.bin/git-repack

   1 #!/usr/bin/python
   2
   3 """Repack git repos fully the way I like them."""
   4
   5 from __future__ import print_function
   6
   7 import argparse
   8 import glob
   9 import os
  10 import shutil
  11 import subprocess
  12 import sys
  13 import tempfile
  14
  15
  16 def mount_settings():
  17     """Return dict mapping path to its type"""
  18     ret = {}
  19     with open('/proc/mounts') as fp:
  20         for line in fp:
  21             ele = line.split()
  22             ret[ele[1]] = ele[2]
  23     return ret
  24
  25
  26 def is_git_dir(path):
  27     """Whether |path| is a .git dir"""
  28     return (os.path.isdir(os.path.join(path, 'refs')) and
  29             os.path.isdir(os.path.join(path, 'objects')) and
  30             os.path.isfile(os.path.join(path, 'config')))
  31
  32
  33 def find_git_dir(path):
  34     """Try to find the .git dir to operate on"""
  35     orig_path = path
  36     real_path = path = os.path.realpath(path)
  37     while True:
  38         curr_path = path
  39         if os.path.isdir(os.path.join(path, '.git')):
  40             curr_path = os.path.join(path, '.git')
  41
  42         if is_git_dir(curr_path):
  43             return curr_path
  44
  45         path = os.path.dirname(path)
  46
  47         if path == '/':
  48             raise ValueError('could not locate .git dir: %s (%s)' %
  49                              (orig_path, real_path))
  50
  51
  52 def find_temp_dir():
  53     """Find a good temp dir (one backed by tmpfs)"""
  54     SEARCH_PATHS = (
  55         '/var/tmp/portage',
  56         '/var/tmp',
  57         '/tmp',
  58         tempfile.gettempdir(),
  59     )
  60     mounts = mount_settings()
  61     for path in SEARCH_PATHS:
  62         if mounts.get(path) == 'tmpfs':
  63             return path
  64     return None
  65
  66
  67 def readfile(path):
  68     """Read |path| and return its data"""
  69     if os.path.isfile(path):
  70         return open(path).read()
  71     return ''
  72
  73
  74 def unlink(path):
  75     """Unlink |path| if it exists else do nothing"""
  76     if os.path.isfile(path):
  77         os.unlink(path)
  78
  79
  80 def clean_hooks(path):
  81     """Strip out sample files from hooks/"""
  82     hooks_path = os.path.join(path, 'hooks')
  83     for hook in glob.glob(os.path.join(hooks_path, '*.sample')):
  84         print('Trimming hook: %s' % hook)
  85         os.unlink(hook)
  86
  87
  88 def clean_packs(path):
  89     """Strip out temp files from objects/packs/"""
  90     packs_path = os.path.join(path, 'objects', 'packs')
  91     for pack in glob.glob(os.path.join(packs_path, 'tmp_pack_*')):
  92         print('Trimming pack: %s' % pack)
  93         os.unlink(pack)
  94
  95
  96 def is_packed(path):
  97     """See if the git repo is already packed"""
  98     obj_path = os.path.join(path, 'objects')
  99     paths = set(os.listdir(obj_path))
 100     if {'info', 'pack'} != paths and {'pack'} != paths:
 101         return False
 102     packs = os.listdir(os.path.join(obj_path, 'pack'))
 103     if len(packs) != 2:
 104         return False
 105     return True
 106
 107
 108 def repack(path):
 109     """Clean up and trim cruft and repack |path|"""
 110     path = find_git_dir(path)
 111     print('Repacking %s' % path)
 112
 113     # Repack any submodules this project might use.
 114     modules_path = os.path.join(path, 'modules')
 115     if os.path.isdir(modules_path):
 116         for root, dirs, _ in os.walk(modules_path):
 117             dirs.sort()
 118             for d in dirs:
 119                 mod_path = os.path.join(root, d)
 120                 if is_git_dir(mod_path):
 121                     repack(mod_path)
 122
 123     tmpdir = find_temp_dir()
 124     if tmpdir:
 125         tmpdir = tempfile.mkdtemp(prefix='git-repack.', dir=tmpdir)
 126         print('Using tempdir: %s' % tmpdir)
 127         os.rmdir(tmpdir)
 128         # Doesn't matter for these needs.
