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