import streamlit as st
from keras.models import load_model
import nltk
import re
from nltk.tokenize import TweetTokenizer
from tensorflow.keras.preprocessing.text import Tokenizer
from tensorflow.keras.preprocessing.sequence import pad_sequences
import subprocess
import numpy as np

# Download NLTK stopwords if not already downloaded
try:
    nltk.data.find('corpora/stopwords')
except LookupError:
    nltk.download('stopwords')

# Additional imports
from nltk.corpus import stopwords

# Download NLTK punkt tokenizer if not already downloaded
try:
    nltk.data.find('tokenizers/punkt/PY3/english.pickle')
except LookupError:
    nltk.download('punkt')

# Additional imports
from nltk.tokenize import word_tokenize

# Load the LSTM model
model_path = "./my_model.h5"  # Set your model path here

def load_lstm_model(model_path):
    return load_model(model_path)



def clean_text(text):
    # Remove stopwords
    stop_words = set(stopwords.words('english'))
    words = word_tokenize(text)
    filtered_words = [word for word in words if word not in stop_words]
    
    # Remove Twitter usernames
    text = re.sub(r'@\w+', '', ' '.join(filtered_words))
    
    # Remove URLs
    text = re.sub(r'http\S+', '', text)
    
    # Tokenize using TweetTokenizer
    tokenizer = TweetTokenizer(preserve_case=True)
    text = tokenizer.tokenize(text)
    
    # Remove hashtag symbols
    text = [word.replace('#', '') for word in text]
    
    # Remove short words
    text = ' '.join([word.lower() for word in text if len(word) > 2])
    
    # Remove digits
    text = re.sub(r'\d+', '', text)
    
    # Remove non-alphanumeric characters
    text = re.sub(r'[^a-zA-Z\s]', '', text)
    
    return text

def preprocess_text(text):
    # Clean the text
    cleaned_text = clean_text(text)
    
    # Tokenize and pad sequences
    token = Tokenizer()
    token.fit_on_texts([cleaned_text])
    text_sequences = token.texts_to_sequences([cleaned_text])
    padded_sequences = pad_sequences(text_sequences, maxlen=100)
    
    return padded_sequences

# Function to predict hate speech
def predict_hate_speech(text, lstm_model):
    # Preprocess the text
    padded_sequences = preprocess_text(text)
    prediction = lstm_model.predict(padded_sequences)
    return prediction

# Main function to run the Streamlit app
def main():
    # Set up Streamlit UI
    st.title("Hate Speech Detection")
    st.write("Enter text below to detect hate speech:")
    input_text = st.text_area("Input Text", "")

    if st.button("Detect Hate Speech"):
        if input_text:
            # Load the model
            lstm_model = load_lstm_model(model_path)
            # Predict hate speech
            prediction = predict_hate_speech(input_text, lstm_model)
            # Convert the list to a numpy array
            arr = np.array(prediction[0])
            max_index = np.argmax(arr)
            if max_index == 1:
                #negative 
                st.error("Hate Speech Detected")
            else:
                st.success("No Hate Speech Detected")
        else:
            st.warning("Please enter some text")

# Run the app
if __name__ == "__main__":
    main()