SharpZipLib make tar.gz with directory name prefixed to filenames

Question

I'm using SharpZipLib to make a tar.gz archive, but I need to prefix a directory name to the filenames. That is, the files going into the archives look like foo.exe, but the archive needs to know them as foo-1.0/foo.exe, so the correct directory structure will be created on unpacking.

The input files are not already in a directory with the correct name, so the archiving process, needs to set the names.

This is what I have so far:

    var archiveName = $"bin/{projectVersion}.tar.gz";
    using var archiveStream = File.Create(archiveName);
    using var gzipStream = new GZipOutputStream(archiveStream);
    using var archive = TarArchive.CreateOutputTarArchive(gzipStream);
    archive.RootPath = projectVersion;
    foreach (var path in Directory.GetFileSystemEntries(publishPath)) {
        var entry = TarEntry.CreateEntryFromFile(path);
        entry.Name = $"{projectVersion}/{Path.GetFileName(path)}";
        archive.WriteEntry(entry, true);
    }

where projectVersion = e.g. "foo-1.0".

I was hoping either setting archive.RootPath, or setting each entry.Name to the appropriately prefixed string, would do the job, but even though I am doing both, when I unpack the resulting archive, the files just get dumped into the current directory.

What am I missing?

Answer 1

You might need to modify the way you set the entry.Name, i.e. setting the entry.Name with the prefixed directory.
The key would be to make sure the full path, including the directory name, is correctly assigned to each entry in the archive.

var archiveName = $"bin/{projectVersion}.tar.gz";
using var archiveStream = File.Create(archiveName);
using var gzipStream = new GZipOutputStream(archiveStream);
using var archive = TarArchive.CreateOutputTarArchive(gzipStream);

foreach (var filePath in Directory.GetFiles(publishPath)) {
    var entry = TarEntry.CreateEntryFromFile(filePath);
    // Make sure the directory structure is preserved in the entry's name
    entry.Name = $"{projectVersion}/{Path.GetFileName(filePath)}";
    archive.WriteEntry(entry, true);
}

That now takes each file in publishPath, prefixes it with the projectVersion directory, and sets the entry.Name.
That makes sure upon extraction, the directory structure foo-1.0/ will be created, and files will be placed inside it.

Plus, Directory.GetFileSystemEntries(publishPath) is changed to Directory.GetFiles(publishPath), to make sure only files are processed, not directories. That should avoid potential issues with directory entries.