您现在的位置是:亿华云 > 应用开发

HashMap的循环姿势你真的使用对了吗?

亿华云2025-10-03 02:34:59【应用开发】7人已围观

简介hashMap 应该是java程序员工作中用的比较多的一个键值对处理的数据的类型了。hashMap 有常见的六七种遍历的方式。这么多的选择,大家平时都是使用哪一种来遍历数据列?欢迎大家在下方留言哦。说

hashMap 应该是的对java程序员工作中用的比较多的一个键值对处理的数据的类型了。hashMap 有常见的循环六七种遍历的方式。这么多的姿势选择,大家平时都是使用使用哪一种来遍历数据列?欢迎大家在下方留言哦。说实话这么多种方式,的对想记也不记不住,循环也不想浪费时间来记这玩意,姿势所以本人在JDK1.8以前基本上都是使用用Map.Entry的方式来遍历,1.8及以后就习惯性用forEach了,的对不过这个不能有continue或者break操作这个有时候还是挺不方便的,其他几种基本上没怎么用过,循环也没太研究这几种方式,姿势哪种性能是使用比较好的。反正就是的对挑自己熟悉的云南idc服务商方式。好了话不多说,循环我们还是姿势直入今天的主题。先来看看每种遍历的方式:

在for循环中使用entries实现Map的遍历

public static void forEachEntries() {          for (Map.Entry<String, String> entry : map.entrySet()) {              String mapKey = entry.getKey();             String mapValue = entry.getValue();         }     } 

在for循环中遍历key

public static void forEachKey() {          for (String key : map.keySet()) {              String mapKey = key;             String mapValue = map.get(mapKey);         }     } 

在for循环中遍历value

public static void forEachValues() {          for (String key : map.values()) {              String val = key;         }     } 

Iterator遍历

public static void forEachIterator() {          Iterator<Entry<String, String>> entries = map.entrySet().iterator();         while (entries.hasNext()) {              Entry<String, String> entry = entries.next();             String key = entry.getKey();             String value = entry.getValue();         }     } 

forEach jdk1.8遍历

public static void forEach() {          map.forEach((key, val) -> {              String key1 = key;             String value = val;         });     } 

Stream jdk1.8遍历

map.entrySet().stream().forEach((entry) -> {              String key = entry.getKey();             String value = entry.getValue();         }); 

Streamparallel jdk1.8遍历

public static void forEachStreamparallel() {          map.entrySet().parallelStream().forEach((entry) -> {              String key = entry.getKey();             String value = entry.getValue();         });     } 

以上就是常见的对于map的一些遍历的方式,下面我们来写个测试用例来看下这些遍历方式,哪些是效率最好的。下面测试用例是基于JMH来测试的 首先引入pom

<dependency>             <groupId>org.openjdk.jmh</groupId>             <artifactId>jmh-core</artifactId>             <version>1.23</version>         </dependency>         <dependency>             <groupId>org.openjdk.jmh</groupId>             <artifactId>jmh-generator-annprocess</artifactId>             <version>1.23</version>             <scope>provided</scope>         </dependency> 

关于jmh测试如可能会影响结果的一些因素这里就不详细介绍了,可以参考文末的第一个链接写的非常详细。以及测试用例为什么要这么写(都是为了消除JIT对测试代码的影响)这是参照官网的链接:编写测试代码如下:

