Update app.py

app.py

@@ -79,12 +79,15 @@ def update_vectors(files, parser):
     return f"Vector store updated successfully. Processed {total_chunks} chunks from {len(files)} files using {parser}."
 
 def generate_chunked_response(prompt, model, max_tokens=1000, num_calls=3, temperature=0.2):
+    print(f"Starting generate_chunked_response with {num_calls} calls")  # Debug log
     client = InferenceClient(model, token=huggingface_token)
     full_responses = []
     messages = [{"role": "user", "content": prompt}]
     
-    for _ in range(num_calls):
-        if stop_clicked:  # Check if stop was clicked
+    for i in range(num_calls):
+        print(f"Starting API call {i+1}")  # Debug log
+        if stop_clicked:
+            print("Stop clicked, breaking loop")  # Debug log
             break
         try:
             response = ""
@@ -94,23 +97,25 @@ def generate_chunked_response(prompt, model, max_tokens=1000, num_calls=3, tempe
                 temperature=temperature,
                 stream=True,
             ):
-                if stop_clicked:  # Check if stop was clicked
+                if stop_clicked:
+                    print("Stop clicked during streaming, breaking")  # Debug log
                     break
                 if message.choices and message.choices[0].delta and message.choices[0].delta.content:
                     chunk = message.choices[0].delta.content
                     response += chunk
+            print(f"API call {i+1} response: {response[:100]}...")  # Debug log
             full_responses.append(response)
         except Exception as e:
             print(f"Error in generating response: {str(e)}")
     
-    # Combine all responses into a single string
     combined_response = " ".join(full_responses)
+    print(f"Combined response: {combined_response[:100]}...")  # Debug log
     
-    # Clean the combined response
     clean_response = re.sub(r'<s>\[INST\].*?\[/INST\]\s*', '', combined_response, flags=re.DOTALL)
     clean_response = clean_response.replace("Using the following context:", "").strip()
     clean_response = clean_response.replace("Using the following context from the PDF documents:", "").strip()
     
+    print(f"Final clean response: {clean_response[:100]}...")  # Debug log
     return clean_response
 
 def duckduckgo_search(query):
@@ -246,29 +251,39 @@ with gr.Blocks() as demo:
         clear_btn = gr.Button("Clear")
 
     def protected_generate_response(message, history, use_web_search, model, temperature, num_calls, is_generating, stop_clicked):
+        print("Starting protected_generate_response")  # Debug log
         if is_generating:
+            print("Already generating, returning")  # Debug log
             return message, history, is_generating, stop_clicked
         is_generating = True
         stop_clicked = False
         
         try:
+            print(f"Generating response for: {message}")  # Debug log
             if use_web_search:
+                print("Using web search")  # Debug log
                 main_content, sources = get_response_with_search(message, model, num_calls=num_calls, temperature=temperature)
                 formatted_response = f"{main_content}\n\nSources:\n{sources}"
             else:
+                print("Using PDF search")  # Debug log
                 response = get_response_from_pdf(message, model, num_calls=num_calls, temperature=temperature)
                 formatted_response = response
     
+            print(f"Generated response: {formatted_response[:100]}...")  # Debug log
+    
             if not stop_clicked:
-                # Only append the final, combined response to the history
+                print("Appending to history")  # Debug log
                 history.append((message, formatted_response))
+            else:
+                print("Stop clicked, not appending to history")  # Debug log
         except Exception as e:
             print(f"Error generating response: {str(e)}")
             history.append((message, "I'm sorry, but I encountered an error while generating the response. Please try again."))
         
         is_generating = False
+        print(f"Returning history with {len(history)} items")  # Debug log
         return "", history, is_generating, stop_clicked
-            
+
     submit_btn.click(
         protected_generate_response,
         inputs=[msg, chatbot, use_web_search, model_dropdown, temperature_slider, num_calls_slider, is_generating, stop_clicked],