import gradio as gr
import numpy as np
import pandas as pd
import pickle
from sklearn.metrics.pairwise import cosine_similarity

pt = pd.read_pickle('pt.pkl')
user_similarity_scores = cosine_similarity(pt.T)
books = pd.read_pickle("books.pkl")

def recommend_books_for_user(user_id):
    
    user_index = pt.columns.get_loc(user_id)

    similar_users = sorted(list(enumerate(user_similarity_scores[user_index])), key=lambda x: x[1], reverse=True)[1:5]

    recommended_books = []

    for similar_user_index, similarity_score in similar_users:
        user_ratings = pt.iloc[:, user_index]
        similar_user_ratings = pt.iloc[:, similar_user_index]

        unrated_books = similar_user_ratings[(user_ratings == 0) & (similar_user_ratings > 0)]

        recommended_books.extend(unrated_books.index)

    recommended_books_set = set(recommended_books)

    ans = [(book_title, image_url) for book_title, image_url in zip(books["Book-Title"], books["Image-URL-M"]) if book_title in recommended_books_set]

    return ans

def recommend_books_gradio(user_id):
    """Recommends books for a user based on collaborative filtering"""
    recommended_books = recommend_books_for_user(int(user_id))
    return [[book] for book in recommended_books]

interface = gr.Interface(fn=recommend_books_gradio, 
                         inputs=gr.Textbox(label="Enter User ID"),
                         outputs=gr.List(label="Recommended Books"),
                         title="Book Recommender System")

interface.launch()