From 35000f728e66533de9b0cf6e8b040f9cdb82c1a7 Mon Sep 17 00:00:00 2001 From: Scott Worley Date: Mon, 21 Jun 2021 14:43:18 -0700 Subject: [PATCH] Refuse force-pushes even if cache is cleared --- Changelog | 2 ++ git_cache.py | 68 ++++++++++++++++++++++++++++++++++++++++++----- test_git_cache.py | 23 +++++++++++++++- 3 files changed, 86 insertions(+), 7 deletions(-) diff --git a/Changelog b/Changelog index 1f6a1be..3e6b108 100644 --- a/Changelog +++ b/Changelog @@ -1,4 +1,6 @@ ## [Unreleased] +### Changed +- Refuse force-push updates even if XDG_CACHE_HOME is cleared ## [1.3.1] - 2021-06-07 diff --git a/git_cache.py b/git_cache.py index 077406c..1363f11 100644 --- a/git_cache.py +++ b/git_cache.py @@ -4,23 +4,37 @@ # implementation details for my comfort. So we re-implement here half of # nix's builtins.fetchGit. :( +import functools import hashlib import logging import os import subprocess import sys +import time -from typing import Tuple, Union +from typing import Iterator, NamedTuple, Optional, TypeVar, Tuple, Union import backoff -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" -Rev = str # eg: "53a27350551844e1ed1a9257690294767389ef0d" +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" +Rev = str # eg: "53a27350551844e1ed1a9257690294767389ef0d" RefOrRev = Union[Ref, Rev] +class _LogEntry(NamedTuple): + ref: Ref + rev: Rev + + +T = TypeVar('T') + + +def _repo_hashname(repo: Repo) -> str: + return hashlib.sha256(repo.encode()).hexdigest() + + def git_cachedir(repo: Repo) -> Path: # Use xdg module when it's less painful to have as a dependency XDG_CACHE_HOME = Path( @@ -29,7 +43,18 @@ def git_cachedir(repo: Repo) -> Path: return Path(os.path.join( XDG_CACHE_HOME, 'git-cache/v1', - hashlib.sha256(repo.encode()).hexdigest())) + _repo_hashname(repo))) + + +def _log_filename(repo: Repo) -> Path: + # Use xdg module when it's less painful to have as a dependency + XDG_DATA_HOME = Path( + os.environ.get('XDG_DATA_HOME', os.path.expanduser('~/.local/share'))) + + return Path(os.path.join( + XDG_DATA_HOME, + 'git-cache/v1', + _repo_hashname(repo))) def is_ancestor(repo: Repo, descendant: RefOrRev, ancestor: RefOrRev) -> bool: @@ -54,6 +79,36 @@ def verify_ancestry( raise Exception('%s is not an ancestor of %s' % (ancestor, 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: + for line in f: + _, _, rev, ref = line.strip().split(maxsplit=3) + yield _LogEntry(ref, rev) + + +def _last(it: Iterator[T]) -> Optional[T]: + return functools.reduce(lambda a, b: b, it, None) + + +def _previous_fetched_rev(repo: Repo, ref: Ref) -> Optional[Rev]: + return _last(entry.rev for entry in _read_fetch_log( + 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) + 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)) + + @backoff.on_exception( backoff.expo, subprocess.CalledProcessError, @@ -78,6 +133,7 @@ def fetch(repo: Repo, ref: Ref) -> Tuple[Path, Rev]: with open(os.path.join(cachedir, 'refs', 'heads', ref)) as rev_file: rev = Rev(rev_file.read(999).strip()) verify_ancestry(repo, ref, rev) + _log_fetch(repo, ref, rev) return cachedir, rev diff --git a/test_git_cache.py b/test_git_cache.py index bbe7118..b7e0ede 100644 --- a/test_git_cache.py +++ b/test_git_cache.py @@ -37,9 +37,14 @@ def _commit_file( class TestGitCache(unittest.TestCase): def setUp(self) -> None: - self.xdgcache = tempfile.TemporaryDirectory(prefix='git_cache_test-') + self.xdgcache = tempfile.TemporaryDirectory( + prefix='git_cache_test-cache-') + self.xdgdata = tempfile.TemporaryDirectory( + prefix='git_cache_test-data-') self.old_XDG_CACHE_HOME = os.environ.get('XDG_CACHE_HOME') + self.old_XDG_DATA_HOME = os.environ.get('XDG_DATA_HOME') _setenv('XDG_CACHE_HOME', self.xdgcache.name) + _setenv('XDG_DATA_HOME', self.xdgdata.name) os.environ['GIT_AUTHOR_NAME'] = 'test_git_cache' os.environ['GIT_COMMITTER_NAME'] = 'test_git_cache' @@ -55,6 +60,7 @@ class TestGitCache(unittest.TestCase): def tearDown(self) -> None: _setenv('XDG_CACHE_HOME', self.old_XDG_CACHE_HOME) + _setenv('XDG_DATA_HOME', self.old_XDG_DATA_HOME) self.tempdir.cleanup() self.xdgcache.cleanup() @@ -204,6 +210,21 @@ class TestGitCache(unittest.TestCase): with self.assertRaises(Exception): git_cache.ensure_rev_available(self.upstream, 'master', rev) + def test_fetch_raises_on_amend_after_cache_deleted(self) -> None: + d, _ = git_cache.fetch(self.upstream, 'master') + shutil.rmtree(d) + _git(self.upstream, 'commit', '--amend', '-m', 'Amended') + with self.assertRaises(Exception): + git_cache.fetch(self.upstream, 'master') + + def test_ensure_raises_on_amend_after_cache_deleted(self) -> None: + d, _ = git_cache.fetch(self.upstream, 'master') + shutil.rmtree(d) + _git(self.upstream, 'commit', '--amend', '-m', 'Amended') + rev = _git(self.upstream, 'log', '--format=%H', '-n1').strip().decode() + with self.assertRaises(Exception): + git_cache.ensure_rev_available(self.upstream, 'master', rev) + if __name__ == '__main__': unittest.main() -- 2.44.1