.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 find_git_dir(path):
  27     """Try to find the .git dir to operate on"""
  28     orig_path = path
  29     real_path = path = os.path.realpath(path)
  30     while True:
  31         curr_path = path
  32         if os.path.isdir(os.path.join(path, '.git')):
  33             curr_path = os.path.join(path, '.git')
  34
  35         if (os.path.isdir(os.path.join(curr_path, 'refs')) and
  36             os.path.isdir(os.path.join(curr_path, 'objects')) and
  37             os.path.isfile(os.path.join(curr_path, 'config'))):
  38             return curr_path
  39
  40         path = os.path.dirname(path)
  41
  42         if path == '/':
  43             raise ValueError('could not locate .git dir: %s (%s)' %
  44                              (orig_path, real_path))
  45
  46
  47 def find_temp_dir():
  48     """Find a good temp dir (one backed by tmpfs)"""
  49     SEARCH_PATHS = (
  50         '/var/tmp/portage',
  51         '/var/tmp',
  52         '/tmp',
  53         tempfile.gettempdir(),
  54     )
  55     mounts = mount_settings()
  56     for path in SEARCH_PATHS:
  57         if mounts.get(path) == 'tmpfs':
  58             return path
  59     return None
  60
  61
  62 def readfile(path):
  63     """Read |path| and return its data"""
  64     if os.path.isfile(path):
  65         return open(path).read()
  66     return ''
  67
  68
  69 def unlink(path):
  70     """Unlink |path| if it exists else do nothing"""
  71     if os.path.isfile(path):
  72         os.unlink(path)
  73
  74
  75 def clean_hooks(path):
  76     """Strip out sample files from hooks/"""
  77     hooks_path = os.path.join(path, 'hooks')
  78     for hook in glob.glob(os.path.join(hooks_path, '*.sample')):
  79         print('Trimming hook: %s' % hook)
  80         os.unlink(hook)
  81
  82
  83 def clean_packs(path):
  84     """Strip out temp files from objects/packs/"""
  85     packs_path = os.path.join(path, 'objects', 'packs')
  86     for pack in glob.glob(os.path.join(packs_path, 'tmp_pack_*')):
  87         print('Trimming pack: %s' % pack)
  88         os.unlink(pack)
  89
  90
  91 def is_packed(path):
  92     """See if the git repo is already packed"""
  93     if set(('info', 'pack')) != set(os.listdir(path)):
  94         return False
  95     packs = os.listdir(os.path.join(path, 'pack'))
  96     if len(packs) != 2:
  97         return False
  98     return True
  99
 100
 101 def repack(path):
 102     """Clean up and trim cruft and repack |path|"""
 103     path = find_git_dir(path)
 104     print('Repacking %s' % path)
 105
 106     tmpdir = find_temp_dir()
 107     if tmpdir:
 108         tmpdir = tempfile.mkdtemp(prefix='git-repack.', dir=tmpdir)
 109         print('Using tempdir: %s' % tmpdir)
 110         os.rmdir(tmpdir)
 111
 112     grafts = alts = None
 113     try:
 114         # Push/pop the graft & alternate paths so we don't read them.
 115         # XXX: In some cases, this is bad, but I don't use them that way ...
 116         graft_file = os.path.join(path, 'info', 'grafts')
 117         grafts = readfile(graft_file)
 118         unlink(graft_file)
 119
 120         alt_file = os.path.join(path, 'objects', 'info', 'alternates')
 121         alts = readfile(alt_file)
 122         unlink(alt_file)
 123
 124         clean_hooks(path)
 125
 126         origin_path = os.path.join(path, 'refs', 'remotes', 'origin')
 127         packed_refs = readfile(os.path.join(path, 'packed-refs'))
 128         if os.path.exists(origin_path) or 'refs/remotes/origin/' in packed_refs:
 129             cmd = ['git', '--git-dir', path, 'remote', 'prune', 'origin']
 130             subprocess.check_call(cmd, cwd='/')
 131
 132         clean_packs(path)
 133
 134         if is_packed(path):
 135             print('Git repo is already packed; nothing to do')
 136             return
 137
 138         if tmpdir:
 139             print('Syncing git repo to tempdir')
 140             shutil.copytree(path, tmpdir, symlinks=True)
 141             rundir = tmpdir
 142         else:
 143             rundir = path
 144
 145         cmd = ['git', '--git-dir', rundir, 'reflog', 'expire', '--all', '--stale-fix']
 146         print('Cleaning reflog: %s' % ' '.join(cmd))
 147         subprocess.check_call(cmd, cwd='/')
 148
 149         # This also packs refs/tags for us.
 150         cmd = ['git', '--git-dir', rundir, 'gc', '--aggressive', '--prune=all']
 151         print('Repacking git repo: %s' % ' '.join(cmd))
 152         subprocess.check_call(cmd, cwd='/')
 153
 154         # Clean empty dirs.
 155         cmd = ['find', rundir, '-depth', '-type', 'd', '-exec', 'rmdir', '{}', '+']
 156         subprocess.call(cmd, stderr=open('/dev/null', 'w'))
 157
 158         if tmpdir:
 159             cmd = ['rsync', '-a', '--delete', tmpdir + '/', path + '/']
 160             print('Syncing back git repo: %s' % ' '.join(cmd))
 161             subprocess.check_call(cmd, cwd='/')
 162             cmd = ['find', path + '/', '-exec', 'chmod', 'u+rw', '{}', '+']
 163             subprocess.check_call(cmd, cwd='/')
 164
 165     finally:
 166         if grafts:
 167             open(graft_file, 'w').write(grafts)
 168         if alts:
 169             open(alt_file, 'w').write(alts)
 170         if tmpdir:
 171             shutil.rmtree(tmpdir)
 172
 173
 174 def get_parser():
 175     """Get the command line parser"""
 176     parser = argparse.ArgumentParser(description=__doc__)
 177     parser.add_argument('dir', help='The git repo to process')
 178     return parser
 179
 180
 181 def main(argv):
 182     """The main script entry point"""
 183     parser = get_parser()
 184     opts = parser.parse_args(argv)
 185     repack(opts.dir)
 186
 187
 188 if __name__ == '__main__':
 189     exit(main(sys.argv[1:]))