import streamlit as st
import pandas as pd
import numpy as np
from sklearn.metrics.pairwise import cosine_similarity

books = pd.read_csv("books.csv")
books["Book-Title"] = books["Book-Title"] + " by " + books["Book-Author"]
weighted_similarity = np.load("weighted_similarity.npy")


def recommend_books(input_books, top_n=5):
    recommended_books_set = set()
    book_vectors = []
    for book_title in input_books:
        if book_title in books['Book-Title'].values:
            book_index = books[books['Book-Title'] == book_title].index[0]
            book_vectors.append(weighted_similarity[book_index])
    # Sum the vectors
    summed_vector = np.sum(book_vectors, axis=0)
    # Calculate cosine similarity
    cosine_sim = cosine_similarity(summed_vector.reshape(1, -1), weighted_similarity)
    # Get top book indices
    top_books_indices = cosine_sim.argsort()[0][-top_n - 1:-1][::-1]
    recommended_books_set.update(books.loc[top_books_indices, 'Book-Title'])
    recommended_books_list = list(recommended_books_set)[:top_n]
    return recommended_books_list


st.title("Multi-Modal Book Recommendation")
st.write("Book Recommender is an open-source application designed to enhance the book selection process. By incorporating user ratings, publication years, author names, book descriptions, and cover photos, Book Recommender employs cosine similarity to provide personalized book recommendations. Interested individuals can access project on [GitHub](https://github.com/egecandrsn/multimodal_book_recommendation) to explore its functionalities. Future plans include considering neural collaborative filtering for improved recommendation accuracy.")

search = st.multiselect("Pick your favorites:", options=books['Book-Title'].unique())

if search:
    cols = st.columns(len(search))
    for i, book_title in enumerate(search):
        with cols[i]:
            try:
                book_image_url = books.loc[books['Book-Title'] == book_title, 'Image-URL-L'].values
                cols[i].image(book_image_url[0], width=150)
            except:
                cols[i].warning(f"No image available for {book_title}")
                
recommend = st.button("Recommend")

if recommend:
    recommended_books = recommend_books(search)
    st.write("Recommended Books:")
    columns = st.columns(5)
    for i, book_title in enumerate(recommended_books, start=1):
        with columns[i-1]:
            book_image_url = books.loc[books['Book-Title'] == book_title, 'Image-URL-L'].values
            try:
                columns[i-1].image(book_image_url[0], caption=book_title, width=150)
            except:
                columns[i-1].warning(f"No image available for {book_title}")