app.py

import os
import gradio as gr
from langchain_groq import ChatGroq
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain.chains.combine_documents import create_stuff_documents_chain
from langchain_core.prompts import ChatPromptTemplate
from langchain.chains import create_retrieval_chain
from langchain_community.vectorstores import FAISS
from langchain_community.document_loaders import PyPDFLoader
from langchain_community.embeddings import HuggingFaceEmbeddings
from langchain.prompts import PromptTemplate
from dotenv import load_dotenv

"""
### How to Use the App:

1. **Upload PDF:** Upload PDF and click 'Process PDF' to add them to the knowledge base.
2. **Ask a Question:** Switch to the 'Q&A System' tab, enter your question, and click 'Ask Question' to get an answer based on the uploaded PDF content.
3. **Clear Data:** Click 'Clear Knowledge Base' to reset and remove all uploaded documents.

Ensure a PDF is uploaded before asking questions.
"""


# Load environment variables
load_dotenv()

# Load the GROQ API key
GROQ_API_KEY = os.environ.get("GROQ_API_KEY")

# Set up the language model
llm = ChatGroq(temperature=0, model_name='llama-3.1-8b-instant', groq_api_key=GROQ_API_KEY)

# Define the prompt template
prompt = ChatPromptTemplate.from_template("""
Answer the questions based on the provided context only.
Please provide the most accurate response based on the question.
<context>{context}</context>
Question: {input}
""")

# Set up embeddings model
embeddings = HuggingFaceEmbeddings(model_name="all-MiniLM-L6-v2")
vectors = None

# Function to process PDF files
def process_pdf(file):
    global vectors
    if file is not None:
        loader = PyPDFLoader(file.name)
        docs = loader.load()
        text_splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=200)
        final_documents = text_splitter.split_documents(docs)
        if vectors is None:
            vectors = FAISS.from_documents(final_documents, embeddings)
        else:
            vectors.add_documents(final_documents)
        return "PDF processed and added to the knowledge base."
    return "No file uploaded."

# Function to clear the knowledge base
def clear_knowledge_base():
    global vectors
    vectors = None  # Reset the vector store
    return "Knowledge base cleared."

def process_question(question):
    global vectors
    if vectors is None:
        return "Please upload a PDF first.", "", 0
    
    # Create document retrieval chain
    retriever = vectors.as_retriever(search_type="similarity", search_kwargs={"k": 5})
    
    # Use the invoke method for retrieving relevant documents
    documents = retriever.invoke(question)
    
    if not documents:
        return "No relevant context found.", "", 0
    
    # Create context from retrieved documents
    context = "\n\n".join([doc.page_content for doc in documents])
    
    # Combine context and question into a single string (formatted input for LLM)
    prompt = f"Answer the question based on the provided context.\n\nContext: {context}\n\nQuestion: {question}"

    # Pass the string to llm.invoke
    response = llm.invoke(prompt)

    # Confidence score as average relevance
    confidence_score = sum([doc.metadata.get('score', 0) for doc in documents]) / len(documents)

    return response, context, round(confidence_score, 2)

# CSS styling
CSS = """
.duplicate-button { margin: auto !important; color: white !important; background: black !important; border-radius: 100vh !important;}
h3, p, h1 { text-align: center; color: white;}
footer { text-align: center; padding: 10px; width: 100%; background-color: rgba(240, 240, 240, 0.8); z-index: 1000; position: relative; margin-top: 10px; color: black;}
"""

# Footer text
FOOTER_TEXT = """
<footer>
    <p>If you enjoyed the functionality of the app, please leave a like!<br>
    Check out more on <a href="https://www.linkedin.com/in/your-linkedin/" target="_blank">LinkedIn</a> | 
    <a href="https://your-portfolio-url.com/" target="_blank">Portfolio</a></p>
</footer>
"""

# Title text
TITLE = "<h1>📚 RAG Document Q&A 📚</h1>"

# Gradio interface
with gr.Blocks(css=CSS, theme="Nymbo/Nymbo_Theme") as demo:
    gr.HTML(TITLE)

    with gr.Tab("PDF Uploader"):
        pdf_file = gr.File(label="Upload PDF")
        upload_button = gr.Button("Process PDF")
        clear_button = gr.Button("Clear Knowledge Base")  # New button to clear the knowledge base
        upload_output = gr.Textbox(label="Upload Status")

    with gr.Tab("Q&A System"):
        question_input = gr.Textbox(lines=2, placeholder="Enter your question here...")
        submit_button = gr.Button("Ask Question")
        answer_output = gr.Textbox(label="Answer")
        context_output = gr.Textbox(label="Relevant Context", lines=10)
        confidence_output = gr.Number(label="Confidence Score")

    # Button actions
    upload_button.click(process_pdf, inputs=[pdf_file], outputs=[upload_output])
    submit_button.click(process_question, inputs=[question_input], outputs=[answer_output, context_output, confidence_output])
    
    # Action to clear the knowledge base
    clear_button.click(clear_knowledge_base, outputs=[upload_output])

    gr.HTML(FOOTER_TEXT)

# Launch the Gradio app
if __name__ == "__main__":
    demo.launch()