survey_agent1.py

import streamlit as st
import openai
import os
from dotenv import load_dotenv
from langchain_openai import ChatOpenAI
from langchain_core.prompts import ChatPromptTemplate
from langchain_core.runnables import RunnablePassthrough
from langchain_openai.embeddings import OpenAIEmbeddings
from langchain_text_splitters import RecursiveCharacterTextSplitter
from langchain_community.vectorstores import Qdrant
from langchain_core.documents import Document
from operator import itemgetter


# Load API key from .env file
load_dotenv()
base_llm = ChatOpenAI(model="gpt-4o")
openai.api_key = os.getenv("OPENAI_API_KEY")
embedding_model = OpenAIEmbeddings(model="text-embedding-3-small")


def analyze_survey_document(survey_text, question):
    """Send entire survey document to OpenAI for RAG-based analysis."""
    prompt = """
    You are an HR analyst answering questions about employee pulse survey responses. Below is the entire survey with questions and corresponding employee answers.

    **context:**
    {context}

    **question:**
    {question}

    Answer all questions based on the context. If you don't know say I don't know.
    """
    rag_prompt = ChatPromptTemplate.from_template(prompt)

# Token length function

   # Text splitter for chunking per page
    text_splitter =  RecursiveCharacterTextSplitter.from_tiktoken_encoder(
        chunk_size=500,  # Adjust chunk size based on contract structure
        chunk_overlap=50
    )


    chunks = text_splitter.split_text(survey_text)
    
    # Initialize Qdrant Vector Store for each page
    qdrant_vectorstore = Qdrant.from_texts(
        chunks,  # Pass as a Document object
        embedding_model,  # Ensure embedding_model is defined
        location=":memory:", 
        collection_name="formatted_text"
    )
    retriever = qdrant_vectorstore.as_retriever()

    # Define retrieval and response generation chain
    lcel_rag_chain = (
        {"context": itemgetter("question") | retriever, "question": itemgetter("question")}
        | RunnablePassthrough.assign(context=itemgetter("context"))
        | {"response": rag_prompt | base_llm, "context": itemgetter("context")}
    )
    response = lcel_rag_chain.invoke({"context": survey_text, "question": question})['response'].content

    return response

def chat_with_ai(chat_history):
    """Chat interface with OpenAI model."""
    try:
        response = openai.chat.completions.create(
            model="gpt-4o",
            messages=chat_history
        )
        return response.choices[0].message.content
    except Exception as e:
        return f"Error: {str(e)}"

def survey_agent(prompt,uploaded_file):
    st.subheader("📊 Employee Pulse Survey Analysis & AI Chat")

    with open('./data/employee_pulse_survey.txt', 'r') as file:
        survey_text_base = file.read()

    #uploaded_file = st.file_uploader("📂 Upload Pulse Survey (.txt)", type="txt")
    if not uploaded_file:
        uploaded_file1 = survey_text_base

    #check1 = st.button(f"Generate summary")

    #if check1:
    if uploaded_file or uploaded_file1:
        st.write("✅ File uploaded successfully! Analyzing responses...")
        if uploaded_file:
            survey_text = uploaded_file.read().decode("utf-8").strip()
        else:
            survey_text = uploaded_file1

        with st.spinner("🔍 Analyzing entire survey..."):
            analysis = analyze_survey_document(survey_text)
            st.session_state["survey_summary"] = analysis
            st.markdown(analysis)

    # AI Chat with the survey analysis
    st.header("💬 Chat with AI about the Survey")
    st.write("Ask questions about the pulse survey insights.")

    if "messages" not in st.session_state:
        st.session_state["messages"] = [
            {"role": "system", "content": "You are an HR expert analyzing a pulse survey. Provide insights based on the summary below."},
            {"role": "user", "content": st.session_state.get("survey_summary", "No survey uploaded yet.")}
        ]

    user_input = st.text_input("🔍 Ask a question about the survey results:")
    
    if st.button("Ask AI"):
        if user_input:
            st.session_state["messages"].append({"role": "user", "content": user_input})
            with st.spinner("🔍 Thinking..."):
                ai_response = chat_with_ai(st.session_state["messages"])
                st.session_state["messages"].append({"role": "assistant", "content": ai_response})
                st.markdown(f"**AI:** {ai_response}")
        else:
            st.warning("⚠️ Please enter a question.")

    return ai_response