-utils' on Linux).", None, failed_records, "0/0"
 
-    # Mock Salesforce object fields (replace with dynamic logic later)
+    # Mock Salesforce object fields
     mock_object_fields = ["Name", "Description", "Amount", "Date"] if object_name else []
     total_files = len(uploaded_files)
     processed_files = 0
     results = []
-    ai_result = None  # Initialize to avoid UnboundLocalError
+    ai_result = None
     with tqdm(total=total_files, desc="Processing PDFs") as pbar:
         for i, file_bytes in enumerate(uploaded_files):
-            # Generate a filename based on index since name is not available with type="binary"
             pdf_name = f"uploaded_file_{i}.pdf"
             if not is_pdf_file(file_bytes):
                 save_failed_record(pdf_name, object_name, "Invalid PDF content", {})
@@ -75,7 +74,7 @@ def process_contract(uploaded_files, object_name, manual_mappings):
             save_uploaded_file_details(pdf_name, temp_path)
 
             try:
-                # EPIC 4: OCR and Data Extraction
+                print(f"Processing {pdf_name} - OCR stage")
                 text_data = extract_text_from_pdf_with_tesseract_or_layoutlm(temp_path)
                 if not text_data:
                     save_failed_record(pdf_name, object_name, "No text extracted from PDF", {})
@@ -84,7 +83,7 @@ def process_contract(uploaded_files, object_name, manual_mappings):
                     pbar.update(1)
                     continue
 
-                # EPIC 3: Auto Mapping via AI
+                print(f"Processing {pdf_name} - AI mapping stage")
                 key_values = extract_key_values_with_layoutlm(text_data, temp_path)
                 ai_result = run_ai_mapping_with_layoutlm(key_values, mock_object_fields, temp_path)
                 if ai_result['status'] == 'failed':
@@ -94,13 +93,11 @@ def process_contract(uploaded_files, object_name, manual_mappings):
                     pbar.update(1)
                     continue
 
-                # Apply manual mappings
                 mappings = {k: v for k, v in ai_result['mappings'].items()}
                 for field, value in manual_mappings.items():
                     if value and field in mock_object_fields:
                         mappings[field] = value
 
-                # Mock record creation success (EPIC 5 placeholder)
                 results.append(f"✅ {pdf_name}: Data processed locally (Mock ID: {hash(pdf_name)})")
                 processed_files += 1
                 pbar.update(1)
@@ -115,6 +112,7 @@ def process_contract(uploaded_files, object_name, manual_mappings):
                 uploaded_file_details[pdf_name]["processed"] = True
 
     progress = f"{processed_files}/{total_files}"
+    print(f"Processing completed - Results: {results}, Progress: {progress}")
     return "\n".join(results), ai_result, failed_records, progress
 
 def retry_failed_record(index, object_name, manual_mappings):
@@ -134,24 +132,21 @@ def retry_failed_record(index, object_name, manual_mappings):
 
 # Gradio UI
 with gr.Blocks(title="Smart Contract Migrator (Local Mode)") as demo:
-    # EPIC 1: PDF Upload Interface
     with gr.Row():
         uploaded_files = gr.File(type="binary", file_types=["pdf"], file_count="multiple", label="Upload Contract PDFs")
         upload_progress = gr.Textbox(label="Upload Progress", value="0/0", interactive=False)
 
-    # EPIC 2: Mock Salesforce Object Mapping (placeholder)
     object_name = gr.Dropdown(choices=["Contract", "Invoice", "Agreement"], label="Select Object Type (Mock)")
 
     def update_fields(selected_object):
         if selected_object:
-            mock_fields = ["Name", "Description", "Amount", "Date"]  # Mock fields
+            mock_fields = ["Name", "Description", "Amount", "Date"]
             return gr.update(visible=True, value="\n".join(mock_fields))
         return gr.update(visible=False)
 
     object_fields_output = gr.Textbox(label="Available Fields (Mock)", interactive=False)
     object_name.change(fn=update_fields, inputs=object_name, outputs=object_fields_output)
 
-    # EPIC 3 & 4: Auto Mapping and OCR
     manual_mapping_inputs = gr.State(value={})
     def update_manual_mappings(selected_object):
         if selected_object:
@@ -166,7 +161,6 @@ with gr.Blocks(title="Smart Contract Migrator (Local Mode)") as demo:
         outputs=manual_mapping_inputs
     )
 
-    # EPIC 5: Mock Record Processing
     process_button = gr.Button("Extract, Map, and Process")
     status_output = gr.Textbox(label="Status", interactive=False)
     ai_result_output = gr.JSON(label="AI Mapping Results (High-Confidence Mappings)")
@@ -185,7 +179,6 @@ with gr.Blocks(title="Smart Contract Migrator (Local Mode)") as demo:
         outputs=[status_output, ai_result_output, upload_progress]
     )
 
-    # EPIC 6: Reconciliation & Retry
     with gr.Tab("Reconciliation & Retry"):
         failed_records_output = gr.Textbox(label="Failed Records", interactive=False, value="No failed records.")