Repository navigation
Add CliRunner.set_filesystem() to use a specific working directory #3123
Description
Activity
I'm not clear why this is needed. I can't think of a good reason for a command to require being run in a directory with a certain name. That said, if you need it for some reason, create it in the isolated context, then cd further.
with runner.isolated_filesystem() as tmp_dir: lib_dir = Path(tmp_dir, "library") lib_dir.mkdir() os.chdir(lib_dir) ...
Nowadays you can also use pytest directly for this, or use
with tempfile.TemporaryDirectory(). I don't think we'd addisolated_filesystemtoday, I think it was added before Python 3 and pytest were in common use.def test_isolated(tmp_path, monkeypatch): lib_dir = tmp_path / "library" lib_dir.mkdir() monkeypatch.chdir(lib_dir)
with tempfile.TemporaryDirectory() as tmp_dir: lib_dir = Path(tmp_dir, "library") lib_dir.mkdir() os.chdir(lib_dir)
In that case, perhaps consider marking
isolated_filesystem()as deprecated. It seemed to me to be the way things should be done, and without knowing otherwise I assumed that it did more than justos.chdir().I am in favor of deprecating
isolated_filesystem()and simplify code. Maybe for Click 9.x.Pytest is maintained and wide-spread. We can simply add a snippet in the documentation to explain how to use the
CliRunnerwith vanilla Pytest.- locked as resolved and limited conversation to collaborators
on Nov 9, 2025 - changed the title
[-]Add CliRunner.set_filesystem() to use a specific working directory[/-][+]Add `CliRunner.set_filesystem()` to use a specific working directory[/+]on May 23, 2026
The
CliRunner.isolated_filesystemis find for testing tools that have no preconceived notion of what the directory structure should look like, or what files must be present. While it's true that you can provide a parent directory toisolated_filesystem(), it will still create a brand new empty directory and operate within it.CliRunner.set_filesystem()operates in the same way, except that you provide it with a pre-existing directory to operate in.Suppose your tool requires that it be run in a directory named 'library' containing certain data files. Your test might look start like this:
With
isolated_filesystem(), running your command in this directory will not work, because it will run not in the library directory just created, but in a sub-directory of that one.With
set_filesystem(), you get what you need:... runner = CliRunner() with runner.set_filesystem(lib_dir): runner.invoke(...)