Sort a Map by values

Question

I am relatively new to Java, and often find that I need to sort a Map<Key, Value> on the values.

Since the values are not unique, I find myself converting the keySet into an array, and sorting that array through array sort with a custom comparator that sorts on the value associated with the key.

Is there an easier way?

Answer 1

Here is an OO solution (i.e., doesn't use static methods):

import java.util.Collections;
import java.util.Comparator;
import java.util.HashMap;
import java.util.Iterator;
import java.util.LinkedList;
import java.util.LinkedHashMap;
import java.util.List;
import java.util.Map;

public class SortableValueMap<K, V extends Comparable<V>>
  extends LinkedHashMap<K, V> {
  public SortableValueMap() { }

  public SortableValueMap( Map<K, V> map ) {
    super( map );
  }

  public void sortByValue() {
    List<Map.Entry<K, V>> list = new LinkedList<Map.Entry<K, V>>( entrySet() );

    Collections.sort( list, new Comparator<Map.Entry<K, V>>() {
      public int compare( Map.Entry<K, V> entry1, Map.Entry<K, V> entry2 ) {
        return entry1.getValue().compareTo( entry2.getValue() );
      }
    });

    clear();

    for( Map.Entry<K, V> entry : list ) {
      put( entry.getKey(), entry.getValue() );
    }
  }

  private static void print( String text, Map<String, Double> map ) {
    System.out.println( text );

    for( String key : map.keySet() ) {
      System.out.println( "key/value: " + key + "/" + map.get( key ) );
    }
  }

  public static void main( String[] args ) {
    SortableValueMap<String, Double> map =
      new SortableValueMap<String, Double>();

    map.put( "A", 67.5 );
    map.put( "B", 99.5 );
    map.put( "C", 82.4 );
    map.put( "D", 42.0 );

    print( "Unsorted map", map );
    map.sortByValue();
    print( "Sorted map", map );
  }
}

Hereby donated to the public domain.

Answer 2

For sure the solution of Stephen is really great, but for those who can't use Guava:

Here's my solution for sorting by value a map. This solution handle the case where there are twice the same value etc...

// If you want to sort a map by value, and if there can be twice the same value:

// here is your original map
Map<String,Integer> mapToSortByValue = new HashMap<String, Integer>();
mapToSortByValue.put("A", 3);
mapToSortByValue.put("B", 1);
mapToSortByValue.put("C", 3);
mapToSortByValue.put("D", 5);
mapToSortByValue.put("E", -1);
mapToSortByValue.put("F", 1000);
mapToSortByValue.put("G", 79);
mapToSortByValue.put("H", 15);

// Sort all the map entries by value
Set<Map.Entry<String,Integer>> set = new TreeSet<Map.Entry<String,Integer>>(
        new Comparator<Map.Entry<String,Integer>>(){
            @Override
            public int compare(Map.Entry<String,Integer> obj1, Map.Entry<String,Integer> obj2) {
                Integer val1 = obj1.getValue();
                Integer val2 = obj2.getValue();
                // DUPLICATE VALUE CASE
                // If the values are equals, we can't return 0 because the 2 entries would be considered
                // as equals and one of them would be deleted (because we use a set, no duplicate, remember!)
                int compareValues = val1.compareTo(val2);
                if ( compareValues == 0 ) {
                    String key1 = obj1.getKey();
                    String key2 = obj2.getKey();
                    int compareKeys = key1.compareTo(key2);
                    if ( compareKeys == 0 ) {
                        // what you return here will tell us if you keep REAL KEY-VALUE duplicates in your set
                        // if you want to, do whatever you want but do not return 0 (but don't break the comparator contract!)
                        return 0;
                    }
                    return compareKeys;
                }
                return compareValues;
            }
        }
);
set.addAll(mapToSortByValue.entrySet());


// OK NOW OUR SET IS SORTED COOL!!!!

// And there's nothing more to do: the entries are sorted by value!
for ( Map.Entry<String,Integer> entry : set ) {
    System.out.println("Set entries: " + entry.getKey() + " -> " + entry.getValue());
}




// But if you add them to an hashmap
Map<String,Integer> myMap = new HashMap<String,Integer>();
// When iterating over the set the order is still good in the println...
for ( Map.Entry<String,Integer> entry : set ) {
    System.out.println("Added to result map entries: " + entry.getKey() + " " + entry.getValue());
    myMap.put(entry.getKey(), entry.getValue());
}

// But once they are in the hashmap, the order is not kept!
for ( Integer value : myMap.values() ) {
    System.out.println("Result map values: " + value);
}
// Also this way doesn't work:
// Logic because the entryset is a hashset for hashmaps and not a treeset
// (and even if it was a treeset, it would be on the keys only)
for ( Map.Entry<String,Integer> entry : myMap.entrySet() ) {
    System.out.println("Result map entries: " + entry.getKey() + " -> " + entry.getValue());
}


// CONCLUSION:
// If you want to iterate on a map ordered by value, you need to remember:
// 1) Maps are only sorted by keys, so you can't sort them directly by value
// 2) So you simply CAN'T return a map to a sortMapByValue function
// 3) You can't reverse the keys and the values because you have duplicate values
//    This also means you can't neither use Guava/Commons bidirectionnal treemaps or stuff like that

