I've seen several other questions similiar to this one but I haven't really been able to find anything that resolves my problem.
My use case is this: user has a list of items initially (listA). They reorder the items and want to persist that order (listB), however, due to restrictions I'm unable persist the order on the backend so I have to sort listA after I retrieve it.
So basically, I have 2 ArrayLists (listA and listB). One with the specific order the lists should be in (listB) and the other has the list of items (listA). I want to sort listA based on listB.
Using Java 8:
Collections.sort(listToSort,
Comparator.comparing(item -> listWithOrder.indexOf(item)));
or better:
listToSort.sort(Comparator.comparingInt(listWithOrder::indexOf));
Collections.sort(listB, new Comparator<Item>() {
public int compare(Item left, Item right) {
return Integer.compare(listA.indexOf(left), listA.indexOf(right));
}
});
This is quite inefficient, though, and you should probably create a Map<Item, Integer> from listA to lookup the positions of the items faster.
Guava has a ready-to-use comparator for doing that: Ordering.explicit()
Let's say you have a listB list that defines the order in which you want to sort listA. This is just an example, but it demonstrates an order that is defined by a list, and not the natural order of the datatype:
List<String> listB = Arrays.asList("Sunday", "Monday", "Tuesday", "Wednesday",
"Thursday", "Friday", "Saturday");
Now, let's say that listA needs to be sorted according to this ordering. It's a List<Item>, and Item has a public String getWeekday() method.
Create a Map<String, Integer> that maps the values of everything in listB to something that can be sorted easily, such as the index, i.e. "Sunday" => 0, ..., "Saturday" => 6. This will provide a quick and easy lookup.
Map<String, Integer> weekdayOrder = new HashMap<String, Integer>();
for (int i = 0; i < listB.size(); i++)
{
String weekday = listB.get(i);
weekdayOrder.put(weekday, i);
}
Then you can create your custom Comparator<Item> that uses the Map to create an order:
public class ItemWeekdayComparator implements Comparator<Item>
{
private Map<String, Integer> sortOrder;
public ItemWeekdayComparator(Map<String, Integer> sortOrder)
{
this.sortOrder = sortOrder;
}
#Override
public int compare(Item i1, Item i2)
{
Integer weekdayPos1 = sortOrder.get(i1.getWeekday());
if (weekdayPos1 == null)
{
throw new IllegalArgumentException("Bad weekday encountered: " +
i1.getWeekday());
}
Integer weekdayPos2 = sortOrder.get(i2.getWeekday());
if (weekdayPos2 == null)
{
throw new IllegalArgumentException("Bad weekday encountered: " +
i2.getWeekday());
}
return weekdayPos1.compareTo(weekdayPos2);
}
}
Then you can sort listA using your custom Comparator.
Collections.sort(listA, new ItemWeekdayComparator(weekdayOrder));
Speed improvement on JB Nizet's answer (from the suggestion he made himself). With this method:
Sorting a 1000 items list 100 times improves speed 10 times on my
unit tests.
Sorting a 10000 items list 100 times improves speed 140 times (265 ms for the whole batch instead of 37 seconds) on my
unit tests.
This method will also work when both lists are not identical:
/**
* Sorts list objectsToOrder based on the order of orderedObjects.
*
* Make sure these objects have good equals() and hashCode() methods or
* that they reference the same objects.
*/
public static void sortList(List<?> objectsToOrder, List<?> orderedObjects) {
HashMap<Object, Integer> indexMap = new HashMap<>();
int index = 0;
for (Object object : orderedObjects) {
indexMap.put(object, index);
index++;
}
Collections.sort(objectsToOrder, new Comparator<Object>() {
public int compare(Object left, Object right) {
Integer leftIndex = indexMap.get(left);
Integer rightIndex = indexMap.get(right);
if (leftIndex == null) {
return -1;
}
if (rightIndex == null) {
return 1;
}
return Integer.compare(leftIndex, rightIndex);
}
});
}
Problem : sorting a list of Pojo on the basis of one of the field's all possible values present in another list.
