X-Git-Url: http://git.scottworley.com/git-cache/blobdiff_plain/d1ab08536175fe3eb4bd655446f6e3f9eb4b97c0..a3835f508d4a441c64156d4669378e080064d4bf:/git_cache.py?ds=inline diff --git a/git_cache.py b/git_cache.py index d43df26..8171b48 100644 --- a/git_cache.py +++ b/git_cache.py @@ -1,3 +1,10 @@ +# git-cache: Cache git content locally +# +# This program is free software: you can redistribute it and/or modify it +# under the terms of the GNU General Public License as published by the +# Free Software Foundation, version 3. + + # It would be nice if we could share the nix git cache, but as of the # time of writing it is transitioning from gitv2 (deprecated) to gitv3 # (not ready yet), and trying to straddle them both is too far into nix @@ -17,6 +24,11 @@ from typing import Iterator, NamedTuple, Optional, TypeVar, Tuple, Union import backoff + +class GitCacheError(Exception): + pass + + Path = str # eg: "/home/user/.cache/git-cache/v1" Repo = str # eg: "https://github.com/NixOS/nixpkgs.git" Ref = str # eg: "master" or "v1.0.0" @@ -75,16 +87,17 @@ def is_ancestor(repo: Repo, descendant: RefOrRev, ancestor: RefOrRev) -> bool: def verify_ancestry( repo: Repo, descendant: RefOrRev, - ancestor: RefOrRev) -> None: - if not is_ancestor(repo, descendant, ancestor): - raise Exception('%s is not an ancestor of %s' % (ancestor, descendant)) + ancestor: RefOrRev, + force: bool = False) -> None: + if not force and not is_ancestor(repo, descendant, ancestor): + raise GitCacheError(f'{ancestor} is not an ancestor of {descendant}') def _read_fetch_log(repo: Repo) -> Iterator[_LogEntry]: filename = _log_filename(repo) if not os.path.exists(filename): return - with open(filename, 'r') as f: + with open(filename, 'r', encoding='utf-8') as f: for line in f: _, _, rev, ref = line.strip().split(maxsplit=3) yield _LogEntry(ref, rev) @@ -99,47 +112,108 @@ def _previous_fetched_rev(repo: Repo, ref: Ref) -> Optional[Rev]: repo) if entry.ref == ref) -def _log_fetch(repo: Repo, ref: Ref, rev: Rev) -> None: - prev_rev = _previous_fetched_rev(repo, ref) - if prev_rev is not None: - verify_ancestry(repo, rev, prev_rev) +def _log_fetch(repo: Repo, ref: Ref, rev: Rev, force: bool = False) -> None: + if not force: + prev_rev = _previous_fetched_rev(repo, ref) + if prev_rev is not None: + verify_ancestry(repo, rev, prev_rev) filename = _log_filename(repo) os.makedirs(os.path.dirname(filename), exist_ok=True) - with open(filename, 'a') as f: - f.write('%s fetch %s %s\n' % - (time.strftime('%Y-%m%d-%H:%M:%S%z'), rev, ref)) + with open(filename, 'a', encoding='utf-8') as f: + f.write( + f'{time.strftime("%Y-%m%d-%H:%M:%S%z")} ' + f'{"FORCEDFETCH" if force else "fetch"} {rev} {ref}\n' + ) + + +def _show_force_warning() -> None: + print(''' +************************************************************************** +* WARNING: git-cache INVOKED WITH --force! * +* * +* This mode allows previously-fetched refs to be overwritten to point to * +* non-descendants -- commits that don't have the previous version of the * +* the ref in their history! * +* * +* This should only be invoked by a human operator who knows what they're * +* doing to correct a specific, known, problem. Care should be taken to * +* prevent recurrence. * +* * +* Press ^C to abort. * +* * +''', end='', file=sys.stderr) + warn_time_override = os.environ.get('FORCE_WARNING_TIME', None) + warn_time: int + if warn_time_override is None: + warn_time = 15 + else: + warn_time = int(warn_time_override) + print( + '''* !