git-migrate

   1 #!/usr/bin/python
   2
   3 import glob
   4 import os
   5 import sys
   6 import subprocess
   7 import shutil
   8
   9 def tagBase(pkg):
  10     p = subprocess.Popen(['git', 'tag',
  11                           '-l',
  12                           'base'],
  13                          cwd='%s.git' % pkg,
  14                          stdout=subprocess.PIPE)
  15     p.wait()
  16     if p.stdout.read().strip() != '':
  17         return
  18
  19     p = subprocess.Popen(['git', 'rev-list',
  20                           '--reverse',
  21                           'master'],
  22                          cwd='%s.git' % pkg,
  23                          stdout=subprocess.PIPE)
  24     p.wait()
  25     base = p.stdout.read().split()[0]
  26
  27     subprocess.check_call(['git', 'tag',
  28                            'base',
  29                            base],
  30                           cwd='%s.git' % pkg)
  31
  32 def clonePackage(base, repo_path):
  33     pkg = os.path.basename(repo_path)
  34
  35     if not os.path.isdir('%s.git' % pkg):
  36         if os.path.isdir(pkg):
  37             shutil.rmtree(pkg)
  38
  39         args = []
  40         args.append('-Ttrunk/%s' % repo_path)
  41         if repo_path.startswith('packages/'):
  42             args.append('-tpackage_tags/%s' % pkg)
  43         args.append(base)
  44         args.append(pkg)
  45
  46         # Use --no-follow-parent because we're going to handle that with
  47         # grafts.
  48         subprocess.check_call(['git', 'svn', 'clone',
  49                                '--no-follow-parent',
  50                                '-Aauthors',
  51                                '-q',
  52                                '--no-metadata'] + args,
  53                               stdout=subprocess.PIPE)
  54
  55         # Then make the repository bare, because git-svn can't do this
  56         shutil.move('%s/.git' % pkg, '%s.git' % pkg)
  57         shutil.rmtree(pkg)
  58         subprocess.check_call(['git', 'config', 'core.bare', 'true'],
  59                               cwd='%s.git' % pkg)
  60
  61     # Some of these repos have a rev where everything was deleted
  62     # as a result of the move. We don't want that rev to exist.
  63     p = subprocess.Popen(['git', 'ls-tree', 'HEAD'],
  64                          cwd='%s.git' % pkg,
  65                          stdout=subprocess.PIPE)
  66     p.wait()
  67     if len(p.stdout.read()) == 0:
  68         subprocess.check_call(['git', 'reset', '--soft', 'HEAD^'],
  69                               cwd='%s.git' % pkg)
  70
  71     # Early in the project's history, there were a bunch of double
  72     # directory trees - i.e. the source was actually in
  73     # trunk/packages/$package/$package. Correct for that
  74     cwd = os.getcwd()
  75     os.environ['PACKAGE'] = pkg
  76     p = subprocess.check_call(['git', 'filter-branch',
  77                                '--commit-filter', '%s "$@"' % os.path.join(cwd, 'filter-subdirs'),
  78                                '--tag-name-filter', 'cat',
  79                                '--',
  80                                '--all'],
  81                               cwd='%s.git' % pkg)
  82
  83     shutil.rmtree('%s.git/refs/original' % pkg, True)
  84
  85     tagBase(pkg)
  86
  87 def cloneAllPackages(base):
  88     for pkg in open('package-list'):
  89         clonePackage(base, pkg.strip())
  90
  91 def mergeHistory(old_pkg, new_pkg, n):
  92     n = int(n)
  93
  94     subprocess.check_call(['git', 'push',
  95                            '../%s.git' % new_pkg,
  96                            'master:refs/heads/%s' % old_pkg],
  97                           cwd='%s.git' % old_pkg)
  98
  99     # Find the merge commit
 100     if n == 0:
 101         p = subprocess.Popen(['git', 'rev-parse',
 102                               'base'],
 103                              cwd='%s.git' % new_pkg,
 104                              stdout=subprocess.PIPE)
 105     else:
 106         p = subprocess.Popen(['git', 'rev-list',
 107                               '--reverse',
 108                               '--boundary',
 109                               '--skip=%s' % (n - 1),
 110                               'base..master'],
 111                              cwd='%s.git' % new_pkg,
 112                              stdout=subprocess.PIPE)
 113     p.wait()
 114     new_rev = p.stdout.read().split()[0].strip('-')
 115
 116     # Find any other parents of the merge commit
 117     p = subprocess.Popen(['git', 'log',
 118                           '-1',
 119                           '--pretty=format:%P',
 120                           new_rev],
 121                          cwd='%s.git' % new_pkg,
 122                          stdout=subprocess.PIPE)
 123     p.wait()
 124     parents = p.stdout.read().split()
 125
 126     # Find the additional parent we're adding
 127     p = subprocess.Popen(['git', 'rev-parse',
 128                           old_pkg],
 129                          cwd='%s.git' % new_pkg,
 130                          stdout=subprocess.PIPE)
 131     p.wait()
 132     parents.append(p.stdout.read().strip())
 133
 134     # Write out the grafts file
 135     f = open('%s.git/info/grafts' % new_pkg, 'a')
 136     print >>f, '%s %s' % (new_rev, ' '.join(parents))
 137     f.close()
 138
 139     # Run filter-branch
 140     subprocess.call(['git', 'filter-branch',
 141                      '--tag-name-filter', 'cat',
 142                      '--',
 143                      '--all'],
 144                     cwd='%s.git' % new_pkg)
 145
 146     subprocess.call(['git', 'branch',
 147                      '-D',
 148                      old_pkg],
 149                     cwd='%s.git' % new_pkg)
 150     shutil.rmtree('%s.git/refs/original' % new_pkg, True)
 151
 152 def mergeHistories():
 153     merges = []
 154     for line in open('merges'):
 155         line = line.strip()
 156         if line == '' or line[0] == '#':
 157             continue
 158
 159         merges.append(line.split())
 160
 161     for merge in merges:
 162         mergeHistory(*merge)
 163
 164     for merge in merges:
 165         shutil.rmtree('%s.git' % merge[0])
 166
 167 def cleanupRepos():
 168     for pkg in glob.glob('*.git'):
 169         subprocess.check_call(['git', 'tag', '-d', 'base'],
 170                               cwd='%s' % pkg)
 171
 172         subprocess.check_call(['git', 'gc'],
 173                               cwd='%s' % pkg)
 174
 175 if __name__ == '__main__':
 176     try:
 177         base = sys.argv[1]
 178     except:
 179         base = 'svn://invirt.mit.edu'
 180
 181     cloneAllPackages(base)
 182     mergeHistories()
 183     cleanupRepos()