Spark笔记-DataSet，DataFrame

DataSet：面向对象的，从JVM进行构建，或从其它格式进行转化

DataFrame：面向SQL查询，从多种数据源进行构建，或从其它格式进行转化

RDD DataSet DataFrame互转

1.RDD -> Dataset
val ds = rdd.toDS()

2.RDD -> DataFrame
val df = spark.read.json(rdd)

3.Dataset -> RDD
val rdd = ds.rdd

4.Dataset -> DataFrame
val df = ds.toDF()

5.DataFrame -> RDD
val rdd = df.toJSON.rdd

6.DataFrame -> Dataset
val ds = df.toJSON 

DataFrameTest1.scala

package com.spark.dataframe

import org.apache.spark.{SparkConf, SparkContext}

class DataFrameTest1 {
}

object DataFrameTest1{

  def main(args : Array[String]): Unit ={
    System.setProperty("hadoop.home.dir", "E:\\spark\\hadoophome\\hadoop-common-2.2.0-bin");
    val logFile = "e://temp.txt"
    val conf = new SparkConf().setAppName("test").setMaster("local[4]")
    val sc = new SparkContext(conf)
    val logData = sc.textFile(logFile,2).cache()

    val numAs = logData.filter(line => line.contains("a")).count()
    val numBs = logData.filter(line => line.contains("b")).count()

    println(s"Lines with a: $numAs , Line with b : $numBs")

    sc.stop()
  }
}

DataFrameTest2.scala

package com.spark.dataframe

import org.apache.spark.sql.SparkSession

class DataFrameTest2 {
}

object DataFrameTest2{

  def main(args : Array[String]): Unit ={
    System.setProperty("hadoop.home.dir", "E:\\spark\\hadoophome\\hadoop-common-2.2.0-bin")
    val spark = SparkSession
      .builder()
      .appName("Spark SQL basic example")
      .master("local[4]")
      .getOrCreate()

    val df = spark.read.json("E:\\spark\\datatemp\\people.json")
    df.show()

    // This import is needed to use the $-notation
    import spark.implicits._
    df.printSchema()
    df.select("name").show()
    df.filter("age>21").show()
    df.select($"name",$"age"+1).show()

    df.groupBy("age").count().show()

  }
}

DataFrameTest3.scala

package com.spark.dataframe

import org.apache.spark.sql.SparkSession

class DataFrameTest3 {
}

object DataFrameTest3{

  def main(args : Array[String]): Unit ={
      System.setProperty("hadoop.home.dir", "E:\\spark\\hadoophome\\hadoop-common-2.2.0-bin")
      val spark = SparkSession
        .builder()
        .appName("Spark SQL basic example")
        .master("local[4]")
        .getOrCreate()

      val df = spark.read.json("E:\\spark\\datatemp\\people.json")
      // 将DataFrame注册为sql temporary view
      df.createOrReplaceTempView("people")

      val sqlDF = spark.sql("select * from people")
      sqlDF.show()
      //spark.sql("select * from global_temp.people").show()

    }
}

DataSetTest1.scala

package com.spark.dataframe

import org.apache.spark.sql.SparkSession

class DataSetTest1 {
}

case class Person(name: String, age: Long)

object DataSetTest1 {
  def main(args : Array[String]): Unit ={

    System.setProperty("hadoop.home.dir", "E:\\spark\\hadoophome\\hadoop-common-2.2.0-bin")
    val spark = SparkSession
      .builder()
      .appName("Spark SQL basic example")
      .master("local[4]")
      .getOrCreate()

    // This import is needed to use the $-notation
    import spark.implicits._

    val caseClassDS = Seq(Person("Andy", 32)).toDS()
    caseClassDS.show()

    val ds = spark.read.json("E:\\spark\\datatemp\\people.json").as[Person]
    ds.show()

