I wonder if it wouldn't be better/(easier on high level libraries that will use it) to assign to 'self.assocs['filename']' as opposed to 'self.assocs['filename*']'?
shouldn't the file name extracted from 'filename*' be 1) urldecoded 2) character decoded from defined encoding? (https://tools.ietf.org/html/rfc5987)
I have a question or perhaps a suggestion to https://github.com/g2p/rfc6266/blob/cad58963ed13f5e1068fcc9e4326123b6b2bdcf8/rfc6266.py#L107