app.py

import threading
import re
import gradio as gr
import os
import google.generativeai as genai
GOOGLE_API_KEY = os.getenv("GOOGLE_API_KEY")


import chromadb
from langchain.document_loaders import PyPDFLoader
from langchain.text_splitter import RecursiveCharacterTextSplitter
from uuid import uuid4
import gradio as gr

# Now you can use hugging_face_api_key in your code

genai.configure(api_key=GOOGLE_API_KEY)
model = genai.GenerativeModel('gemini-pro')  # Load the model

def get_Answer(query):
    res = collection.query(  # Assuming `collection` is defined elsewhere
        query_texts=query,
        n_results=2
    )
    system = f"""You are a teacher. You will be provided some context, 
    your task is to analyze the relevant context and answer the below question:
    - {query}
    """
    context = " ".join([re.sub(r'[^\x00-\x7F]+', ' ', r) for r in res['documents'][0]])
    prompt = f"### System: {system} \n\n ###: User: {context} \n\n ### Assistant:\n"
    answer = model.generate_content(prompt).text
    return answer

# # Define the Gradio interface
# iface = gr.Interface(
#     fn=get_Answer,
#     inputs=gr.Textbox(lines=5, placeholder="Ask a question"),  # Textbox for query
#     outputs="textbox",  # Display the generated answer in a textbox
#     title="Answer Questions with Gemini-Pro",
#     description="Ask a question and get an answer based on context from a ChromaDB collection.",
# )




text_splitter = RecursiveCharacterTextSplitter(
    chunk_size=800,
    chunk_overlap=50
)
client = chromadb.PersistentClient("test")
collection = client.create_collection("test_data")

def upload_pdf(file_path):
    loader = PyPDFLoader(file_path)
    pages = loader.load()
    documents = []
    for page in pages:
        docs = text_splitter.split_text(page.page_content)
        for doc in docs:
            documents.append({
                "text": docs, "meta_data": page.metadata,
            })
    collection.add(
        ids=[str(uuid4()) for _ in range(len(documents))],
        documents=[doc['text'][0] for doc in documents],
        metadatas=[doc['meta_data'] for doc in documents]
    )
    return f"PDF Uploaded Successfully. {collection.count()} chunks stored in ChromaDB"

# # Define the Gradio interface
# iface = gr.Interface(
#     fn=upload_pdf,
#     inputs=["file"],  # Specify a file input component
#     outputs="textbox",  # Display the output text in a textbox
#     title="Upload PDF to ChromaDB",
#     description="Upload a PDF file and store its text chunks in ChromaDB.",
# )

# Gradio interfaces
iface1 = gr.Interface(
    fn=get_Answer,
    inputs=gr.Textbox(lines=5, placeholder="Ask a question"),
    outputs="textbox",
    title="Answer Questions with Gemini-Pro",
    description="Ask a question and get an answer based on context from a ChromaDB collection.",
)
iface2 = gr.Interface(
    fn=upload_pdf,
    inputs=["file"],
    outputs="textbox",
    title="Upload PDF to ChromaDB",
    description="Upload a PDF file and store its text chunks in ChromaDB.",
)



# Launch the apps in separate threads
thread1 = threading.Thread(target=iface1.launch, args=(debug=True, share=True))
thread2 = threading.Thread(target=iface2.launch, args=(debug=True, share=True))

thread1.start()
thread2.start()