app.py

import numpy as np
import pandas as pd
import ast
from sklearn.feature_extraction.text import CountVectorizer
from sklearn.metrics.pairwise import cosine_similarity
from nltk.stem.porter import PorterStemmer
import gradio as gr
import nltk
import difflib
nltk.download('punkt')

# Load data
movies = pd.read_csv('tmdb_5000_movies.csv')
credits = pd.read_csv('tmdb_5000_credits.csv')
movies = movies.merge(credits, on='title')
movies = movies[['movie_id', 'title', 'overview', 'genres', 'keywords', 'cast', 'crew']]
movies.dropna(inplace=True)


# Process genres, keywords
def convert(obj):
    return [i['name'].replace(" ", "") for i in ast.literal_eval(obj)]


movies['genres'] = movies['genres'].apply(convert)
movies['keywords'] = movies['keywords'].apply(convert)


# Top 3 cast
def convert3(obj):
    return [i['name'].replace(" ", "") for i in ast.literal_eval(obj)[:3]]


movies['cast'] = movies['cast'].apply(convert3)


# Director
def fetch_director(obj):
    for i in ast.literal_eval(obj):
        if i['job'] == 'Director':
            return [i['name'].replace(" ", "")]
    return []


movies['crew'] = movies['crew'].apply(fetch_director)

# Overview processing
movies['overview'] = movies['overview'].apply(lambda x: x.split())

# Create tags
movies['tags'] = movies['overview'] + movies['genres'] + movies['keywords'] + movies['cast'] + movies['crew']
new_df = movies[['movie_id', 'title', 'tags']]
new_df['tags'] = new_df['tags'].apply(lambda x: " ".join(x).lower())

# Stemming
ps = PorterStemmer()


def stem(text):
    return " ".join([ps.stem(word) for word in text.split()])


new_df['tags'] = new_df['tags'].apply(stem)

# Vectorization
cv = CountVectorizer(max_features=5000, stop_words='english')
vectors = cv.fit_transform(new_df['tags']).toarray()

# Similarity
similarity = cosine_similarity(vectors)


# Recommendation function
def recommend(movie):
    movie = movie.lower()

    titles = new_df['title'].str.lower().tolist()
    # close_matches = difflib.get_close_matches(movie, titles, n=10, cutoff=0.4)
    #
    # if not close_matches:
    #     return ["Movie not found"], []
    #
    # if movie not in new_df['title'].str.lower().values:
    #     return ["Movie not found in database :( "]
    # # movie = close_matches[0]
    index = new_df[new_df['title'].str.lower() == movie].index[0]
    distances = similarity[index]
    movie_list = sorted(enumerate(distances), reverse=True, key=lambda x: x[1])[1:6]
    return [new_df.iloc[i[0]].title for i in movie_list]

movie_list = new_df['title'].tolist()
# Gradio interface
def recommend_interface(movie_name):
    return recommend(movie_name)


demo = gr.Interface(fn=recommend_interface,
                    inputs=gr.Dropdown(movie_list, label="Select a movie..."),
                    outputs=gr.List(label="Top 5 Recommendations"),
                    title="Movie Recommender")

if __name__ == "__main__":
    demo.launch()