https://github.com/twitter/finatra/blob/develop/http/src/test/scala/com/twitter/finatra/http/tests/integration/doeverything/test/DoEverythingServerFeatureTest.scala#L243
If you just want to test json mapping, then yeah you can follow what happens in the Jackson tests:
https://github.com/twitter/finatra/blob/develop/jackson/src/test/scala/com/twitter/finatra/jackson/ScalaObjectMapperTest.scala
Or if you want to test how object mapping interacts with HTTP requests, then you can build a minimal injector like in the MessageBodyManagerTest (https://github.com/twitter/finatra/blob/develop/http/src/test/scala/com/twitter/finatra/http/tests/marshalling/MessageBodyManagerTest.scala#L72) and get references to the DefaultMessageBodyReader and DefaultMessageBodyWriter (which are supersets of logic and end up deferring to the ScalaObjectMapper for JSON serialization): https://github.com/twitter/finatra/blob/develop/http/src/main/scala/com/twitter/finatra/http/internal/marshalling/DefaultMessageBodyWriterImpl.scala#L68
But I’m not sure that tells you anything about your code built on top of this.
Oy Folks,
Happy user of Finatra here for some time now. Just asking for an advice. We use finatra for some REST services let's say /api/A and /api/B with some custom request.
And that works great, but all theses services has some common parameters let's say userId so it looks like that :
trait UserTrait{ @QueryParam("user.id") val userId:String}
case class GetAReQuest(@QueryParam a:String, @QueryParam("user.id") userId:String) extends UserTrait
get("/api/A") { aRequest: GetARequest => /* Magic here*/ }
case class GetBReQuest(@QueryParam b:Int, @QueryParam("user.id") userId:String) extends UserTrait
get("/api/B") { bRequest: GetBRequest => /* Magic here*/ }It works but I'm not fan of doing copy paste of the trait all the time (and I don't know how to make the @QueryParam used automatically if I extends the case class ) and would like to have a better mechanism to "compose" the requests object.
TL;DR I tried Request Context but at the end if I understand I need to "parse manually" the requests to my object. But I'm wondering if it would be possible to useobjectMapper.parseMessageBody[UserCaseClass](request)
without the exception
Injecting request attributes (e.g. QueryParam, Header, etc) not supported when explicitly calling ScalaObjectMapper.parse. Instead use a 'case class' input parameter on a Controller callback (e.g. get('/') { r: ClassWithRequestAttributes => ... } ).
so I just declare the case class and parsing is done automatically ?
Thanks by advance !
@totetmatt defining the @QueryParam should only be necessary on the field in the trait. You should not need to redefine the field nor the annotation in the case class implementations like in your example. E.g. your classes should just be defined like so:
trait UserTrait{ @QueryParam("user.id") val userId:String}
case class GetAReQuest(@QueryParam a:String) extends UserTrait
case class GetBReQuest(@QueryParam b:Int) extends UserTraitHowever, if you wanted to do this manually, you would use the DefaultMessageBodyReader#parse method here (not the ScalaObjectMapper#parseMessageBody): https://github.com/twitter/finatra/blob/develop/http/src/main/scala/com/twitter/finatra/http/marshalling/DefaultMessageBodyReader.scala#L59
parseMessageBody function — that doesn’t know how to interpet the annotations. Only the MessageBodyReader in the HTTP package has that ability.
Hey @cacoco thanks for the insights !
I actually tried what you explain before but if I do so, scala complain as
class GetAReQuest needs to be abstract, since value userId in trait Usertrait of type String is not definedAnd then, even if I add userId:String to the case class, the @QueryParam won't be propagated : "user_id: field is required"
But, if on the trait I declare the attributes as def then it works, Needs to repeat the field, but then @QueryParam looks "propagated".
trait UserTrait { @QueryParam("user.id") def userId:String}
// ^---- use of def instead of val
case class GetAReQuest(@QueryParam a:String, userId:String) extends UserTrait
case class GetBReQuest(@QueryParam b:Int, userId:String) extends UserTrait
// ^----- Repeat the trait attributes but no need to repeat @QueryParam or any other @AnnotationI don't really know what are the implication or even if it's correct to do this way, but so far it's acceptable for my use case.
But then if I understand there is no real way to parse a request "ad-hoc" like :
case class UserData([...])
case class QueryData([...])
case class XYZData([...])
get("/api/") { req:Requests =>
val userData:UserData = req.asInstanceOf[UserData]
val queryData:QueryData = req.asInstanceOf[QueryData]
val xyzData:XYZData = req.asInstanceOf[XYZData]
^----- Or any other magical method ?
// Yeah I know feels weird but just want to know for knowledge
}def in the trait and in the case class repeat it as a field (but you don’t need to redefine the annotation). This just Scala. For your second question, I’m not sure I understand what you mean by “ad-hoc” here. You have a type so it’s not really ad-hoc? If you manually want to parse the JSON body of a request, then you would just use an http Request as the callback type and then parse the content string into some object like here.
inject-mdc and MDCInitializer from finatra in a finagle(finch) service, but for some reason the mdc keys aren't showing up in logs. anyone have an idea? the filter is pretty much the exact same as the LoggingMDCFilter and it's the very first filter.
4 replies
We want to introduce
finatra in the-benchmarker/web-frameworks#3835@cacoco do you (or any core team member) consent (or is no opposed) on such a thing ?
I noticed a little typo
https://github.com/the-benchmarker/web-frameworks/pull/4146#issuecomment-814612219
9 replies
