git.scottworley.com Git - git-cache/blob - git_cache.py

   1 # It would be nice if we could share the nix git cache, but as of the
   2 # time of writing it is transitioning from gitv2 (deprecated) to gitv3
   3 # (not ready yet), and trying to straddle them both is too far into nix
   4 # implementation details for my comfort.  So we re-implement here half of
   5 # nix's builtins.fetchGit.  :(
   6
   7 import hashlib
   8 import logging
   9 import os
  10 import subprocess
  11 import sys
  12
  13 from typing import Tuple
  14
  15 import backoff
  16
  17 Path = str  # eg: "/home/user/.cache/git-cache/v1"
  18 Repo = str  # eg: "https://github.com/NixOS/nixpkgs.git"
  19 Ref = str   # eg: "master" or "v1.0.0"
  20 Rev = str   # eg: "53a27350551844e1ed1a9257690294767389ef0d"
  21
  22
  23 def git_cachedir(repo: Repo) -> Path:
  24     # Use xdg module when it's less painful to have as a dependency
  25     XDG_CACHE_HOME = Path(
  26         os.environ.get('XDG_CACHE_HOME', os.path.expanduser('~/.cache')))
  27
  28     return Path(os.path.join(
  29         XDG_CACHE_HOME,
  30         'git-cache/v1',
  31         hashlib.sha256(repo.encode()).hexdigest()))
  32
  33
  34 def is_ancestor(repo: Repo, ref: Ref, rev: Rev) -> bool:
  35     cachedir = git_cachedir(repo)
  36     logging.debug('Checking if rev %s is an ancestor of ref "%s"', rev, ref)
  37     process = subprocess.run(
  38         ['git', '-C', cachedir, 'merge-base', '--is-ancestor', rev, ref], check=False)
  39     return process.returncode == 0
  40
  41
  42 def verify_ancestry(repo: Repo, ref: Ref, rev: Rev) -> None:
  43     if not is_ancestor(repo, ref, rev):
  44         raise Exception('Rev %s is not an ancestor of ref "%s"' % (rev, ref))
  45
  46
  47 @backoff.on_exception(
  48     backoff.expo,
  49     subprocess.CalledProcessError,
  50     max_time=lambda: int(os.environ.get('BACKOFF_MAX_TIME', '30')))
  51 def _git_fetch(cachedir: Path, repo: Repo, ref: Ref) -> None:
  52     # We don't use --force here because we want to abort and freak out if forced
  53     # updates are happening.
  54     subprocess.run(['git', '-C', cachedir, 'fetch', repo,
  55                     '%s:%s' % (ref, ref)], check=True)
  56
  57
  58 def fetch(repo: Repo, ref: Ref) -> Tuple[Path, Rev]:
  59     cachedir = git_cachedir(repo)
  60     if not os.path.exists(cachedir):
  61         logging.debug("Initializing git repo")
  62         subprocess.run(['git', 'init', '--bare', cachedir],
  63                        check=True, stdout=sys.stderr)
  64
  65     logging.debug('Fetching ref "%s" from %s', ref, repo)
  66     _git_fetch(cachedir, repo, ref)
  67
  68     with open(os.path.join(cachedir, 'refs', 'heads', ref)) as rev_file:
  69         rev = Rev(rev_file.read(999).strip())
  70     verify_ancestry(repo, ref, rev)
  71
  72     return cachedir, rev
  73
  74
  75 def ensure_rev_available(repo: Repo, ref: Ref, rev: Rev) -> Path:
  76     cachedir = git_cachedir(repo)
  77     if os.path.exists(cachedir) and is_ancestor(repo, ref, rev):
  78         return cachedir
  79
  80     logging.debug(
  81         'We do not have rev %s.  We will fetch ref "%s" and hope it appears.',
  82         rev, ref)
  83     fetch(repo, ref)
  84     logging.debug('Verifying that fetch retrieved rev %s', rev)
  85     subprocess.run(['git', '-C', cachedir, 'cat-file', '-e', rev], check=True)
  86     verify_ancestry(repo, ref, rev)
  87
  88     return cachedir
  89
  90
  91 def _main() -> None:
  92     if len(sys.argv) == 3:
  93         print('{1} {0}'.format(*fetch(Repo(sys.argv[1]), Ref(sys.argv[2]))))
  94     elif len(sys.argv) == 4:
  95         print(ensure_rev_available(
  96             Repo(sys.argv[1]), Ref(sys.argv[2]), Rev(sys.argv[3])))
  97     else:
  98         usage = '''usage: git-cache repo ref [rev]
  99 example: git-cache https://github.com/NixOS/nixpkgs.git master'''
 100         print(usage, file=sys.stderr)
 101         sys.exit(1)