Spark Hive SQL实例

def initDimFrontCate(sqlContext: HiveContext): mutable.HashMap[String, String] =
  {
    var dimValues = new mutable.HashMap[String, String]
    val sql = s"""select front_cate_id, page_level_id
                          | from dw.dim_front_cate
                          | order by front_cate_id""".stripMargin

    val dimData = sqlContext.sql(sql).persist(StorageLevel.MEMORY_AND_DISK)

    dimData.map(line => {
      val front_cate_id = line.getAs[Int]("front_cate_id")
      val level_id = line.getAs[Int]("page_level_id")

      val key = front_cate_id.toString
      (key, level_id.toString)
    })
      .collect()
      .foreach( items => {
        val value = items._2
        val key = items._1
        dimValues += (key -> value)
      })

    dimData.unpersist(true)
    dimValues
  }