ZipReader ignored DirectoryEntry completely

[zhuxb711]

ZipReader -> ignored DirectoryEntry completely | Could not extract the zip using ZipReader

sharpcompress/src/SharpCompress/Readers/Zip/ZipReader.cs

Lines 63 to 85 in 095c871

	protected override IEnumerable<ZipEntry> GetEntries(Stream stream)
	{
	foreach (var h in _headerFactory.ReadStreamHeader(stream))
	{
	if (h != null)
	{
	switch (h.ZipHeaderType)
	{
	case ZipHeaderType.LocalEntry:
	{
	yield return new ZipEntry(
	new StreamingZipFilePart((LocalEntryHeader)h, stream)
	);
	}
	break;
	case ZipHeaderType.DirectoryEnd:
	{
	yield break;
	}
	}
	}
	}
	}

ZipArchive -> processed DirectoryEntry | Could extract the zip using ZipArchive

sharpcompress/src/SharpCompress/Archives/Zip/ZipArchive.cs

Lines 261 to 296 in 095c871

	switch (h.ZipHeaderType)
	{
	case ZipHeaderType.DirectoryEntry:
	{
	var deh = (DirectoryEntryHeader)h;
	Stream s;
	if (
	deh.RelativeOffsetOfEntryHeader + deh.CompressedSize
	> vols[deh.DiskNumberStart].Stream.Length
	)
	{
	var v = vols.Skip(deh.DiskNumberStart).ToArray();
	s = new SourceStream(
	v[0].Stream,
	i => i < v.Length ? v[i].Stream : null,
	new ReaderOptions() { LeaveStreamOpen = true }
	);
	}
	else
	{
	s = vols[deh.DiskNumberStart].Stream;
	}
	yield return new ZipArchiveEntry(
	this,
	new SeekableZipFilePart(headerFactory.NotNull(), deh, s)
	);
	}
	break;
	case ZipHeaderType.DirectoryEnd:
	{
	var bytes = ((DirectoryEndHeader)h).Comment ?? Array.Empty<byte>();
	vols.Last().Comment = ReaderOptions.ArchiveEncoding.Decode(bytes);
	yield break;
	}
	}

[zhuxb711]

@adamhathcock Would you please take a look about this issue？

[adamhathcock]

I think I'm intentionally skipping them or used to. Will get copilot to check

[zhuxb711]

I think I'm intentionally skipping them or used to. Will get copilot to check

Why will ignore the directory entry intentionally? It doesn't make sense. This behavior makes it hard to extract a zip with nested directory inside. All I got is an empty output~ Sad...

[adamhathcock]

I think I'm intentionally skipping them or used to. Will get copilot to check

Why will ignore the directory entry intentionally? It doesn't make sense. This behavior makes it hard to extract a zip with nested directory inside. All I got is an empty output~ Sad...

I first wrote this for myself and skipped directory entries as I thought they didn't make sense with paths. I've since walked back on that. Formats are inconsistent though

[adamhathcock]

Ah, right. Here's the real reason as found by AI:
// DirectoryEntry headers in the central directory are intentionally skipped.
// In streaming mode, we can only read forward, and DirectoryEntry headers
// reference LocalEntry headers that have already been processed. The file
// data comes from LocalEntry headers, not DirectoryEntry headers.
// For multi-volume ZIPs where file data spans multiple files, use ZipArchive
// instead, which requires seekable streams.