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! * +* !! WARNING DISPLAY TIME OVERRIDDEN !! * +* !! !! * +* !! This message is intended to be displayed long enough for a !! * +* !! human operator to read it and have a chance to abort. An !! * +* !! override for the delay time is provided FOR THE UNIT TESTS !! * +* !! to avoid delaying software builds unnecessarily. This is !! * +* !! INTENDED FOR USE IN UNIT TESTS ONLY; THIS MESSAGE SHOULD !! * +* !! NEVER BE SEEN OUTSIDE BUILD LOGS! !! * +* !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! * +* * +''', end='', file=sys.stderr) + + for i in range(warn_time, 0, -1): + msg = f'* {f"Continuing in {i} seconds...":-70s} *' + print(msg, file=sys.stderr) + time.sleep(1) + print('*' * 74, file=sys.stderr) @backoff.on_exception( backoff.expo, subprocess.CalledProcessError, max_time=lambda: int(os.environ.get('BACKOFF_MAX_TIME', '30'))) -def _git_fetch(cachedir: Path, repo: Repo, ref: Ref) -> None: - # We don't use --force here because we want to abort and freak out if forced - # updates are happening. - subprocess.run(['git', '-C', cachedir, 'fetch', repo, - '%s:%s' % (ref, ref)], check=True) +def _git_fetch( + cachedir: Path, + repo: Repo, + ref: Ref, + force: bool = False) -> None: + subprocess.run(['git', '-C', cachedir, 'fetch'] + + (['--force'] if force else []) + + [repo, f'{ref}:{ref}'], check=True) -def fetch(repo: Repo, ref: Ref) -> Tuple[Path, Rev]: +def fetch(repo: Repo, ref: Ref, force: bool = False) -> Tuple[Path, Rev]: + if force: + _show_force_warning() cachedir = git_cachedir(repo) if not os.path.exists(cachedir): logging.debug("Initializing git repo") - subprocess.run(['git', 'init', '--bare', cachedir], - check=True, stdout=sys.stderr) + subprocess.run(['git', + '-c', + 'init.defaultBranch=git-cache--no-default-branch', + 'init', + '--bare', + cachedir], + check=True, + stdout=sys.stderr) logging.debug('Fetching ref "%s" from %s', ref, repo) - _git_fetch(cachedir, repo, ref) + _git_fetch(cachedir, repo, ref, force=force) - with open(os.path.join(cachedir, 'refs', 'heads', ref)) as rev_file: + with open(os.path.join(cachedir, 'refs', 'heads', ref), encoding='utf-8') as rev_file: rev = Rev(rev_file.read(999).strip()) - verify_ancestry(repo, ref, rev) - _log_fetch(repo, ref, rev) + verify_ancestry(repo, ref, rev, force=force) + _log_fetch(repo, ref, rev, force=force) return cachedir, rev -def ensure_rev_available(repo: Repo, ref: Ref, rev: Rev) -> Path: +def ensure_rev_available( + repo: Repo, + ref: Ref, + rev: Rev, + force: bool = False) -> Path: cachedir = git_cachedir(repo) if os.path.exists(cachedir) and is_ancestor(repo, ref, rev): return cachedir @@ -147,10 +221,10 @@ def ensure_rev_available(repo: Repo, ref: Ref, rev: Rev) -> Path: logging.debug( 'We do not have rev %s. We will fetch ref "%s" and hope it appears.', rev, ref) - fetch(repo, ref) + fetch(repo, ref, force=force) logging.debug('Verifying that fetch retrieved rev %s', rev) subprocess.run(['git', '-C', cachedir, 'cat-file', '-e', rev], check=True) - verify_ancestry(repo, ref, rev) + verify_ancestry(repo, ref, rev, force=force) return cachedir @@ -159,6 +233,10 @@ def _main() -> None: parser = argparse.ArgumentParser( description='Cache remote git repositories locally.', epilog='example usage: git-cache https://github.com/NixOS/nixpkgs.git master') + parser.add_argument( + '--force', + action='store_true', + help='Recover from a force-push in the remote repo') parser.add_argument( 'repo', metavar='Repo', @@ -179,6 +257,12 @@ def _main() -> None: args = parser.parse_args() if args.rev is None: - print('{1} {0}'.format(*fetch(args.repo, args.ref))) + cachedir, rev = fetch(args.repo, args.ref, force=args.force) + print(f'{rev} {cachedir}') else: - print(ensure_rev_available(args.repo, args.ref, args.rev)) + print( + ensure_rev_available( + args.repo, + args.ref, + args.rev, + force=args.force))