之前也写过很多篇关于Java8使用的文章了，但是回顾一下，好像还没介绍过Java8 Stream的flatMap操作，昨天刚好在工作中遇到一个场景，发现flatMap简直太方便了，这里总结一下flatMap的常规使用。附带讲一下，使用Java8实现集合的并、交、差操作，其实之前也讲过一种使用Guava的实现方式，具体请参考Guava集合工具

• flatMap

首先看一下一种场景，存在一个Map<Integer， ListContainer>，ListContainer中存在一个List<AClass>成员变量。有这样一个需求，讲Map中values中所有的List<AClass>组合成一个List<AClass>。我们也许会这样操作：

List<AClass> resultAClassList = Lists.newArrayList();
for (ListContainer tmp : map.values()){
    resultAClassList.addAll(tmp.getLst());
}

这还是只存在一层List的情况，如果存在多层，for还需要嵌套，使用起来很不方便。后来查到，Java8 Stream的flatMap操作可以很好地适用这种场景，首先看一下flatMap方法定义：

<R> Stream<R> flatMap(Function<? super T, ? extends Stream<? extends R>> mapper);

参数是一个Function函数式接口，提供T到Stram的转换。其实参考方法实现，flatMap就是将Function转化后的Stram合并成一个Stream。下面看一下使用示例，完成上述相同的功能：

@Test
public void mergeMapValuesTest(){
    Map<Integer, ListContainer> map = Maps.newHashMap();
    List<AClass> aClassList1 = Lists.newArrayList();
    AClass aClass = new AClass(1, "zhuoli1", "haha1");
    aClassList1.add(aClass);
    aClassList1.add(new AClass(2, "zhuoli2", "haha2"));
    aClassList1.add(new AClass(3, "zhuoli3", "haha3"));

    List<AClass> aClassList2 = Lists.newArrayList();
    aClassList2.add(aClass);
    aClassList2.add(new AClass(5, "zhuoli5", "haha5"));
    aClassList2.add(new AClass(6, "zhuoli6", "haha6"));

    /*交集*/
    /*[AClass(id=1, name=zhuoli1, description=haha1)]*/
    List<AClass> intersectResult = aClassList1.stream().filter(aClassList2::contains).collect(Collectors.toList());
    System.out.println(intersectResult);

    /*并集*/
    List<AClass> unionResult = Stream.of(aClassList1, aClassList2).flatMap(Collection::stream).distinct().collect(Collectors.toList());
    assertEquals(unionResult.size(), 5);
    System.out.println(unionResult);

    /*差集*/
    /*[AClass(id=2, name=zhuoli2, description=haha2), AClass(id=3, name=zhuoli3, description=haha3)]*/
    List<AClass> differenceResult = aClassList1.stream().filter(x -> !aClassList2.contains(x)).collect(Collectors.toList());
    System.out.println(differenceResult);

    map.put(1, new ListContainer(aClassList1));
    map.put(2, new ListContainer(aClassList2));

    /*合并多个list*/
    List<AClass> aClassListResult = map.values().stream().flatMap(listContainer -> listContainer.getLst().stream()).collect(Collectors.toList());
    /*注意跟并集的区别*/
    assertEquals(aClassListResult.size(), 6);
    System.out.println(aClassListResult);
}

分享一个flatMap的复杂操作，实现List<Data1>和List<Data2>根据Id进行连接，将连接结果输出为一个List<OutputData>：

@Data
@AllArgsConstructor
public class Data1 {
    private int id;
    private String name;
    private int amount;
}

@Data
@AllArgsConstructor
public class Data2 {
    private int id;
    private String name;
    private String type;
}

@Data
@AllArgsConstructor
public class OutputData {
    private int id;
    private String name;
    private String type;
    private int amount;
}


@Test
public void intersectByKeyTest(){
    List<Data2> listOfData2 = new ArrayList<Data2>();

    listOfData2.add(new Data2(10501, "JOE"  , "Type1"));
    listOfData2.add(new Data2(10603, "SAL"  , "Type5"));
    listOfData2.add(new Data2(40514, "PETER", "Type4"));
    listOfData2.add(new Data2(59562, "JIM"  , "Type2"));
    listOfData2.add(new Data2(29415, "BOB"  , "Type1"));
    listOfData2.add(new Data2(61812, "JOE"  , "Type9"));
    listOfData2.add(new Data2(98432, "JOE"  , "Type7"));
    listOfData2.add(new Data2(62556, "JEFF" , "Type1"));
    listOfData2.add(new Data2(10599, "TOM"  , "Type4"));


    List<Data1> listOfData1 = new ArrayList<Data1>();

    listOfData1.add(new Data1(10501, "JOE"    ,3000000));
    listOfData1.add(new Data1(10603, "SAL"    ,6225000));
    listOfData1.add(new Data1(40514, "PETER"  ,2005000));
    listOfData1.add(new Data1(59562, "JIM"    ,3000000));
    listOfData1.add(new Data1(29415, "BOB"    ,3000000));

    List<OutputData> result = listOfData1.stream()
            .flatMap(x -> listOfData2.stream()
                    .filter(y -> x.getId() == y.getId())
                    .map(y -> new OutputData(y.getId(), x.getName(), y.getType(), x.getAmount())))
            .collect(Collectors.toList());
    System.out.println(result);

    /*difference by key*/
    List<Data1> data1IntersectResult = listOfData1.stream().filter(data1 -> listOfData2.stream().map(Data2::getId).collect(Collectors.toList()).contains(data1.getId())).collect(Collectors.toList());
    System.out.println(data1IntersectResult);
}

• faltMapToInt

首先看一下flatMapToInt方法定义：

IntStream flatMapToInt(Function<? super T, ? extends IntStream> mapper);

跟flatMap不同的是，参数Function函数式接口提供由T到IntStream的转化，方法返回值是IntStream。

@Test
public void flatMapToIntTest() {
    List<List<String>> listOfLists = Arrays.asList(
            Arrays.asList("1", "2"),
            Arrays.asList("5", "6"),
            Arrays.asList("3", "4")
    );

    IntStream intStream =
            listOfLists.stream()
                    .flatMapToInt(childList ->
                            childList.stream()
                                    .mapToInt(Integer::new));

    int sum = intStream.peek(System.out::println).sum();
    System.out.println("sum: " + sum);
}

Stream接口中还存在类似的方法flatMapToDouble、flatMapToLong，使用方法跟flatMapToInt是一样的，这里就不多介绍了，仅罗列一下方法定义：

DoubleStream flatMapToDouble(Function<? super T, ? extends DoubleStream> mapper);
LongStream flatMapToLong(Function<? super T, ? extends LongStream> mapper);

示例代码：码云 – 卓立 – Java8 flatMap示例