  }
}

RDDToDataFrame.scala

package com.spark.dataframe

import org.apache.spark.sql.{Row, SparkSession}

class RDDToDataFrame {
}

//介绍两种将RDD转换为DataFrame的方式
object RDDToDataFrame{
  def main(args : Array[String]): Unit ={
    System.setProperty("hadoop.home.dir", "E:\\spark\\hadoophome\\hadoop-common-2.2.0-bin")
    val spark = SparkSession
      .builder()
      .appName("Rdd to DataFrame")
      .master("local[4]")
      .getOrCreate()

    // This import is needed to use the $-notation
    import spark.implicits._

    // 数据读取类可以提前定义，Person
    val peopleDF =spark.sparkContext
      .textFile("E:\\spark\\datatemp\\people.txt")
      .map(_.split(","))
      .map(attribute => Person(attribute(0),attribute(1).trim.toInt))
      .toDF()

    peopleDF.createOrReplaceTempView("people")

    val teenagerDF = spark.sql("select name, age from people where age between 13 and 19")
    teenagerDF.map(teenager=> "name:"+teenager(0)).show()
    teenagerDF.map(teenager => "Name: "+teenager.getAs[String]("name")).show()

    // No pre-defined encoders for Dataset[Map[K,V]], define explicitly

    //隐式参数，后面需要Encoder类型的参数时时候则自动调用
    implicit val mapEncoder = org.apache.spark.sql.Encoders.kryo[Map[String,Any]]
    // Primitive types and case classes can be also defined as
    // implicit val stringIntMapEncoder: Encoder[Map[String, Any]] = ExpressionEncoder()

    // row.getValuesMap[T] retrieves multiple columns at once into a Map[String, T]
    teenagerDF.map(teenager => teenager.getValuesMap[Any](List("name","age"))).collect().foreach(println(_))
    // Array(Map("name" -> "Justin", "age" -> 19))

    //////////////////////////////////////////
    //case classes 不能提前定义
    /*
    * When case classes cannot be defined ahead of time
    * (for example, the structure of records is encoded in a string,
    * or a text dataset will be parsed and fields will be projected differently for different users),
    * a DataFrame can be created programmatically with three steps.
    * 1. Create an RDD of Rows from the original RDD;
    * 2. Create the schema represented by a StructType matching the structure of Rows in the RDD created in Step 1.
    * 3. Apply the schema to the RDD of Rows via createDataFrame method provided by SparkSession.
    * */
    import org.apache.spark.sql.types._

    //1. 创建RDD
    val peopleRDD = spark.sparkContext.textFile("e:\\spark\\datatemp\\people.txt")
    //2.1 创建和RDD相匹配的schema
    val schemaString = "name age"
    val fields = schemaString.split(" ")
      .map(fieldName => StructField(fieldName, StringType, nullable = true))
    val schema = StructType(fields)

    //2.2. 将RDD进行格式化
    val rowRDD = peopleRDD
      .map(_.split(","))
      .map(attributes => Row(attributes(0),attributes(1).trim))

    //3. 将RDD转换为DF
    val peopleDF2 = spark.createDataFrame(rowRDD, schema)
    peopleDF2.createOrReplaceTempView("people")
    val results = spark.sql("select name from people")

    results.show()

  }
}

GenericLoadAndSave.scala

package com.spark.dataframe
import org.apache.spark.sql.{SaveMode, SparkSession}
class GenericLoadAndSave {
}

object GenericLoadAndSave{
  def main(args: Array[String]): Unit ={
    System.setProperty("hadoop.home.dir", "E:\\spark\\hadoophome\\hadoop-common-2.2.0-bin")
    val spark = SparkSession
      .builder()
      .appName("Rdd to DataFrame")
      .master("local[4]")
      .getOrCreate()

    // This import is needed to use the $-notation
    import spark.implicits._

    //保存为parquet格式的数据
    val userDF = spark.read.json("e:\\spark\\datatemp\\people.json")
    //userDF.select("name","age").write.save("e:\\spark\\datasave\\nameAndAge.parquet")
    //数据保存时的模式设置为append
    userDF.select("name","age").write.mode(SaveMode.Overwrite).save("e:\\spark\\datasave\\nameAndAge.parquet")

