gate_twitter_test.scala

import gate._
import gate.creole._
import gate.util.persistence._
import gate.corpora._

import scala.collection.JavaConverters._
import scala.collection.mutable.HashSet

object TwitterPrototype extends App {

  Gate.init()

  val anniePlugin = new Plugin.Maven("uk.ac.gate.plugins", "annie", "8.6-SNAPSHOT")
  val annieFrenchPlugin = new Plugin.Maven("uk.ac.gate.plugins", "lang-french", "8.6-SNAPSHOT")
  val twitterJsonPlugin = new Plugin.Maven("uk.ac.gate.plugins", "format-twitter", "8.6-SNAPSHOT")
  val twitterPlugin = new Plugin.Maven("uk.ac.gate.plugins", "twitter", "8.6-SNAPSHOT")

  //Gate.getCreoleRegister.registerPlugin(anniePlugin)
  //Gate.getCreoleRegister.registerPlugin(annieFrenchPlugin)
  Gate.getCreoleRegister.registerPlugin(twitterJsonPlugin)
  Gate.getCreoleRegister.registerPlugin(twitterPlugin)

  val resourceDef = new ResourceReference(twitterPlugin,"resources/twitie-english-only.gapp").toURL
  val controller = PersistenceManager.loadObjectFromUrl(resourceDef).asInstanceOf[ConditionalSerialAnalyserController]

  //val annieController=  Factory.createResource("gate.creole.SerialAnalyserController", Factory.newFeatureMap(), Factory.newFeatureMap(), "ANNIE").asInstanceOf[SerialAnalyserController]


  def infoPRS (controller:CorpusController) ={
    val PRNames = controller.getPRs().asScala.map{
      pr => pr.getName()
    }

  println(PRNames.mkString(","))
  }

  infoPRS(controller)

  def processString(controller:CorpusController, text:String) = {

    val corpus = Factory.newCorpus("TwitIE Corpus")
    val doc = Factory.newDocument(text)

    corpus.add(doc)

    controller.setCorpus(corpus)
    controller.execute()
    processResults(doc,corpus)
    Factory.deleteResource(doc)
    Factory.deleteResource(corpus)
  }

  def processResults(doc:Document, corpus:Corpus): Unit = {
    val annotationTypesRequired= HashSet("Person","Location","Organization","UserID", "Emoticon","Hashtag").asJava

    val c = corpus.iterator().asScala

    val listOfAnnotationSet :Seq[AnnotationSet] = c.map{ c =>
      val doc = c.asInstanceOf[Document]
      doc.setPreserveOriginalContent(true)
      doc.getAnnotations().get(annotationTypesRequired)

    }.toSeq

    listOfAnnotationSet.map{ aset =>
      aset.asScala.map{ a =>
        println("Type = " + a.getType() + " = " + Utils.stringFor(doc,a))
      }
    }

  }

  processString(controller,"Hi @seb :) , i'm really happy to see you with @paul and @axt in #paris #eiffeltower ! ")

}

Nice, just a comment on lines 43-44 - don't use new DocumentImpl and setContent, instead use Factory.newDocument(text), and then make sure you pass both the document and the corpus to Factory.deleteResource before the method returns.

Thanks @ianroberts, code corrected :)