问题
I have the following problem: Given these classes,
class Person {
private String zip;
...
public String getZip(){
return zip;
}
}
class Region {
private List<String> zipCodes;
...
public List<String> getZipCodes() {
return zipCodes;
}
}
using the Java 8 Stream API, how do I obtain a Map<Person, List<Region>> based on whether the Region contains that Person's zip code? In other words how do I group the regions by the people whose zip codes belong to those regions?
I've done it in Java 7 the old fashioned way, but now I have to migrate the code to take advantage of the new features in Java 8.
Thank you,
Impeto
回答1:
I suspect the cleanest way to do this -- I'm not quite happy with the other answers posted -- would be
persons.stream().collect(Collectors.toMap(
person -> person,
person -> regions.stream()
.filter(region -> region.getZipCodes().contains(person.getZip()))
.collect(Collectors.toList())));
回答2:
The original answer does an unnecessary mapping with tuples, so you see there the final solution. You could remove the mapping, and simply filter directly the regions list:
//A Set<Region> is more appropriate, IMO
.stream()
.collect(toMap(p -> p,
p -> regions.stream()
.filter(r -> r.getZipCodes().contains(p.getZip()))
.collect(toSet())));
If I understand well, you could do something like this:
import java.util.AbstractMap.SimpleEntry;
import static java.util.stream.Collectors.toMap;
import static java.util.stream.Collectors.toList;
...
List<Person> persons = ...;
List<Region> regions = ...;
Map<Person, List<Region>> map =
persons.stream()
.map(p -> new SimpleEntry<>(p, regions))
.collect(toMap(SimpleEntry::getKey,
e -> e.getValue().stream()
.filter(r -> r.getZipCodes().contains(e.getKey().getZip()))
.collect(toList())));
From the List<Person> you get a Stream<Person>. Then you map each instance to a tuple <Person, List<Region>> that contains all the regions. From there, you collect the data in a map with the toMap collector and, for each person, you build a List of Region that contains the zip code of that person.
For example, given the input:
List<Person> persons = Arrays.asList(new Person("A"), new Person("B"), new Person("C"));
List<Region> regions =
Arrays.asList(new Region(Arrays.asList("A", "B")), new Region(Arrays.asList("A")));
It outputs:
Person{zip='A'} => [Region{zipCodes=[A, B]}, Region{zipCodes=[A]}]
Person{zip='B'} => [Region{zipCodes=[A, B]}]
Person{zip='C'} => []
Also I guess the zipCodes for each Region could be a Set.
回答3:
I have not done any testing of this code, but it compiles so it must be right (:eyeroll:).
public Map<Person,List<Region>> mapPeopleToRegion(List<Person> people, List<Region> regions){
final Map<Person,List<Region>> personToRegion = new HashMap<>();
people.forEach(person ->
personToRegion.put(
person,regions.stream().filter(
region -> region.getZipCodes().contains(person.getZip()))
.collect(Collectors.toList())));
return personToRegion;
}
回答4:
It's still pretty ugly, and I think it would be improved by changing how you model things a bit, but I've only managed to come up with the following so far:
public static void main(String[] args) {
Person[] people = {new Person("00001"), new Person("00002"), new Person("00005")};
Region[] regions = {
new Region("Region 1", Arrays.asList("00001", "00002", "00003")),
new Region("Region 2", Arrays.asList("00002", "00003", "00004")),
new Region("Region 3", Arrays.asList("00001", "00002", "00005"))
};
Map<Person, List<Region>> result = Stream.of(regions)
.flatMap(region -> region.getZipCodes().stream()
.map(zip -> new SimpleEntry<>(zip, region)))
.flatMap(entry -> Stream.of(people)
.filter(person -> person.getZip().equals(entry.getKey()))
.map(person -> new SimpleEntry<>(person, entry.getValue())))
.collect(Collectors.groupingBy(Entry::getKey, Collectors.mapping(Entry::getValue, Collectors.toList())));
result.entrySet().forEach(entry -> System.out.printf("[%s]: {%s}\n", entry.getKey(), entry.getValue()));
// Output:
// [Person: 0]: {[name: Region 1, name: Region 3]}
// [Person: 1]: {[name: Region 1, name: Region 2, name: Region 3]}
// [Person: 2]: {[name: Region 3]}
}
Having a ZipCode class that contained the mapping and could be keyed on would make things cleaner:
public static void main(String[] args) {
Region r1 = new Region("Region 1");
Region r2 = new Region("Region 2");
Region r3 = new Region("Region 3");
ZipCode zipCode1 = new ZipCode("00001", Arrays.asList(r1, r3));
ZipCode zipCode2 = new ZipCode("00002", Arrays.asList(r1, r2, r3));
ZipCode zipCode3 = new ZipCode("00003", Arrays.asList());
ZipCode zipCode4 = new ZipCode("00004", Arrays.asList());
ZipCode zipCode5 = new ZipCode("00005", Arrays.asList(r3));
Person[] people = {
new Person(zipCode1),
new Person(zipCode2),
new Person(zipCode5)
};
Map<Person, List<Region>> result = Stream.of(people)
.collect(Collectors.toMap(person -> person,
person -> person.getZip().getRegions()));
result.entrySet().forEach(entry -> System.out.printf("[%s]: {%s}\n", entry.getKey(), entry.getValue()));
// Output:
// [Person: 0]: {[name: Region 1, name: Region 3]}
// [Person: 1]: {[name: Region 1, name: Region 2, name: Region 3]}
// [Person: 2]: {[name: Region 3]}
}
回答5:
Some of the other answers contain code that does a lot of linear searching through lists. I think the Java 8 Stream solution should not be much slower than the classical variant. So here is a solution that takes advantage of Streams without sacrificing much performance.
List<Person> people = ...
List<Region> regions = ...
Map<String, List<Region>> zipToRegions =
regions.stream().collect(
() -> new HashMap<>(),
(map, region) -> {
for(String zipCode: region.getZipCodes()) {
List<Region> list = map.get(zipCode);
if(list == null) list = new ArrayList<>();
list.add(region);
map.put(zipCode, list);
}
},
(m1, m2) -> m1.putAll(m2)
);
Map<Person, List<Region>> personToRegions =
people.stream().collect(
Collectors.toMap(person -> person,
person -> zipToRegions.get(person.getZip()))
);
来源：https://stackoverflow.com/questions/30245142/how-to-group-elements-of-a-list-by-elements-of-another-in-java-8