package com.workit.autoconfigure.autoconfigure.controller; import org.openjdk.jmh.annotations.*; import org.openjdk.jmh.infra.Blackhole; import org.openjdk.jmh.results.format.ResultFormatType; import org.openjdk.jmh.runner.Runner; import org.openjdk.jmh.runner.RunnerException; import org.openjdk.jmh.runner.options.Options; import org.openjdk.jmh.runner.options.OptionsBuilder; import java.util.HashMap; import java.util.Iterator; import java.util.Map; import java.util.Map.Entry; import java.util.UUID; import java.util.concurrent.TimeUnit; /**  * @author:公众号:java金融  * @Date:   * @Description:微信搜一搜【java金融】回复666  */ @State(Scope.Thread) @Warmup(iterations = 5, time = 1, timeUnit = TimeUnit.SECONDS) @Measurement(iterations = 5, time = 1, timeUnit = TimeUnit.SECONDS) @Fork(1) @BenchmarkMode(Mode.AverageTime) @OutputTimeUnit(TimeUnit.NANOSECONDS) public class InstructionsBenchmark {      public static void main(String[] args) throws RunnerException {          Options opt = new OptionsBuilder().include(InstructionsBenchmark.class.getSimpleName()).result("result.json").resultFormat(ResultFormatType.JSON).build();         new Runner(opt).run();     }     static final int BASE = 42;     static int add(int key,int val) {        return  BASE + key +val;     }     @Param({ "1", "10", "100", "1000","10000","100000"})     int size;     private static Map<Integer, Integer>  map;     // 初始化方法,在全部Benchmark运行之前进行     @Setup(Level.Trial)     public void init() {          map = new HashMap<>(size);         for (int i = 0; i < size; i++) {              map.put(i, i);         }     }     /**      * 在for循环中使用entries实现Map的遍历:      */     @Benchmark     public static void forEachEntries(Blackhole blackhole) {          for (Map.Entry<Integer, Integer> entry : map.entrySet()) {              Integer mapKey = entry.getKey();             Integer mapValue = entry.getValue();             blackhole.consume(add(mapKey,mapValue));         }     }     /**      * 在for循环中遍历key      */     @Benchmark     public static StringBuffer forEachKey(Blackhole blackhole) {          StringBuffer stringBuffer = new StringBuffer();         for (Integer key : map.keySet()) {            //  Integer mapValue = map.get(key);             blackhole.consume(add(key,key));         }         return stringBuffer;     }     /**      * 在for循环中遍历value      */     @Benchmark     public static void forEachValues(Blackhole blackhole) {          for (Integer key : map.values()) {              blackhole.consume(add(key,key));         }     }     /**      * Iterator遍历;      */     @Benchmark     public static void forEachIterator(Blackhole blackhole) {          Iterator<Entry<Integer, Integer>> entries = map.entrySet().iterator();         while (entries.hasNext()) {              Entry<Integer, Integer> entry = entries.next();             Integer key = entry.getKey();             Integer value = entry.getValue();             blackhole.consume(add(key,value));         }     }     /**      * forEach jdk1.8遍历      */     @Benchmark     public static void forEachLamada(Blackhole blackhole) {          map.forEach((key, value) -> {              blackhole.consume(add(key,value));         });     }     /**      * forEach jdk1.8遍历      */     @Benchmark     public static void forEachStream(Blackhole blackhole) {          map.entrySet().stream().forEach((entry) -> {              Integer key = entry.getKey();             Integer value = entry.getValue();             blackhole.consume(add(key,value));         });     }     @Benchmark     public static void forEachStreamparallel(Blackhole blackhole) {          map.entrySet().parallelStream().forEach((entry) -> {              Integer key = entry.getKey();             Integer value = entry.getValue();             blackhole.consume(add(key,value));         });     } } 

