join(1) in scala

AA.txt

AA 100円
BB 300円
CC 200円
DD 400円
EE 500円

BB.txt

labelA AA Category1
labelB BB Category2
labelC CC Category3
labelD DD Category4

join.scala

import java.io.File
import scala.io.Source
import scala.sys.process._

def join(file1AndColumn: (File, Int),
         file2AndColumn: (File, Int),
         separator: String = "\t"): Iterator[List[String]] = {
  val (file1, col1) = file1AndColumn
  val (file2, col2) = file2AndColumn
  join(
    Source.fromFile(file1).getLines.map(_.split(separator).toList) -> col1,
    Source.fromFile(file2).getLines.map(_.split(separator).toList) -> col2
  )
}

def join(
  line1AndKeyColumn: (Iterator[List[String]], Int),
  line2AndKeyColumn: (Iterator[List[String]], Int)
): Iterator[List[String]] = new Iterator[List[String]] {

  var it1: Iterator[List[String]] = line1AndKeyColumn._1

  var it2: Iterator[List[String]] = line2AndKeyColumn._1

  var _nextLine1: List[String] = null

  val keyColumn1 = line1AndKeyColumn._2

  val keyColumn2 = line2AndKeyColumn._2

  def hasNext: Boolean = {
    (! it1.isEmpty) && {
      _nextLine1 = it1.next
      it2 = it2.dropWhile(x => x(keyColumn2 - 1) != _nextLine1(keyColumn1 - 1))
      it2.hasNext
    }
  }

  def next: List[String] = {
    val _line2Next = it2.next
    val result = if (_nextLine1 != null) {
      _nextLine1 ::: _line2Next.filterNot(_ == _line2Next(keyColumn2 - 1))
    } else {
      it1.next ::: _line2Next.filterNot(_ == _line2Next(keyColumn2 - 1))
    }
    _nextLine1 = null
    result
  }

}

for {
  line <- join(new File("AA.txt") -> 1, new File("BB.txt") -> 2, separator = " ")
} {
  println(line.mkString("\t"))
}

/*
 $ scala join.scala
AA      100円   labelA  Category1
BB      300円   labelB  Category2
CC      200円   labelC  Category3
DD      400円   labelD  Category4
*/