feat: initial commit

llm/scripts/backend.py

@@ -0,0 +1,141 @@
+#!/usr/bin/env python3
+"""Odczyt llm/models.yml: wykrycie backendu i renderowanie zmiennych środowiskowych.
+
+Jedno miejsce decyduje, jaki backend jest właściwy dla platformy i jakie zmienne
+dostaje silnik - Taskfile tylko o to pyta, nie duplikuje wiedzy.
+
+Użycie:
+    backend.py detect                 # nazwa backendu dla tej maszyny
+    backend.py env [--backend NAME]   # linie `export ...` do eval
+    backend.py get serve.model [--backend NAME]
+    backend.py list
+"""
+
+from __future__ import annotations
+
+import argparse
+import platform
+import shutil
+import subprocess
+import sys
+from pathlib import Path
+from typing import Any
+
+import yaml
+
+MODELS_FILE = Path(__file__).resolve().parent.parent / "models.yml"
+
+
+def load() -> dict[str, Any]:
+    return yaml.safe_load(MODELS_FILE.read_text(encoding="utf-8")) or {}
+
+
+def _has_nvidia_gpu() -> bool:
+    if not shutil.which("nvidia-smi"):
+        return False
+    try:
+        return subprocess.run(["nvidia-smi", "-L"], capture_output=True, timeout=10).returncode == 0
+    except (OSError, subprocess.SubprocessError):
+        return False
+
+
+def _has_docker() -> bool:
+    if not shutil.which("docker"):
+        return False
+    try:
+        return subprocess.run(["docker", "info"], capture_output=True, timeout=20).returncode == 0
+    except (OSError, subprocess.SubprocessError):
+        return False
+
+
+def detect() -> str:
+    """Kolejność jest świadoma: najpierw akceleracja sprzętowa, potem cokolwiek, co działa."""
+    system = platform.system().lower()
+
+    if system == "darwin" and platform.machine() == "arm64":
+        return "vllm-metal"
+    if system == "linux" and _has_nvidia_gpu() and _has_docker():
+        return "vllm-gpu"
+    if shutil.which("ollama") or _has_docker():
+        return "ollama"
+    # Brak GPU, Metala i Ollamy: realnego modelu i tak nie ma sensu tu uruchamiać.
+    return "mock"
+
+
+def resolve(name: str | None) -> tuple[str, dict[str, Any]]:
+    data = load()
+    backends = data.get("backends") or {}
+    if not name or name == "auto":
+        name = detect()
+    if name not in backends:
+        raise SystemExit(f"Nieznany backend '{name}'. Dostępne: {', '.join(sorted(backends))}")
+    return name, backends[name]
+
+
+def _dotted(config: dict[str, Any], path: str) -> Any:
+    value: Any = config
+    for part in path.split("."):
+        if not isinstance(value, dict) or part not in value:
+            raise SystemExit(f"Brak klucza '{path}' w konfiguracji backendu")
+        value = value[part]
+    return value
+
+
+def cmd_env(args: argparse.Namespace) -> None:
+    name, config = resolve(args.backend)
+    models = config.get("models") or {}
+    lines = [
+        f"export CODEMOD_LLM_BACKEND={name}",
+        f"export CODEMOD_MODEL_PROVIDER={config.get('provider', 'openai_like')}",
+        f"export CODEMOD_LLM_BASE_URL={config['base_url']}",
+        f"export CODEMOD_LLM_API_KEY={config.get('api_key', 'not-required')}",
+    ]
+    lines += [f"export CODEMOD_MODEL_{role.upper()}={model}" for role, model in sorted(models.items())]
+    print("\n".join(lines))
+
+
+def cmd_detect(args: argparse.Namespace) -> None:
+    requested = (args.requested or "auto").strip()
+    print(detect() if requested in ("", "auto") else requested)
+
+
+def cmd_get(args: argparse.Namespace) -> None:
+    _, config = resolve(args.backend)
+    value = _dotted(config, args.key)
+    print(" ".join(str(v) for v in value) if isinstance(value, list) else value)
+
+
+def cmd_list(_: argparse.Namespace) -> None:
+    data = load()
+    current = detect()
+    for name, config in (data.get("backends") or {}).items():
+        marker = "*" if name == current else " "
+        platforms = ", ".join(config.get("platforms", []))
+        print(f" {marker} {name:<12} {config['base_url']:<32} [{platforms}]")
+    print("\n * = wykryty jako właściwy dla tej maszyny")
+
+
+def main() -> None:
+    parser = argparse.ArgumentParser(description=__doc__, formatter_class=argparse.RawDescriptionHelpFormatter)
+    sub = parser.add_subparsers(dest="command", required=True)
+
+    det = sub.add_parser("detect")
+    det.add_argument("--requested", default="auto", help="jawnie wskazany backend; 'auto' = wykrycie")
+    det.set_defaults(func=cmd_detect)
+    sub.add_parser("list").set_defaults(func=cmd_list)
+
+    env = sub.add_parser("env")
+    env.add_argument("--backend")
+    env.set_defaults(func=cmd_env)
+
+    get = sub.add_parser("get")
+    get.add_argument("key")
+    get.add_argument("--backend")
+    get.set_defaults(func=cmd_get)
+
+    args = parser.parse_args()
+    args.func(args)
+
+
+if __name__ == "__main__":
+    sys.exit(main())