运行结果如下:「注:运行环境idea 2019.3,jdk1.8,windows7 64位。亿华云

Benchmark                                    (size)  Mode  Cnt        Score        Error  Units InstructionsBenchmark.forEachEntries              1  avgt    5       10.021 ±      0.224  ns/op InstructionsBenchmark.forEachEntries             10  avgt    5       71.709 ±      2.537  ns/op InstructionsBenchmark.forEachEntries            100  avgt    5      738.873 ±     12.132  ns/op InstructionsBenchmark.forEachEntries           1000  avgt    5     7804.431 ±    136.635  ns/op InstructionsBenchmark.forEachEntries          10000  avgt    5    88540.345 ±  14915.682  ns/op InstructionsBenchmark.forEachEntries         100000  avgt    5  1083347.001 ± 136865.960  ns/op InstructionsBenchmark.forEachIterator             1  avgt    5       10.675 ±      2.532  ns/op InstructionsBenchmark.forEachIterator            10  avgt    5       73.934 ±      4.517  ns/op InstructionsBenchmark.forEachIterator           100  avgt    5      775.847 ±    198.806  ns/op InstructionsBenchmark.forEachIterator          1000  avgt    5     8905.041 ±   1294.618  ns/op InstructionsBenchmark.forEachIterator         10000  avgt    5    98686.478 ±  10944.570  ns/op InstructionsBenchmark.forEachIterator        100000  avgt    5  1045309.216 ±  36957.608  ns/op InstructionsBenchmark.forEachKey                  1  avgt    5       18.478 ±      1.344  ns/op InstructionsBenchmark.forEachKey                 10  avgt    5       76.398 ±     12.179  ns/op InstructionsBenchmark.forEachKey                100  avgt    5      768.507 ±     23.892  ns/op InstructionsBenchmark.forEachKey               1000  avgt    5    11117.896 ±   1665.021  ns/op InstructionsBenchmark.forEachKey              10000  avgt    5    84871.880 ±  12056.592  ns/op InstructionsBenchmark.forEachKey             100000  avgt    5  1114948.566 ±  65582.709  ns/op InstructionsBenchmark.forEachLamada               1  avgt    5        9.444 ±      0.607  ns/op InstructionsBenchmark.forEachLamada              10  avgt    5       76.125 ±      5.640  ns/op InstructionsBenchmark.forEachLamada             100  avgt    5      861.601 ±     98.045  ns/op InstructionsBenchmark.forEachLamada            1000  avgt    5     7769.714 ±   1663.914  ns/op InstructionsBenchmark.forEachLamada           10000  avgt    5    73250.238 ±   6032.161  ns/op InstructionsBenchmark.forEachLamada          100000  avgt    5   836781.987 ±  72125.745  ns/op InstructionsBenchmark.forEachStream               1  avgt    5       29.113 ±      3.275  ns/op InstructionsBenchmark.forEachStream              10  avgt    5      117.951 ±     13.755  ns/op InstructionsBenchmark.forEachStream             100  avgt    5     1064.767 ±     66.869  ns/op InstructionsBenchmark.forEachStream            1000  avgt    5     9969.549 ±    342.483  ns/op InstructionsBenchmark.forEachStream           10000  avgt    5    93154.061 ±   7638.122  ns/op InstructionsBenchmark.forEachStream          100000  avgt    5  1113961.590 ± 218662.668  ns/op InstructionsBenchmark.forEachStreamparallel       1  avgt    5       65.466 ±      5.519  ns/op InstructionsBenchmark.forEachStreamparallel      10  avgt    5     2298.999 ±    721.455  ns/op InstructionsBenchmark.forEachStreamparallel     100  avgt    5     8270.759 ±   1801.082  ns/op InstructionsBenchmark.forEachStreamparallel    1000  avgt    5    16049.564 ±   1972.856  ns/op InstructionsBenchmark.forEachStreamparallel   10000  avgt    5    69230.849 ±  12169.260  ns/op InstructionsBenchmark.forEachStreamparallel  100000  avgt    5   638129.559 ±  14885.962  ns/op InstructionsBenchmark.forEachValues               1  avgt    5        9.743 ±      2.770  ns/op InstructionsBenchmark.forEachValues              10  avgt    5       70.761 ±     16.574  ns/op InstructionsBenchmark.forEachValues             100  avgt    5      745.069 ±    329.548  ns/op InstructionsBenchmark.forEachValues            1000  avgt    5     7772.584 ±   1702.295  ns/op InstructionsBenchmark.forEachValues           10000  avgt    5    74063.468 ±  23752.678  ns/op InstructionsBenchmark.forEachValues          100000  avgt    5   994057.370 ± 279310.867  ns/op 

通过上述的图我们可以发现,数据量较小的时候forEachEntries和forEachIterator、以及lamada循环效率都差不多forEachStreamarallel的效率反而较低,只有当数据量达到10000以上parallelStream的优势就体现出来了。所以平时选择使用哪种循环方式的时候没必要太纠结哪一种方式,其实每种方式之间的效率还是微乎其微的。选择适合自己的就好。为什么parallelStream在数据量较小的时候效率反而不行?这个大家可以在下方留言哦。

总结

上面小实验只是在我机器上跑出来的结果,可能放到不同的机器运行结果有不一样哦,大家感兴趣的同学可以把代码贴到自己的机器上跑一跑,也许我这这个结论就不适用了。

本文转载自微信公众号「 java金融」,可以通过以下二维码关注。转载本文请联系 java金融公众号。

很赞哦!(15736)