前言

提到Group By，首先想到的往往是sql中的group by操作，对搜索结果进行分组。其实Java8 Streams API中的Collector也支持流中的数据进行分组和分区操作，本片文章讲简单介绍一下，如何使用groupingBy 和 partitioningBy来对流中的元素进行分组和分区。

groupingBy

首先看一下Java8之前如果想对一个List做分组操作，我们需要如下代码操作：

@Test
public void groupListBeforeJava8() {
    Map<String, List<Employee>> result = new HashMap<>();
    for (Employee e : employees) {
        String city = e.getCity();
        List<Employee> empsInCity = result.get(city);
        if (empsInCity == null) {
            empsInCity = new ArrayList<>();
            result.put(city, empsInCity);
        }
        empsInCity.add(e);
    }
    System.out.println(result);
    assertEquals(result.get("London").size(), 2);
}

而如果使用Java8中Stream的groupingBy分组器，就可以这样操作：


/**
 * 使用java8 stream groupingBy操作,按城市分组list
 */
@Test
public void groupingByTest() {
    Map<String, List<Employee>> employeesByCity =
            employees.stream().collect(Collectors.groupingBy(Employee::getCity));
    System.out.println(employeesByCity);
    assertEquals(employeesByCity.get("London").size(), 2);
}

上面是groupingBy分组器最常见的一个用法，下面简单介绍一下其他用法：

统计每个分组的count

/**
 * 使用java8 stream groupingBy操作,按城市分组list统计count
 */
@Test
public void groupingByCountTest() {
    Map<String, Long> employeesByCity =
            employees.stream().collect(Collectors.groupingBy(Employee::getCity, Collectors.counting()));
    System.out.println(employeesByCity);
    assertEquals(employeesByCity.get("London").longValue(), 2L);
}

统计分组平均值:

/**
 * 使用java8 stream groupingBy操作,按城市分组list并计算分组销售平均值
 */
@Test
public void groupingByAverageTest() {
    Map<String, Double> employeesByCity =
            employees.stream().collect(Collectors.groupingBy(Employee::getCity, Collectors.averagingInt(Employee::getSales)));
    System.out.println(employeesByCity);
    assertEquals(employeesByCity.get("London").intValue(), 175);
}

统计分组总值:

/**
 * 使用java8 stream groupingBy操作,按城市分组list并计算分组销售总值
 */
@Test
public void groupingBySumTest() {
    Map<String, Long> employeesByCity =
            employees.stream().collect(Collectors.groupingBy(Employee::getCity, Collectors.summingLong(Employee::getSales)));
 
    //对Map按照分组销售总值逆序排序
    Map<String, Long> finalMap = new LinkedHashMap<>();
    employeesByCity.entrySet().stream()
            .sorted(Map.Entry.<String, Long>comparingByValue()
                    .reversed()).forEachOrdered(e -> finalMap.put(e.getKey(), e.getValue()));
 
    System.out.println(finalMap);
    assertEquals(finalMap.get("London").longValue(), 350);
}

Join分组List:

/**
 * 通过type分组list，通过join操作连接分组list
 */
@Test
public void groupingByConvertResultTest(){
    List<BlogPost> blogPostList = Lists.newArrayList();
    blogPostList.add(new BlogPost("post1", "zhuoli", 1, 30));
    blogPostList.add(new BlogPost("post2", "zhuoli", 1, 40));
    blogPostList.add(new BlogPost("post3", "zhuoli", 2, 15));
    blogPostList.add(new BlogPost("post4", "zhuoli", 3, 33));
    blogPostList.add(new BlogPost("post5", "Alice", 1, 99));
    blogPostList.add(new BlogPost("post6", "Michael", 3, 65));
 
    Map<Integer, String> postsPerType = blogPostList.stream()
            .collect(Collectors.groupingBy(BlogPost::getType,
                    Collectors.mapping(BlogPost::getTitle, Collectors.joining(", ", "Post titles: [", "]"))));
    System.out.println(postsPerType);
}

转换分组结果List -> List:

/**
 * 使用java8 stream groupingBy操作,按城市分组list,将List转化为name的List
 */
@Test
public void groupingByCityMapList(){
    Map<String, List<String>> namesByCity =
            employees.stream().collect(Collectors.groupingBy(Employee::getCity, Collectors.mapping(Employee::getName, Collectors.toList())));
    System.out.println(namesByCity);
    assertThat(namesByCity.get("London"), contains("Alice", "Bob"));
}

转换分组结果List -> Set:

/**
 * 使用java8 stream groupingBy操作,按城市分组list,将List转化为name的Set
 */
@Test
public void groupingByCityMapListToSet(){
    Map<String, Set<String>> namesByCity =
            employees.stream().collect(Collectors.groupingBy(Employee::getCity, Collectors.mapping(Employee::getName, Collectors.toSet())));
    System.out.println(namesByCity);
    assertThat(namesByCity.get("London"), containsInAnyOrder("Alice", "Bob"));
}

