Only attempt extraction if a known payload compressor is used.

[jordansissel]

This change is to improve safety primarily to reject malicious rpms with a custom PAYLOADCOMPRESSOR tag causing arr-pm to execute arbitrary code.

The security issue was reported by Joern Schneeweisz.

An alternate implementation could be to do what rpm does in its rpmio library which is to use the C implementations of various compressor libraries. However, doing this in Ruby can cause problems because it's not always possible to compile C Ruby extensions or even install Ruby's FFI library.

Test coverage checks for accepting known payload compressors and also rejecting unknown compressor values.

Context: https://gitlab.com/gitlab-org/gitlab/-/merge_requests/97602

[jordansissel]

% bundle exec rspec spec/rpm/file_spec.rb -e compressor -f d
Run options: include {:full_description=>/compressor/}

RPM::File
  #extract
    with an invalid payloadcompressor
      should raise an error
    with a 'gzip' payloadcompressor
      should succeed
    with a 'bzip2' payloadcompressor
bzip2: (stdin) is not a bzip2 file.
cpio: premature end of archive
      should succeed
    with a 'xz' payloadcompressor
xz: (stdin): File format not recognized
cpio: premature end of archive
      should succeed
    with a 'lzma' payloadcompressor
lzma: (stdin): File format not recognized
cpio: premature end of archive
      should succeed
    with a 'zstd' payloadcompressor
      should succeed

Finished in 0.01085 seconds (files took 0.06391 seconds to load)
6 examples, 0 failures

[alperkaya0]

Using $(command..) in place of PAYLOADCOMPRESSOR also works.