instant-bom/bom_assistant/ai/client.py

from __future__ import annotations

import os

from dotenv import load_dotenv
from openai import OpenAI

load_dotenv()

_DEFAULT_MODEL = "google/gemini-flash-1.5"


def complete(prompt: str, max_tokens: int = 200) -> str:
    """Call OpenRouter with prompt, return response text. Raises on failure."""
    client = OpenAI(
        base_url="https://openrouter.ai/api/v1",
        api_key=os.environ["OPENROUTER_API_KEY"],
    )
    model = os.environ.get("OPENROUTER_MODEL", _DEFAULT_MODEL)
    resp = client.chat.completions.create(
        model=model,
        max_tokens=max_tokens,
        messages=[{"role": "user", "content": prompt}],
        extra_body={"reasoning": {"enabled": False}},
    )
    content = resp.choices[0].message.content
    if content is None:
        finish_reason = resp.choices[0].finish_reason
        raise RuntimeError(
            f"OpenRouter returned empty content for model {model!r} "
            f"(finish_reason={finish_reason!r}, raw={resp.model_dump_json()})"
        )
    return content.strip()