Convert PyPE's svn repository to git

simple_svn2git.py

'''
simple_svn2git.py
Written November 2010 by Josiah Carlson, released into the public domain.

This script was used to convert PyPE's svn repository on sourceforge to git.

Setup included checking out/cloning the two repos:

svn co https://pype.svn.sourceforge.net/svnroot/pype pype
git clone ssh://josiahcarlson@pype.git.sourceforge.net/gitroot/pype/pype pype.git

I then added the README file to the git repo, made a small modification, and
added it again... just in case I needed to do a "git rebase -i" to blow
everything away.  The 'git reset' in case of exception worked to reset the
repo in case of error.

After the setup was completed, I ran this script in the parent of both of the
local paths, performed some verification, then did 'git push origin master' in
the git path.


This was written because both the standard 'git svn' and 'svn2git' options
would result in a git repo with no HEAD, no files, no history, etc.  Since
PyPE didn't use tags, branches, etc., this script worked just fine.  For other
repositories with only trunk, no tags, etc., a suitably modified version of
this script should "just work", though I make no promises and offer no
support.

'''

import os
import shutil
import stat
import subprocess

def get_output(args):
    p = subprocess.Popen(args, shell=False, bufsize=4096,
        stdin=subprocess.PIPE, stdout=subprocess.PIPE, stderr=subprocess.STDOUT)
    r = p.communicate()[0]
    if 'cannot create' in r:
        print r
        raise Exception
    if 'fatal:' in r:
        print r
        raise Exception
    return r

def get_files_and_changelog(revision):
    flist = [line.split(None, 1) for line in get_output(['svn', 'up', '-r', str(revision)]).split('\n')]
    flist.pop()
    flist.pop()
    log = get_output(['svn', 'log', '-c', str(revision)]).strip('-\n ')
    return flist, log

def isdir(path):
    return stat.S_ISDIR(os.stat(path).st_mode)

def handle_file_changes(cwd, flist, message):
    for ch, filename in flist:
        dfn = filename if not filename.startswith('pype/') else filename[5:]
        if ch in ('A', 'U'):
            if isdir(os.path.join(cwd, 'pype', filename)):
                continue
            real_dest = os.path.join(cwd, 'pype.git', dfn)
            try:
                os.makedirs(os.path.split(real_dest)[0])
            except OSError:
                pass
            mode = 'r'
            if filename.split('.')[-1] in ('py', 'txt', 'cfg'):
                # get rid of Windows line endings :P
                inp = open(filename, 'rU')
                out = open(real_dest, 'w')
                shutil.copyfileobj(inp, out)
                inp.close()
                out.close()
            else:
                print get_output(['cp', filename, real_dest])
    os.chdir(os.path.join(cwd, 'pype.git'))
    for ch, filename in flist:
        dfn = filename if not filename.startswith('pype/') else filename[5:]
        if ch in ('A', 'U'):
            if isdir(os.path.join(cwd, 'pype', filename)):
                continue
            print get_output(['git', 'add', dfn])
        else:
            try:
                os.stat(dfn)
            except OSError:
                continue
            print get_output(['git', 'rm', dfn])
    print get_output(['git', 'commit', '-m', message])

def handle_one_revision(revision):
    cwd = os.getcwd()
    os.chdir(os.path.join(cwd, 'pype'))
    files, log = get_files_and_changelog(revision)
    handle_file_changes(cwd, files, log)
    os.chdir(cwd)

cwd = os.getcwd()
try:
    for i in xrange(2, 76):
        handle_one_revision(i)
except:
    os.chdir(os.path.join(cwd, 'pype'))
    print get_output(['svn', 'up', '-r', '1'])
    os.chdir(os.path.join(cwd, 'pype.git'))
    print get_output(['git', 'reset', '--hard', '1dc57e0167a882af2821a0a13a2e142e0ac7fbc1'])
    os.chdir(cwd)
    raise