app.py

import gradio as gr
from datasets import load_dataset
from transformers import T5ForConditionalGeneration, T5Tokenizer, AutoTokenizer, AutoModelForSequenceClassification
import random
import torch
import groq  # Assuming you are using the Groq library
import os
from dotenv import load_dotenv
from huggingface_hub import login

# Load environment variables from .env file
load_dotenv()
HUGGING_FACE_TOKEN = os.getenv("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")

# Authenticate with Hugging Face (use your token)
login(HUGGING_FACE_TOKEN)

# Load the mental health counseling conversations dataset
ds = load_dataset("Amod/mental_health_counseling_conversations")
context = ds["train"]["Context"]
response = ds["train"]["Response"]

GROQ_API_KEY = "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"

client = groq.Groq(api_key=GROQ_API_KEY)

# Load FLAN-T5 model and tokenizer for primary RAG
flan_tokenizer = T5Tokenizer.from_pretrained("google/flan-t5-small")
flan_model = T5ForConditionalGeneration.from_pretrained("google/flan-t5-small")

# Load sentiment analysis model
sentiment_tokenizer = AutoTokenizer.from_pretrained("distilbert-base-uncased-finetuned-sst-2-english")
sentiment_model = AutoModelForSequenceClassification.from_pretrained("distilbert-base-uncased-finetuned-sst-2-english")

# Groq client setup (assuming you have an API key)
client = groq.Groq(api_key=GROQ_API_KEY)  # Corrected Groq client initialization
# Function for sentiment analysis
def analyze_sentiment(text):
    inputs = sentiment_tokenizer(text, return_tensors="pt")
    outputs = sentiment_model(**inputs)
    probs = torch.nn.functional.softmax(outputs.logits, dim=-1)
    sentiment = "positive" if torch.argmax(probs) == 1 else "negative"
    confidence = probs.max().item()
    return sentiment, confidence

# Function to generate response based on sentiment and user input
def generate_response(sentiment, user_input):
    prompt = f"The user feels {sentiment}. Respond with supportive advice based on: {user_input}"
    inputs = flan_tokenizer(prompt, return_tensors="pt")
    response = flan_model.generate(**inputs, max_length=150)
    return flan_tokenizer.decode(response[0], skip_special_tokens=True)
# Main chatbot function
def chatbot(user_input):
    if not user_input.strip():
        return "Please enter a question or concern to receive guidance."

    # Word count limit
    word_count = len(user_input.split())
    max_words = 50
    remaining_words = max_words - word_count
    if remaining_words < 0:
        return f"Your input is too long. Please limit it to {max_words} words."

    # Sentiment analysis
    sentiment, confidence = analyze_sentiment(user_input)

    # Groq API fallback for a personalized response
    try:
        brief_response = client.chat.completions.create(
            messages=[{
                "role": "user",
                "content": user_input,
            }],
            model="llama3-8b-8192",  # Change model if needed
        )
        brief_response = brief_response.choices[0].message.content
    except Exception as e:
        brief_response = None

    if brief_response:
        return f"**Personalized Response from Groq:** {brief_response}"

    # Fallback to FLAN-T5 model for response generation
    response_text = generate_response(sentiment, user_input)
def generate_response(user_input):
    # Generate response using FLAN-T5
    inputs = flan_tokenizer.encode("summarize: " + user_input, return_tensors="pt", max_length=512, truncation=True)
    summary_ids = flan_model.generate(inputs, max_length=100, num_beams=4, early_stopping=True)
    generated_response = flan_tokenizer.decode(summary_ids[0], skip_special_tokens=True)

    if not generated_response:
        return "I'm sorry, I don't have information specific to your concern. Please consult a professional."

    # Final response with different sources
    complete_response = (
        f"**Sentiment Analysis:** {sentiment} (Confidence: {confidence:.2f})\n\n"
        f"**Generated Response:**\n{generated_response}\n\n"
        f"**Contextual Information:**\n{context_text}\n\n"
        f"**Additional Dataset Response:**\n{dataset_response}\n\n"
        f"Words entered: {word_count}, Words remaining: {remaining_words}"
    )

    return complete_response

# Example call to the function
response = generate_response("This is an example input.")
print(response)
# Set up Gradio interface
interface = gr.Interface(
    fn=chatbot,
    inputs=gr.Textbox(
        label="Ask your question:",
        placeholder="Describe how you're feeling today...",
        lines=4
    ),
    outputs=gr.Markdown(label="Psychologist Assistant Response"),
    title="Virtual Psychiatrist Assistant",
    description="Enter your mental health concerns, and receive guidance and responses from a trained assistant.",
    theme="huggingface"
)

# Launch the app
interface.launch()