import gradio as gr
import torch
import re
from underthesea import word_tokenize
# Load model directly
from transformers import AutoTokenizer, AutoModelForSequenceClassification

tokenizer = AutoTokenizer.from_pretrained("VietTung04/videberta-base-topic-classification")
model = AutoModelForSequenceClassification.from_pretrained("VietTung04/videberta-base-topic-classification")

# Check if GPU is available and set the device accordingly
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
model.to(device)

def preprocess_fn(text):
    stopword_path = 'vietnamese-stopwords.txt'
    
    with open(stopword_path, 'r', encoding='utf-8') as file:
        stopwords = file.read().splitlines()
    
    def remove_stopwords(tokens):
        return [word for word in tokens if word not in stopwords]
        
    text = re.sub(r'http\S+', ' ', text)  # Remove URLs
    text = re.sub(r'#\w+', ' ', text)  # Remove hashtags
    text = re.sub(r'@\w+', ' ', text)  # Remove mentions
    text = re.sub(r'\d+', ' ', text)  # Remove numbers
    text = re.sub(r'[^\w\sđĐàÀảẢãÃáÁạẠăĂằẰẳẲẵẴắẮặẶâÂầẦẩẨẫẪấẤậẬèÈẻẺẽẼéÉẹẸêÊềỀểỂễỄếẾệỆìÌỉỈĩĨíÍịỊòÒỏỎõÕóÓọỌôÔồỒổỔỗỖốỐộỘơƠờỜởỞỡỠớỚợỢùÙủỦũŨúÚụỤưƯừỪửỬữỮứỨựỰỳỲỷỶỹỸýÝỵỴ]', ' ', text)  # Remove special characters
    # Tokenize Vietnamese text
    tokens = word_tokenize(' '.join(text.split()).lower())
    
    # Remove stop words
    tokens = remove_stopwords(tokens)

    return ' '.join(tokens)

def predict_topic(text):
    inputs = tokenizer(
        preprocess_fn(text),
        truncation=True,
        padding='max_length',
        max_length=512,
        add_special_tokens=True,
        return_tensors='pt'
    )
    inputs = {key: value.to(device) for key, value in inputs.items()}
    with torch.no_grad():
        outputs = model(**inputs)
        logits = outputs.logits
        probabilities = torch.softmax(logits, dim=1).cpu().numpy()[0]

    # Get the top 3 classes
    top3_indices = probabilities.argsort()[-3:][::-1]
    top3_probabilities = probabilities[top3_indices]
    top3_classes = [model.config.id2label[idx] for idx in top3_indices]  # Assuming your model has this attribute

    return {top3_classes[i]: float(top3_probabilities[i]) for i in range(3)}

# Define the Gradio interface
iface = gr.Interface(
    fn=predict_topic,
    inputs=gr.Textbox(lines=2, placeholder="Enter your text here..."),
    outputs=gr.Label(num_top_classes=3),
    title="Text Classification",
    description="Enter text to classify it into different categories and get the probability for each class."
)

# Launch the interface
iface.launch()