modules/rag/rag_chain.py

from langchain.chains import RetrievalQA
from langchain.llms import OpenAI
from langchain.prompts import PromptTemplate
from modules.rag.vector_store import VectorStore
from modules.analyzer import client
import os

class RAGChain:
    def __init__(self):
        self.vector_store = VectorStore()
        self.retriever = self.vector_store.vector_store.as_retriever(
            search_type="similarity",
            search_kwargs={"k": 5}
        )
        
        # Custom prompt template optimized for your model
        self.prompt_template = """
        You are an expert research assistant with access to relevant documents. 
        Use the following context to answer the question accurately.
        If the context doesn't contain enough information, say so.
        Always cite specific parts of the context in your response.
        
        Context: {context}
        
        Question: {question}
        
        Answer:
        """
        
        self.prompt = PromptTemplate(
            template=self.prompt_template,
            input_variables=["context", "question"]
        )
    
    def query(self, question):
        """Query the RAG system"""
        try:
            # Search for relevant documents
            search_result = self.vector_store.search(question)
            if search_result["status"] != "success":
                return {"status": "error", "message": search_result["message"]}
            
            # Format context
            context = "\n\n".join([doc.page_content for doc in search_result["documents"]])
            
            # Create enhanced prompt
            enhanced_prompt = self.prompt.format(context=context, question=question)
            
            # For streaming, we'll return the prompt for the analyzer to handle
            return {
                "status": "success", 
                "prompt": enhanced_prompt,
                "context_docs": search_result["documents"]
            }
        except Exception as e:
            return {"status": "error", "message": str(e)}