Split a monolithic git repo into several small ones

splitrepos.py

# -*- coding: utf-8 -*-
"""
I used the following script to convert each subdirectory of work.git into its own git repo. The one exception was paper-2013.01.16_nea_collector which I did by hand using essentially the same method as below, except I skipped the part where I removed the tags. I used the procedure found at [1] and slightly modified it. I also used the solution in [2] to convert non-bare repos to bare. Note also that I first cloned work from its bare git repo on the SD card, removed the remote, and tarred the result to work.tar.

[1] http://stackoverflow.com/questions/359424/detach-subdirectory-into-separate-git-repository
[2] http://stackoverflow.com/questions/1784506/when-creating-a-git-repository-that-will-be-on-the-server-can-i-convert-it-to-a
"""

import os
import subprocess
import shutil

subprocess.call(["tar","xf","work.tar"])
contents = os.listdir("work")

for dirname in contents:
  if dirname[0] is not ".":
    os.chdir("work")

    # Remove the tags.
    print("=" * 50)
    print "="*5, dirname, "="*5
    print("removing tags")
    git_tag = subprocess.Popen(["git","tag","-l"],stdout=subprocess.PIPE)
    git_tag.wait()
    xargs = subprocess.Popen(["xargs","git","tag","-d"],stdin=git_tag.stdout)
    xargs.wait()

    print("git filter-branch")
    subprocess.call(["git","filter-branch","--tag-name-filter","cat","--prune-empty","--subdirectory-filter",dirname,"--","--all"])

    print("deleting the backup reflogs")
    subprocess.call(["git","reset","--hard"])
    git_foreach = subprocess.Popen(["git","for-each-ref",'--format="%(refname)"',"refs/original/"], stdout=subprocess.PIPE)
    xargs = subprocess.Popen(["xargs","-n","1","git","update-ref","-d"], stdin = git_foreach.stdout)
    subprocess.call(["git","reflog","expire","--expire=now","--all"])
    subprocess.call(["git","gc","--aggressive","--prune=now"])

    print("Moving newly filter-branched repo to its own folder")
    os.chdir("..")
    shutil.move("work","split_repos/"+dirname)
    
    print("Turning repo to bare, then removing the non-bare copy.")
    os.chdir("split_repos")
    subprocess.call(["git", "clone", "--bare", "-l", dirname, dirname + ".git"])
    shutil.rmtree(dirname)
    os.chdir("..")
    
    print("Un-taring the work git repo again.")
    subprocess.call(["tar","xf","work.tar"])