Convert folders named d1, d2, d3, etc into git repository

d2git.py

#!/usr/bin/env python
#
# Convert "versioning" with folders called d1, d2, d3.. into a git repository in the directory
# containing those folders. Written for use on *nix systems for people who weren't using version
# control software for a long time.

import os
import re
import time
import sys
import argparse
import tempfile
from datetime import datetime
from subprocess import check_output, check_call

# Set up command line interface
parser = argparse.ArgumentParser(description='Convert d1, d2, d3.. style versioning into git repositories')
parser.add_argument('--path', '-p', default=os.getcwd(), help='Directory to run conversion in. Defaults to current working directory')
parser.add_argument('--author', '-a', default='Git Robot <[email protected]>', help='Git commit author')
parser.add_argument('-v', '--verbose', action='store_true')
args = parser.parse_args()

GIT_AUTHOR = args.author

# Print a message for debugging
def debug(message):
    if args.verbose:
        print "%s: %s" % (datetime.now().strftime('%Y-%m-%d %H:%M:%S'), message)


# Ensure the given path exists
if not os.path.isdir(args.path):
    print '%s is not a directory' % (args.path)
    sys.exit(1)

debug('Working with %s' %os.path.abspath(args.path))

# Get dirs named in draft style
draft_folder = re.compile('d[0-9]+.*',re.IGNORECASE)
dirs = os.walk(args.path).next()[1]
dirs = filter(draft_folder.match, dirs)
dirs.sort()

debug("Matching directories: %s" % str(dirs))

# Bail out if not a draft-style directory parent
if len(dirs) == 0:
    print 'No draft-style directories found in %s' % (args.path)
    sys.exit(1)

# Set up temp directory
TEMP_DIR = tempfile.mkdtemp('', 'd2git-')
check_call(['git', 'init', TEMP_DIR])

# Loop through all d## folders (ascending) and commit each version to the temp repository
to_remove = []
commits = 0
for name in dirs:
    path = os.path.abspath( os.path.join(args.path, name) )
    date = time.strftime('%a %B %d %H:%M:%S %Y %z', time.localtime(os.path.getmtime(path)))
    commit_msg = "%s (backdated, automated)" % (name)
    debug('COMMIT (%s): %s' % (date, commit_msg))

    # Copy directory to working copy
    check_call(['rsync', '-a', '--delete', '--exclude=.git', path + "/", TEMP_DIR])
    check_call(['git', '-C', TEMP_DIR, 'add', '--all', '.'])

    try:
        check_call(['git', '-C', TEMP_DIR, 'commit', '-a', '-m', commit_msg, '--author='+GIT_AUTHOR, '--date='+date])
    except:
        # Git writes to stderr, so the cause of this error will be known.
        # This is usually caused by having no change in git's working dir
        doContinue = raw_input("Git returned a non-zero code. Continue? y/N: ")
        if(doContinue != "y" and doContinue !="Y"):
            exit(1)
        else:
            pass

    # Add the successfully processed dir to a removal list
    to_remove.append(path)

    # Delete any zip of the version that exists (these are unecessary and waste space)
    zip_path = os.path.join(os.path.abspath(args.path), name + '.zip')

    if os.path.isfile(zip_path):
        to_remove.append(zip_path)

# Clean up and compact the new repo
check_call(['git', '-C', TEMP_DIR, 'gc'])

for path in to_remove:
    debug('removing %s' % path)
    check_call(['rm', '-rf', path])

# Move repository back into original directory and clean up the temp directory
debug('Syncing repository back to source')
check_call(['rsync', '-a', TEMP_DIR + '/', args.path])

debug('removing temp directory')
check_call(['rm', '-rf', TEMP_DIR])