Jtdylktuy

Objects are used in example are from package org.jsoup.nodes

import org.jsoup.nodes.Attribute;

import org.jsoup.nodes.Element;

import org.jsoup.select.Elements;

I need group attributes by key with resulting value Set.

Optional<Element> buttonOpt = ...;

Map<String, Set<String>> stringStringMap =

    buttonOpt.map(button -> button.attributes().asList().stream()

            .collect(groupingBy(Attribute::getKey, 

                  mapping(attribute -> attribute.getValue(), toSet()))))

            .orElse(new HashMap<>());

It seems collected correctly, but value all the time is single string (because of library implementation) that contains different values split by space. Trying to improve solution:

Map<String, Set<HashSet<String>>> stringSetMap = buttonOpt.map(

        button -> button.attributes()

            .asList()

            .stream()

            .collect(groupingBy(Attribute::getKey, 

                        mapping(attribute -> 

                          new HashSet<String>(Arrays.asList(attribute.getValue()

                                                                .split(" "))),

                   toSet()))))

  .orElse(new HashMap<>());

In result i've got different structure Map<String, Set<HashSet<String>>> but i need Map<String, Set<String>>

I've checked some collectors but have not managed my issue.

Question is:

How to merge all sets that related to the same attribute key?

You can split your attributes with flatMap and create new entries to group:

Optional<Element> buttonOpt = ...

Map<String, Set<String>> stringStringMap =

        buttonOpt.map(button -> 

            button.attributes()

                  .asList()

                  .stream()

                  .flatMap(at -> Arrays.stream(at.getValue().split(" "))

                                       .map(v -> new SimpleEntry<>(at.getKey(),v)))

                  .collect(groupingBy(Map.Entry::getKey, 

                                      mapping(Map.Entry::getValue, toSet()))))

                .orElse(new HashMap<>());

Here's a Java9 way of doing it,

Map<String, Set<String>> stringSetMap = buttonOpt

    .map(button -> button.attributes().asList().stream()

        .collect(Collectors.groupingBy(Attribute::getKey, Collectors.flatMapping(

            attribute -> Arrays.stream(attribute.getValue().split(" ")), Collectors.toSet()))))

    .orElse(Collections.emptyMap());

This becomes less complicated if you use a more suitable data structure for it, namely a multimap.

Multimaps are present e.g. in Guava, where you can do this as follows:

SetMultimap<String, String> stringMultimap = buttonOpt

        .map(button -> button.attributes().asList().stream()

                .collect(ImmutableSetMultimap.flatteningToImmutableSetMultimap(

                        Attribute::getKey,

                        attribute -> Arrays.stream(attribute.getValue().split(" "))

                ))

        ).orElse(ImmutableSetMultimap.of());

I made it immutable (ImmutableSetMultimap), but a mutable version can also be obtained using Multimaps.flatteningToMultimap.