import streamlit as st
from transformers import pipeline
import pandas as pd
from datetime import datetime
from PIL import Image
import requests
from io import BytesIO

# Constants
RATING_MAP = {
    0: "Negative (⭐)",
    1: "Neutral (⭐⭐)",
    2: "Positive (⭐⭐⭐)"
}

# Emoji mapping based on ratings
EMOJI_MAP = {
    0: "😠",  # Angry face for negative
    1: "😐",  # Neutral face
    2: "😊"   # Happy face
}

@st.cache_resource
def load_models():
    # Load sentiment analysis model
    sentiment_model = pipeline(
        "text-classification", 
        model="AndrewLi403/CustomModel_tripadvisor_finetuned"
    )
    # Load text-to-emoji model
    emoji_pipe = pipeline("text-classification", model="j-hartmann/emotion-english-roberta-large")
    return sentiment_model, emoji_pipe

def analyze_review(text, sentiment_model, emoji_pipe):
    # Sentiment analysis
    sentiment_result = sentiment_model(text)[0]
    rating = int(sentiment_result['label'].split('_')[-1])
    
    # Emoji analysis
    emoji_result = emoji_pipe(text)[0]
    dominant_emoji = emoji_result['label']
    
    return {
        'sentiment': RATING_MAP[rating],
        'sentiment_score': sentiment_result['score'],
        'rating': rating,
        'dominant_emoji': dominant_emoji,
        'emoji_confidence': emoji_result['score']
    }

def main():
    st.title("Hotel Review Analyzer")
    st.markdown("Analyze sentiment and detect emotional tone")
    
    # Load models
    sentiment_model, emoji_pipe = load_models()
    
    # Input
    review_text = st.text_area("Paste your hotel review here:", height=150)
    
    if st.button("Analyze"):
        if review_text:
            with st.spinner("Analyzing emotions..."):
                # Get analysis results
                results = analyze_review(review_text, sentiment_model, emoji_pipe)
                
                # Display results
                st.subheader("Analysis Results")
                
                # First row: Rating and Emoji
                col1, col2 = st.columns(2)
                with col1:
                    st.metric("Sentiment Rating", 
                            results['sentiment'],
                            delta=f"{results['sentiment_score']:.2f}")
                
                with col2:
                    # Display both system emoji and detected emoji
                    st.metric("Emotional Tone",
                            f"{EMOJI_MAP[results['rating']]} ",
                            delta=f"Confidence: {results['emoji_confidence']:.2f}")
                
                # Visual emoji display
                st.subheader("Emotional Response")
                cols = st.columns(3)
                with cols[1]:
                    st.header(EMOJI_MAP[results['rating']] * 5)  # Repeat emoji for visual impact
                    st.caption("Based on your star rating")
                
                # Emotion breakdown
                with st.expander("Detailed Emotion Analysis"):
                    full_emoji_results = emoji_pipe(review_text, top_k=5)
                    for emotion in full_emoji_results:
                        st.progress(
                            int(emotion['score'] * 100),
                            text=f"{emotion['label']}: {emotion['score']:.2f}"
                        )
        else:
            st.error("Please enter a review to analyze")

if __name__ == "__main__":
    main()