Initial SQL-only 1C adapter baseline

scripts/check_inference_endpoint.py

@@ -0,0 +1,84 @@
+from __future__ import annotations
+
+import argparse
+import datetime as dt
+import json
+import sys
+import time
+import urllib.error
+import urllib.request
+from pathlib import Path
+
+
+ROOT = Path(__file__).resolve().parents[1]
+DEFAULT_REPORT = ROOT / "reports" / "inference-endpoint-check.json"
+
+
+def fetch_models(base_url: str, timeout: int) -> tuple[list[str], int]:
+    started_at = time.perf_counter()
+    request = urllib.request.Request(f"{base_url.rstrip('/')}/v1/models", method="GET")
+    with urllib.request.urlopen(request, timeout=timeout) as response:
+        payload = json.loads(response.read().decode("utf-8"))
+    latency_ms = round((time.perf_counter() - started_at) * 1000)
+    rows = payload.get("data") if isinstance(payload, dict) else []
+    if not isinstance(rows, list):
+        return [], latency_ms
+    return [str(row.get("id")) for row in rows if isinstance(row, dict) and row.get("id")], latency_ms
+
+
+def check_endpoint(base_url: str, expected_model: str | None, timeout: int) -> dict:
+    result = {
+        "base_url": base_url,
+        "expected_model": expected_model,
+        "status": "failed",
+        "models": [],
+        "available": None,
+        "latency_ms": None,
+        "error": None,
+    }
+    try:
+        models, latency_ms = fetch_models(base_url, timeout)
+    except (TimeoutError, OSError, urllib.error.URLError, ValueError) as exc:
+        result["error"] = str(exc)
+        return result
+
+    result["models"] = models
+    result["latency_ms"] = latency_ms
+    result["available"] = expected_model in models if expected_model else None
+    result["status"] = "ok" if models and (expected_model is None or result["available"]) else "failed"
+    return result
+
+
+def main() -> int:
+    parser = argparse.ArgumentParser(description="Check an OpenAI-compatible inference endpoint.")
+    parser.add_argument("--base-url", default="http://docker-gpu.cin.su:8000")
+    parser.add_argument("--expected-model")
+    parser.add_argument("--timeout", type=int, default=10)
+    parser.add_argument("--report", type=Path, default=DEFAULT_REPORT)
+    parser.add_argument("--print", action="store_true")
+    parser.add_argument("--no-report", action="store_true")
+    args = parser.parse_args()
+
+    result = check_endpoint(args.base_url, args.expected_model, args.timeout)
+    report = {
+        "created_at": dt.datetime.now(dt.UTC).isoformat(),
+        "status": result["status"],
+        "checks": [result],
+    }
+
+    if not args.no_report:
+        args.report.parent.mkdir(parents=True, exist_ok=True)
+        args.report.write_text(json.dumps(report, ensure_ascii=False, indent=2) + "\n", encoding="utf-8")
+
+    if args.print:
+        print(json.dumps(report, ensure_ascii=False, indent=2))
+    else:
+        print(f"Endpoint status: {result['status']} ({args.base_url})")
+        if result["error"]:
+            print(result["error"], file=sys.stderr)
+
+    return 0 if result["status"] == "ok" else 1
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())