app.py

import gradio as gr
from langchain import HuggingFacePipeline
from transformers import pipeline
from langchain.document_loaders import PyPDFLoader
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain_community.vectorstores import Chroma
from langchain.chains import RetrievalQA
from langchain.embeddings import HuggingFaceEmbeddings

# --- Load free Hugging Face LLM ---
# You can switch to another free model if you like
hf_pipeline = pipeline(
    "text-generation",
    model="tiiuae/falcon-7b-instruct",
    max_new_tokens=256,
    temperature=0.5
)
llm = HuggingFacePipeline(pipeline=hf_pipeline)

# --- PDF Loader ---
def document_loader(file):
    loader = PyPDFLoader(file.name)
    return loader.load()

# --- Text Splitter ---
def text_splitter(documents):
    splitter = RecursiveCharacterTextSplitter(
        chunk_size=1000,
        chunk_overlap=20
    )
    return splitter.split_documents(documents)

# --- Vector Database using embeddings ---
def vector_database(chunks):
    embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2")
    vectordb = Chroma.from_documents(chunks, embeddings)
    return vectordb

# --- Build retriever ---
def retriever(file):
    chunks = text_splitter(document_loader(file))
    vectordb = vector_database(chunks)
    return vectordb.as_retriever()

# --- RetrievalQA ---
def retriever_qa(file, query):
    retriever_obj = retriever(file)
    qa_chain = RetrievalQA.from_chain_type(
        llm=llm,
        chain_type="stuff",
        retriever=retriever_obj,
        return_source_documents=False
    )
    return qa_chain.run(query)

# --- Gradio Interface ---
app = gr.Interface(
    fn=retriever_qa,
    inputs=[
        gr.File(label="Upload PDF", file_types=['.pdf']),
        gr.Textbox(label="Ask a question about the PDF", lines=2, placeholder="Type your question here...")
    ],
    outputs=gr.Textbox(label="Answer"),
    title="Free LLM PDF Q&A Bot",
    description="Upload a PDF document and ask any question. This bot uses a free open-source LLM and vector search to answer your questions."
)

if __name__ == "__main__":
    app.launch()