new.scala

  import scala.collection.mutable.ArrayBuffer
  val name = """\[\d+,\d+,\d+\]""".r
  val name2 = """\[(\d+),(\d+),\d+]""".r
  def parseLine(line: String): ArrayBuffer[String] = {
    val jsonstr = line.split("\t")(1)
    val result = ArrayBuffer[String]()
    val m = name.findAllIn(jsonstr)
    val itemIdfArray = ArrayBuffer[Pair[Int, Int]]()
    m.foreach(a => { val name2(item, idfnum) = a; val p: Pair[Int, Int] = Pair(item.toInt, idfnum.toInt); itemIdfArray += p; })
    if (itemIdfArray.length > 1) {
      for (i <- 0 until itemIdfArray.length - 1) {
        for (j <- 1 until itemIdfArray.length) {
          val one = itemIdfArray(i)
          val two = itemIdfArray(j)
          if (one._2 < two._2) {
            result += one._1.toString + "-" + two._1.toString
          } else if (one._2 > two._2) {
            result += two._1.toString + "-" + one._1.toString
          }
        }
      }
    }
    result
  }
val file=sc.textFile("/home/mps/md5_labels.txt")
val comatrix=file.flatMap(parseLine).map(item=>(item,1)).reduceByKey((a,b)=>(a+b))
comatrix.saveAsTextFile("/home/mps/comatrix")