src/streamlit_app.py

"""
Streamlit Dashboard for DLRM Book Recommendation System - Hugging Face Space Compatible
Simple interface for DLRM-based book recommendations optimized for HF Spaces
"""

import os
import sys
import streamlit as st

# Force CPU-only mode for Hugging Face Spaces
os.environ['CPU_ONLY'] = 'true'
os.environ['CUDA_VISIBLE_DEVICES'] = ''

# Disable Streamlit telemetry for HF Spaces
os.environ['STREAMLIT_BROWSER_GATHER_USAGE_STATS'] = 'false'

import pandas as pd
import numpy as np
import warnings
warnings.filterwarnings('ignore')

# Page configuration
st.set_page_config(
    page_title="DLRM Book Recommendations",
    page_icon="📚",
    layout="wide",
    initial_sidebar_state="expanded"
)

# Custom CSS
st.markdown("""
<style>
    .main-header {
        font-size: 3rem;
        color: #1f77b4;
        text-align: center;
        margin-bottom: 2rem;
    }
    .cpu-mode-banner {
        background-color: #d4edda;
        color: #155724;
        padding: 0.75rem;
        border-radius: 0.5rem;
        border-left: 4px solid #28a745;
        margin: 1rem 0;
        text-align: center;
    }
    .book-card {
        background-color: #ffffff;
        padding: 1rem;
        border-radius: 0.5rem;
        border: 1px solid #e1e5eb;
        margin-bottom: 1rem;
    }
</style>
""", unsafe_allow_html=True)

@st.cache_data
def load_sample_data():
    """Load sample data for demo purposes"""
    # Sample book data
    sample_books = {
        'ISBN': ['0439023483', '0439358078', '0316666343', '0452264464', '0061120081'],
        'Book-Title': [
            'The Hunger Games',
            'Harry Potter and the Chamber of Secrets', 
            'The Catcher in the Rye',
            '1984',
            'To Kill a Mockingbird'
        ],
        'Book-Author': [
            'Suzanne Collins',
            'J.K. Rowling',
            'J.D. Salinger', 
            'George Orwell',
            'Harper Lee'
        ],
        'Year-Of-Publication': [2008, 1999, 1951, 1949, 1960],
        'Publisher': ['Scholastic', 'Scholastic', 'Little, Brown', 'Signet', 'Harper']
    }
    
    # Sample users
    sample_users = {
        'User-ID': [1, 2, 3, 4, 5],
        'Age': [25, 32, 19, 45, 28],
        'Location': ['New York, USA', 'London, UK', 'Tokyo, Japan', 'Berlin, Germany', 'Toronto, Canada']
    }
    
    # Sample ratings
    sample_ratings = {
        'User-ID': [1, 1, 2, 2, 3, 3, 4, 4, 5, 5],
        'ISBN': ['0439023483', '0439358078', '0316666343', '0452264464', '0061120081', '0439023483', '0316666343', '0439358078', '0452264464', '0061120081'],
        'Book-Rating': [9, 8, 7, 10, 8, 6, 9, 7, 8, 9]
    }
    
    books_df = pd.DataFrame(sample_books)
    users_df = pd.DataFrame(sample_users)  
    ratings_df = pd.DataFrame(sample_ratings)
    
    return books_df, users_df, ratings_df

def simulate_dlrm_prediction(user_id, book_isbn, user_data=None, book_data=None):
    """Simulate DLRM prediction for demo purposes"""
    # Simple heuristic-based simulation
    np.random.seed(hash(f"{user_id}_{book_isbn}") % 2**32)
    
    base_score = 0.5
    
    # User preferences (simulated)
    user_bias = np.random.uniform(-0.2, 0.2)
    
    # Book popularity (simulated)  
    book_bias = np.random.uniform(-0.1, 0.1)
    
    # Add some randomness
    noise = np.random.uniform(-0.05, 0.05)
    
    final_score = base_score + user_bias + book_bias + noise
    final_score = max(0.0, min(1.0, final_score))  # Clamp to [0,1]
    
    return final_score

def display_book_info(book_isbn, books_df, show_rating=None):
    """Display book information"""
    book_info = books_df[books_df['ISBN'] == book_isbn]
    
    if len(book_info) == 0:
        st.write(f"Book with ISBN {book_isbn} not found")
        return
    
    book = book_info.iloc[0]
    
    col1, col2 = st.columns([1, 3])
    
    with col1:
        # Placeholder book cover
        st.image("https://via.placeholder.com/150x200?text=📚&color=1f77b4&bg=f0f2f6", width=150)
    
    with col2:
        st.markdown(f"**{book['Book-Title']}**")
        st.write(f"*by {book['Book-Author']}*")
        st.write(f"📅 Published: {book.get('Year-Of-Publication', 'Unknown')}")
        st.write(f"🏢 Publisher: {book.get('Publisher', 'Unknown')}")
        st.write(f"📖 ISBN: {book['ISBN']}")
        
        if show_rating is not None:
            st.markdown(f"**🎯 DLRM Score: {show_rating:.4f}**")

def main():
    # Header
    st.markdown('<h1 class="main-header">📚 DLRM Book Recommendation System</h1>', unsafe_allow_html=True)
    st.markdown("### Deep Learning Recommendation Model for Personalized Book Suggestions")
    
