import os

from llm import stream_openai_response
from prompts import assemble_prompt
from prompts.types import Stack


async def generate_code_core(image_url: str, stack: Stack) -> str:
    prompt_messages = assemble_prompt(image_url, stack)
    openai_api_key = os.environ.get("OPENAI_API_KEY")
    openai_base_url = None

    async def process_chunk(content: str):
        pass

    if not openai_api_key:
        raise Exception("OpenAI API key not found")

    completion = await stream_openai_response(
        prompt_messages,
        api_key=openai_api_key,
        base_url=openai_base_url,
        callback=lambda x: process_chunk(x),
    )

    return completion