Remove all directories with a given name

Today I had the task to delete all folders with a given name in a directory tree. Seems easy:

find src -type d -name test -exec rm -r {} \;    

In the src folder, find all subfolders named test, and remove them together with all their content. This actually kinda works, but the output was not as expected:

find: ‘src/Api/Car/test’: File or folder not found
find: ‘src/Business/CalculationParameter/test’: File or folder not found
find: ‘src/Infrastructure/RestClient/test’: File or folder not found
... many more

What happens here?

Find finds the test folder ‘src/Api/Car/test‘, executes the rm -r on it, and then wants to continue by descending into the folder – that was just deleted.

So we need to remember all the folders first, and delete them in a second step. I first thought about creating a temp file, piping all the folders to delete into it and feed that to rm afterwards. But there is an easier way:

find src -type d -name test -print0 | xargs -0 rm -r

With the print0 option, find will write all output in the form ‘filename\0filename‘ – meaning with a zero-byte as separator. This we can pipe to xargs, who will split at the zero-byte because of the -0-option.

Basically the same solution, but in-memory instead of using a temp file.

So now our test code is no longer deployed to production any more, saving us a little disk space and releasing us from that nagging feeling that we should not put code in production that is not needed there.