scripts/test_api_smoke.py

#!/usr/bin/env python
"""Minimal API smoke test for fic-agent.

Checks:
1) LLM chat completion
2) Embedding API call

Exit code:
0 = all requested checks passed
1 = at least one requested check failed
"""

from __future__ import annotations

import argparse
import sys

from fic_agent.config import RuntimeConfig


def _mask_len(value: str | None) -> str:
    if not value:
        return "0"
    return str(len(value))


def _test_llm(cfg: RuntimeConfig) -> bool:
    if not cfg.llm_api_key:
        print("[LLM] FAIL: missing llm_api_key")
        return False
    try:
        from openai import OpenAI  # type: ignore
    except Exception as e:
        print(f"[LLM] FAIL: openai import error: {e}")
        return False

    try:
        client = OpenAI(base_url=cfg.llm_base_url, api_key=cfg.llm_api_key)
        resp = client.chat.completions.create(
            model=cfg.llm_model,
            messages=[
                {"role": "system", "content": "You are a concise assistant."},
                {"role": "user", "content": "Reply with exactly: API_OK"},
            ],
            temperature=0.0,
            max_tokens=20,
        )
        text = (resp.choices[0].message.content or "").strip()
        usage = getattr(resp, "usage", None)
        total = getattr(usage, "total_tokens", None) if usage is not None else None
        print(f"[LLM] PASS: model={cfg.llm_model} total_tokens={total} reply={text!r}")
        return True
    except Exception as e:
        print(f"[LLM] FAIL: {type(e).__name__}: {e}")
        return False


def _test_embedding(cfg: RuntimeConfig) -> bool:
    if not cfg.embedding_api_key:
        print("[EMBED] FAIL: missing embedding_api_key")
        return False
    try:
        from openai import OpenAI  # type: ignore
    except Exception as e:
        print(f"[EMBED] FAIL: openai import error: {e}")
        return False

    try:
        client = OpenAI(base_url=cfg.embedding_base_url, api_key=cfg.embedding_api_key)
        resp = client.embeddings.create(
            model=cfg.embedding_model,
            input=["api smoke test"],
        )
        data = getattr(resp, "data", None) or []
        if not data:
            print("[EMBED] FAIL: empty data")
            return False
        vec = getattr(data[0], "embedding", None)
        dim = len(vec) if isinstance(vec, list) else 0
        usage = getattr(resp, "usage", None)
        total = getattr(usage, "total_tokens", None) if usage is not None else None
        print(f"[EMBED] PASS: model={cfg.embedding_model} dim={dim} total_tokens={total}")
        return True
    except Exception as e:
        print(f"[EMBED] FAIL: {type(e).__name__}: {e}")
        return False


def main() -> int:
    parser = argparse.ArgumentParser(description="Minimal API smoke test for fic-agent")
    parser.add_argument("--skip-llm", action="store_true", help="Skip LLM chat test")
    parser.add_argument("--skip-embedding", action="store_true", help="Skip embedding test")
    args = parser.parse_args()

    cfg = RuntimeConfig()
    print(
        "[CFG] "
        f"llm_base_url={cfg.llm_base_url} llm_model={cfg.llm_model} llm_key_len={_mask_len(cfg.llm_api_key)}"
    )
    print(
        "[CFG] "
        f"embedding_base_url={cfg.embedding_base_url} embedding_model={cfg.embedding_model} "
        f"embedding_key_len={_mask_len(cfg.embedding_api_key)}"
    )

    ok = True
    if not args.skip_llm:
        ok = _test_llm(cfg) and ok
    if not args.skip_embedding:
        ok = _test_embedding(cfg) and ok

    if ok:
        print("API smoke test: PASS")
        return 0
    print("API smoke test: FAIL")
    return 1


if __name__ == "__main__":
    sys.exit(main())