ParseJsonExampleJob.scala

import com.twitter.scalding._
import scala.util.parsing.json._

class ParseJsonJob(args: Args) extends Job(args) {
  TextLine(args("input"))
    .map(('line) -> ('parseStatus, 'uri)) {
      line: String => {
        JSON.parseFull(line) match {
          case Some(data: Map[String, String]) => ("success", data("uri"))
          case None => ("failed", "")
        }
      }
    }
    .filter('parseStatus) { status: String => status != "failed" }
    .project('uri)
    .write(Tsv(args("output")))
}