    # HF Space optimized banner
    st.markdown('''
    <div class="cpu-mode-banner">
        🚀 Optimized for Hugging Face Spaces - CPU-only mode with simulated DLRM predictions
    </div>
    ''', unsafe_allow_html=True)
    
    st.markdown("---")

    # Load sample data
    with st.spinner("Loading sample data..."):
        books_df, users_df, ratings_df = load_sample_data()
    
    # Sidebar info
    st.sidebar.title("📊 Demo Dataset")
    st.sidebar.metric("📚 Sample Books", len(books_df))
    st.sidebar.metric("👥 Sample Users", len(users_df))
    st.sidebar.metric("⭐ Sample Ratings", len(ratings_df))
    
    st.sidebar.markdown("---")
    st.sidebar.markdown("""
    ### 🔧 HF Space Features:
    - CPU-only processing
    - Simulated DLRM predictions
    - Sample dataset demo
    - No GPU dependencies
    """)
    
    # Main interface
    tab1, tab2, tab3 = st.tabs(["🎯 Get Recommendations", "📊 How DLRM Works", "🔍 Book Explorer"])
    
    with tab1:
        st.header("🎯 DLRM Book Recommendations (Simulated)")
        st.info("Demo of DLRM-based recommendations using simulated predictions")
        
        # User selection
        col1, col2 = st.columns([2, 1])
        
        with col1:
            selected_user_id = st.selectbox("Select a user", users_df['User-ID'].tolist())
        
        with col2:
            num_recommendations = st.slider("Number of recommendations", 3, 5, 5)
        
        # Show user info
        user_info = users_df[users_df['User-ID'] == selected_user_id]
        if len(user_info) > 0:
            user = user_info.iloc[0]
            st.markdown(f"**User Info**: Age: {user.get('Age', 'Unknown')}, Location: {user.get('Location', 'Unknown')}")
        
        # User's reading history
        user_ratings = ratings_df[ratings_df['User-ID'] == selected_user_id]
        if len(user_ratings) > 0:
            with st.expander(f"📖 User's Reading History ({len(user_ratings)} books)", expanded=True):
                for _, rating in user_ratings.iterrows():
                    book_info = books_df[books_df['ISBN'] == rating['ISBN']]
                    if len(book_info) > 0:
                        book = book_info.iloc[0]
                        st.write(f"• **{book['Book-Title']}** by {book['Book-Author']} - {rating['Book-Rating']}/10 ⭐")
        
        if st.button("🚀 Get Simulated DLRM Recommendations", type="primary"):
            with st.spinner("🤖 Simulating DLRM analysis..."):
                
                # Get books not rated by user
                user_rated_books = set(user_ratings['ISBN']) if len(user_ratings) > 0 else set()
                candidate_books = [isbn for isbn in books_df['ISBN'] if isbn not in user_rated_books]
                
                # Get simulated recommendations
                recommendations = []
                for book_isbn in candidate_books:
                    score = simulate_dlrm_prediction(selected_user_id, book_isbn)
                    recommendations.append((book_isbn, score))
                
                # Sort and take top recommendations
                recommendations.sort(key=lambda x: x[1], reverse=True)
                recommendations = recommendations[:num_recommendations]
            
            st.success(f"Generated {len(recommendations)} simulated DLRM recommendations!")
            
            st.subheader("🎯 Simulated DLRM Recommendations")
            
            for i, (book_isbn, score) in enumerate(recommendations, 1):
                with st.expander(f"{i}. Recommendation (Simulated DLRM Score: {score:.4f})", expanded=(i <= 2)):
                    display_book_info(book_isbn, books_df, show_rating=score)
                    
                    # Additional info
                    st.markdown(f"""
                    **📊 Prediction Details:**
                    - User ID: {selected_user_id}
                    - Book ISBN: {book_isbn}
                    - Simulated DLRM Confidence: {score:.1%}
                    - Recommendation Rank: #{i}
                    """)
    
    with tab2:
        st.header("📊 How DLRM Works for Book Recommendations")
        
        st.markdown("""
        ## 🤖 Deep Learning Recommendation Model (DLRM)
        
        DLRM is specifically designed for recommendation systems and offers several advantages over traditional approaches:
        
