Mapping values to case class in json4s - scala

Question

Say I have the following:
[ {
"job_id": "1",
"status": "running"
},
{
"job_id": "0",
"status": "finished"
}]
Could I somehow do the following with json4s:
case class Job(job_id: Int, status: JobStatus)
abstract class JobStatus
case class JobFinished extends JobStatus
case class JobRunning extends JobStatus
... some magic is probably needed here
Such that extracting the first snippet would result in:
[ Job(1, JobRunning()), Job(0, JobFinished())]

Answer 1

I think that the best way of creating a scala case class based on the JSON, is using this site, this add the magic, I normally use this site, you can even change the name of the clases so in your case, you can use the site and then manage the relations within classes:
JSON to Scala

Answer 2

This is possible, but needs some coding.
I will try to break it down in some smaller steps.
Defintion of types / model
// Types
case class Job(jobId: Int, status: JobStatus)
// Sealed trait to make match exhaustive in helper functions
sealed trait JobStatus
// use case object to not create uneeded instances, also case class without () no longer allowed
case object JobFinished extends JobStatus
case object JobRunning extends JobStatus
The "magic"
Needed imports
import org.json4s._
import org.json4s.native.Serialization
import org.json4s.native.Serialization.{read, write}
Helper Functions
// helper functions, could be improved by having a mapping
implicit def stringToJobStatus(in: String) : JobStatus = in match {
case "running" => JobRunning
case "finished" => JobFinished
}
implicit def jobStatusToString(jobStatus: JobStatus) : String = jobStatus match {
case JobRunning => "running"
case JobFinished => "finished"
}
Custom Serializer
// here is the "magic" a custom serializer
class JobSerializer extends CustomSerializer[Job](format => (
// unmarshal Function
{
case JObject( JField("job_id", JString(jobId)) :: JField("status", JString(status)) :: Nil ) => {
new Job(jobId.toInt, status)
}
},
// masrshal Function
{
case Job(jobId, status) => JObject(
JField("job_id", JString(jobId.toString)) ::
JField("status", JString(status)) :: Nil)
}
))
make Serializer acessible
// Implicit formats for serialization and deserialization
implicit val formats = Serialization.formats(NoTypeHints) + new JobSerializer
Example in REPL
val data = """
[
{
"job_id": "1",
"status": "running"
},
{
"job_id": "0",
"status": "finished"
}
]
"""
read[List[Job]](data)
res3: List[Job] = List(Job(1,JobRunning), Job(0,JobFinished))

Answer 3

Although #Andres Neumann's answer is pretty good it does require reimplementing the serialization of the whole Job class (which might be a lot larger than the dumbed-down Job class in the example), whereas the only serialization which is actually needed is for the status. Based on #Andreas' answer the actual code needed is somewhat shorter and does not require every field in the Job to be serialized manually.
// here is the "magic" a custom serializer
class JobStatusSerializer extends CustomSerializer[JobStatus](format => (
// unmarshal Function
{
case JString(status) => {
// helper functions, could be improved by having a mapping
def stringToJobStatus(in: String): JobStatus = in match {
case "running" => JobRunning
case "finished" => JobFinished
}
stringToJobStatus(status)
}
},
// marshal Function
{
case status: JobStatus => {
def jobStatusToString(jobStatus: JobStatus): String = jobStatus match {
case JobRunning => "running"
case JobFinished => "finished"
}
JString(jobStatusToString(status))
}
}
)
)

Answer 4

You could use an enum and add EnumSerializer to your formats through json4s-ext.
However, your enum would be serialized as an int (0 or 1 in your case).
Adding to my own answer, you can also use EnumNameSerializer which would serialize to the enum values you specified (like "running" or "finished").