Take a look at this solution, may be this is what you are trying to achieve:
import java.util.ArrayList;
import java.util.Collections;
import java.util.Comparator;
import java.util.List;
public class Test {
public static void main(String[] args) {
List<Employee> listToSort = new ArrayList<>();
listToSort.add(new Employee("a", "age11"));
listToSort.add(new Employee("c", "age33"));
listToSort.add(new Employee("b", "age22"));
listToSort.add(new Employee("a", "age111"));
listToSort.add(new Employee("c", "age3"));
listToSort.add(new Employee("b", "age2"));
listToSort.add(new Employee("a", "age1"));
List<String> listWithOrder = new ArrayList<>();
listWithOrder.add("a");
listWithOrder.add("b");
listWithOrder.add("c");
Collections.sort(listToSort, Comparator.comparing(item ->
listWithOrder.indexOf(item.getName())));
System.out.println(listToSort);
}
}
class Employee {
String name;
String age;
public Employee(String name, String age) {
super();
this.name = name;
this.age = age;
}
public String getName() {
return name;
}
public String getAge() {
return age;
}
#Override
public String toString() {
return "[name=" + name + ", age=" + age + "]";
}
}
O U T P U T
[[name=a, age=age11], [name=a, age=age111], [name=a, age=age1], [name=b, age=age22], [name=b, age=age2], [name=c, age=age33], [name=c, age=age3]]
Here is a solution that increases the time complexity by 2n, but accomplishes what you want. It also doesn't care if the List R you want to sort contains Comparable elements so long as the other List L you use to sort them by is uniformly Comparable.
public class HeavyPair<L extends Comparable<L>, R> implements Comparable<HeavyPair<L, ?>> {
public final L left;
public final R right;
public HeavyPair(L left, R right) {
this.left = left;
this.right = right;
}
public compareTo(HeavyPair<L, ?> o) {
return this.left.compareTo(o.left);
}
public static <L extends Comparable<L>, R> List<R> sort(List<L> weights, List<R> toSort) {
assert(weights.size() == toSort.size());
List<R> output = new ArrayList<>(toSort.size());
List<HeavyPair<L, R>> workHorse = new ArrayList<>(toSort.size());
for(int i = 0; i < toSort.size(); i++) {
workHorse.add(new HeavyPair(weights.get(i), toSort.get(i)))
}
Collections.sort(workHorse);
for(int i = 0; i < workHorse.size(); i++) {
output.add(workHorse.get(i).right);
}
return output;
}
}
Excuse any terrible practices I used while writing this code, though. I was in a rush.
Just call HeavyPair.sort(listB, listA);
Edit: Fixed this line return this.left.compareTo(o.left);. Now it actually works.
Here is an example of how to sort a list and then make the changes in another list according to the changes exactly made to first array list. This trick will never fails and ensures the mapping between the items in list. The size of both list must be same to use this trick.
ArrayList<String> listA = new ArrayList<String>();
ArrayList<String> listB = new ArrayList<String>();
int j = 0;
// list of returns of the compare method which will be used to manipulate
// the another comparator according to the sorting of previous listA
ArrayList<Integer> sortingMethodReturns = new ArrayList<Integer>();
public void addItemstoLists() {
listA.add("Value of Z");
listA.add("Value of C");
listA.add("Value of F");
listA.add("Value of A");
listA.add("Value of Y");
listB.add("this is the value of Z");
listB.add("this is the value off C");
listB.add("this is the value off F");
listB.add("this is the value off A");
listB.add("this is the value off Y");
Collections.sort(listA, new Comparator<String>() {
#Override
public int compare(String lhs, String rhs) {
// TODO Auto-generated method stub
int returning = lhs.compareTo(rhs);
sortingMethodReturns.add(returning);
return returning;
}
});
// now sort the list B according to the changes made with the order of
// items in listA
Collections.sort(listB, new Comparator<String>() {
#Override
public int compare(String lhs, String rhs) {
// TODO Auto-generated method stub
// comparator method will sort the second list also according to
// the changes made with list a
int returning = sortingMethodReturns.get(j);
j++;
return returning;
}
});
}
try this for java 8:
listB.sort((left, right) -> Integer.compare(list.indexOf(left), list.indexOf(right)));
or
listB.sort(Comparator.comparingInt(item -> list.indexOf(item)));
import java.util.Comparator;
import java.util.List;
public class ListComparator implements Comparator<String> {
private final List<String> orderedList;
private boolean appendFirst;
public ListComparator(List<String> orderedList, boolean appendFirst) {
this.orderedList = orderedList;
this.appendFirst = appendFirst;
}
#Override
public int compare(String o1, String o2) {
if (orderedList.contains(o1) && orderedList.contains(o2))
return orderedList.indexOf(o1) - orderedList.indexOf(o2);
else if (orderedList.contains(o1))
return (appendFirst) ? 1 : -1;
else if (orderedList.contains(o2))
return (appendFirst) ? -1 : 1;
return 0;
}
}
You can use this generic comparator to sort list based on the the other list.
For example, when appendFirst is false below will be the output.
Ordered list: [a, b]
Un-ordered List: [d, a, b, c, e]
Output:
[a, b, d, c, e]
One way of doing this is looping through listB and adding the items to a temporary list if listA contains them:
List<?> tempList = new ArrayList<?>();
for(Object o : listB) {
if(listA.contains(o)) {
tempList.add(o);
}
}
listA.removeAll(listB);
tempList.addAll(listA);
return tempList;
Not completely clear what you want, but if this is the situation:
A:[c,b,a]
B:[2,1,0]
And you want to load them both and then produce:
C:[a,b,c]
Then maybe this?
