Data structures that can map a range of keys to a value

Question

I am trying to find a data structure that takes in a particular value from a range of values and map it to a key.

For example, I have the following conditions:

1. From 1 to 2.9, I want to map it to A.
2. From 4 to 6, I want to map it to B.
3. From 6.5 to 10, I want to map it to C.

I have a value of 5 and I would like to map it to a key. So based on the above conditions, I should map it to B.

Is there any data structure in Java that anyone can recommend to me to solve the problem?

Currently I am using a hashtable that can only map a value to a key. I tried to map the range of values to a particular value that exists in the hashtable. However, I got stuck in the mapping of the range of values to a particular value. So now I am trying to do another way of mapping the ranges of values to a key. Does anyone have any idea how I can solve this problem?

EDIT:

Thanks to Martin Ellis, I decided to use TreeMap to solve the problem.

Answer 1

Are your ranges non-overlapping? If so you could use a TreeMap:

TreeMap<Double, Character> m = new TreeMap<Double, Character>();
m.put(1.0, 'A');
m.put(2.9, null);
m.put(4.0, 'B');
m.put(6.0, null);
m.put(6.5, 'C');
m.put(10.0, null);

The lookup logic is a bit complicated by the fact that you probably want an inclusive lookup (i.e. 2.9 maps to 'A', and not undefined):

private static <K, V> V mappedValue(TreeMap<K, V> map, K key) {
    Entry<K, V> e = map.floorEntry(key);
    if (e != null && e.getValue() == null) {
        e = map.lowerEntry(key);
    }
    return e == null ? null : e.getValue();
}

Example:

mappedValue(m, 5) == 'B'

More results include:

0.9 null
1.0 A
1.1 A
2.8 A
2.9 A
3.0 null
6.4 null
6.5 C
6.6 C
9.9 C
10.0 C
10.1 null

Answer 2

Guava RangeMap provides specialized solution out of the box:

RangeMap<Integer, String> rangeMap = TreeRangeMap.create();
rangeMap.put(Range.closed(1, 100), "foo"); // {[1, 100] => "foo"}
rangeMap.put(Range.open(3, 6), "bar"); // {[1, 3] => "foo", (3, 6) => "bar", [6, 100] => "foo"}

rangeMap.get(42); // returns "foo"

Answer 3

A HashMap will not work for mapping ranges to values unless you find a way to generate a hashcode for ranges and single values in there that matches. But below approach could be what you are looking for

public class RangeMap {
    static class RangeEntry {
        private final double lower;
        private final double upper;
        private final Object value;
        public RangeEntry(double lower, double upper, Object mappedValue) {
            this.lower = lower;
            this.upper = upper;
            this.value = mappedValue;
        }
        public boolean matches(double value) {
            return value >= lower && value <= upper;
        }
        public Object getValue() { return value; }
    }

    private final List<RangeEntry> entries = new ArrayList<RangeEntry>();
    public void put(double lower, double upper, Object mappedValue) {
        entries.add(new RangeEntry(lower, upper, mappedValue));
    }
    public Object getValueFor(double key) {
        for (RangeEntry entry : entries) {
            if (entry.matches(key))
                return entry.getValue();
        }
        return null;
    }
}

You could do

RangeMap map = new RangeMap();
map.put(1, 2.9, "A");
map.put(4, 6, "B");

map.getValueFor(1.5); // = "A"
map.getValueFor(3.5); // = null

It's not very efficient since it's just iterating over a list and it will in that state not complain if you put conflicting ranges in there. Will just return the first it finds.

P.S.: mapping like this would be mapping a range of keys to a value

Answer 4

This seems like a natural situation to use a tree structure.

Unfortunately it won't be practical to implement the java.util.Map interface because it specifies a method to return all of the keys, and in your situation you theoretically have an impractically large number of keys.

Each node of your tree should have a minimum key, a maximum key, and a value associated with that range. You can then have links to the nodes representing the next higher and next lower range (if they exist). Something like:

public class RangeMap<K extends Comparable<K>, V> {
    protected boolean empty;
    protected K lower, upper;
    protected V value;
    protected RangeMap<K, V> left, right;

    public V get(K key) {
        if (empty) {
            return null;
        }

        if (key.compareTo(lower) < 0) {
            return left.get(key);
        }

        if (key.compareTo(upper) > 0) {
            return right.get(key);
        }

        /* if we get here it is in the range */
        return value;
    }

    public void put(K from, K to, V val) {
        if (empty) {
            lower = from;
            upper = to;
            value = val;
            empty = false;
            left = new RangeMap<K,V>();
            right = new RangeMap<K,V>();
            return;
        }

        if (from.compareTo(lower) < 0) {
            left.put(from, to, val);
            return;
        }

        if (to.compareTo(upper) > 0) {
            right.put(from, to, val);
            return;
        }

        /* here you'd have to put the code to deal with adding an overlapping range,
           however you want to handle that. */
    }

    public RangeMap() {
        empty = true;
    }
}

If you need faster lookups than the tree can provide, you may want to look into something like a skip list or developing your own hash function.

Answer 5

This type of data structure is called an Interval Tree. (The Wikipedia page only presents the case where intervals may overlap, but one can imagine a case where you want to remove mappings for any overlapped intervals when you add a new interval. Do a Google search for implementations and see if any fit your needs.

Answer 6

just have an List as a value in your Map.

Map<String, List<Double>> map = new HashMap<String, List<Double>>();

and also one Suggustion:

do not use Hashtable unless you want synchronized access.

EDIT:

Hashtable methods are synchronized, i.e., no two threads can access those methods at a single point of time. HashMap menthods are not Synchronized. If you use Hashtable there will be a performance hit. use HashMap for better performance.

Answer 7

I think Apache commons includes some solution for this issue.

http://commons.apache.org/proper/commons-net/apidocs/index.html

SubnetUtils with usage of SubnetUtils.SubnetInfo

Answer 8

One of the way would be, use one of list implementation as value for key.

map.put("A", ArrayList<Integer>);