mymecab.scala

mymecab.scala

import java.text.Normalizer
import scala.collection.mutable.ArrayBuffer
import org.chasen.mecab.{MeCab, Tagger, Node}

/*
 scalac mymecab.scala
 scala MyMeCabTest

 ArrayBuffer(サイト, リンクフリー, リンクフリー)

 */

class MyMeCab {
  System.loadLibrary("MeCab")
  val tagger = new Tagger
  def parse(s: String) : ArrayBuffer[String] = {
    var termbuffer = new ArrayBuffer[String]
    var node:Node = tagger.parseToNode(normalize(s))
    while(node != null){
      val feature = node.getFeature()
      val ft = feature.split(",")
      if(ft(0) == "名詞"){
	if( ft(1) == "一般" || ft(1) == "固有名詞"){
	  termbuffer += node.getSurface()
	}
      }
      node = node.getNext()
    }
    termbuffer
  }
  def normalize(s:String):String = {
    // unicode正規化 NFKC
    var str = s.toLowerCase()
    str = str.replace("~", "-").replace("〜", "-")
    str = str.replace("−", "-").replace("‐", "-").replace("─", "-")
    str = Normalizer.normalize(s, Normalizer.Form.NFKC)
    str
  }
}

object MyMeCabTest{
  def main(args:Array[String]){
    val m = new MyMeCab
    println(m.parse("当サイトはリンクフリーですが、ﾘﾝｸﾌﾘーご一報いただけると幸いです。"))
  }
}