import streamlit as st
from transformers import pipeline

# Initialize sentiment analysis pipeline
sentiment_pipeline = pipeline("sentiment-analysis")

# Use a sarcasm detection model
sarcasm_model_name = "mrm8488/t5-base-finetuned-sarcasm-twitter"  # Correct model name for sarcasm detection

# Create the sarcasm detection pipeline
sarcasm_pipeline = pipeline("text2text-generation", model=sarcasm_model_name)

def classify_sentence(sentence):
    # Detect sarcasm
    sarcasm_result = sarcasm_pipeline(sentence)[0]['generated_text']
    is_sarcastic = sarcasm_result.strip().lower() == 'true'

    # Detect sentiment
    sentiment_result = sentiment_pipeline(sentence)[0]
    sentiment_label = sentiment_result['label']
    sentiment_score = sentiment_result['score']

    # Determine sentiment
    if sentiment_label == "NEGATIVE":
        sentiment = "negative"
    elif sentiment_label == "POSITIVE":
        sentiment = "positive"
    else:
        sentiment = "neutral"

    # Handle sarcasm
    if is_sarcastic:
        sentiment += " (sarcastic)"

    return sentiment

# Streamlit app
st.title("Sentence Analyzer")

# User input
sentence = st.text_input("Enter a sentence:", "")

if st.button("Analyze"):
    if sentence:
        classification = classify_sentence(sentence)
        st.write(f"Sentence: {sentence}")
        st.write(f"Classification: {classification}")
    else:
        st.write("Please enter a sentence to analyze.")

# Example sentences
st.subheader("Example Sentences")
example_sentences = [
    "they are so beautiful",
    "This is the best day of my life.",
    "I'm not happy with your work.",
    "Yeah,you are not a good person!"
]

if st.button("Analyze Example Sentences"):
    for sentence in example_sentences:
        st.write(f"Sentence: {sentence}")
        st.write(f"Classification: {classify_sentence(sentence)}")
        st.write()