    //数据源的格式可以指定为 (json, parquet, jdbc, orc, libsvm, csv, text)
    val peopleDF = spark.read.format("json").load("e:\\spark\\datatemp\\people.json")
    //peopleDF.select("name","age").write.format("json").save("e:\\spark\\datasave\\peopleNameAndAge.json")
    //数据保存时的模式设置为overwrite
    peopleDF.select("name","age").write.mode(SaveMode.Overwrite).format("json").save("e:\\spark\\datasave\\peopleNameAndAge.json")

    //从parquet格式的数据源中读取数据构建DataFrame
    val peopleDF2 = spark.read.format("parquet").load("E:\\spark\\datasave\\nameAndAge.parquet\\")
      //+"part-00000-*.snappy.parquet") //这行加上便于精准定位。事实上parquet可以根据文件路径自行发现和推断分区信息
    System.out.println("------------------")
    peopleDF2.select("name","age").show()

    //userDF.select("name","age").write.saveAsTable("e:\\spark\\datasave\\peopleSaveAsTable")

    //代码有错误，原因暂时未知
    //val sqlDF = spark.sql("SELECT * FROM parquet.'E:\\spark\\datasave\\nameAndAge.parquet\\part-00000-c8740fc5-cba8-4ebe-a7a8-9cec3da7dfa2.snappy.parquet'")
    //sqlDF.show()
  }
}

ReadFromParquet.scala

package com.spark.dataframe
import org.apache.spark.sql.{SaveMode, SparkSession}

class ReadFromParquet {
}

object  ReadFromParquet{
  def main(args: Array[String]): Unit ={
    System.setProperty("hadoop.home.dir", "E:\\spark\\hadoophome\\hadoop-common-2.2.0-bin")
    val spark = SparkSession
      .builder()
      .appName("Rdd to DataFrame")
      .master("local[4]")
      .getOrCreate()

    // This import is needed to use the $-notation
    import spark.implicits._
    //从parquet格式的数据源中读取数据构建DataFrame
    val peopleDF2 = spark.read.format("parquet").load("E:\\spark\\datasave\\people")

    /*
    * 目录结构为：
    * people
    *   |- country=china
    *           |-data.parquet
    *   |- country=us
    *           |-data.parquet
    *
    * data.parquet内包含people的name和age。加上文件路径中的country信息，最终得到的表结构为：
    * +-------+----+-------+
    * |   name| age|country|
    * +-------+----+-------+
    * */
    peopleDF2.show()
  }
}

SchemaMerge.scala

package com.spark.dataframe
import org.apache.spark.sql.{SaveMode, SparkSession}

class SchemaMerge {
}

object SchemaMerge{
  def main(args: Array[String]) {
    System.setProperty("hadoop.home.dir", "E:\\spark\\hadoophome\\hadoop-common-2.2.0-bin")
    val spark = SparkSession
      .builder()
      .appName("Rdd to DataFrame")
      .master("local[4]")
      .getOrCreate()

    // This import is needed to use the $-notation
    import spark.implicits._

    val squaresDF = spark.sparkContext.makeRDD(1 to 5)
      .map(i=>(i,i*i))
      .toDF("value","square")

    squaresDF.write.mode(SaveMode.Overwrite).parquet("E:\\spark\\datasave\\schemamerge\\test_table\\key=1")

    val cubesDF = spark.sparkContext.makeRDD(1 to 5)
      .map(i => (i,i*i*i))
      .toDF("value","cube")
    cubesDF.write.mode(SaveMode.Overwrite).parquet("E:\\spark\\datasave\\schemamerge\\test_table\\key=2")

    val mergedDF = spark.read.option("mergeSchema","true")
      .parquet("E:\\spark\\datasave\\schemamerge\\test_table\\")

    mergedDF.printSchema()
    mergedDF.show()
  }
}

结果：