app.py

import streamlit as st
from PyPDF2 import PdfReader
import docx
from sentence_transformers import SentenceTransformer
import faiss
from transformers import pipeline

# Load and cache models
@st.cache_resource
def load_resources():
    embedder = SentenceTransformer('sentence-transformers/all-MiniLM-L6-v2')
    chat_gen = pipeline(
        'text2text-generation',
        model='google/flan-t5-base',
        tokenizer='google/flan-t5-base',
        device=-1,
        # enforce deterministic decoding and low temperature to reduce hallucinations
        do_sample=False,
        temperature=0.0,
    )
    return embedder, chat_gen

# Extract text from uploaded file
def extract_text(uploaded):
    name = uploaded.name.lower()
    if name.endswith('.pdf'):
        reader = PdfReader(uploaded)
        return ''.join(page.extract_text() or '' for page in reader.pages)
    if name.endswith('.docx'):
        doc = docx.Document(uploaded)
        return '\n'.join(para.text for para in doc.paragraphs)
    return uploaded.getvalue().decode('utf-8', errors='ignore')

# Chunking helper
def chunk_text(text, size=500, overlap=50):
    words = text.split()
    chunks, start = [], 0
    while start < len(words):
        end = min(start + size, len(words))
        chunks.append(' '.join(words[start:end]))
        start += size - overlap
    return chunks

# Build FAISS index
@st.cache_resource
def build_index(chunks, _embedder):  # underscore avoids hashing
    embs = _embedder.encode(chunks, convert_to_numpy=True)
    dim = embs.shape[1]
    index = faiss.IndexFlatL2(dim)
    index.add(embs)
    return index

# Compose prompt
def make_prompt(system_prompt, context, history, question):
    prompt = system_prompt + "\n\n"
    if context:
        prompt += f"Document Context:\n{context}\n\n"
    for msg in history:
        role = 'User' if msg['role']=='User' else 'Assistant'
        prompt += f"{role}: {msg['text']}\n"
    prompt += f"User: {question}\nAssistant:"
    return prompt

# Main app
def main():
    st.set_page_config(page_title='📄 RagBot Chat+RAG', layout='wide')
    st.title('🤖 RagBot')
    st.sidebar.header('📂 Optional: Upload Document')

    # Initialize state
    if 'history' not in st.session_state:
        st.session_state.history = []
    if 'chunks' not in st.session_state:
        st.session_state.chunks = []
    if 'index' not in st.session_state:
        st.session_state.index = None

    # Document upload
    uploaded = st.sidebar.file_uploader('Upload PDF, DOCX or TXT', type=['pdf','docx','txt'])
    if uploaded and (st.session_state.get('uploaded_name') != uploaded.name):
        text = extract_text(uploaded)
        st.session_state.chunks = chunk_text(text)
        st.session_state.embedder, st.session_state.chat_gen = load_resources()
        st.session_state.index = build_index(st.session_state.chunks, st.session_state.embedder)
        st.session_state.uploaded_name = uploaded.name
        st.session_state.history = []
    # Load models if missing
    if 'embedder' not in st.session_state or 'chat_gen' not in st.session_state:
        st.session_state.embedder, st.session_state.chat_gen = load_resources()

    # Display chat history
    for msg in st.session_state.history:
        with st.chat_message('user' if msg['role']=='User' else 'assistant'):
            st.markdown(f"**{msg['role']}:** {msg['text']}")

    # Chat input always available
    question = st.chat_input('Ask a question—general or document-specific...')
    if question:
        # Retrieve context
        context = ''
        if st.session_state.index is not None:
            q_emb = st.session_state.embedder.encode([question], convert_to_numpy=True)
            _, idxs = st.session_state.index.search(q_emb, k=3)
            context = '\n\n'.join(st.session_state.chunks[i] for i in idxs[0])

        # Build prompt with hallucination guard
        system_prompt = (
            "You are RagBot, an AI assistant. "
            "You must ONLY use the document context provided to answer document-specific questions. "
            "If the answer is not contained in the context, respond with: "
            "\"I’m sorry, I don’t know based on the document.\" "
            "For general knowledge questions, answer using your training knowledge without hallucinating."
        )
        prompt = make_prompt(system_prompt, context, st.session_state.history, question)

        # Generate answer
        response = st.session_state.chat_gen(prompt, max_new_tokens=200)
        answer = response[0]['generated_text'].strip()

        # Save & display
        st.session_state.history.append({'role':'User','text':question})
        st.session_state.history.append({'role':'Assistant','text':answer})
        with st.chat_message('user'):
            st.markdown(f"**You:** {question}")
        with st.chat_message('assistant'):
            st.markdown(f"**RagBot:** {answer}")

if __name__ == '__main__':
    main()