spark是大数据领域近几年比较火的编程开发语言。有众多的好处,比如速度快,基于内存式计算框架。
创新互联建站服务项目包括乐至网站建设、乐至网站制作、乐至网页制作以及乐至网络营销策划等。多年来,我们专注于互联网行业,利用自身积累的技术优势、行业经验、深度合作伙伴关系等,向广大中小型企业、政府机构等提供互联网行业的解决方案,乐至网站推广取得了明显的社会效益与经济效益。目前,我们服务的客户以成都为中心已经辐射到乐至省份的部分城市,未来相信会继续扩大服务区域并继续获得客户的支持与信任!不多说直接讲 spark的RDD 算子的使用。
如果有spark环境搭建等问题,请自行查找资料。本文不做讲述。
spark rdd的创建有两种方式:
1>从集合创建。也就是从父rdd继承过来
2>从外部创建。
import java.util.Arrays; import java.util.Iterator; import java.util.List; import org.apache.spark.SparkConf; import org.apache.spark.api.java.JavaPairRDD; import org.apache.spark.api.java.JavaRDD; import org.apache.spark.api.java.JavaSparkContext; import org.apache.spark.api.java.function.FlatMapFunction; import org.apache.spark.api.java.function.Function; import org.apache.spark.api.java.function.Function2; import org.apache.spark.api.java.function.PairFunction; import org.apache.spark.api.java.function.VoidFunction; import com.google.common.base.Optional; import scala.Tuple2; public class Demo01 { public static void main(String[] args) { SparkConf conf = new SparkConf().setAppName("Demo01").setMaster("local"); JavaSparkContext jsc = new JavaSparkContext(conf); //map(jsc); //filter(jsc); // flatMap(jsc); //groupByKey(jsc); //reduceByKey(jsc); //sortByKey(jsc); //join(jsc); leftOutJoin(jsc); jsc.stop(); } //每一条元素 都乘以2,并且打印 private static void map(JavaSparkContext jsc) { //数据源 Listlst = Arrays.asList(1,2,3,4,5,6,7,8); JavaRDD numRDD = jsc.parallelize(lst); JavaRDD resultRDD = numRDD.map(new Function () { private static final long serialVersionUID = 1L; @Override public Integer call(Integer num) throws Exception { return num * 2; } }); resultRDD.foreach(new VoidFunction () { private static final long serialVersionUID = 1L; @Override public void call(Integer num) throws Exception { System.out.println(num); } }); } // 把集合中的偶数过滤出来 private static void filter(JavaSparkContext jsc) { //数据源 List lst = Arrays.asList(1,2,3,4,5,6,7,8); JavaRDD numRDD = jsc.parallelize(lst); System.out.println(numRDD.filter(new Function () { private static final long serialVersionUID = 1L; @Override public Boolean call(Integer num) throws Exception { return num % 2 ==0; } }).collect()); } //将一行行数据的单词拆分为一个个单词 private static void flatMap(JavaSparkContext jsc) { List lst = Arrays.asList("hi tim ","hello girl","hello spark"); JavaRDD lines = jsc.parallelize(lst); JavaRDD resultRDD = lines.flatMap(new FlatMapFunction () { private static final long serialVersionUID = 1L; @Override public Iterable call(String line) throws Exception { return Arrays.asList(line.split(" ")); } }); System.out.println(resultRDD.collect()); } // 根据班级进行分组 private static void groupByKey(JavaSparkContext jsc) { // int ,Integer // scala 里面的类型,没有像Java这样分为基本类型和包装类,因为scala是一种更加强的面向对象语言, //一切皆对象,里面的类型,也有对应的方法可以调用,隐式转换 // 模拟数据 @SuppressWarnings("unchecked") List > lst = Arrays.asList( new Tuple2 ("class01", 100), new Tuple2 ("class02",101), new Tuple2 ("class01",199), new Tuple2 ("class02",121), new Tuple2 ("class02",120)); JavaPairRDD cla***DD = jsc.parallelizePairs(lst); JavaPairRDD > groupedRDD = cla***DD.groupByKey(); groupedRDD.foreach(new VoidFunction >>() { private static final long serialVersionUID = 1L; @Override public void call(Tuple2 > tuple) throws Exception { String classKey = tuple._1; Iterator values = tuple._2.iterator(); while (values.hasNext()) { Integer value = values.next(); System.out.println("key:" + classKey + "\t" + "value:" + value); } } }); } private static void reduceByKey(JavaSparkContext jsc) { @SuppressWarnings("unchecked") List > lst = Arrays.asList( new Tuple2 ("class01", 100), new Tuple2 ("class02",101), new Tuple2 ("class01",199), new Tuple2 ("class02",121), new Tuple2 ("class02",120)); JavaPairRDD cla***DD = jsc.parallelizePairs(lst); JavaPairRDD resultRDD = cla***DD.reduceByKey(new Function2 () { private static final long serialVersionUID = 1L; @Override public Integer call(Integer v1, Integer v2) throws Exception { return v1 + v2; } }); resultRDD.foreach(new VoidFunction >() { private static final long serialVersionUID = 1L; @Override public void call(Tuple2 tuple) throws Exception { System.out.println("key:" + tuple._1 + "\t" + "value:" + tuple._2); } }); } // 把学生的成绩前3名取出来,并打印 // 1.