app.py

import gradio as gr
from transformers import AutoTokenizer, AutoModelForCausalLM, GenerationConfig
import torch
import time

# =======================================================
# Global session state for multi-step questioning
# =======================================================
session_answers = {}

# =======================================================
# Load Model
# =======================================================
model_name = "augtoma/qCammel-13"
print("Loading tokenizer and model...")

tokenizer = AutoTokenizer.from_pretrained(model_name)
if tokenizer.pad_token is None:
    tokenizer.pad_token = tokenizer.eos_token

model = AutoModelForCausalLM.from_pretrained(
    model_name,
    device_map="auto",
    torch_dtype=torch.float16,
    trust_remote_code=True,
    low_cpu_mem_usage=True
)
model.eval()

print("Model loaded successfully!")
print(f"Device map: {model.hf_device_map}")
print(f"Model device: {next(model.parameters()).device}")

# =======================================================
# Generate Doctor Response
# =======================================================
def generate_doctor_response(history):
    global session_answers
    user_message = history[-1]["content"]

    if not user_message.strip():
        history.append({"role": "assistant", "content": "⚠️ Please describe your symptoms or ask a question."})
        yield history
        return

    # Build prompt with context
    prompt = """# 🩺 You are a highly knowledgeable Medical Expert

        **Roles:**
        - **Doctor:** Diagnose symptoms, ask relevant follow-up questions, and provide medical advice.
        - **Nutritionist:** Give diet and lifestyle recommendations.
        - **Medical Teacher:** Explain complex medical terms and conditions in simple, understandable language.
        
        **Guidelines:**
        1. Ask **only one follow-up question at a time** to gather necessary information.
        2. Provide advice only after collecting enough information.
        3. Be **empathetic, professional, and conversational**.
        4. Provide disclaimers when needed:  
           ⚕️ *This is AI-generated information and not a substitute for professional medical advice. Please consult a healthcare provider for proper diagnosis and treatment.*
        5. Keep answers concise, clear, and actionable.
        
        Use this expertise to respond naturally to any patient message, balancing teaching, advice, and medical guidance.
        """

    recent_history = history[-10:-1] if len(history) > 10 else history[:-1]
    for msg in recent_history:
        role = "Patient" if msg["role"] == "user" else "Doctor"
        content = msg['content'].replace("⚕️ *Note: This is AI-generated information*", "").strip()
        prompt += f"{role}: {content}\n"
    prompt += f"Patient: {user_message}\nDoctor:"

    # Tokenize input
    inputs = tokenizer(prompt, return_tensors="pt").to(model.device)

    # Generation configuration for concise, interactive answers
    gen_config = GenerationConfig(
        temperature=0.7,
        top_p=0.9,
        do_sample=True,
        max_new_tokens=500,  # short answers
        pad_token_id=tokenizer.pad_token_id,
        eos_token_id=tokenizer.eos_token_id,
        repetition_penalty=1.2
    )

    input_len = inputs["input_ids"].shape[1]

    with torch.no_grad():
        output_ids = model.generate(**inputs, generation_config=gen_config)

    generated_ids = output_ids[0][input_len:]
    response = tokenizer.decode(generated_ids, skip_special_tokens=True).strip()

    # Take only first 2-3 sentences to make it concise
    response = ". ".join(response.split(". ")[:3]).strip()
    if response.lower().startswith("doctor:"):
        response = response[7:].strip()
    if len(response) < 10:
        response = "I understand your concern. Could you please provide more details about your symptoms?"

    # Add assistant placeholder for streaming
    history.append({"role": "assistant", "content": ""})

    # Stream response token by token
    for i in range(0, len(response), 4):
        chunk = response[:i+4]
        history[-1]["content"] = chunk + "▌"
        yield history.copy()
        time.sleep(0.015)

    # Final response
    history[-1]["content"] = response
    yield history

# =======================================================
# Gradio Interface
# =======================================================
with gr.Blocks(theme=gr.themes.Soft()) as demo:
    gr.Markdown("# 🩺 AI Doctor Chat Assistant")

    chatbot = gr.Chatbot(
        label="💬 Doctor Consultation",
        type='messages',
        avatar_images=(
            "https://cdn-icons-png.flaticon.com/512/706/706830.png",  # Patient
            "https://cdn-icons-png.flaticon.com/512/3774/3774299.png"   # Doctor
        ),
        height=500
    )

    with gr.Row():
        user_input = gr.Textbox(
            placeholder="Type your symptoms or question here...",
            label="🧍 Your Message",
            lines=2,
            scale=4
        )

    with gr.Row():
        send_btn = gr.Button("💬 Send", variant="primary", scale=1)
        clear_btn = gr.Button("🧹 Clear Chat", scale=1)

    gr.Examples(
        examples=[
            "I have a fever of 102°F since yesterday",
            "I've been having headaches for the past week",
            "I feel very tired all the time",
            "I have a sore throat and body aches",
        ],
        inputs=user_input,
        label="💡 Example Questions"
    )

    def respond(message, history):
        if history is None:
            history = []
        if not message.strip():
            return "", history
        history.append({"role": "user", "content": message})
        for updated_history in generate_doctor_response(history):
            yield "", updated_history

    send_btn.click(respond, [user_input, chatbot], [user_input, chatbot])
    user_input.submit(respond, [user_input, chatbot], [user_input, chatbot])
    clear_btn.click(lambda: [], None, chatbot, queue=False)

# Launch
if __name__ == "__main__":
    demo.queue()
    demo.launch(share=True)