使用对象分组List:

/**
 * 使用java8 stream groupingBy操作,通过Object对象的成员分组List
 */
@Test
public void groupingByObjectTest(){
    List<BlogPost> blogPostList = Lists.newArrayList();
    blogPostList.add(new BlogPost("post1", "zhuoli", 1, 30));
    blogPostList.add(new BlogPost("post2", "zhuoli", 1, 40));
    blogPostList.add(new BlogPost("post3", "zhuoli", 2, 15));
    blogPostList.add(new BlogPost("post4", "zhuoli", 3, 33));
    blogPostList.add(new BlogPost("post5", "Alice", 1, 99));
    blogPostList.add(new BlogPost("post6", "Michael", 3, 65));
 
    Map<Tuple, List<BlogPost>> postsPerTypeAndAuthor = blogPostList.stream()
            .collect(Collectors.groupingBy(post -> new Tuple(post.getAuthor(), post.getType())));
 
    System.out.println(postsPerTypeAndAuthor);
}

使用两个成员分组List:

/**
 * 通过author和type分组list
 */
@Test
public void groupingByMultiItemTest(){
    List<BlogPost> blogPostList = Lists.newArrayList();
    blogPostList.add(new BlogPost("post1", "zhuoli", 1, 30));
    blogPostList.add(new BlogPost("post2", "zhuoli", 1, 40));
    blogPostList.add(new BlogPost("post3", "zhuoli", 2, 15));
    blogPostList.add(new BlogPost("post4", "zhuoli", 3, 33));
    blogPostList.add(new BlogPost("post5", "Alice", 1, 99));
    blogPostList.add(new BlogPost("post6", "Michael", 3, 65));
 
    Map<String, Map<Integer, List<BlogPost>>> map = blogPostList.stream()
            .collect(Collectors.groupingBy(BlogPost::getAuthor, Collectors.groupingBy(BlogPost::getType)));
    System.out.println(map);
}

自定义DistinctBy对分组结果去重:

使用groupingBy源于工作的一个需求，存在如下数据结构：

@Data
@AllArgsConstructor
public class TestData {
    private Integer scene;
    private Integer placement;
    private Long bid;
}

对TestData的List分组，统计每个sene已被占用的placement，我当时直接使用groupIngBy进行分组，得到了一个Map<Integer， List的map，看似完成了目标需求，但当我审查结果的时候，发现List中存在重复现象。比如List中存在多个Scene为1，placement也为1的元素，目标Map中key为1的value List中就会存在多个1，而实际上我们只需要一个1就能说明placement 1已经被占用了，所以我又希望通过distinct进行去重。但是Stream的distinct只能根据元素去重，并不能根据元素的某个成员去重，即distinctBy操作。经过一番折腾，最终实现了功能，看一下示例代码：

public class DistinctByKey {
    @Test
    public void distinctByKeyTest() {
        TestData testData1 = new TestData(1, 1, 100L);
        TestData testData2 = new TestData(1, 2, 1000L);
        TestData testData3 = new TestData(1, 3, 100L);
        TestData testData4 = new TestData(1, 1, 80L);
 
        TestData testData5 = new TestData(2, 1, 1600L);
        TestData testData6 = new TestData(2, 2, 1030L);
        TestData testData7 = new TestData(2, 2, 1001L);
        TestData testData8 = new TestData(2, 2, 1500L);
 
        TestData testData9 = new TestData(3, 5, 1500L);
 
        List<TestData> testDataList = Stream.of(testData1, testData2, testData3, testData4, testData5, testData6, testData7, testData8, testData9).collect(Collectors.toList());
 
        /*直接按照placement去重，scene为2的placement为1和2的元素被去掉*/
        List<TestData> distinctBykeyList = testDataList.stream().filter(distinctByKey(TestData::getPlacement)).collect(Collectors.toList());
        System.out.println(distinctBykeyList);
 
        Map<Integer, List<Integer>> resultMap = testDataList.stream().collect(Collectors.groupingBy(TestData::getScene)).entrySet().stream()
                .collect(Collectors.toMap(Map.Entry::getKey,
                        entry -> entry.getValue().stream().filter(distinctByKey(TestData::getPlacement)).map(TestData::getPlacement).collect(Collectors.toList())));
        System.out.println(resultMap);
    }
 
    private static <T> Predicate<T> distinctByKey(Function<? super T, ?> keyExtractor) {
        Set<Object> seen = ConcurrentHashMap.newKeySet();
        return t -> seen.add(keyExtractor.apply(t));
    }
}

其实Stream的distinct方法，也是filter的一个特别实现。上述示例中distinctByKey也是通过filter，实现的，不多说，看一下示例代码就可以看明白。

示例代码：卓立 – 码云 – groupingBy操作

Q.E.D.

曙光在头上，不抬起头，便永远只能看见物质的闪光。