大橙子网站建设,新征程启航
为企业提供网站建设、域名注册、服务器等服务
一:准备数据源
创新互联是少有的成都网站建设、成都做网站、营销型企业网站、小程序设计、手机APP,开发、制作、设计、买链接、推广优化一站式服务网络公司,从2013年成立,坚持透明化,价格低,无套路经营理念。让网页惊喜每一位访客多年来深受用户好评
在项目下新建一个student.txt文件,里面的内容为:
1,zhangsan,20 2,lisi,21 3,wanger,19 4,fangliu,18
二:实现
Java版:
1.首先新建一个student的Bean对象,实现序列化和toString()方法,具体代码如下:
package com.cxd.sql; import java.io.Serializable; @SuppressWarnings("serial") public class Student implements Serializable { String sid; String sname; int sage; public String getSid() { return sid; } public void setSid(String sid) { this.sid = sid; } public String getSname() { return sname; } public void setSname(String sname) { this.sname = sname; } public int getSage() { return sage; } public void setSage(int sage) { this.sage = sage; } @Override public String toString() { return "Student [sid=" + sid + ", sname=" + sname + ", sage=" + sage + "]"; } }
2.转换,具体代码如下
package com.cxd.sql; import java.util.ArrayList; import org.apache.spark.SparkConf; import org.apache.spark.api.java.JavaRDD; import org.apache.spark.sql.Dataset; import org.apache.spark.sql.Row; import org.apache.spark.sql.RowFactory; import org.apache.spark.sql.SaveMode; import org.apache.spark.sql.SparkSession; import org.apache.spark.sql.types.DataTypes; import org.apache.spark.sql.types.StructField; import org.apache.spark.sql.types.StructType; public class TxtToParquetDemo { public static void main(String[] args) { SparkConf conf = new SparkConf().setAppName("TxtToParquet").setMaster("local"); SparkSession spark = SparkSession.builder().config(conf).getOrCreate(); reflectTransform(spark);//Java反射 dynamicTransform(spark);//动态转换 } /** * 通过Java反射转换 * @param spark */ private static void reflectTransform(SparkSession spark) { JavaRDDsource = spark.read().textFile("stuInfo.txt").javaRDD(); JavaRDD rowRDD = source.map(line -> { String parts[] = line.split(","); Student stu = new Student(); stu.setSid(parts[0]); stu.setSname(parts[1]); stu.setSage(Integer.valueOf(parts[2])); return stu; }); Dataset df = spark.createDataFrame(rowRDD, Student.class); df.select("sid", "sname", "sage"). coalesce(1).write().mode(SaveMode.Append).parquet("parquet.res"); } /** * 动态转换 * @param spark */ private static void dynamicTransform(SparkSession spark) { JavaRDD
source = spark.read().textFile("stuInfo.txt").javaRDD(); JavaRDD rowRDD = source.map( line -> { String[] parts = line.split(","); String sid = parts[0]; String sname = parts[1]; int sage = Integer.parseInt(parts[2]); return RowFactory.create( sid, sname, sage ); }); ArrayList
fields = new ArrayList (); StructField field = null; field = DataTypes.createStructField("sid", DataTypes.StringType, true); fields.add(field); field = DataTypes.createStructField("sname", DataTypes.StringType, true); fields.add(field); field = DataTypes.createStructField("sage", DataTypes.IntegerType, true); fields.add(field); StructType schema = DataTypes.createStructType(fields); Dataset df = spark.createDataFrame(rowRDD, schema); df.coalesce(1).write().mode(SaveMode.Append).parquet("parquet.res1"); } }
scala版本:
import org.apache.spark.sql.SparkSession import org.apache.spark.sql.types.StringType import org.apache.spark.sql.types.StructField import org.apache.spark.sql.types.StructType import org.apache.spark.sql.Row import org.apache.spark.sql.types.IntegerType object RDD2Dataset { case class Student(id:Int,name:String,age:Int) def main(args:Array[String]) { val spark=SparkSession.builder().master("local").appName("RDD2Dataset").getOrCreate() import spark.implicits._ reflectCreate(spark) dynamicCreate(spark) } /** * 通过Java反射转换 * @param spark */ private def reflectCreate(spark:SparkSession):Unit={ import spark.implicits._ val stuRDD=spark.sparkContext.textFile("student2.txt") //toDF()为隐式转换 val stuDf=stuRDD.map(_.split(",")).map(parts⇒Student(parts(0).trim.toInt,parts(1),parts(2).trim.toInt)).toDF() //stuDf.select("id","name","age").write.text("result") //对写入文件指定列名 stuDf.printSchema() stuDf.createOrReplaceTempView("student") val nameDf=spark.sql("select name from student where age<20") //nameDf.write.text("result") //将查询结果写入一个文件 nameDf.show() } /** * 动态转换 * @param spark */ private def dynamicCreate(spark:SparkSession):Unit={ val stuRDD=spark.sparkContext.textFile("student.txt") import spark.implicits._ val schemaString="id,name,age" val fields=schemaString.split(",").map(fieldName => StructField(fieldName, StringType, nullable = true)) val schema=StructType(fields) val rowRDD=stuRDD.map(_.split(",")).map(parts⇒Row(parts(0),parts(1),parts(2))) val stuDf=spark.createDataFrame(rowRDD, schema) stuDf.printSchema() val tmpView=stuDf.createOrReplaceTempView("student") val nameDf=spark.sql("select name from student where age<20") //nameDf.write.text("result") //将查询结果写入一个文件 nameDf.show() } }
注:
1.上面代码全都已经测试通过,测试的环境为spark2.1.0,jdk1.8。
2.此代码不适用于spark2.0以前的版本。
以上这篇Java和scala实现 Spark RDD转换成DataFrame的两种方法小结就是小编分享给大家的全部内容了,希望能给大家一个参考,也希望大家多多支持创新互联。