I would like to generate an index of a tar file with the start and end offset of each file in the tarball, such that I can mmap or extract a single file later on. Is this possible with tar-stream?

The documentation of headers only mentions the size of each file, but I would also need the offset within the tar.

From hacking it looks like the global property extract._buffer.shifted contains what I need but this is mostly a guess. It would be nice if the header callback could include the offset property for each entry.