TopNPricedProducts.scala

// Function to get top n priced products using Scala collections API

val products = sc.textFile("/public/retail_db/products")
val productsMap = products.
  filter(product => product.split(",")(4) != "").
  map(product => (product.split(",")(1).toInt, product))
val productsGroupByCategory = productsMap.groupByKey

def getTopNPricedProducts(productsIterable: Iterable[String], topN: Int): Iterable[String] = {
  val productPrices = productsIterable.map(p => p.split(",")(4).toFloat).toSet
  val topNPrices = productPrices.toList.sortBy(p => -p).take(topN)

  val productsSorted = productsIterable.toList.sortBy(product => -product.split(",")(4).toFloat)
  val minOfTopNPrices = topNPrices.min

  val topNPricedProducts = productsSorted.takeWhile(product => product.split(",")(4).toFloat >= minOfTopNPrices)

  topNPricedProducts
}

val productsIterable = productsGroupByCategory.first._2
getTopNPricedProducts(productsIterable, 3).foreach(println)