Fix: Fixed an issue where archives with reserved item names wouldn't extract correctly

[Lamparter]

Resolved / Related Issues

• Closes Bug: Smart extracting a .tar file with "." as the only directory in root is not properly handled #18035

Steps used to test these changes

1. Opened Files
2. Create an archive in a directory using tar --exclude=archive.tar -cf archive.tar . in Powershell
3. Do the 'Extract (smart)' action on the newly created tar archive
4. See that the folder '.' in the archive gets renamed to the archive name 'archive'

[Copilot]

Pull request overview

This PR fixes a bug where "." and ".." directory entries in archives were being incorrectly treated as meaningful top-level items during smart extraction. The fix ensures that when using "Extract (smart)" on archives containing "." (current directory) or ".." (parent directory) entries, these special directory references are properly ignored when determining whether to create a subfolder.

Changes:

• Refactored the logic for detecting multiple top-level items in archives to filter out "." and ".." entries
• Introduced a new helper function GetFirstMeaningfulSegment that skips special directory references when parsing archive entry paths

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Lamparter]

@yaira2 requesting your review

[yair100]

Noted. I'd like to complete your earlier PRs before reviewing this one.

[yair100]

I can confirm these changes are working as expected.

[0x5bfa]

Using LINQ isn't performant and allocates memory a lot.

Consider this.

static ReadOnlySpan<char> GetFirstMeaningfulSegment(ReadOnlySpan<char> path)
{
    while (!path.IsEmpty)
    {
        while (!path.IsEmpty && (path[0] == '/' || path[0] == '\\'))
            path = path[1..];

        if (path.IsEmpty) break;

        int sep = path.IndexOfAny('/', '\\');
        ReadOnlySpan<char> seg = sep < 0 ? path : path[..sep];

        path = sep < 0 ? ReadOnlySpan<char>.Empty : path[(sep + 1)..];

        if (seg.SequenceEqual("."u8) || seg.SequenceEqual(".."u8)) // Comparison of ASCII bytes
            continue;

        return seg;
    }

    return default;
}

[yair100]

@Lamparter are you still planning to work on this? It's most of the way there but needs the changes @0x5bfa requested.

[Lamparter]

I sent a comment on the discord server about an alternative solution but received no response, that's why this PR is stale
Either way, I will go with 0x5BFA's implementation for that method for now

[yair100]

Can you please post the alternative solution here?

[Lamparter]

A friend of mine suggested using ZLinq (which doesn't allocate memory at all) instead of Microsoft's System.Linq implementation (that allocates memory a lot)
https://github.com/Cysharp/ZLinq
It might also save performance elsewhere in the app if fully replaced

[yair100]

I would like to avoid third party dependencies at this point.

[0x5bfa]

LGTM, code wise.