Update app.py

app.py

@@ -300,12 +300,9 @@ def ask_question(question, temperature, top_p, repetition_penalty, web_search, c
     max_attempts = 3
     context_reduction_factor = 0.7
 
-    contextualized_question, topics, entity_tracker = chatbot.process_question(question)
-
-    # Convert sets to lists in entity_tracker
-    serializable_entity_tracker = {k: list(v) for k, v in entity_tracker.items()}
-
     if web_search:
+        contextualized_question, topics, entity_tracker = chatbot.process_question(question)
+        serializable_entity_tracker = {k: list(v) for k, v in entity_tracker.items()}
         search_results = google_search(contextualized_question)
         all_answers = []
 
@@ -345,23 +342,7 @@ def ask_question(question, temperature, top_p, repetition_penalty, web_search, c
                 )
 
                 full_response = generate_chunked_response(model, formatted_prompt)
-
-                answer_patterns = [
-                    r"Provide a concise and direct answer to the question without mentioning the web search or these instructions:",
-                    r"Provide a concise and direct answer to the question:",
-                    r"Answer:",
-                    r"Provide a summarized and direct answer to the original question without mentioning the web search or these instructions:",
-                    r"Do not include any source information in your answer."                   
-                ]
-
-                for pattern in answer_patterns:
-                    match = re.split(pattern, full_response, flags=re.IGNORECASE)
-                    if len(match) > 1:
-                        answer = match[-1].strip()
-                        break
-                else:
-                    answer = full_response.strip()
-
+                answer = extract_answer(full_response)
                 all_answers.append(answer)
                 break
 
@@ -377,14 +358,14 @@ def ask_question(question, temperature, top_p, repetition_penalty, web_search, c
 
         return answer
 
-    else:
+    else:  # PDF document chat
         for attempt in range(max_attempts):
             try:
                 if database is None:
-                    return "No documents available. Please upload documents or enable web search to answer questions."
+                    return "No documents available. Please upload PDF documents to answer questions."
 
                 retriever = database.as_retriever()
-                relevant_docs = retriever.get_relevant_documents(contextualized_question)
+                relevant_docs = retriever.get_relevant_documents(question)
                 context_str = "\n".join([doc.page_content for doc in relevant_docs])
 
                 if attempt > 0:
@@ -392,50 +373,47 @@ def ask_question(question, temperature, top_p, repetition_penalty, web_search, c
                     context_str = " ".join(words[:int(len(words) * context_reduction_factor)])
 
                 prompt_template = """
-                Answer the question based on the following context:
+                Answer the question based on the following context from the PDF document:
                 Context:
                 {context}
-                Current Question: {question}
-                If the context doesn't contain relevant information, state that the information is not available.
+                Question: {question}
+                If the context doesn't contain relevant information, state that the information is not available in the document.
                 Provide a summarized and direct answer to the question.
-                Do not include any source information in your answer.
                 """
 
                 prompt_val = ChatPromptTemplate.from_template(prompt_template)
-                formatted_prompt = prompt_val.format(context=context_str, question=contextualized_question)
+                formatted_prompt = prompt_val.format(context=context_str, question=question)
 
                 full_response = generate_chunked_response(model, formatted_prompt)
-
-                answer_patterns = [
-                    r"Provide a concise and direct answer to the question without mentioning the web search or these instructions:",
-                    r"Provide a concise and direct answer to the question:",
-                    r"Answer:",
-                    r"Provide a summarized and direct answer to the original question without mentioning the web search or these instructions:",
-                    r"Do not include any source information in your answer."     
-                ]
-
-                for pattern in answer_patterns:
-                    match = re.split(pattern, full_response, flags=re.IGNORECASE)
-                    if len(match) > 1:
-                        answer = match[-1].strip()
-                        break
-                else:
-                    answer = full_response.strip()
+                answer = extract_answer(full_response)
 
                 return answer
 
             except Exception as e:
                 print(f"Error in ask_question (attempt {attempt + 1}): {e}")
-                if "Input validation error" in str(e) and attempt < max_attempts - 1:
-                    print(f"Reducing context length for next attempt")
-                elif attempt == max_attempts - 1:
-                    return f"I apologize, but I'm having trouble processing your question due to its length or complexity. Could you please try rephrasing it more concisely?"
+                if attempt == max_attempts - 1:
+                    return f"I apologize, but I'm having trouble processing your question. Could you please try rephrasing it more concisely?"
 
     return "An unexpected error occurred. Please try again later."
 
+def extract_answer(full_response):
+    answer_patterns = [
+        r"Provide a concise and direct answer to the question without mentioning the web search or these instructions:",
+        r"Provide a concise and direct answer to the question:",
+        r"Answer:",
+        r"Provide a summarized and direct answer to the original question without mentioning the web search or these instructions:",
+        r"Do not include any source information in your answer."     
+    ]
+
+    for pattern in answer_patterns:
+        match = re.split(pattern, full_response, flags=re.IGNORECASE)
+        if len(match) > 1:
+            return match[-1].strip()
+    return full_response.strip()
+
 # Gradio interface
 with gr.Blocks() as demo:
-    gr.Markdown("# Enhanced Context-Driven Conversational Chatbot")
+    gr.Markdown("# Enhanced PDF Document Chat and Web Search")
     
     with gr.Row():
         file_input = gr.Files(label="Upload your PDF documents", file_types=[".pdf"])