Fix for ogg trailing null bytes

[kism]

Fixes #591

So I have encountered some ogg files from some encoder that seems to append one or two null bytes to the file.

All existing tests pass.

[phw]

Seems useful. Is there some example file to test this? Would also be good to have tests covering this case.

[kism]

Found an encoding of a CC-BY-SA song that has the issue, Bad Sign by Brad Sucks
https://autist.network/adhoc/test.ogg
md5: 8e847801d93ecb8643f50a52e0e99f86

import mutagen
audio = mutagen.File("test.ogg")
audio['artist'] = "Brad Sucks"
audio.save()

Will cause

mutagen.oggvorbis.OggVorbisHeaderError: unable to read full header; got b'\x00'

In this case, the file has one null byte at the end, if I remove it with a hex editor and run that script, its happy.

If I just append null bytes to a good ogg file, then test with that script, it doesn't cause the error. So maybe it's todo with the length of the file per the ogg page headers.

[phw]

Thanks for the example file. The fix mostly seems fine, with some minor comments.

But we really should have a test for this. Have a look at test_ogg.py, specifically the test_renumber* tests there. There is already a test_renumber_extradata, which tests failing for some garbage data at the end.

There could be a similar test_renumber_trailing_null_bytes that adds some null bytes (26, to test for the edge case) and checks that the call to renumber succeeded, but the trailing bytes got preserved.

[kism]

Yep, good suggestions.

I'll work on a test when I have some time.

[kism]

Added test test_renumber_reread_trailing_bytes

It fails without my change, passes with my change.

Narrowing down what the issue is, it happens specifically when cls.renumber(fileobj, serial, sequence) is run, and the file has trailing bytes.

I do not have enough knowledge to know what that actually means, my test file needed to be renubered, and the test files I tried to generate to trigger the issue didn't.

[phw]

Thanks again