Create app.py

app.py

@@ -0,0 +1,228 @@
+import streamlit as st
+import requests
+from transformers import pipeline
+
+# Enter your TMDb API key here
+TMDB_API_KEY = '2cc77b08a475b510f54a0dcca2c5c0c7'
+IMAGE_BASE_URL = "https://image.tmdb.org/t/p/w200"  # TMDb base URL for poster images (width 200)
+
+@st.cache_resource
+def load_emotion_analyzer():
+    return pipeline("text-classification", model="j-hartmann/emotion-english-distilroberta-base")
+
+emotion_analyzer = load_emotion_analyzer()
+
+def analyze_emotion(description):
+    """Analyze emotion of a movie's description."""
+    emotions = emotion_analyzer(description)
+    return emotions[0]['label']  # Return the top emotion
+
+def get_movie_genres():
+    """Fetches movie genres from TMDb API and returns a dictionary of genres."""
+    url = f'https://api.themoviedb.org/3/genre/movie/list?api_key={TMDB_API_KEY}&language=en-US'
+    response = requests.get(url)
+    if response.status_code == 200:
+        genres = response.json().get('genres', [])
+        return {genre['name']: genre['id'] for genre in genres}
+    else:
+        return None  # Return None if the request failed
+
+def get_top_movies_by_genre(genre_id):
+    """Fetches the top 10 highest-rated movies for a given genre ID from TMDb, sorted by title length."""
+    url = f'https://api.themoviedb.org/3/discover/movie?api_key={TMDB_API_KEY}&language=en-US&sort_by=vote_average.desc&vote_count.gte=1000&with_genres={genre_id}'
+    response = requests.get(url)
+    if response.status_code == 200:
+        movies = response.json().get('results', [])[:10]  # Get the top 10 results
+        # Sort movies by title length, shortest to longest
+        return sorted([(movie['title'], movie['vote_average'], movie['id'], movie['poster_path']) for movie in movies], key=lambda x: len(x[0]))
+    else:
+        return None
+
+def get_similar_movies(movie_id, genre_id):
+    """Fetches similar movies for a given movie ID from TMDb and filters by the selected genre."""
+    url = f'https://api.themoviedb.org/3/movie/{movie_id}/similar?api_key={TMDB_API_KEY}&language=en-US'
+    response = requests.get(url)
+    if response.status_code == 200:
+        movies = response.json().get('results', [])[:15]
+        # Include the movie description (overview) in the returned values
+        return [
+            (movie['title'], movie['vote_average'], movie['poster_path'], movie['id'], movie['overview'])
+            for movie in movies if genre_id in movie['genre_ids'] and 'overview' in movie
+        ]
+    else:
+        return None
+
+def get_movie_details(movie_id):
+    """Fetches detailed information for a given movie ID from TMDb, including description, release date, director, and cast."""
+    url = f'https://api.themoviedb.org/3/movie/{movie_id}?api_key={TMDB_API_KEY}&language=en-US&append_to_response=credits'
+    response = requests.get(url)
+    if response.status_code == 200:
+        movie = response.json()
+        directors = [member['name'] for member in movie['credits']['crew'] if member['job'] == 'Director']
+        cast = [member['name'] for member in movie['credits']['cast'][:5]]  # Get top 5 cast members
+        return {
+            "title": movie['title'],
+            "rating": movie['vote_average'],
+            "description": movie['overview'],
+            "poster_path": movie['poster_path'],
+            "release_date": movie['release_date'],
+            "director": ", ".join(directors),
+            "cast": ", ".join(cast)
+        }
+    else:
+        return None
+
+# Initialize session state to store selected movies and recommendations
+if 'selected_movies' not in st.session_state:
+    st.session_state['selected_movies'] = set()
+if 'recommendations' not in st.session_state:
+    st.session_state['recommendations'] = []
+if 'last_genre' not in st.session_state:
+    st.session_state['last_genre'] = None
+
+# Streamlit app title
+st.title("Movie Recommender")
+
+# CSS for alignment and button styling
+st.markdown("""
+    <style>
+    .movie-container {
+        display: flex;
+        flex-direction: column;
+        align-items: center;
+        justify-content: flex-start;
+    }
+    .movie-caption {
+        min-height: 50px;  /* Set a consistent height for caption sections */
+        display: flex;
+        align-items: center;
+        justify-content: center;
+        text-align: center;
+    }
+    .movie-caption button {
+        width: 100%;
+    }
+    </style>
+""", unsafe_allow_html=True)
+
+# Fetch genres from TMDb and display in a dropdown
+genres = get_movie_genres()
+if genres is None:
+    st.error("Failed to load movie genres. Please check your API key or connection.")
+else:
+    genre_name = st.selectbox("Select a movie genre:", [""] + list(genres.keys()))  # Include an empty option
+
+    # Check if genre has changed, and reset selections if so
+    if genre_name != st.session_state['last_genre']:
+        st.session_state['selected_movies'].clear()
+        st.session_state['recommendations'].clear()
+        st.session_state['last_genre'] = genre_name
+
+    # Only fetch and display movies if a genre is selected
+    if genre_name:
+        genre_id = genres[genre_name]  # Get the genre ID based on the selected genre name
+        top_movies = get_top_movies_by_genre(genre_id)  # Fetch top 10 movies in this genre, sorted by title length
+
+        # Check if movies were successfully loaded
+        if top_movies is None:
+            st.error("Failed to load movies. Please check your API key or connection.")
