上面的是SparkSQL的API操作。
1. 将RDD转化为DataFrame对象
DataFrame：
DataFrame是一种以RDD为基础的分布式数据集，类似于传统数据库中的二维表格。带有schema元信息，即DataFrame所表示的二维表数据集的每一列都带有名称和类型。这样的数据集可以用SQL查询。
创建方式
准备数据
1 zhangsan 20 male
2 lisi 30 female
3 wangwu 35 male
4 zhaosi 40 femaletoDF方式。
package com.hainiu.sparkimport org.apache.spark.sql.SQLContext
import org.apache.spark.{SparkConf, SparkContext}object TestSparkSql{def main(args: Array[String]): Unit = {val conf = new SparkConf()conf.setAppName("test sql")conf.setMaster("local[*]")val sc = new SparkContext(conf)val sqlSc = new SQLContext(sc)//环境对象包装import sqlSc.implicits._//引入环境信息val rdd = sc.textFile("data/a.txt").map(t => {val strs = t.split(" ")(strs(0).toInt, strs(1), strs(2).toInt)})//增加字段信息val df = rdd.toDF("id", "name", "age")df.show() //展示表数据df.printSchema() //展示表格字段信息}
}使用样例类定义schema：
object TestSparkSql{def main(args: Array[String]): Unit = {val conf = new SparkConf()conf.setAppName("test sql")conf.setMaster("local[*]")val sc = new SparkContext(conf)val sqlSc = new SQLContext(sc)import sqlSc.implicits._val rdd = sc.textFile("data/a.txt").map(t => {val strs = t.split(" ")Student(strs(0).toInt, strs(1), strs(2).toInt)})// val df = rdd.toDF("id", "name", "age")val df = rdd.toDF()df.show() //打印数据，以表格的形式打印数据df.printSchema() //打印表的结构信息}
}
case class Student(id:Int,name:String,age:Int)createDataFrame方式
这种方式需要将rdd和schema信息进行合并，得出一个新的DataFrame对象
package com.hainiu.sparkimport org.apache.spark.sql.types.{IntegerType, StringType, StructField, StructType}
import org.apache.spark.sql.{Row, SQLContext}
import org.apache.spark.{SparkConf, SparkContext}object TestSparkSqlWithCreate {def main(args: Array[String]): Unit = {val conf = new SparkConf()conf.setAppName("test create")conf.setMaster("local[*]")val sc = new SparkContext(conf)val sqlSc = new SQLContext(sc)val rdd = sc.textFile("data/a.txt").map(t => {val strs = t.split(" ")Row(strs(0).toInt, strs(1), strs(2).toInt)})
// rdd + schemaval schema = StructType(Array(StructField("id",IntegerType),StructField("name",StringType),StructField("age",IntegerType)))val df = sqlSc.createDataFrame(rdd, schema)df.show()df.printSchema()}
}2. SparkSQL的查询方式（推荐第二种写法）
第二个部分关于df的查询
第一种sql api的方式查询
- 使用的方式方法的形式编程
- 但是思想还是sql形式
- 和rdd编程特别相似的一种写法
object TestSql {def main(args: Array[String]): Unit = {val conf = new SparkConf()conf.setAppName("test sql")conf.setMaster("local[*]")val sc = new SparkContext(conf)val sqlSc = new SQLContext(sc)import sqlSc.implicits._val rdd = sc.textFile("data/a.txt").map(t => {val strs = t.split(" ")(strs(0).toInt, strs(1), strs(2).toInt,strs(3))})val df = rdd.toDF("id", "name", "age","gender")//select * from student where age >20//df.where("age >20")//分组聚合//df.groupby("gender").sum("age")//几个问题//聚合函数不能增加别名 聚合函数不能多次聚合 orderby不识别desc // df.groupBy("gender").agg(count("id").as("id"),sum("age").as("age")).orderBy($"age".desc) //字段标识可以是字符串，也可以是字段对象//df.orderBy($"age".desc) //df.orderBy(col("age").desc) //df.orderBy(df("age").desc) //增加字段对象可以实现高端操作//df.select($"age".+(1)) //join问题//val df1 = sc.makeRDD(Array(// (1,100,98),// (2,100,95),// (3,90,92),//(4,90,93)//)).toDF("id","chinese","math")//df.join(df1,"id") //字段相同 //df.join(df1,df("id")===df1("id")) //窗口函数//普通函数 聚合函数 窗口函数 sum|count|rowkey over (partition by gender order by age desc)//按照条件分割完毕进行数据截取//班级的前两名 每个性别年龄最高的前两个//select *,row_number() over (partition by gender order by age desc) rn from tableimport sqlSc.implicits._import org.apache.spark.sql.functions._df.withColumn("rn",row_number().over(Window.partitionBy("gender").orderBy($"age".desc))).where("rn = 1").show()}
}第二种纯sql形式的查询
- 首先注册表
- 然后使用sql查询
- 最终得出的还是dataFrame的对象
- 其中和rdd的编程没有任何的区别，只不过现在使用sql形式进行处理了而已
package com.hainiu.sparkimport org.apache.spark.sql.types.{IntegerType, StringType, StructField, StructType}
import org.apache.spark.sql.{Row, SQLContext}
import org.apache.spark.{SparkConf, SparkContext}object TestSparkSqlWithCreate {def main(args: Array[String]): Unit = {val conf = new SparkConf()conf.setAppName("test create")conf.setMaster("local[*]")val sc = new SparkContext(conf)val sqlSc = new SQLContext(sc)val rdd = sc.textFile("data/a.txt").map(t => {val strs = t.split(" ")Row(strs(0).toInt, strs(1), strs(2).toInt,strs(3))})
// rdd + schemaval schema = StructType(Array(StructField("id",IntegerType),StructField("name",StringType),StructField("age",IntegerType),StructField("gender",StringType),))val df = sqlSc.createDataFrame(rdd, schema)//sql形式查询//select col from tabledf.createTempView("student")val df1 = sqlSc.sql("""|select count(1) cnt,gender from student group by gender|""".stripMargin)df1.createTempView("student1")val df2 = sqlSc.sql("""|select * from student1 where cnt>1|""".stripMargin)df2.show()df2.printSchema()}
}