app.py

import streamlit as st
from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer
from threading import Thread
import torch
import re

# --- SETTINGS ---
MODEL_ID = "Qwen/Qwen2.5-Coder-0.5B-Instruct"
DEVICE = "cpu"

st.set_page_config(page_title="Smol Claude", layout="wide", page_icon="🤖")

# --- CUSTOM CSS FOR CLAUDE VIBE ---
st.markdown("""
    <style>
    .stChatMessage { border-radius: 10px; padding: 10px; margin-bottom: 10px; }
    .artifact-container { background-color: #f8f9fa; border-left: 2px solid #ddd; padding: 20px; height: 100vh; overflow-y: auto; }
    </style>
""", unsafe_allow_html=True)

# --- MODEL LOADING ---
@st.cache_resource
def load_llm():
    tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
    model = AutoModelForCausalLM.from_pretrained(MODEL_ID, torch_dtype="auto").to(DEVICE)
    return tokenizer, model

tokenizer, model = load_llm()

# --- HELPER: EXTRACT CODE ---
def extract_code(text):
    match = re.search(r"```(?:\w+)?\n([\s\S]*?)\n```", text)
    return match.group(1) if match else None

# --- UI SETUP ---
if "messages" not in st.session_state:
    st.session_state.messages = []
if "current_code" not in st.session_state:
    st.session_state.current_code = ""

col_chat, col_artifact = st.columns([1, 1])

# --- CHAT COLUMN ---
with col_chat:
    st.subheader("💬 Chat")
    for msg in st.session_state.messages:
        with st.chat_message(msg["role"]):
            st.markdown(msg["content"])

    if prompt := st.chat_input("What should we build?"):
        st.session_state.messages.append({"role": "user", "content": prompt})
        with st.chat_message("user"):
            st.markdown(prompt)

        with st.chat_message("assistant"):
            # Prepare conversation
            input_msgs = [{"role": "system", "content": "You are a professional coder. Output code in markdown blocks."}] + st.session_state.messages
            text = tokenizer.apply_chat_template(input_msgs, tokenize=False, add_generation_prompt=True)
            inputs = tokenizer([text], return_tensors="pt").to(DEVICE)
            
            # Streaming setup
            streamer = TextIteratorStreamer(tokenizer, skip_prompt=True, skip_special_tokens=True)
            generation_kwargs = dict(inputs, streamer=streamer, max_new_tokens=1024, temperature=0.1)
            
            thread = Thread(target=model.generate, kwargs=generation_kwargs)
            thread.start()
            
            # Stream the response
            full_response = ""
            message_placeholder = st.empty()
            for new_text in streamer:
                full_response += new_text
                message_placeholder.markdown(full_response + "▌")
            
            message_placeholder.markdown(full_response)
            st.session_state.messages.append({"role": "assistant", "content": full_response})
            
            # Update Artifact
            code_snippet = extract_code(full_response)
            if code_snippet:
                st.session_state.current_code = code_snippet
                st.rerun()

# --- ARTIFACT COLUMN (The "Claude" Sidebar) ---
with col_artifact:
    st.subheader("🚀 Code Artifact")
    if st.session_state.current_code:
        st.code(st.session_state.current_code, language="python", line_numbers=True)
        st.download_button("Download File", st.session_state.current_code, file_name="artifact.py")
    else:
        st.info("Ask the AI to write some code to see it appear here.")