        ### 🏗️ Architecture Benefits:
        """)
        
        col1, col2 = st.columns(2)
        
        with col1:
            st.markdown("""
            **🔧 Technical Features:**
            - Multi-feature processing
            - Embedding tables for categorical features
            - Cross-feature interactions
            - Scalable design for large datasets
            - Real-time inference capability
            """)
        
        with col2:
            st.markdown("""
            **📊 Input Features:**
            - User ID, Age, Location
            - Book ID, Publisher, Publication Year
            - Rating patterns and user activity
            - Cross-feature interactions
            """)
        
        st.markdown("""
        ### 🎯 Why DLRM vs Traditional Methods:
        
        | Feature | DLRM | Traditional CF | Content-Based |
        |---------|------|----------------|---------------|
        | **Feature Integration** | ✅ Excellent | ❌ Limited | ⚠️ Moderate |
        | **Cold Start Problem** | ✅ Handles well | ❌ Poor | ✅ Good |
        | **Scalability** | ✅ Highly scalable | ⚠️ Moderate | ✅ Good |
        | **Accuracy** | ✅ High | ⚠️ Moderate | ⚠️ Moderate |
        | **Real-time Inference** | ✅ Fast | ⚠️ Slow | ✅ Fast |
        
        ### 💡 Best Use Cases:
        - **E-commerce**: Product recommendations
        - **Streaming**: Content recommendations  
        - **Publishing**: Book/article suggestions
        - **Social Media**: Feed optimization
        """)
        
        # Demo architecture visualization
        st.subheader("🏗️ DLRM Architecture Overview")
        
        st.markdown("""
        ```
        User Features     Book Features
        ┌─────────────┐   ┌─────────────┐
        │ User ID     │   │ Book ID     │
        │ Age Group   │   │ Publisher   │  
        │ Location    │   │ Decade      │
        └─────────────┘   └─────────────┘
               │                 │
               ▼                 ▼
        ┌─────────────────────────────┐
        │    Embedding Tables         │
        └─────────────────────────────┘
               │
               ▼
        ┌─────────────────────────────┐
        │   Cross-Feature Network     │
        └─────────────────────────────┘
               │
               ▼
        ┌─────────────────────────────┐
        │   Rating Prediction         │
        │   (0.0 - 1.0 score)         │
        └─────────────────────────────┘
        ```
        """)
    
    with tab3:
        st.header("🔍 Book Explorer")
        st.info("Browse sample books and see simulated DLRM predictions")
        
        # Book selection
        selected_book_isbn = st.selectbox("Select a book", books_df['ISBN'].tolist())
        selected_user_for_prediction = st.selectbox("Select user for prediction", users_df['User-ID'].tolist(), key="pred_user")
        
        # Display selected book
        st.subheader("📚 Selected Book")
        display_book_info(selected_book_isbn, books_df)
        
        # Show prediction
        if st.button("🎯 Get Simulated DLRM Prediction"):
            with st.spinner("Calculating simulated prediction..."):
                prediction_score = simulate_dlrm_prediction(selected_user_for_prediction, selected_book_isbn)
            
            st.success(f"Simulated DLRM Prediction: {prediction_score:.4f}")
            
            # Interpretation
            if prediction_score > 0.7:
                st.success("🎯 High recommendation confidence - User likely to enjoy this book!")
            elif prediction_score > 0.5:
                st.info("⚖️ Moderate recommendation confidence - Could be interesting for user")
            else:
                st.warning("📉 Low recommendation confidence - May not match user preferences")
        
        # All books overview
        st.subheader("📚 All Sample Books")
        
        for _, book in books_df.iterrows():
            with st.expander(f"{book['Book-Title']} by {book['Book-Author']}"):
                col1, col2 = st.columns([2, 1])
                
                with col1:
                    st.write(f"**ISBN:** {book['ISBN']}")
                    st.write(f"**Publisher:** {book['Publisher']}")
                    st.write(f"**Year:** {book['Year-Of-Publication']}")
                
                with col2:
                    # Show ratings from sample users
                    book_ratings = ratings_df[ratings_df['ISBN'] == book['ISBN']]
                    if len(book_ratings) > 0:
                        avg_rating = book_ratings['Book-Rating'].mean()
                        st.metric("Avg Rating", f"{avg_rating:.1f}/10")
                        st.metric("# Ratings", len(book_ratings))
    
    # Footer
    st.markdown("---")
    st.markdown("""
    ### 🚀 About this Demo
    
    This is a **Hugging Face Space** compatible version of a DLRM Book Recommendation System:
    
    - **CPU-only processing**: No GPU or NVIDIA drivers required
    - **Simulated predictions**: Demonstrates DLRM concept with heuristic-based scoring
    - **Sample dataset**: 5 popular books and 5 sample users
    - **Educational purpose**: Shows how DLRM would work in production
    
    **For production use:**
    - Train actual DLRM model with PyTorch/TorchRec
    - Use full book datasets (millions of books/users)
    - Deploy on GPU infrastructure for better performance
    - Implement proper feature engineering and preprocessing
    
    **🔗 Learn more about DLRM:** [Facebook Research DLRM](https://github.com/facebookresearch/dlrm)
    """)

if __name__ == "__main__":
    main()