app.py

import os
import PyPDF2
import faiss
import numpy as np
from sentence_transformers import SentenceTransformer
import gradio as gr
import requests

# Load your Anthropic API key from environment variable or set it here
ANTHROPIC_API_KEY = os.getenv("ANTHROPIC_API_KEY")  # Make sure to set your key
# Alternatively, set it directly in the code for testing purposes:
# ANTHROPIC_API_KEY = "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"

# Step 1: Extract text from PDFs
def extract_text_from_pdf(pdf_file):
    reader = PyPDF2.PdfReader(pdf_file)
    text = ""
    for page in reader.pages:
        text += page.extract_text()
    return text

# Step 2: Generate embeddings
def create_embeddings(text):
    model = SentenceTransformer('all-MiniLM-L6-v2')
    return model.encode(text.split('\n'), convert_to_tensor=True)

# Step 3: Create FAISS index
def create_faiss_index(embeddings):
    dim = embeddings.shape[1]
    index = faiss.IndexFlatL2(dim)
    index.add(np.array(embeddings))
    return index

# Step 4: Query with Anthropic
def query_anthropic(prompt):
    headers = {
        "Authorization": f"Bearer {ANTHROPIC_API_KEY}",
        "Content-Type": "application/json"
    }
    data = {
        "prompt": prompt,
        "max_tokens": 150,
        "stop": None
    }
    response = requests.post('https://api.anthropic.com/v1/complete', json=data, headers=headers)
    return response.json().get('completion', 'No response from model')

# Step 5: Complete workflow
def process_pdf_and_query(pdf_file, user_query):
    # Step 1: Extract text
    pdf_text = extract_text_from_pdf(pdf_file)
    
    # Step 2: Generate embeddings
    embeddings = create_embeddings(pdf_text)
    
    # Step 3: Create FAISS index
    faiss_index = create_faiss_index(embeddings)
    
    # Step 4: Query with Anthropic
    index_query_embedding = create_embeddings(user_query)
    D, I = faiss_index.search(np.array([index_query_embedding]), k=1)  # Searching for the closest match
    closest_text = pdf_text.split('\n')[I[0][0]]  # Get the closest text based on the index
    response = query_anthropic(f"Answer the question based on this context: {closest_text}. Question: {user_query}")
    
    return response

# Gradio interface
def run_gradio():
    iface = gr.Interface(
        fn=process_pdf_and_query,
        inputs=[gr.File(label="Upload PDF File"), gr.Textbox(label="Ask a question")],
        outputs="text",
        title="PDF Query with Anthropic",
        description="Upload a PDF file and ask questions related to its content."
    )
    iface.launch()

if __name__ == "__main__":
    run_gradio()