# isort: split ignored if in trailing comment

[ndevenish]

test.py:

import x
import y

import f # isort: split

import a
import b

isort is a noop:

% isort --version
                 _                 _
                (_) ___  ___  _ __| |_
                | |/ _/ / _ \/ '__  _/
                | |\__ \/\_\/| |  | |_
                |_|\___/\___/\_/   \_/

      isort your imports, so you don't have to.

                    VERSION 5.13.2
% isort --diff test.py
%

ruff ignores the marker:

% ruff --version
ruff 0.4.8
% ruff check --select=I --diff test.py
--- test.py
+++ test.py
@@ -1,8 +1,7 @@

-import x
-import y
-
-import f # isort: split
+import f  # isort: split

 import a
 import b
+import x
+import y

Would fix 3 errors.

[charliermarsh]

I find it a bit confusing. How should the import f be treated in such a case? Is it "before" or "after" the split, or neither?

[charliermarsh]

I'm tempted to call this unsupported. Do you have a use-case for the inline split?

[ndevenish]

I find it a bit confusing. How should the import f be treated in such a case? Is it "before" or "after" the split, or neither?

It's effectively treated as

import x
import y

# isort: split
import f
# isort: split

import a
import b

This is explicitly documented in isort: https://pycqa.github.io/isort/docs/configuration/action_comments.html#isort-split

You can also use it inline to keep an import from having imports above or below it swap position:

---

I'm tempted to call this unsupported. Do you have a use-case for the inline split?

We've used this in the past for cases where upstream dependencies have not handled arbitrary-import-order very well, and so a specific import needs to be "First". We can work around by using normal split/on/off features.

Perhaps the fact that we're the only ones to notice this in so long is perhaps evidence that supporting it isn't very important.

[seproDev]

I am not sure if this is 100% related, but we also ran in to differences with trailing # isort: split. This is a noop in isort 5.13.2:

from x import a  # isort: split
from x import b

While ruff 0.4.8 changes it to:

from x import a  # isort: split

from x import b

When running

ruff check --select I --fix --isolated

[charliermarsh]

Yeah I think we just don't really support inline # isort: split right now.