Question

在一次采访中，技术负责人说我的scala代码就像java代码，但是使用scala api，他希望我改进它。

我是一名3年的Java开发人员，我在Coursera上跟随MOOC开始进行scala编码。

有谁能告诉我这是什么问题，我该如何改进呢？

我得到了这份工作，因为我的Java认可，但这项工作是基于scala，编码风格是在试用期间要解决的一件事。

object Extraction {

// IntelliJ use .idea/modules as current working directory
  val FilePathPre = "../../src/main/resources/"
  val UserIdFile = "lookup_user.csv"
  val ProductIdFile = "lookup_product.csv"
  val RatingFile = "agg_ratings.csv"

  def readFile(file: String): Iterator[((String, String), String, String)] = {
val Splitter = ","
Source.fromInputStream(this.getClass.getResourceAsStream(file)).getLines()
      .map(_.split(Splitter))
      .filter(_.size >= 4) // in case line is not valid
      .map(x => ((x(0), x(1)), x(2), x(3))) // (userId, ItemId), rating, time
  }

  def filePrinter(fileName: String, lines: mutable.Map[String, Int]) = {
val file = new File(fileName)
val bw = new BufferedWriter(new FileWriter(file))
lines.toArray.sortWith((a, b) => a._2 < b._2)
     .map(x => x._1 + "," + x._2.toString + "\n")
     .foreach(bw.write)
bw.close()
  }

  def aggFilePrinter(fileName: String, lines: mutable.Map[(Int, Int), Float]) = {
val file = new File(fileName)
val bw = new BufferedWriter(new FileWriter(file))
lines.foreach(x => {
     val line = x._1._1.toString + "," + x._1._2.toString + "," + (math.round(x._2 * 100.0) / 100.0).toFloat + "\n"
     bw.write(line)
   })
bw.close()
  }

  /**
* * une pénalité multiplicative de 0.95 est appliquée au rating
*   pour chaque jour d'écart avec le timestamp maximal de input.csv
*
* @param nowTime maximal timestamp at input.csv
* @param pastTime    current rating time
* @param rating  original rating
* @return final rating multiplied by 0.95 for every day interval from the maximal timestamp
*/
  def finalRating(nowTime: String, pastTime: String, rating: String): Float = {
val now =
  LocalDateTime.ofInstant(Instant.ofEpochMilli(nowTime.toLong), ZoneId.systemDefault())
val past =
  LocalDateTime.ofInstant(Instant.ofEpochMilli(pastTime.toLong), ZoneId.systemDefault())
val diff = ChronoUnit.DAYS.between(past, now)
(math.pow(0.95, diff) * rating.toFloat).toFloat
  }

  /**
*
* @param file file to extract
*/
  def fileDispatcher(file: String) = {

    /**
  * get idIndice or increment to idIndice and put it to id map
  * @param id  id in String
  * @param idIndice id in Int
  * @param idMap userIdMap or productIdMap
  * @return (indice for id, max idIndice)
  */
def getIndice(id: String, idIndice: Int, idMap: mutable.Map[String, Int]): (Int, Int) = {
  idMap.get(id) match {
    case Some(i) => (i, idIndice)
    case None => {
      val indice = idIndice + 1
      idMap += (id -> indice)
      (indice, indice)
    }
  }
}

// 1. scan the file the find the max time
val maxTime = readFile(file).reduce((a, b) => if(a._3 > b._3) a else b)._3

// 2. apply rating condition, calculate rating and return only valid rating lines
val validLines = readFile(file).map(x => (x._1, finalRating(maxTime.toString, x._3, x._2))).filter(_._2 > 0.01)

// 3. loop file lines, sum ratings by (userId, productId), and combine id_String and id_Int
val userIdMap = mutable.Map[String, Int]() // (userId, userIdAsInt)
val productIdMap = mutable.Map[String, Int]() // (productId, productIdAsInt)
val userProductRatingMap = mutable.Map[(Int, Int), Float]() // (userIdAsInt, productIdAsInt, ratingSum)

var userIdIndice = -1
var productIdIndice = -1

validLines.foreach(x => {
  val userIdString = x._1._1
  val userId = getIndice(userIdString, userIdIndice, userIdMap)
  userIdIndice = userId._2

  val productIdString = x._1._2
  val productId = getIndice(productIdString, productIdIndice, productIdMap)
  productIdIndice = productId._2

  val key = (userId._1, productId._1)
  userProductRatingMap.get(key) match {
    case Some(i) => userProductRatingMap += (key -> (i + x._2))
    case None => userProductRatingMap += (key -> x._2)
  }
})

filePrinter(FilePathPre + UserIdFile, userIdMap)
filePrinter(FilePathPre + ProductIdFile, productIdMap)
aggFilePrinter(FilePathPre + RatingFile, userProductRatingMap)
  }
}

Answer 1

除了 javish 代码之外，您还有代码样式问题，建议在开始时阅读https://docs.scala-lang.org/style/（这不是最终指南，但是开始就可以了）。避免在元组上使用match { case (a, b, c) => ... }，而是使用mutable.。

主要问题是你使用了可变结构，因此在scala中，默认情况下每个结构都是不可变的，除非你有充分的理由让它变得可变，否则它应该保持不变。它更多地是关于函数式编程，从一个角度来试图避免可变性和副作用，你可以更多地谷歌搜索这个主题。

从代码中删除foreach并将foldLeft替换为例如。 immutable.Map在每次迭代时获得新创建的NULL，而不是修改现有的。{/ p>

如何编写不像java代码的scala代码？

1 个答案: