Add HMAC digest to libpurple

[QuLogic]

So khc wanted to take a look at the MSNP15 stuff, so I'm posting these tickets.

This patch adds the HMAC digest to libpurple. A few points:

• The underlying hash function can be set via purple_cipher_context_set_option(cipher, "hash", hash). The blocksize isn't really available, so I just assumed a size of 64 bytes. Fortunately, the supplied hashes all have a blocksize of 64 bytes, but this will need to be fixed if a different hash is added.
• The key can be set with purple_cipher_context_set_key. Because of the way it's implemented, this acts almost like a reset (but without changing the hash).
• Since the key can be of any length, perhaps with NULLs, and _set_key doesn't accept a length, I've had to add purple_cipher_context_set_option(cipher, "keyLen", key_len) so that things would work properly.
• Finally, I've added some tests with both MD5 and SHA1 as underlying hashes from RFC2202.

[grim]

We have 4 padded memebers in the PurpleCipherOps? structure, and I see no reason why we shouldn't use them for the majority of this.

Adding a _get_block_size op seems totally reasonable, we may also want to add a _set_block_size, but I can't come up with a case were it would be useful. Mind you my crypto in general isn't that great :).

We could add a purple_cipher_context_set_key_with_len function, and use another op.

I'd like to see these changes use the padding rather than the option system, since that seems like it'd be a lot harder to debug and what not.

[QuLogic]

OK, updated this patch as well. It adds purple_cipher_context_get_block_size and purple_cipher_context_set_key_with_len.

Since I've added to the PurpleCipherOps structure in both #4380 and here, this one is probably dependant on you applying the previous patch first. I've noticed some fuzz when applying it, but it still seems to work alright.

[QuLogic]

Adds HMAC digest to libpurple. Also adds _get_block_size and _set_key_with_len to PurpleCipherOps? structure.

[grim]

This has my blessing, good work!

What should I use for the author on the commit?

[QuLogic]

Cool. You can use Elliott Sales de Andrade, it should be in the Authors list already.

[grim]

Commited as c00604416a0d01f2fbf75b5cc6b6d4f9371e7a67