|
| 1 | +#!/usr/bin/env python3 |
| 2 | +"""backup-repos — mirror all GitHub repos to a timestamped snapshot dir. |
| 3 | +
|
| 4 | +Python port of backup-repos.sh. Uses the `gh` CLI (already authenticated) to |
| 5 | +list repos and `git clone --mirror` to snapshot each one. |
| 6 | +
|
| 7 | +Usage: |
| 8 | + backup-repos.py [--out DIR] [--account OWNER] [--limit N] |
| 9 | +""" |
| 10 | + |
| 11 | +from __future__ import annotations |
| 12 | + |
| 13 | +import argparse |
| 14 | +import datetime |
| 15 | +import re |
| 16 | +import sys |
| 17 | +from pathlib import Path |
| 18 | + |
| 19 | +_SELF_DIR = Path(__file__).resolve().parent |
| 20 | +for _c in (_SELF_DIR, _SELF_DIR.parent / "scripts"): |
| 21 | + if (_c / "localpibox").is_dir(): |
| 22 | + sys.path.insert(0, str(_c)) |
| 23 | + break |
| 24 | + |
| 25 | +from localpibox import log # noqa: E402 |
| 26 | +from localpibox.cli import add_common_args, console_from_args, install_sigpipe_handler # noqa: E402 |
| 27 | +from localpibox.run import run_cmd, which # noqa: E402 |
| 28 | + |
| 29 | +_REPO_RE = re.compile(r"[\w.-]+/[\w.-]+") |
| 30 | + |
| 31 | + |
| 32 | +def parse_repo_list(text: str) -> list[str]: |
| 33 | + """Parse `gh repo list` TSV output -> ``owner/name`` strings. |
| 34 | +
|
| 35 | + Ignores blank lines and any row whose first column is not a valid |
| 36 | + ``owner/name`` pair (headers, error text, extra columns). |
| 37 | + """ |
| 38 | + repos: list[str] = [] |
| 39 | + for line in text.splitlines(): |
| 40 | + if not line.strip(): |
| 41 | + continue |
| 42 | + first = line.split("\t", 1)[0].strip() |
| 43 | + if _REPO_RE.fullmatch(first): |
| 44 | + repos.append(first) |
| 45 | + return repos |
| 46 | + |
| 47 | + |
| 48 | +def list_repos( |
| 49 | + account: str | None = None, |
| 50 | + *, |
| 51 | + limit: int = 500, |
| 52 | + gh: str = "gh", |
| 53 | + runner=None, |
| 54 | +) -> list[str]: |
| 55 | + """List repos for *account* (or the authenticated user when None).""" |
| 56 | + runner = runner or run_cmd |
| 57 | + cmd = [gh, "repo", "list", "--limit", str(limit)] |
| 58 | + if account: |
| 59 | + cmd.insert(3, account) |
| 60 | + out, err, code = runner(cmd, timeout=120) |
| 61 | + if code: |
| 62 | + raise RuntimeError(f"`gh repo list` failed: {err.strip() or out.strip()}") |
| 63 | + return parse_repo_list(out) |
| 64 | + |
| 65 | + |
| 66 | +def mirror_repo( |
| 67 | + repo: str, |
| 68 | + dest: str | Path, |
| 69 | + *, |
| 70 | + git: str = "git", |
| 71 | + runner=None, |
| 72 | +) -> bool: |
| 73 | + """Mirror-clone *repo* into *dest*; True on success.""" |
| 74 | + runner = runner or run_cmd |
| 75 | + out, err, code = runner( |
| 76 | + [git, "clone", "--mirror", f"https://github.com/{repo}", str(dest)], |
| 77 | + timeout=600, |
| 78 | + ) |
| 79 | + if code: |
| 80 | + log.warn(f" failed to clone {repo}: {err.strip() or out.strip()}") |
| 81 | + return False |
| 82 | + return True |
| 83 | + |
| 84 | + |
| 85 | +def default_out_dir() -> Path: |
| 86 | + """Default snapshot root: ``~/.localpibox/backups`` (matches lpb state dirs).""" |
| 87 | + return Path.home() / ".localpibox" / "backups" |
| 88 | + |
| 89 | + |
| 90 | +def main(argv: list[str] | None = None) -> int: |
| 91 | + install_sigpipe_handler() |
| 92 | + parser = argparse.ArgumentParser(prog="backup-repos", description=__doc__) |
| 93 | + add_common_args(parser) |
| 94 | + parser.add_argument( |
| 95 | + "--out", default=None, |
| 96 | + help="snapshot dir (default: ~/.localpibox/backups/snapshot_<timestamp>)", |
| 97 | + ) |
| 98 | + parser.add_argument( |
| 99 | + "--account", default=None, |
| 100 | + help="GitHub owner to back up (default: authenticated user)", |
| 101 | + ) |
| 102 | + parser.add_argument("--limit", type=int, default=500, help="max repos to fetch") |
| 103 | + args = parser.parse_args(argv) |
| 104 | + cons = console_from_args(args) |
| 105 | + |
| 106 | + if not which("gh"): |
| 107 | + log.error("gh CLI not found — install it and run `gh auth login` first") |
| 108 | + return 1 |
| 109 | + if not which("git"): |
| 110 | + log.error("git not found") |
| 111 | + return 1 |
| 112 | + |
| 113 | + out_dir = ( |
| 114 | + Path(args.out) |
| 115 | + if args.out |
| 116 | + else default_out_dir() |
| 117 | + / f"snapshot_{datetime.datetime.now().strftime('%Y%m%d_%H%M%S')}" |
| 118 | + ) |
| 119 | + out_dir.mkdir(parents=True, exist_ok=True) |
| 120 | + cons.info(f"Backing up repos to {out_dir}") |
| 121 | + |
| 122 | + try: |
| 123 | + repos = list_repos(args.account, limit=args.limit) |
| 124 | + except RuntimeError as exc: |
| 125 | + log.error(str(exc)) |
| 126 | + return 1 |
| 127 | + |
| 128 | + ok = failed = 0 |
| 129 | + for repo in repos: |
| 130 | + dest = out_dir / f"{repo.split('/', 1)[1]}.git" |
| 131 | + cons.info(f"Backing up: {repo}") |
| 132 | + if mirror_repo(repo, dest): |
| 133 | + ok += 1 |
| 134 | + else: |
| 135 | + failed += 1 |
| 136 | + |
| 137 | + cons.done(f"Backup complete: {ok} backed up, {failed} failed -> {out_dir}") |
| 138 | + return 1 if failed else 0 |
| 139 | + |
| 140 | + |
| 141 | +if __name__ == "__main__": |
| 142 | + raise SystemExit(main()) |
0 commit comments