Separate verification into its own function

[quantumsheep]

Hello!

I have this use-case where I want to parse the JWT, fetch the secret elsewhere and then verify the JWT.

The current available functions forces me to parse the JWT another time. I want to validate the token, not parse it again.

Thanks,

[oxisto]

Maybe a trivial question, but can't you do this logic inside the keyfunc? You have access to the *Token which has been parsed (but not validated) up to this point. You can then fetch the secret and then return it. This is basically what libraries such as https://github.com/MicahParks/keyfunc do.

[quantumsheep]

I sometimes only parse the JWT to speedup the process when it comes from a fully trusted source (from internal code). I could duplicate some code to make it work but separating the functions costs nothing and fits my use-case.

[oxisto]

I sometimes only parse the JWT to speedup the process when it comes from a fully trusted source (from internal code). I could duplicate some code to make it work but separating the functions costs nothing and fits my use-case.

We are extremely careful about introducing new public functions because we need to maintain them in a way that we cannot break their function signature for quite a long time (since we tend to stick with major versions for quite a while). So yes, separating these functions actually does costs something: the time of a maintainer ;)

We intentionally did not expose any of these functions to not confuse people who might not be as experienced as you and might be confused, whether a simple Parse is enough or if VerifyToken is also needed; probably further complicated through the fact that we also have now a "validator".

As a bare minimum this function needs a godoc string and we probably would need to have an additional though about its function signature, because as I said before we need to stick with it for quite a while.

[oxisto]

any though on this @mfridman ?

[mfridman]

Yeah I'm okay with this. Needs a godoc comment though.

[oxisto]

@quantumsheep Can you supply a good for this function? Then we can accept the merge.