大数据Spark“蘑菇云”行动-第13课Scala模式匹配实战和Spark源码鉴赏

Scala的模式匹配，可以对：值、类型、集合、Option、case class 、函数等进行匹配。

首先来鉴赏Spark源代码中的模式匹配示例：

    val classNameV1_0 = "org.apache.spark.mllib.classification.LogisticRegressionModel"
    (loadedClassName, version) match {
      case (className, "1.0") if className == classNameV1_0 =>
        val (numFeatures, numClasses) = ClassificationModel.getNumFeaturesClasses(metadata)
        val data = GLMClassificationModel.SaveLoadV1_0.loadData(sc, path, classNameV1_0)
        // numFeatures, numClasses, weights are checked in model initialization
        val model =
          new LogisticRegressionModel(data.weights, data.intercept, numFeatures, numClasses)
        data.threshold match {
          case Some(t) => model.setThreshold(t)
          case None => model.clearThreshold()
        }
        model
      case _ => throw new Exception(
        s"LogisticRegressionModel.load did not recognize model with (className, format version):" +
        s"($loadedClassName, $version).  Supported:\n" +
        s"  ($classNameV1_0, 1.0)")
    }

以上示例了一个模式匹配，传入一个Tuple类型，进行模式匹配，而且模式匹配中还嵌套了一个模式匹配。

每个case里面都是一个函数体， 通过=>符号来表示，左边是输入，右边是函数体。

下面详细讲解各种模式匹配情况。

---

对值进行模式匹配：

scala>   def getSalary(name: String, age: Int) {
     |     name match {
     |       case "Spark" => println(name + "：$150000/year")
     |       case "Hadoop" => println(name + "：$100000/year")
     |       case _ if name == "Scala" => println(name + "：$140000/year")
     |       case _ if name == "C" => println(name + "：$90000/year")
     |       case bbb if age >= 5 => println("name: " + bbb + ", age: " + age + "，$120000/year")
     |       case _ => println(name + "：$80000/year")
     |     }
     |   }
getSalary: (name: String, age: Int)Unit

scala> getSalary("Java",10)
name: Java, age: 10，$120000/year

scala> getSalary("Java",2)
Java：$80000/year

scala> getSalary("Spark",2)
Spark：$150000/year

scala> getSalary("Hadoop",2)
Hadoop：$100000/year

scala> getSalary("Scala",2)
Scala：$140000/year

scala> getSalary("C",2)
C：$90000/year

scala> getSalary("R",2)
R：$80000/year

解释：
case bbb if age >= 5 这句代码中的bbb变量值为传入参数name。

---

对类型进行模式匹配：

scala> def getMatchType(msg: Any) {
     |     msg match {
     |       case i: Int => println("Integer")
     |       case s: String => println("String")
     |       case d: Double => println("Double")
     |       case array: Array[Int] => println("Array")
     |       case _ => println("Unkown type")
     |     }
     |   }
getMatchType: (msg: Any)Unit

scala> getMatchType(10)
Integer

scala> getMatchType(10.000)
Double

scala> getMatchType("Good!")
String

scala> getMatchType(Array(1,2,3))
Array

scala> getMatchType(true)
Unkown type

---

对集合进行模式匹配：

scala>   def getMatchCollection(msg: Array[String]) {
     |     msg match {
     |       case Array("Scala") => println("One element")
     |       case Array("Scala", "Java") => println("Two elements")
     |       case Array("Spark", _*) => println("Many elements begins with Spark")
     |       case _ => println("Unkown type")
     |     }
     |   }
getMatchCollection: (msg: Array[String])Unit

scala> getMatchCollection(Array("Scala", "Java"))
Two elements

scala> getMatchCollection(Array("Scala", "Java","aa"))
Unkown type

scala> getMatchCollection(Array("Spark","Scala", "Java","aa"))
Many elements begins with Spark

scala> getMatchCollection(Array("Spark"))
Many elements begins with Spark

scala> getMatchCollection(Array("Scala"))
One element

scala> getMatchCollection(Array("Java"))
Unkown type

---

对 case class进行模式匹配：

scala> class DataFrameWork
defined class DataFrameWork

scala> case class ComputationFramework(name: String, popular: Boolean) extends DataFrameWork
defined class ComputationFramework

scala> case class StorageFramework(name: String, popular: Boolean) extends DataFrameWork
defined class StorageFramework

scala>   def getBigDataType(data: DataFrameWork) {
     |     data match {
     |       case ComputationFramework(name, popular) =>
     |         println("ComputationFramework :" + "name : " + name + " , popular :" + popular)
     |       case StorageFramework(name, popular) =>
     |         println("StorageFramework :" + "name : " + name + " , popular :" + popular)
     |       case _ => println("Some other type")
     |     }
     |   }
getBigDataType: (data: DataFrameWork)Unit

scala> getBigDataType(ComputationFramework("Spark", true))
ComputationFramework :name : Spark , popular :true

scala> getBigDataType(ComputationFramework("Spark", false))
ComputationFramework :name : Spark , popular :false

scala> getBigDataType(StorageFramework("HDFS", true))
StorageFramework :name : HDFS , popular :true

scala> getBigDataType(null)
Some other type

scala> getBigDataType(new StorageFramework("HDFS", true))
StorageFramework :name : HDFS , popular :true

注意在：ComputationFramework(“Spark”, true)这行代码中，直接写类名然后传递值，
case class默认有apply类的工厂方法，因为case class也是类，scala编译器会自动生成类的伴生对象，自动生成apply方法。

---

对Option进行模式匹配：

如果值存在， Option[T] 就是一个 Some[T] ，如果不存在， Option[T] 就是对象 None ，建议都用这种方式取值。

scala>   def getValue(key: String, content: Map[String, String]) {
     |     content.get(key) match {
     |       case Some(value) => println(value)
     |       case None => println("Not Found!!!")
     |     }
     |   }
getValue: (key: String, content: Map[String,String])Unit

scala> getValue("Spark",Map("Spark"->"100","Hadoop"->"90"))
100

scala> getValue("scala",Map("Spark"->"100","Hadoop"->"90"))
Not Found!!!

scala> getValue("Hadoop",Map("Spark"->"100","Hadoop"->"90"))
90

---

本博声明：
博文内容源自DT大数据梦工厂大数据Spark“蘑菇云”前置课程。相关课程内容视频可以参考：
百度网盘链接：http://pan.baidu.com/s/1cFqjQu（如果链接失效或需要后续的更多资源，请联系QQ460507491或者微信号：DT1219477246 获取上述资料，或者直接拨打 18610086859咨询）。