The current tar code doesn't put the subject into a directory, guaranteeing that the resulting archive is a tar bomb.
The easiest way to do this might be to normalise the corpus being tarred such that each subject naturally gets directed to its own directory (but this would need us to make sure the normaliser can separate its mappings per subject).
The current tar code doesn't put the subject into a directory, guaranteeing that the resulting archive is a tar bomb.
The easiest way to do this might be to normalise the corpus being tarred such that each subject naturally gets directed to its own directory (but this would need us to make sure the normaliser can separate its mappings per subject).