app.py

import os
from dotenv import load_dotenv
load_dotenv(".env")

os.environ['USER_AGENT'] = os.getenv("USER_AGENT")
os.environ["GROQ_API_KEY"] = os.getenv("GROQ_API_KEY")
os.environ["TOKENIZERS_PARALLELISM"]='true'

import nltk
nltk.download('punkt_tab')

from langchain.chains import create_history_aware_retriever, create_retrieval_chain
from langchain.chains.combine_documents import create_stuff_documents_chain
from langchain_community.chat_message_histories import ChatMessageHistory
from langchain_community.document_loaders import WebBaseLoader
from langchain_core.chat_history import BaseChatMessageHistory
from langchain_core.prompts import ChatPromptTemplate, MessagesPlaceholder
from langchain_core.runnables.history import RunnableWithMessageHistory

from pinecone import Pinecone
from pinecone_text.sparse import BM25Encoder

from langchain_huggingface import HuggingFaceEmbeddings
from langchain_community.retrievers import PineconeHybridSearchRetriever

from langchain_groq import ChatGroq

import gradio as gr
import spaces
import torch


try:
    pc = Pinecone(api_key=os.getenv("PINECONE_API_KEY"))
    index_name = "traveler-demo-website-vectorstore"
    # connect to index
    pinecone_index = pc.Index(index_name)
except:
    pc = Pinecone(api_key=os.getenv("PINECONE_API_KEY"))
    index_name = "traveler-demo-website-vectorstore"
    # connect to index
    pinecone_index = pc.Index(index_name)

bm25 = BM25Encoder().load("./bm25_traveler_website.json")

embed_model = HuggingFaceEmbeddings(model_name="Alibaba-NLP/gte-large-en-v1.5", model_kwargs={"trust_remote_code":True, 'device': 'cuda'})

retriever = PineconeHybridSearchRetriever(
    embeddings=embed_model, 
    sparse_encoder=bm25, 
    index=pinecone_index, 
    top_k=20, 
    alpha=0.5, 
)

llm = ChatGroq(model="llama-3.1-70b-versatile", temperature=0.1, max_tokens=1024, max_retries=2)

### Contextualize question ###
contextualize_q_system_prompt = """Given a chat history and the latest user question \
which might reference context in the chat history, formulate a standalone question \
which can be understood without the chat history. Do NOT answer the question, \
just reformulate it if needed and otherwise return it as is.
"""
contextualize_q_prompt = ChatPromptTemplate.from_messages(
    [
        ("system", contextualize_q_system_prompt),
        MessagesPlaceholder("chat_history"),
        ("human", "{input}")
    ]
)

history_aware_retriever = create_history_aware_retriever(
    llm, retriever, contextualize_q_prompt
)


qa_system_prompt = """You are a highly skilled information retrieval assistant. Use the following pieces of retrieved context to answer the question. \
Provide links to sources provided in the answer. \
If you don't know the answer, just say that you don't know. \
Do not give extra long answers. \
When responding to queries, your responses should be comprehensive and well-organized. For each response: \
    1. Provide Clear Answers \
    2. Include Detailed References: \
        - Include links to sources and any links or sites where there is a mentioned in the answer.
        - Links to Sources: Provide URLs to credible sources where users can verify the information or explore further. \
        - Downloadable Materials: Include links to any relevant downloadable resources if applicable. \
        - Reference Sites: Mention specific websites or platforms that offer additional information. \
    3. Formatting for Readability: \
        - Bullet Points or Lists: Where applicable, use bullet points or numbered lists to present information clearly. \
        - Emphasize Important Information: Use bold or italics to highlight key details. \
    4. Organize Content Logically \
Do not include anything about context in the answer. \
{context}
"""
qa_prompt = ChatPromptTemplate.from_messages(
    [
        ("system", qa_system_prompt),
        MessagesPlaceholder("chat_history"),
        ("human", "{input}")
    ]
)
question_answer_chain = create_stuff_documents_chain(llm, qa_prompt)

rag_chain = create_retrieval_chain(history_aware_retriever, question_answer_chain)

### Statefully manage chat history ###
store = {}

def get_session_history(session_id: str) -> BaseChatMessageHistory:
    if session_id not in store:
        store[session_id] = ChatMessageHistory()
    return store[session_id]


conversational_rag_chain = RunnableWithMessageHistory(
    rag_chain,
    get_session_history,
    input_messages_key="input",
    history_messages_key="chat_history",
    output_messages_key="answer",
)

@spaces.GPU(duration=5)
def handle_message(question, history={}):
    response = ''
    chain = conversational_rag_chain.pick("answer")
    for chunk in chain.stream(
         {"input": question},
        config={
            "configurable": {"session_id": "abc123"}
        },
    ):
        response += chunk
        yield response

if __name__ == '__main__': 
    demo = gr.ChatInterface(fn=handle_message)
    demo.launch()