header->buf also needs to be freed

Add DKIM_UNKNOWN check for whitelines backlog as well

Fix calculating subdomain matching for i tag

Fix type-O

rewrite dkim_body_parse, so that we don't copy the line. This means that we don't limit the line length anymore, but might cost a little extra computing overhead. Let's see how bad this can get.

And of course I tested with the wrong file. v=-tags should also increment key, similar to the other tags.

Fix syntax validator in t=-tag (used wrong function) While here fix unfold some line wrappings no longer needed

For the key h tag we should skip over the alg, not the entire value when inside the loop. While here reorder some line folding since the mheader->ltok API name change.

Put v=-tag handling for key into switch. Shorter and better legibility

Always compile test failure issue 1000000

Add more error checking in key parsing

Shrink unknown key tag parsing, similar to what happened in unknown sig tag

Rework the entire mheader file - Rename mheader to ltok (lexical token) - Make all skip functions const - Add a few missing functions

Make sure we don't crash on invalid headers

Always move to dkim_message_verify from dkim_rr_resolve to avoid hangs

t tag must be smaller than x tag

Make sure that l-tag is smaller than body

Check that i-tag is (sub)domain of d-tag

return early if there are missing tags in DKIM-Signature

Don't forget to reset bctx in between calls

Fix some inverted logic in the key's s-tag

Fix pointer forward movement for simple case of fws range

Fix key h tag: - inverted logic for key-h-tag-value - restore temporary NUL-character

Initial implementation for filter-dkimverify. Probably still contains quite a few bugs and does't yet support ed25519.