feat: HMAC support in Crypto APIs

[jonbonazza]

fixes: godotengine/godot-proposals#1098

Couple different ways to use this. The more low level approach (especially useful for HMACing streamed content:

extends Node

var hmac = HMACContext.new()

func _ready():
	var key = "supersecretkey" # Key can be of any size greater than 0.
	var data = "Return of the MAC!" # Data size can be any size greater than 0 as well.
	hmac.start(HashingContext.HASH_SHA256, key.to_utf8())
	hmac.update(data.to_utf8())
	var digest = hmac.finish()
        print_hex(digest)

Or the more high level approach that is going to be most useful 99% of the time:

extends Node

var crypto = Crypto.new()

func _ready():
    var key = "supersecretkey" # Key can be of any size greater than 0.
    var data = "Return of the MAC!" # Data size can be any size greater than 0 as well.
    var digest = crypto.hmac_digest(HashingContext.HASH_SHA256, key.to_utf8(), data.to_utf8())
    print_hex(digest)

I've also provided a convenience function for comparing two HMAC digests in a way that is impervious to timing attacks:

extends Node

var crypto = Crypto.new()

func _ready():
    var hm1 # PackedByteArray containing an HMACdigest
    var hm2 # PackedByteArray containing a second HMAC digest
    var equal = crypto.hmac_equals(hm1, hm2)
    print(equal)

[jonbonazza]

Needs some testing still before merge but this is the general approach I am going for @Faless

[Calinou]

Some stylistic changes for documentation.

[jonbonazza]

@Calinou thanks. I'm going through and trying to address all of the static check issues from CI. Once that's done, I'll accept your suggestions.

[fire]

Can you provide a test project? It'll make testing faster.

[jonbonazza]

@fire yep i def plan to. Still need to do some testing myself though as mentioned above. Pulled an all nighter to implement this though, so ill do so after i rest a bit more haha

[jonbonazza]

@fire I opted instead to create unit tests for everything. If you feel an example project is still useful, I can whip one together though. let me know.

[jonbonazza]

@akien-mga @Faless oh cool. Ill work on adding thise soon then

[jonbonazza]

@Calinou @akien-mga @Faless this is ready again for y'all.

[jonbonazza]

@Calinou @akien-mga I'm not super familiar with Godot's C# jazz. Mind checking my work here?

[jonbonazza]

@Faless I've addressed all of your feedback. This is ready for you again. :)

[Faless]

See my comments

[Faless]

I was wondering if we should rename the parameters to something more sensible. Would byte_array/other_byte_array make sense?
There is also the case where the 2 sizes do not match, so a timing attack could still successfully discover the size of the compared value.
Would it make sense to call the first parameter trusted_bytes and the second one received_bytes, so we can, in the future maybe, improve it knowing which one the user consider trusted (which I think is how safe_memcmps work, but I'm no expert, so the function is fine for now).

[jonbonazza]

I renamed the parameters for the function. I didn't do anything about the size comparison shortcut though, as we discussed on IRC.

[Faless]

It seems you changed the name of the function parameters, not the one in the method_bind (which are the names used by the doc generation tool).
rename hm1 and hm2 in bind_method to: trusted, received.
Possibly rename trusted, received in the method definition to p_trusted, p_received.

[Faless]

And remember to re-run the doctool to update the XML

[jonbonazza]

@Faless done. Let me know if you see anything else. Thanks!

[jonbonazza]

Ready again for you @Faless

[Faless]

Almost ready. See my comment about the parameter renames.
Looks good otherwise.
Not sure if the @godotengine/documentation team has some other remarks.
Please remember to squash your commits afterwards: https://docs.godotengine.org/en/stable/community/contributing/pr_workflow.html#the-interactive-rebase

[Faless]

Great work! 🥇

resolved

[Faless]

Thanks!