Windows path handling (tar.gz, probably others, too)

[seizmo]

What version of the package or command are you using?

v4 lastest (d777df9)

What are you trying to do?

Create a tar.gz file under windows.

What steps did you take?

       files, err := archiver.FilesFromDisk(nil, map[string]string{
		".\res\test\tenant_files\tenant_1": "", // contents added recursively
	})
	if err != nil {
		return err
	}

	archiveDir, err := job.ArchiveDirectory(true)

	// create the output file we'll write to
	out, err := os.Create(filepath.Join(archiveDir, "files.tar.gz"))
	if err != nil {
		return err
	}
	defer out.Close()

	// we can use the CompressedArchive type to gzip a tarball
	// (compression is not required; you could use Tar directly)
	format := archiver.CompressedArchive{
		Compression: archiver.Gz{},
		Archival:    archiver.Tar{},
	}

	// create the archive
	err = format.Archive(context.Background(), out, files)
	if err != nil {
		return err
	}

What did you expect to happen, and what actually happened instead?

Opening the file with 7-zip should extract to the same directory structure of the original files

How do you think this should be fixed?

https://github.com/mholt/archiver/blob/master/archiver.go should probably use filepath instead of path since path only supports slashes and no backslashes

Bonus: What do you use archiver for, and do you find it useful?

Just got started but really like the API (and the author, thanks for Caddy!!!) :)

[mholt]

Thanks; I don't have Windows to test this. Can you give more information? What exactly is broken? What is in the created archive?

archiver.go should probably use filepath instead of path since path only supports slashes and no backslashes

We do use filepath. The only time we use path is to join the components of the path in the archive, where it is required to use slash separator.

[seizmo]

The problem seems to be that the Windows backlash is not removed/replaced with slashes but included in the archive entries in escaped form:

tenant_1
tenant_1/\\1
tenant_1/\\1\\file-example_PDF_1MB.pdf
tenant_1/\\2
tenant_1/\\2\\file-example_PDF_500_kB.pdf
tenant_1/\\3
tenant_1/\\3\\file-sample_150kB.pdf

This is what I get if I unpack that file on Linux (using tar xzvf)

[mholt]

Ah, indeed, path.Join() calls path.Clean() which does NOT call ToSlash(). Only filepath.Join() calls filepath.Clean() which DOES call FromSlash(). 🙄 I get why, but that was a silly oversight on my part.

Can you test what I push to a branch? (Until I have a chance to write actual automated tests, heh. Just been busy lately.)

[seizmo]

I changed the two lines and it worked for me on Windows with Go 1.18. 7zip on windows now also opens and extracts the archive to a Windows folder with correctly named files.

Thanks a lot!

[mholt]

Great, thanks for confirming!