spark-row-level-transformations.scala

// Row level transformations using map

val orders = sc.textFile("/public/retail_db/orders")
// 21,2013-07-25 00:00:00.0,11599,CLOSED -> 20130725 as Int
val str = orders.first
str.split(",")(1).substring(0, 10).replace("-", "").toInt

val orderDates = orders.map((str: String) => {
  str.split(",")(1).substring(0, 10).replace("-", "").toInt
})

val ordersPairedRDD = orders.map(order => {
  val o = order.split(",")
  (o(0).toInt, o(1).substring(0, 10).replace("-", "").toInt)
})

val orderItems = sc.textFile("/public/retail_db/order_items")
val orderItemsPairedRDD = orderItems.map(orderItem => {
  (orderItem.split(",")(1).toInt, orderItem)
})