add scripts/_scratch/_retry_runner.py

scripts/_scratch/_retry_runner.py

@@ -0,0 +1,63 @@
+"""自动重试跑一个命令，直到退出码为 0 或达到上限。
+
+为什么不用 PowerShell 包一层
+---------------------------
+试过了，**嵌套引号在 cmd → powershell → python 三层之间会被吃掉**，导致内层命令
+每次都以退出码 1 立刻失败（日志里能看到 attempt 2/30、3/30 都是秒退）。
+Python 的 `subprocess` 收的是**参数列表**，不经过 shell 解析，没有这个坑。
+
+用法：
+    python scripts/_retry_runner.py --tries 30 --sleep 20 --log logs/x.log -- <命令> [参数...]
+"""
+
+from __future__ import annotations
+
+import argparse
+import subprocess
+import sys
+import time
+from pathlib import Path
+
+
+def main() -> int:
+    ap = argparse.ArgumentParser()
+    ap.add_argument("--tries", type=int, default=30)
+    ap.add_argument("--sleep", type=int, default=20)
+    ap.add_argument("--log", type=Path, default=None)
+    ap.add_argument("cmd", nargs=argparse.REMAINDER)
+    args = ap.parse_args()
+    cmd = args.cmd[1:] if args.cmd and args.cmd[0] == "--" else args.cmd
+    if not cmd:
+        print("[retry] 没给命令")
+        return 2
+
+    log = args.log
+    if log:
+        log.parent.mkdir(parents=True, exist_ok=True)
+
+    def note(msg: str) -> None:
+        line = f"[retry] {msg}  {time.strftime('%Y-%m-%d %H:%M:%S')}"
+        print(line, flush=True)
+        if log:
+            with log.open("a", encoding="utf-8") as fh:
+                fh.write(line + "\n")
+
+    note("cmd = " + " ".join(cmd))
+    for i in range(1, args.tries + 1):
+        note(f"attempt {i}/{args.tries}")
+        try:
+            rc = subprocess.run(cmd, cwd=str(Path(__file__).resolve().parents[1])).returncode
+        except Exception as exc:  # noqa: BLE001
+            note(f"spawn failed: {type(exc).__name__}: {exc}")
+            rc = -1
+        note(f"exit code {rc}")
+        if rc == 0:
+            note("completed successfully")
+            return 0
+        time.sleep(args.sleep)
+    note("gave up after max tries")
+    return 1
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())