先排序sortByKey,然后take(3),再foreach private static void sortByKey(JavaSparkContext jsc) { @SuppressWarnings("unchecked") List > lst = Arrays.asList( new Tuple2 ("tom", 60), new Tuple2 ("kate",80), new Tuple2 ("kobe",100), new Tuple2 ("马蓉",4), new Tuple2 ("宋哲",2), new Tuple2 ("白百合",3), new Tuple2 ("隔壁老王",1)); JavaPairRDD cla***DD = jsc.parallelizePairs(lst); JavaPairRDD pairRDD = cla***DD.mapToPair(new PairFunction ,Integer , String>() { private static final long serialVersionUID = 1L; @Override public Tuple2 call(Tuple2 tuple) throws Exception { return new Tuple2 (tuple._2, tuple._1); } }); //do no JavaPairRDD sortedRDD = pairRDD.sortByKey(); JavaPairRDD sortedRDD01 = sortedRDD.mapToPair(new PairFunction , String, Integer>() { private static final long serialVersionUID = 1L; @Override public Tuple2 call(Tuple2 tuple) throws Exception { return new Tuple2 (tuple._2, tuple._1); } } ); // take 也是一个action操作 List > result = sortedRDD01.take(3); System.out.println(result); } private static void join(JavaSparkContext jsc) { // 模拟数据 @SuppressWarnings("unchecked") List > names =Arrays.asList( new Tuple2 (1,"jack"), new Tuple2 (2,"rose"), new Tuple2 (3,"tom"), new Tuple2 (4,"赵丽颖")); JavaPairRDD num2NamesRDD = jsc.parallelizePairs(names); List > scores = Arrays.asList( new Tuple2 (1,60), new Tuple2 (4,100), new Tuple2 (2,30)); JavaPairRDD num2scoresRDD = jsc.parallelizePairs(scores); JavaPairRDD > joinedRDD = num2scoresRDD.join(num2NamesRDD); //姓名成绩排序,取前2名 JavaPairRDD score2NameRDD = joinedRDD.mapToPair(new PairFunction >,Integer, String>() { private static final long serialVersionUID = 1L; @Override public Tuple2 call( Tuple2 > tuple) throws Exception { Integer score = tuple._2._1; String name = tuple._2._2; return new Tuple2 (score,name); } }); // sortByKey之后,你可以执行一个maptoPair的操作,转换为 System.out.println(score2NameRDD.sortByKey(false).take(2)); } // 学生成绩改良版 private static void leftOutJoin(JavaSparkContext jsc) { // 模拟数据 @SuppressWarnings("unchecked") List > names =Arrays.asList( new Tuple2 (1,"jack"), new Tuple2 (2,"rose"), new Tuple2 (3,"tom"), new Tuple2 (4,"赵丽颖")); JavaPairRDD num2NamesRDD = jsc.parallelizePairs(names); List > scores = Arrays.asList( new Tuple2 (1,60), new Tuple2 (4,100), new Tuple2 (2,30)); JavaPairRDD num2scoresRDD = jsc.parallelizePairs(scores); // num2scoresRDD num2NamesRDD //JavaPairRDD >> joinedRDD = num2NamesRDD.leftOuterJoin(num2scoresRDD); // 注意join,谁join谁,没区别,但是leftoutjoin 是有顺序的 JavaPairRDD >> joinedRDD = num2NamesRDD.leftOuterJoin(num2scoresRDD); JavaPairRDD pairRDD = joinedRDD.mapToPair(new PairFunction >>, Integer, String>() { private static final long serialVersionUID = 1L; @Override public Tuple2 call( Tuple2 >> tuple) throws Exception { String name = tuple._2._1; Optional scoreOptional = tuple._2._2; Integer score = null; if(scoreOptional.isPresent()){ score= scoreOptional.get(); }else { score = 0; } return new Tuple2 (score, name); } }); JavaPairRDD sortedRDD = pairRDD.sortByKey(false); sortedRDD.foreach(new VoidFunction >() { private static final long serialVersionUID = 1L; @Override public void call(Tuple2 tuple) throws Exception { if(tuple._1 == 0){ System.out.println("name:" + tuple._2 + "\t" + "要努力了,你的成绩0分" ); }else{ System.out.println("姓名:" + tuple._2 + "\t" + "分数:" + tuple._1); } } }); } }
如有疑问可跟帖讨论。欢迎拍砖
另外有需要云服务器可以了解下创新互联scvps.cn,海内外云服务器15元起步,三天无理由+7*72小时售后在线,公司持有idc许可证,提供“云服务器、裸金属服务器、高防服务器、香港服务器、美国服务器、虚拟主机、免备案服务器”等云主机租用服务以及企业上云的综合解决方案,具有“安全稳定、简单易用、服务可用性高、性价比高”等特点与优势,专为企业上云打造定制,能够满足用户丰富、多元化的应用场景需求。