src/streamlit_app.py

import os
os.environ["STREAMLIT_BROWSER_GATHER_USAGE_STATS"] = "false"


import pickle
import numpy as np
import streamlit as st
import warnings
from transformers import pipeline, AutoTokenizer, AutoModelForSeq2SeqLM
import os

from pinecone import Pinecone
from langchain_pinecone import PineconeVectorStore
from langchain.chains import RetrievalQA
from langchain_community.embeddings import HuggingFaceEmbeddings
from langchain_community.llms import HuggingFacePipeline

warnings.filterwarnings("ignore")

# ==================== Load Embeddings & Docs ====================



try:
    embeddings = np.load("src/embeddings.npy")
    with open("src/documents.pkl", "rb") as f:
        documents = pickle.load(f)  
except Exception as e:
    st.error(f"❌ Error loading embeddings or documents: {e}")
    st.stop()


# ==================== Setup Pinecone ====================
try:
    pc = Pinecone(api_key=os.environ["PINECONE_API_KEY"])
    index = pc.Index("changi-rag-384")
except Exception as e:
    st.error(f"❌ Error connecting to Pinecone: {e}")
    st.stop()

# ==================== Embedding Model ====================
from sentence_transformers import SentenceTransformer
from langchain_community.embeddings import HuggingFaceEmbeddings

model = SentenceTransformer("sentence-transformers/all-MiniLM-L6-v2")
embed_model = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2")


# ==================== Vector Store & Retriever ====================
vectorstore = PineconeVectorStore(
    index=index,
    embedding=embed_model,
    text_key="page_content"
)
retriever = vectorstore.as_retriever()

# ==================== HuggingFace QA Model ====================
model_name = "google/flan-t5-base"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForSeq2SeqLM.from_pretrained(model_name)

qa_pipeline = pipeline("text2text-generation", model=model, tokenizer=tokenizer)
llm = HuggingFacePipeline(pipeline=qa_pipeline)

qa = RetrievalQA.from_chain_type(llm=llm, retriever=retriever)

# ==================== Streamlit UI ====================
st.set_page_config(page_title="Changi RAG Chatbot", layout="wide")
st.title("🛫 Changi Airport RAG Chatbot")

query = st.text_input("Ask me anything about Changi Airport facilities:")

if query:
    with st.spinner("Thinking..."):
        try:
            response = qa.run(query)
            st.write("### ✈️ Answer:")
            st.success(response)
        except Exception as e:
            st.error(f"⚠️ Failed to generate answer: {e}")