source: trunk/scripts/git-migration/git-migrate @ 1938

#!/usr/bin/python

import glob
import os
import sys
import subprocess
import shutil

def tagBase(pkg):
    p = subprocess.Popen(['git', 'tag',
                          '-l',
                          'base'],
                         cwd='%s.git' % pkg,
                         stdout=subprocess.PIPE)
    p.wait()
    if p.stdout.read().strip() != '':
        return
    
    p = subprocess.Popen(['git', 'rev-list',
                          '--reverse',
                          'master'],
                         cwd='%s.git' % pkg,
                         stdout=subprocess.PIPE)
    p.wait()
    base = p.stdout.read().split()[0]
    
    subprocess.check_call(['git', 'tag',
                           'base',
                           base],
                          cwd='%s.git' % pkg)

def clonePackage(base, repo_path):
    pkg = os.path.basename(repo_path)
    
    if not os.path.isdir('%s.git' % pkg):
        if os.path.isdir(pkg):
            shutil.rmtree(pkg)
        
        args = []
        args.append('-Ttrunk/%s' % repo_path)
        if repo_path.startswith('packages/'):
            args.append('-tpackage_tags/%s' % pkg)
        args.append(base)
        args.append(pkg)
        
        # Use --no-follow-parent because we're going to handle that with
        # grafts.
        subprocess.check_call(['git', 'svn', 'clone',
                               '--no-follow-parent',
                               '-Aauthors',
                               '-q',
                               '--no-metadata'] + args,
                              stdout=subprocess.PIPE)
        
        # Then make the repository bare, because git-svn can't do this
        shutil.move('%s/.git' % pkg, '%s.git' % pkg)
        shutil.rmtree(pkg)
        subprocess.check_call(['git', 'config', 'core.bare', 'true'],
                              cwd='%s.git' % pkg)
        
    # Some of these repos have a rev where everything was deleted
    # as a result of the move. We don't want that rev to exist.
    p = subprocess.Popen(['git', 'ls-tree', 'HEAD'],
                         cwd='%s.git' % pkg,
                         stdout=subprocess.PIPE)
    p.wait()
    if len(p.stdout.read()) == 0:
        subprocess.check_call(['git', 'reset', '--soft', 'HEAD^'],
                              cwd='%s.git' % pkg)
    
    # Early in the project's history, there were a bunch of double
    # directory trees - i.e. the source was actually in
    # trunk/packages/$package/$package. Correct for that
    cwd = os.getcwd()
    os.environ['PACKAGE'] = pkg
    p = subprocess.check_call(['git', 'filter-branch',
                               '--commit-filter', '%s "$@"' % os.path.join(cwd, 'filter-subdirs'),
                               '--tag-name-filter', 'cat',
                               '--',
                               '--all'],
                              cwd='%s.git' % pkg)
    
    shutil.rmtree('%s.git/refs/original' % pkg, True)
    
    tagBase(pkg)

def cloneAllPackages(base):
    for pkg in open('package-list'):
        clonePackage(base, pkg.strip())

def mergeHistory(old_pkg, new_pkg, n):
    n = int(n)
    
    subprocess.check_call(['git', 'push',
                           '../%s.git' % new_pkg,
                           'master:refs/heads/%s' % old_pkg],
                          cwd='%s.git' % old_pkg)
    
    # Find the merge commit
    if n == 0:
        p = subprocess.Popen(['git', 'rev-parse',
                              'base'],
                             cwd='%s.git' % new_pkg,
                             stdout=subprocess.PIPE)
    else:
        p = subprocess.Popen(['git', 'rev-list',
                              '--reverse',
                              '--boundary',
                              '--skip=%s' % (n - 1),
                              'base..master'],
                             cwd='%s.git' % new_pkg,
                             stdout=subprocess.PIPE)
    p.wait()
    new_rev = p.stdout.read().split()[0].strip('-')
    
    # Find any other parents of the merge commit
    p = subprocess.Popen(['git', 'log',
                          '-1',
                          '--pretty=format:%P',
                          new_rev],
                         cwd='%s.git' % new_pkg,
                         stdout=subprocess.PIPE)
    p.wait()
    parents = p.stdout.read().split()
    
    # Find the additional parent we're adding
    p = subprocess.Popen(['git', 'rev-parse',
                          old_pkg],
                         cwd='%s.git' % new_pkg,
                         stdout=subprocess.PIPE)
    p.wait()
    parents.append(p.stdout.read().strip())
    
    # Write out the grafts file
    f = open('%s.git/info/grafts' % new_pkg, 'a')
    print >>f, '%s %s' % (new_rev, ' '.join(parents))
    f.close()
    
    # Run filter-branch
    subprocess.call(['git', 'filter-branch',
                     '--tag-name-filter', 'cat',
                     '--',
                     '--all'],
                    cwd='%s.git' % new_pkg)
    
    subprocess.call(['git', 'branch',
                     '-D',
                     old_pkg],
                    cwd='%s.git' % new_pkg)
    shutil.rmtree('%s.git/refs/original' % new_pkg, True)

def mergeHistories():
    merges = []
    for line in open('merges'):
        line = line.strip()
        if line == '' or line[0] == '#':
            continue
        
        merges.append(line.split())
    
    for merge in merges:
        mergeHistory(*merge)
    
    for merge in merges:
        shutil.rmtree('%s.git' % merge[0])

def cleanupRepos():
    for pkg in glob.glob('*.git'):
        subprocess.check_call(['git', 'tag', '-d', 'base'],
                              cwd='%s' % pkg)
        
        subprocess.check_call(['git', 'gc'],
                              cwd='%s' % pkg)

if __name__ == '__main__':
    try:
        base = sys.argv[1]
    except:
        base = 'svn://invirt.mit.edu'
    
    cloneAllPackages(base)
    mergeHistories()
    cleanupRepos()