app.py

import gradio as gr
import torch
import torch.nn as nn
import pandas as pd
import numpy as np
import re
from sklearn.preprocessing import LabelEncoder
from collections import Counter
from textblob import TextBlob
import random

# 1. Sample Data
data = {
    'text': [
        "I'm so happy today!",
        "This makes me really angry",
        "Feeling the love",
        "I'm so sad right now",
        "Just a normal day"
    ],
    'label': ["joy", "anger", "love", "sadness", "neutral"]
}
df = pd.DataFrame(data)

# 2. Simple Model
class EmotionModel(nn.Module):
    def __init__(self, vocab_size=100, embed_dim=32, num_classes=5):
        super().__init__()
        self.embedding = nn.Embedding(vocab_size, embed_dim)
        self.fc = nn.Linear(embed_dim, num_classes)
    
    def forward(self, x):
        x = self.embedding(x)
        return self.fc(x.mean(1))

# 3. Initialize Components
def clean_text(text):
    return re.sub(r'[^a-z\s]', '', str(text).lower())

vocab = {word:i+2 for i,word in enumerate(set(word for text in df['text'] for word in clean_text(text).split()))}
vocab['<PAD>'] = 0
vocab['<UNK>'] = 1

le = LabelEncoder()
le.fit(df['label'])

model = EmotionModel(len(vocab), num_classes=len(le.classes_))
model.eval()

# 4. Prediction
def predict(text):
    tokens = [vocab.get(word, vocab['<UNK>']) for word in clean_text(text).split()[:32]]
    inp = torch.tensor([tokens + [0]*(32-len(tokens))], dtype=torch.long)
    with torch.no_grad():
        return le.classes_[model(inp).argmax().item()]

# 5. Chat Interface
def chat(message, history, waiting):
    if isinstance(waiting, bool) and waiting:
        if message.lower() == "continue":
            return "", history + [(message, "Ready to continue!")], False
        return "", history + [(message, "Still waiting... say 'continue'")], True
    
    if message.lower() in ["wait", "pause"]:
        return "", history + [(message, "I'll wait. Say 'continue' when ready")], True
    
    emotion = predict(message)
    responses = {
        "joy": ["Great to hear you're happy!", "😊"],
        "anger": ["I hear your frustration", "Would you like to talk about it?"],
        "love": ["Love is wonderful!", "❤️"],
        "sadness": ["I'm here for you", "It's okay to feel sad"],
        "neutral": ["I see", "Tell me more"]
    }
    reply = random.choice(responses.get(emotion, ["I understand"]))
    return "", history + [(message, reply)], False

# 6. Gradio App
with gr.Blocks() as app:
    gr.Markdown("# Emotion Chatbot")
    
    chatbot = gr.Chatbot()
    message = gr.Textbox(label="Your Message")
    state = gr.State(value=False)  # Initialize with explicit value
    
    message.submit(
        chat,
        inputs=[message, chatbot, state],
        outputs=[message, chatbot, state]
    )
    gr.Button("Send").click(
        chat,
        inputs=[message, chatbot, state],
        outputs=[message, chatbot, state]
    )

app.launch()