 129         os.environ['GIT_WORK_TREE'] = tmpdir
 130
 131     grafts = alts = None
 132     try:
 133         # Push/pop the graft & alternate paths so we don't read them.
 134         # XXX: In some cases, this is bad, but I don't use them that way ...
 135         graft_file = os.path.join(path, 'info', 'grafts')
 136         grafts = readfile(graft_file)
 137         unlink(graft_file)
 138
 139         alt_file = os.path.join(path, 'objects', 'info', 'alternates')
 140         alts = readfile(alt_file)
 141         unlink(alt_file)
 142
 143         clean_hooks(path)
 144
 145         # XXX: Should do this for all remotes?
 146         origin_path = os.path.join(path, 'refs', 'remotes', 'origin')
 147         packed_refs = readfile(os.path.join(path, 'packed-refs'))
 148         if os.path.exists(origin_path) or 'refs/remotes/origin/' in packed_refs:
 149             cmd = ['git', '--git-dir', path, 'remote', 'prune', 'origin']
 150             subprocess.check_call(cmd, cwd='/')
 151
 152         clean_packs(path)
 153
 154         if is_packed(path):
 155             print('Git repo is already packed; nothing to do')
 156             return
 157
 158         if tmpdir:
 159             print('Syncing git repo to tempdir')
 160             shutil.copytree(path, tmpdir, symlinks=True)
 161             rundir = tmpdir
 162         else:
 163             rundir = path
 164
 165         cmd = ['git', '--git-dir', rundir, 'reflog', 'expire', '--all', '--stale-fix']
 166         print('Cleaning reflog: %s' % ' '.join(cmd))
 167         subprocess.check_call(cmd, cwd='/')
 168
 169         # This also packs refs/tags for us.
 170         cmd = ['git', '--git-dir', rundir, 'gc', '--aggressive', '--prune=all']
 171         print('Repacking git repo: %s' % ' '.join(cmd))
 172         subprocess.check_call(cmd, cwd='/')
 173
 174         # Clean empty dirs.
 175         cmd = ['find', rundir, '-depth', '-type', 'd', '-exec', 'rmdir', '{}', '+']
 176         subprocess.call(cmd, stderr=open('/dev/null', 'w'))
 177
 178         # There's a few dirs we need to exist even if they're empty.
 179         refdir = os.path.join(rundir, 'refs')
 180         if not os.path.isdir(refdir):
 181             os.mkdir(refdir)
 182
 183         if tmpdir:
 184             cmd = ['rsync', '-a', '--delete', tmpdir + '/', path + '/']
 185             print('Syncing back git repo: %s' % ' '.join(cmd))
 186             subprocess.check_call(cmd, cwd='/')
 187             cmd = ['find', path + '/', '-exec', 'chmod', 'u+rw', '{}', '+']
 188             subprocess.check_call(cmd, cwd='/')
 189
 190     finally:
 191         if grafts:
 192             open(graft_file, 'w').write(grafts)
 193         if alts:
 194             open(alt_file, 'w').write(alts)
 195         if tmpdir and os.path.exists(tmpdir):
 196             shutil.rmtree(tmpdir)
 197
 198
 199 def get_parser():
 200     """Get the command line parser"""
 201     parser = argparse.ArgumentParser(description=__doc__)
 202     parser.add_argument('dir', help='The git repo to process')
 203     return parser
 204
 205
 206 def main(argv):
 207     """The main script entry point"""
 208     parser = get_parser()
 209     opts = parser.parse_args(argv)
 210     repack(opts.dir)
 211
 212
 213 if __name__ == '__main__':
 214     exit(main(sys.argv[1:]))