Fix TemporaryFileSwap regression where file_path could not be Path #1776

This is a general test for TemporaryFileSwap, but by being parametrized by the type of file_path, it reveals a regression introduced in 9e86053 (gitpython-developers#1770). TemporaryFileSwap still works when file_path is a string, but is now broken when it is a Path. That worked before, and the type annotations document that it should be able to work. This is at least a bug because TemporaryFileSwap is public. (I am unsure whether, in practice, GitPython itself uses it in a way that sometimes passes a Path object as file_path. But code that uses GitPython may call it directly and pass Path.)

This fixes the regression introduced in 9e86053 (gitpython-developers#1770) where the file_path argument to TemporaryFileSwap.__init__ could no longer be a Path object. The change also makes this truer to the code from before gitpython-developers#1770, still without the race condition fixed there, in that str was called on file_path then as well. However, it is not clear that this is a good thing, because this is not an idiomatic use of mkstemp. The reason the `prefix` cannot be a Path is that it is expected to be a filename prefix, with leading directories given in the `dir` argument.

This changes the mkstemp call TemporaryFileSwap uses to atomically create (and thus reserve) the temporary file, so that it passes only a filename (i.e., basename) prefix as `prefix`, and passes all other path components (i.e., the directory) as `dir`. (If the original path has no directory, then `dir` is "" as before.) This makes the mkstemp call *slightly* more idiomatic. This also makes it clearer, as it is no longer using `prefix` for something that feels like it should be possible to pass as a Path object. Although mkstemp does not accept a Path as `prefix`, it does (as expected) accept one as `dir`. However, to keep the code simple, I'm passing str for both. The os.path.split function accepts both str and Path (since Python 3.6), and returns str objects, which are now used for the `dir` and `prefix` arguments to mkstemp. For unusual cases, this may technically not be a refactoring. For example, a file_path of "a/b//c" will be split into "a/b" and "c". If the automatically generated temporary file suffix is "xyz", then that results in a tmp_file_path of "a/b/cxyz" where "a/b//cxyz" would have been used before. The tmp_file_path attribute of a TemporaryFileSwap object is public (and used in filesystem calls). However, no guarantee has ever been given that the temporary file path have the original path as an exact string prefix. I believe the slightly weaker relationship I expressed in the recently introduced test_temporary_file_swap -- another file in the same directory, named with the original filename with more characters, consisting of equivalent path components in the same order -- has always been the intended one. Note that this slight possible variation does not apply to the file_path attribute. That attribute is always kept exactly as it was, both in its type and its value, and it always used unmodified in calls that access the filesystem.

This removes the "fmt: off" / "fmt: on" directives around the `@pytest.mark.parametrize` decoration on test_blob_filter, and reformats it with black, for consistency with other such decorations. The style used there, *if* it could be specified as a rule and thus used without "fmt:" directives, may be nicer than how black formats multi-line mark decorations. However, since that decoration was written, there have been a number of other such decorations, which have been in black style. This also removes the only (or only remaining?) "fmt:" directive in the codebase. As such, it should possibly have been done in gitpython-developers#1760.

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Fix TemporaryFileSwap regression where file_path could not be Path #1776

Fix TemporaryFileSwap regression where file_path could not be Path #1776

Commits on Dec 21, 2023