Igor Kromin |   Consultant. Coder. Blogger. Tinkerer. Gamer.

I've written previously about how to do a one time non-standard SVN repository migration to Git which works fine if you want to move your entire SVN repo, but what if you want to move just a subfolder from SVN to Git? This is the guide for you.

Moving a single subfolder is significantly more work than moving the entire repository. Overall the steps are:
  1. Create author mapping file
  2. Clone SVN into a temporary Git repository
  3. Convert svn:ignore properties to .gitignore file
  4. Rewrite the Git index file to update subfolder paths (this is required to preserve file history)
  5. Add a remote origin in the Git repository that files are being merged into
  6. Fetch and merge the remote Git repository files from the temporary Git repository

So lets see this step-by-step. I'm going to assume the following:
SVN subfolder URL: http://localhost/repos/mysvn/tags/mytag/SubFolder
Working directory: ~/temp
Git target directory: ~/mygit

Create the author mapping file. This will generate a file called authors-transform.txt which contains SVN authors with their Git author equivalents. The default value for the Git author is set to 'User <[email protected]>' so adjust that if you care about having real author information.
cd ~/temp
svn log --quiet http://localhost/repos/mysvn/tags/mytag/SubFolder | grep "^r" | awk '{print $3" = User <[email protected]>"}' | sort | uniq > authors-transform.txt

Clone the SVN repository into a temporary Git repository and fix up remote references. Note that I set the trunk to the current directory since we are only cloning a single folder from SVN.
git svn clone http://localhost/repos/mysvn/tags/mytag/SubFolder --trunk=. -A authors-transform.txt svngit
cd svngit
sed -i .bak -e "s|refs/remotes/origin/trunk|refs/remotes/git-svn|g" .git/config

Transform svn:ignore properties to the .gitignore file. Note this is put into a new directory that has the same name as the subfolder that was cloned.
mkdir SubFolder
git svn show-ignore > SubFolder/.gitignore
git add SubFolder/.gitignore
git commit -m 'Convert svn:ignore properties to .gitignore.'

Rewrite the Git index so that the SubFolder appears in the correct location while preserving file history. This essentially moves all files that were cloned from SVN into a new directory called SubFolder and fixes Git history to point to the right files there. I know the documentation provides a sed example of this, however I was doing it on OS X so had to use awk instead.
echo '{print $1"\tSubFolder/"$2}' > /tmp/prog.awk
git filter-branch --index-filter \
'git ls-files -s | awk -F"\t" -f /tmp/prog.awk | \
git update-index --index-info && \
if [ -f "$GIT_INDEX_FILE.new" ]; then mv "$GIT_INDEX_FILE.new" "$GIT_INDEX_FILE"; fi' \
rm /tmp/prog.awk

In the Git directory add a remote to the temporary Git directory we just created, fetch and merge the files, then remove the remote reference.
cd ~/mygit
git remote add SubFolder ~/temp/svngit
git fetch SubFolder
git merge SubFolder/master
git remote remove SubFolder

Now you can finally push the change to the target Git repository.
git push

The temporary Git repository can be removed at this point.


Have comments or feedback on what I wrote? Please share them below!
comments powered by Disqus
Other posts you may like...