Move files from temporary directory changer

[mgjarrett]

What is the change?

Use shutil.move / mv instead of shutil.copy / cp to move files from temporary directory back to the original working directory on __exit__.

Why is the change being made?

Copying files from one location to another keeps the original file intact, but the original files in a temporary directory do not need to be kept around after the __exit__ from the TemporaryDirectoryChanger. These files are immediately deleted after the copy operation is completed.

Using shutil.move() is significantly faster than shutil.copy() when the source and destination locations are on the same drive, since the data doesn't have to physically be copied to a new place in memory.

---

Checklist

• This PR has only one purpose or idea.
• Tests have been added/updated to verify any new/changed code.

• The code style follows good practices.
• The commit message(s) follow good practices.

• The release notes have been updated if necessary.
• The documentation is still up-to-date in the doc folder.
• The dependencies are still up-to-date in pyproject.toml.

[opotowsky]

Looks like the move doesn't work with windows/mac.

But I'd like to make a higher level comment. The secret overwriting of shutil.copy is something I want to remove from ARMI (i.e., I want to go change all the instances of shutil.copy to explicitly use safeCopy instead, then remove the overwrite. See #1690.

Maybe for this update it makes sense to keep the same scheme and we just tack this onto that tech debt payoff since it won't really be causing any extra work.

But I'd also be happy if the shutil.move calls were just safeMove instead. :-)

[mgjarrett]

I didn't touch safeCopy at all, but I added some assert statements to verify that the safeCopy call is actually copying the files. The unit test for safeCopy passes locally for me, but here it is now failing on those assertions, which is strange.

It's also weird that the existing unit test of safeCopy doesn't actually assert that the files are copied successfully. Makes me wonder if there is a known issue here.

Edit: It looks like the problem was the "windows-like" file path dir1\\file2.txt, which doesn't get picked up by safeCopy. I removed the assertion of this file existing and now the test passes.

[opotowsky]

This looks great!

[john-science]

The secret overwriting of shutil.copy is something I want to remove from ARMI

I note we already have a ticket for this: #1690

But I think Michael's PR doesn't touch that, because he didn't alter safeCopy(), but added a new saveMove(). Am I wrong?

[opotowsky]

The secret overwriting of shutil.copy is something I want to remove from ARMI

I note we already have a ticket for this: #1690

But I think Michael's PR doesn't touch that, because he didn't alter safeCopy(), but added a new saveMove(). Am I wrong?

Yep! the first iteration in this PR had a similar setup, (where shutil.move was overwritten with safeMove) but that was undone in favor of not overwriting shutil.move.