Tutorial on how to transform a repository directory into a submodule

Dir2Submodule.md

Directory to submodule

Sometimes you realize a part of your current project (like a library for example) should have it's own repository. Sure, you could simply copy and paste the files into a newly created repository, but you would lose all the previous commit network. Not loosing this network while isolating a part of your porject is the goal of this tutorial.

Let ~root/project be the absolute path of your project and path/to/subproject the path to your subproject directory (the directory you want to isolate) relatively to the project path, so that ~root/project/path/to/subproject is the absolute path of the subproject.

Since we never modify to the actual project until the last step, it's all safe. However, always keep a copy of your project somewhere.

1. Isolate the submodule in its own directory.

   ~root/project$ git clone .git ../subprojectname
   ~root/project$ cd ../subprojectname
   ~root/subprojectname$ git remote rm origin
   ~root/subprojectname$ git filter-branch --subdirectory-filter 'path/to/subproject' --prune-empty -- --all

2. Add ~root/subprojectname to your GitHub repositories. I prefer to do so through GitHub Desktop. Say you published [email protected]:USERNAME/subprojectname.git.

   Feel free to rebase the commit network of the submodule before you publish it.

   If you don't wish to publish it, you're done. Great job 👍

   If you don't want or cannot modify the whole project, stop here as well. 🚧

3. Now, we want to replace the subproject directory within the project by a link to the newly created repository in order to support future versions. First, to delete the previous version, you have two options :

   1. Delete manually the subproject directory. It will let a trace of the subproject in the project commit network.

   2. Remove any trace of the subproject from the project commit network. Don't do this if a lot of, or even a few, people use the repository.

      ~root/subprojectname$ cd ../project
      ~root/project$ git clone .git ../copy
      ~root/project$ cd ../copy
      ~root/copy$ git remote rm origin
      ~root/copy$ git filter-branch -f --index-filter 'git rm -rf --cached --ignore-unmatch path/to/subproject' HEAD
      ~root/copy$ git filter-branch -f --commit-filter 'git_commit_non_empty_tree "$@"' HEAD

      Feel free to clean up manually the commit network after this operation.

4. Add the subproject directory to the project as a submodule.

   ~root/copy$ cd path/to/
   ~root/copy/path/to$ git submodule add [email protected]:USERNAME/subprojectname.git

5. There you go, ~root/copy is the new version of your project. Nevertheless, you still need to push it to GitHub. Say your project repository is [email protected]:USERNAME/project.git.

   ⚠️ If your are not satisfied with the result, or if anything went wrong during the process, don't push it. Try again or stop to the second step if so.

   ~root/copy$ git push -f [email protected]:USERNAME/project.git

And you are done ! Congratulations 👏👏