SparkSQL and CTE for increased readability

sparkdemo.scala

    val df = spark.read.text(inputFile)
    df.createOrReplaceTempView("data")

    val query =
      """
        | WITH loglevel AS (SELECT SPLIT(value, ' ')[0] AS level FROM data WHERE LENGTH(value) > 0),
        |      levelcount AS (SELECT level, COUNT(*) as count FROM loglevel GROUP BY level)
        | SELECT level, count FROM levelcount ORDER BY count DESC
      """.stripMargin

    val result = spark.sql(query)
    println(result.rdd.toDebugString)
    result.show()