import gradio as gr
import numpy as np
from tensorflow.keras.preprocessing.text import text_to_word_sequence
from tensorflow.keras.preprocessing.sequence import pad_sequences
from gensim.models import KeyedVectors
from tensorflow.keras.models import load_model
import gensim.downloader as api

# Load the pre-trained Word2Vec model
word2vec_transfer = api.load("glove-wiki-gigaword-100")

# Define the function to embed a sentence with the pre-trained Word2Vec model
def embed_sentence_with_TF(word2vec, sentence):
    embedded_sentence = []
    for word in sentence:
        if word in word2vec:
            embedded_sentence.append(word2vec[word])
    return np.array(embedded_sentence)

# Define the function to preprocess a new movie review
def preprocess_review(review):
    # Tokenize the review
    review = text_to_word_sequence(review)
    # Embed the review with the pre-trained Word2Vec model
    review_embedded = embed_sentence_with_TF(word2vec_transfer, review)
    # Pad the embedded review
    review_padded = pad_sequences([review_embedded], dtype='float32', padding='post', maxlen=200)
    return review_padded

# Load the trained model
model = load_model('my_model.h5')

def predict_sentiment(review):
    # Preprocess the review
    review_padded = preprocess_review(review)
    # Predict the sentiment
    sentiment = model.predict(review_padded)[0][0]
    if sentiment > 0.5:
        return "Positive"
    elif sentiment == 0.5:
        return "Neutral"
    else:
        return "Negative"

# Create a Gradio interface
inputs = gr.inputs.Textbox(lines=5, label="Input Text")
outputs = gr.outputs.Textbox(label="Sentiment")
title = "Sentiment Analysis"
description = "Enter a text and get the sentiment prediction."
gr.Interface(fn=predict_sentiment, inputs=inputs, outputs=outputs, title=title, description=description).launch(share=True)