// SOLUTIONS
// So you can:
// 1) only sort the values which is easy, but you loose the key/value link (since you have duplicate values)
// 2) sort the map entries, but don't forget to handle the duplicate value case (like i did)
// 3) if you really need to return a map, use a LinkedHashMap which keep the insertion order

The exec: http://www.ideone.com/dq3Lu

The output:

Set entries: E -> -1
Set entries: B -> 1
Set entries: A -> 3
Set entries: C -> 3
Set entries: D -> 5
Set entries: H -> 15
Set entries: G -> 79
Set entries: F -> 1000
Added to result map entries: E -1
Added to result map entries: B 1
Added to result map entries: A 3
Added to result map entries: C 3
Added to result map entries: D 5
Added to result map entries: H 15
Added to result map entries: G 79
Added to result map entries: F 1000
Result map values: 5
Result map values: -1
Result map values: 1000
Result map values: 79
Result map values: 3
Result map values: 1
Result map values: 3
Result map values: 15
Result map entries: D -> 5
Result map entries: E -> -1
Result map entries: F -> 1000
Result map entries: G -> 79
Result map entries: A -> 3
Result map entries: B -> 1
Result map entries: C -> 3
Result map entries: H -> 15

Hope it will help some folks

Answer 3

If you have duplicate keys and only a small set of data (<1000) and your code is not performance critical you can just do the following:

Map<String,Integer> tempMap=new HashMap<String,Integer>(inputUnsortedMap);
LinkedHashMap<String,Integer> sortedOutputMap=new LinkedHashMap<String,Integer>();

for(int i=0;i<inputUnsortedMap.size();i++){
    Map.Entry<String,Integer> maxEntry=null;
    Integer maxValue=-1;
    for(Map.Entry<String,Integer> entry:tempMap.entrySet()){
        if(entry.getValue()>maxValue){
            maxValue=entry.getValue();
            maxEntry=entry;
        }
    }
    tempMap.remove(maxEntry.getKey());
    sortedOutputMap.put(maxEntry.getKey(),maxEntry.getValue());
}

inputUnsortedMap is the input to the code.

The variable sortedOutputMap will contain the data in decending order when iterated over. To change order just change > to a < in the if statement.

Is not the fastest sort but does the job without any additional dependencies.

Answer 4

Instead of using Collections.sort as some do I'd suggest using Arrays.sort. Actually what Collections.sort does is something like this:

public static <T extends Comparable<? super T>> void sort(List<T> list) {
    Object[] a = list.toArray();
    Arrays.sort(a);
    ListIterator<T> i = list.listIterator();
    for (int j=0; j<a.length; j++) {
        i.next();
        i.set((T)a[j]);
    }
}

It just calls toArray on the list and then uses Arrays.sort. This way all the map entries will be copied three times: once from the map to the temporary list (be it a LinkedList or ArrayList), then to the temporary array and finally to the new map.

My solution ommits this one step as it does not create unnecessary LinkedList. Here is the code, generic-friendly and performance-optimal:

public static <K, V extends Comparable<? super V>> Map<K, V> sortByValue(Map<K, V> map) 
{
    @SuppressWarnings("unchecked")
    Map.Entry<K,V>[] array = map.entrySet().toArray(new Map.Entry[map.size()]);

    Arrays.sort(array, new Comparator<Map.Entry<K, V>>() 
    {
        public int compare(Map.Entry<K, V> e1, Map.Entry<K, V> e2) 
        {
            return e1.getValue().compareTo(e2.getValue());
        }
    });

    Map<K, V> result = new LinkedHashMap<K, V>();
    for (Map.Entry<K, V> entry : array)
        result.put(entry.getKey(), entry.getValue());

    return result;
}

Answer 5

Create customized comparator and use it while creating new TreeMap object.

class MyComparator implements Comparator<Object> {

    Map<String, Integer> map;

    public MyComparator(Map<String, Integer> map) {
        this.map = map;
    }

    public int compare(Object o1, Object o2) {

        if (map.get(o2) == map.get(o1))
            return 1;
        else
            return ((Integer) map.get(o2)).compareTo((Integer)     
                                                            map.get(o1));

    }
}

Use the below code in your main func

    Map<String, Integer> lMap = new HashMap<String, Integer>();
    lMap.put("A", 35);
    lMap.put("B", 75);
    lMap.put("C", 50);
    lMap.put("D", 50);

    MyComparator comparator = new MyComparator(lMap);

    Map<String, Integer> newMap = new TreeMap<String, Integer>(comparator);
    newMap.putAll(lMap);
    System.out.println(newMap);

Output:

{B=75, D=50, C=50, A=35}

Answer 6

From http://www.programmersheaven.com/download/49349/download.aspx

private static <K, V> Map<K, V> sortByValue(Map<K, V> map) {
    List<Entry<K, V>> list = new LinkedList<>(map.entrySet());
    Collections.sort(list, new Comparator<Object>() {
        @SuppressWarnings("unchecked")
        public int compare(Object o1, Object o2) {
            return ((Comparable<V>) ((Map.Entry<K, V>) (o1)).getValue()).compareTo(((Map.Entry<K, V>) (o2)).getValue());
        }
    });

    Map<K, V> result = new LinkedHashMap<>();
    for (Iterator<Entry<K, V>> it = list.iterator(); it.hasNext();) {
        Map.Entry<K, V> entry = (Map.Entry<K, V>) it.next();
        result.put(entry.getKey(), entry.getValue());
    }

    return result;
}