Source:のAnalyzerをScalaで実装してみる

SimpleAnalyzer.scala

import scala.util.parsing.combinator.RegexParsers
import scala.util.parsing.input.Positional

object SimpleAnalyzer {
  trait SearchKeyword extends Positional
  case class KeywordIgnore(content:String) extends SearchKeyword
  case class KeywordDefPackage(packageName:List[String]) extends SearchKeyword
  case class KeywordDefClass(className:List[String]) extends SearchKeyword
  case class KeywordRefSymbol(packageName:List[String]) extends SearchKeyword
  case class KeywordComment(content:String) extends SearchKeyword

  class JavaKeywordParsers extends RegexParsers {
    def compilationUnit : Parser[List[SearchKeyword]] = rep(packageDeclaration | importDeclaration | comment)

    def packageDeclaration : Parser[SearchKeyword] = "package" ~> positioned(packageName) <~ ";"
    def packageName : Parser[SearchKeyword] =
      (rep1(identifier <~ ".") ~ identifier) ^^ { p => KeywordDefPackage(p._1 :+ p._2) } |
        identifier ^^ { p => KeywordDefPackage(List(p)) }

    def importDeclaration : Parser[SearchKeyword] = "import" ~> positioned(importClassName) <~ ";"
    def importClassName : Parser[SearchKeyword] =
      (rep1(identifier <~ ".") ~ identifier) ^^ { c => KeywordRefSymbol(c._1 :+ c._2)} |
        (rep1(identifier <~ ".") <~ "*") ^^ KeywordRefSymbol |
        identifier ^^ { c => KeywordRefSymbol(List(c))}

    def comment : Parser[SearchKeyword] = positioned(blockComment)

    // regex: http://stackoverflow.com/questions/462843/improving-fixing-a-regex-for-c-style-block-comments
    def blockComment : Parser[SearchKeyword] = "/*" ~> """(?>(?:(?>[^*]+)|\*(?!/))*)""".r <~ "*/" ^^ KeywordComment

    def identifier = """[_a-zA-Z$][_a-zA-Z$]*""".r

    def parse(input : String) = parseAll(compilationUnit, input)
  }


  def main(args: Array[String]) : Unit = {
    val source =
      """
        |/*
        | * test
        | */
        |
        |package org.watermint.sourcecolon.indexer;
        |import org.watermint.sourcecolon.analyzer.*;
        |import org.watermint.sourcecolon.Main;
        |
      """.stripMargin

    val parser = new JavaKeywordParsers
    val keywords = parser.parse(source)

    keywords.get.foreach((k : SearchKeyword) => println(k.pos.line, k))
  }
}