search

ory / hydra

The most scalable and customizable OpenID Certified™ OpenID Connect and OAuth Provider on the market. Become an OpenID Connect and OAuth2 Provider over night. Broad support for related RFCs. Written in Go, cloud native, headless, API-first. Available as a service on Ory Network and for self-hosters.
https://www.ory.sh/?utm_source=github&utm_medium=banner&utm_campaign=hydra
Apache License 2.0
15.64k stars 1.5k forks source link

fix: broken JSON round-tripping for custom claims #3819

Closed alnr closed 3 months ago

alnr commented 3 months ago

Adding custom claims with numerical types (think JavaScript Number) previously did not round-trip through Hydra correctly. For example, passing UNIX timestamps in custom claims would end up as floating points in exponential notation in the final token. That, in turn, confused or broke downstream consumers of the token, including Kratos.

Ref https://github.com/go-jose/go-jose/issues/144

aeneasr commented 3 months ago

There was this PR: https://github.com/ory/hydra/pull/3722

Is that the same issue, and if yes, #3722 seems to use the number decoding in more places.

aeneasr commented 3 months ago

Also, are there any implications for backwards compatibility with prior consent sessions?

alnr commented 3 months ago

Yes I did see https://github.com/ory/hydra/pull/3722. As the author discovered, using json.Number does not fix this issue. The underlying problem is described here: https://github.com/go-jose/go-jose/issues/144

I don't think this will cause any regressions. This change mostly or exclusively impacts how numbers are serialized into the final token, it doesn't really change any underlying data. While we're handling the custom claims in hydra, we serialize+deserialize them to/from JSON a couple of times. Each time, any "type" information about whether the number is a float or integer is lost since this distinction doesn't exist in JSON.