[Akashijk]

Summary

Fix fatx_open_dir() incorrectly resolving an intermediate directory component as a longer sibling that shares the same prefix.

For example, with sibling directories:

wolf/
wosk/
w/

[xmutantson]

Confirming this one on a real drive. rsync sorts directories as if they had a trailing /, so BloodRayne 2/ is created before BloodRayne/, and every later lookup of BloodRayne/... resolved into BloodRayne 2. The two games' files ended up merged in one directory. With an exact-length comparison like the one here, each directory keeps its own files. In a small test that creates Game 2/video/a.txt and then Game/video/b.txt, on master both Game 2/video and Game/video list only b.txt; with an exact-length comparison, Game 2/video has a.txt and Game/video has b.txt.

It'd be great to see this merged; it's an easy one to hit when copying a game collection.

[Copilot]

🟢 Approval recommended

No active review issues were identified for the change preventing intermediate path components from matching longer sibling names.

No open issues 🎉

What changed in this PR

Fixes intermediate directory matching in fatx_open_dir() so a path component does not resolve to a longer sibling with the same prefix.

Changes:

• Preserves the original component length while scanning sibling entries.
• Requires an exact filename length for slash-terminated components before comparing their contents.

File	Description
libfatx/​fatx_dir.c	Uses a separate comparison length and checks intermediate component lengths to prevent prefix matches.

_{Review effort: Balanced}

[mborgerson]

Could you please add some test cases for this while we are at it?

[xmutantson]

@Akashijk here are two test cases for this, written for pyfatx/tests/test.py, in case they're useful. On current master (now with #106 merged) both fail; with this PR applied on top they pass, along with the rest of the suite (21/21). This PR also still applies cleanly on top of current master.

	@with_formatted_disk
	def test_path_component_does_not_match_longer_sibling(self, path):
		fs = Fatx(path)

		# '/Game 2' is created first, so a lookup of the intermediate
		# component 'Game/' that matched by prefix would resolve to it.
		fs.mkdir('/Game 2')
		fs.mkdir('/Game 2/video')
		fs.write('/Game 2/video/a.txt', b'two')
		fs.mkdir('/Game')
		fs.mkdir('/Game/video')
		fs.write('/Game/video/b.txt', b'one')

		assert [a.filename for a in fs.listdir('/Game 2/video')] == ['a.txt']
		assert [a.filename for a in fs.listdir('/Game/video')] == ['b.txt']
		assert fs.read('/Game 2/video/a.txt') == b'two'
		assert fs.read('/Game/video/b.txt') == b'one'


	@with_formatted_disk
	def test_path_component_prefix_is_not_found(self, path):
		fs = Fatx(path)
		fs.mkdir('/top')
		fs.mkdir('/top/wolf')
		fs.mkdir('/top/wolf/inner')
		fs.write('/top/wolf/inner/x', b'x')

		# Prefixes of an existing directory name, used as intermediate
		# components, are not that directory
		for missing in ('/top/w/inner/x', '/top/wol/inner/x'):
			found = False
			try:
				fs.get_attr(missing)
				found = True
			except AssertionError:
				pass
			assert not found, missing

		assert fs.read('/top/wolf/inner/x') == b'x'

• The first is the case I hit on a real drive: Game 2/ is created before Game/, so on master Game/video resolves to Game 2/video and both files end up in one directory.
• The second checks that a prefix of a directory name used as an intermediate component (/top/w/inner/x) doesn't resolve into /top/wolf/. The last component of the parent path is already compared exactly, so the prefix has to sit in the middle of the path to hit this.

[mborgerson]

@xmutantson Thanks for the tests. I'll merge this PR now, would you then please submit a PR with the tests you created in a follow up?