+        else:
+            # Display the top 10 movies as clickable titles in a 5x2 grid layout with aligned rows
+            st.write(f"Top 10 movies in the {genre_name} genre:")
+            columns = st.columns(5)  # Create 5 columns for each row
+
+            # First row of 5 movies
+            for index, (title, _, movie_id, poster_path) in enumerate(top_movies[:5]):
+                col = columns[index]
+                with col:
+                    st.markdown("<div class='movie-container'>", unsafe_allow_html=True)
+
+                    # Display image
+                    st.image(f"{IMAGE_BASE_URL}{poster_path}", use_container_width=True)
+
+                    # Display clickable title as a button to toggle selection, with consistent height
+                    if st.button(f"{title} {'✅' if movie_id in st.session_state['selected_movies'] else ''}", key=f"title_{movie_id}"):
+                        if movie_id in st.session_state['selected_movies']:
+                            st.session_state['selected_movies'].remove(movie_id)
+                        else:
+                            st.session_state['selected_movies'].add(movie_id)
+
+                    st.markdown("</div>", unsafe_allow_html=True)
+
+            # Second row of 5 movies
+            for index, (title, _, movie_id, poster_path) in enumerate(top_movies[5:]):
+                col = columns[index]
+                with col:
+                    st.markdown("<div class='movie-container'>", unsafe_allow_html=True)
+
+                    # Display image
+                    st.image(f"{IMAGE_BASE_URL}{poster_path}", use_container_width=True)
+
+                    # Display clickable title as a button to toggle selection, with consistent height
+                    if st.button(f"{title} {'✅' if movie_id in st.session_state['selected_movies'] else ''}", key=f"title_2_{movie_id}"):
+                        if movie_id in st.session_state['selected_movies']:
+                            st.session_state['selected_movies'].remove(movie_id)
+                        else:
+                            st.session_state['selected_movies'].add(movie_id)
+
+                    st.markdown("</div>", unsafe_allow_html=True)
+
+    # Button to generate recommendations based on selected movies
+    if st.button("Show Recommendations"):
+        if st.session_state['selected_movies']:
+            # Step 1: Collect emotions of user-selected movies
+            st.write("Analyzing emotions of selected movies...")
+            selected_emotions = []
+            for movie_id in st.session_state['selected_movies']:
+                movie_details = get_movie_details(movie_id)
+                if movie_details:
+                    emotion = analyze_emotion(movie_details["description"])
+                    selected_emotions.append(emotion)
+                    st.write(f"**{movie_details['title']}** - Emotion: {emotion}")
+
+            # Step 2: Collect similar movies for all selected movies
+            all_similar_movies = []
+            for movie_id in st.session_state['selected_movies']:
+                similar_movies = get_similar_movies(movie_id, genre_id)
+                if similar_movies:
+                    all_similar_movies.extend(similar_movies)
+
+            # Step 3: Filter similar movies by matching emotions and remove duplicates
+            recommendations = []
+            seen_movie_ids = set(st.session_state['selected_movies'])  # Start with user-selected movies to exclude them
+
+            with st.spinner("Filtering movies by matching emotions..."):
+                for title, rating, poster_path, movie_id, description in all_similar_movies:
+                    if movie_id not in seen_movie_ids:  # Exclude already-seen movies
+                        movie_emotion = analyze_emotion(description)
+                        if movie_emotion in selected_emotions:
+                            seen_movie_ids.add(movie_id)  # Mark this movie as seen to prevent duplicates
+                            recommendations.append((title, rating, poster_path, movie_id, movie_emotion))
+
+            # Store unique recommendations in session state
+            st.session_state['recommendations'] = recommendations
+
+            # Step 4: Display final recommendations
+            if st.session_state['recommendations']:
+                st.write("Based on your selections, you might also enjoy these top 5 movies:")
+                for title, rating, poster_path, movie_id, emotion in sorted(st.session_state['recommendations'], key=lambda x: x[1], reverse=True)[:5]:
+                    movie_details = get_movie_details(movie_id)
+                    if movie_details:
+                        col1, col2 = st.columns([1, 3])
+                        with col1:
+                            st.image(f"{IMAGE_BASE_URL}{movie_details['poster_path']}", use_container_width=True)
+                        with col2:
+                            st.markdown(f"**Title:** {movie_details['title']}")
+                            st.markdown(f"**TMDb Rating:** {movie_details['rating']}")
+                            st.markdown(f"**Emotion:** {emotion}")
+                            st.markdown(f"**Description:** {movie_details['description']}")
+
+                            # Additional Information button with overlay using expander
+                            with st.expander("Additional Information"):
+                                st.markdown(f"**Release Date:** {movie_details['release_date']}")
+                                st.markdown(f"**Director:** {movie_details['director']}")
+                                st.markdown(f"**Cast:** {movie_details['cast']}")
+            else:
+                st.warning("No movies matched your selected emotions.")
+      else:
+          st.warning("Please select at least one movie to generate recommendations.")