Summary/save conversation using autogen

[glarunsingh]

Objective: I want to ask ask follow up questions to user to understand the process. once I understand the process, I want to save the conversation and summary of it in a variable to pass it to another agent. i figure out to have conversation and when I exit a can't find a way to view the conversation and summary of the process. conversation_history - returns null.

The "user_proxy_agent_result" doesn't have any attribute when I click ctrl+space in vs code. Help me in generating response and save it in a variable to pass it to another agent using langraph.

import os
import json
from langchain_openai import AzureChatOpenAI
from langchain_core.prompts import ChatPromptTemplate
from docx import Document as DocxDocument
from PyPDF2 import PdfReader
from dotenv import load_dotenv
from autogen import AssistantAgent, UserProxyAgent
import autogen

-------------------- Load Environment Variables --------------------

load_dotenv()

api_version = os.getenv("AZURE_OPENAI_API_VERSION")
endpoint = os.getenv("AZURE_OPENAI_ENDPOINT")
api_key = os.getenv("AZURE_OPENAI_API_KEY")
deployment_name = os.getenv("AZURE_OPENAI_CHAT_DEPLOYMENT_NAME")
llm_model = os.getenv("LLM_MODEL")

if not all([api_version, endpoint, api_key, deployment_name, llm_model]):
raise ValueError("Some environment variables are missing. Check your .env file.")

Configure Autogen LLM

llm_config = {
"timeout": 600,
"config_list": autogen.config_list_from_json("OAI_CONFIG_LIST"),
"temperature": 0,
}

Initialize Autogen Agents

input_assistant = AssistantAgent(
name="input_assistant",
system_message="""You are a helpful AI assistant.
Your task is to analyze the content provided by the user, ask clarifying questions, and collect responses one by one.
Use follow-up questions to understand requirements thoroughly. Type 'TERMINATE' when the data collection is complete.
""",
llm_config=llm_config,
)

user_proxy = UserProxyAgent(
"user_proxy",
human_input_mode="ALWAYS",
llm_config=llm_config,
code_execution_config=False,
system_message="You are a helpful assistant."
)

-------------------- Helper Functions --------------------

def extract_text_from_docx(docx_path):
doc = DocxDocument(docx_path)
return "\n".join([para.text for para in doc.paragraphs])

def extract_text_from_pdf(pdf_path):
reader = PdfReader(pdf_path)
return "\n".join(page.extract_text() for page in reader.pages)

def read_document(file_path):
if not file_path or not os.path.exists(file_path):
return None

if file_path.endswith(".txt"):
    with open(file_path, 'r', encoding='utf-8') as f:
        return f.read()
elif file_path.endswith(".docx"):
    return extract_text_from_docx(file_path)
elif file_path.endswith(".pdf"):
    return extract_text_from_pdf(file_path)
else:
    return "Unsupported file format. Please provide a valid TXT, DOCX, or PDF file."

-------------------- Main Logic --------------------

def analyze_and_start_autogen_qa(content):
"""Use Autogen to analyze content and initiate Q&A."""
user_input = content
conversation_history = [] # Store the entire conversation

# Start Autogen Q&A loop
user_proxy_agent_result = user_proxy.initiate_chat(
    recipient=input_assistant,
    message=user_input,
)
conversation_history.append({"role": "user", "message": user_input})

print(user_proxy_agent_result)
#print(user_proxy_agent_result.content)

# Collect the conversation history from the user_proxy_agent_result
for message in user_proxy_agent_result.messages:
    conversation_history.append({
        "role": message["role"],
        "message": message["content"]
    })

return conversation_history, user_proxy_agent_result

def process_document_or_summary(doc_path=None):
"""Process document or initiate with user-provided summary."""
content = read_document(doc_path)

if content:
    print(f"\nExtracted Content from '{doc_path}':\n{content}\n")
else:
    print("No document provided. Please enter a summary.\n")
    content = input("Enter a summary of the process: ")

conversation_history, user_proxy_agent_result = analyze_and_start_autogen_qa(content)

# Pass the collected data to the next module or store it
collected_data = {
    "summary": content,
    "conversation_history": conversation_history
}
print("\nFinal Collected Data:\n", json.dumps(collected_data, indent=2))
print("\nUser Proxy Agent Result:\n", user_proxy_agent_result)
return collected_data

-------------------- Execution --------------------

if name == "main":
doc_path = input("Enter the document path (or press Enter to skip): ").strip()
process_document_or_summary(doc_path)