List c = new ArrayList(b.size());
for(int i=0;i<b.size();i++) {
c.set(b.get(i),a.get(i));
}
that requires an extra copy, but I think to to it in place is a lot less efficient, and all kinds of not clear:
for(int i=0;i<b.size();i++){
int from = b.get(i);
if(from == i) continue;
T tmp = a.get(i);
a.set(i,a.get(from));
a.set(from,tmp);
b.set(b.lastIndexOf(i),from);
}
Note I didn't test either, maybe got a sign flipped.
Another solution that may work depending on your setting is not storing instances in listB but instead indices from listA. This could be done by wrapping listA inside a custom sorted list like so:
public static class SortedDependingList<E> extends AbstractList<E> implements List<E>{
private final List<E> dependingList;
private final List<Integer> indices;
public SortedDependingList(List<E> dependingList) {
super();
this.dependingList = dependingList;
indices = new ArrayList<>();
}
#Override
public boolean add(E e) {
int index = dependingList.indexOf(e);
if (index != -1) {
return addSorted(index);
}
return false;
}
/**
* Adds to this list the element of the depending list at the given
* original index.
* #param index The index of the element to add.
*
*/
public boolean addByIndex(int index){
if (index < 0 || index >= this.dependingList.size()) {
throw new IllegalArgumentException();
}
return addSorted(index);
}
/**
* Returns true if this list contains the element at the
* index of the depending list.
*/
public boolean containsIndex(int index){
int i = Collections.binarySearch(indices, index);
return i >= 0;
}
private boolean addSorted(int index){
int insertIndex = Collections.binarySearch(indices, index);
if (insertIndex < 0){
insertIndex = -insertIndex-1;
this.indices.add(insertIndex, index);
return true;
}
return false;
}
#Override
public E get(int index) {
return dependingList.get(indices.get(index));
}
#Override
public int size() {
return indices.size();
}
}
Then you can use this custom list as follows:
public static void main(String[] args) {
class SomeClass{
int index;
public SomeClass(int index) {
super();
this.index = index;
}
#Override
public String toString() {
return ""+index;
}
}
List<SomeClass> listA = new ArrayList<>();
for (int i = 0; i < 100; i++) {
listA.add(new SomeClass(i));
}
SortedDependingList<SomeClass> listB = new SortedDependingList<>(listA);
Random rand = new Random();
// add elements by index:
for (int i = 0; i < 5; i++) {
int index = rand.nextInt(listA.size());
listB.addByIndex(index);
}
System.out.println(listB);
// add elements by identity:
for (int i = 0; i < 5; i++) {
int index = rand.nextInt(listA.size());
SomeClass o = listA.get(index);
listB.add(o);
}
System.out.println(listB);
}
Of course, this custom list will only be valid as long as the elements in the original list do not change. If changes are possible, you would need to somehow listen for changes to the original list and update the indices inside the custom list.
Note also, that the SortedDependingList does currently not allow to add an element from listA a second time - in this respect it actually works like a set of elements from listA because this is usually what you want in such a setting.
The preferred way to add something to SortedDependingList is by already knowing the index of an element and adding it by calling sortedList.addByIndex(index);
If the two lists are guaranteed to contain the same elements, just in a different order, you can use List<T> listA = new ArrayList<>(listB) and this will be O(n) time complexity. Otherwise, I see a lot of answers here using Collections.sort(), however there is an alternative method which is guaranteed O(2n) runtime, which should theoretically be faster than sort's worst time complexity of O(nlog(n)), at the cost of 2n storage
Set<T> validItems = new HashSet<>(listB);
listA.clear();
listB.forEach(item -> {
if(validItems.contains(item)) {
listA.add(item);
}
});
List<String> listA;
Comparator<B> comparator = Comparator.comparing(e -> listA.indexOf(e.getValue()));
//call your comparator inside your list to be sorted
listB.stream().sorted(comparator)..
Like Tim Herold wrote, if the object references should be the same, you can just copy listB to listA, either:
listA = new ArrayList(listB);
Or this if you don't want to change the List that listA refers to:
listA.clear();
listA.addAll(listB);
If the references are not the same but there is some equivalence relationship between objects in listA and listB, you could sort listA using a custom Comparator that finds the object in listB and uses its index in listB as the sort key. The naive implementation that brute force searches listB would not be the best performance-wise, but would be functionally sufficient.
IMO, you need to persist something else. May be not the full listB, but something. May be just the indexes of the items that the user changed.
Try this. The code below is general purpose for a scenario where listA is a list of Objects since you did not indicate a particular type.
Object[] orderedArray = new Object[listA.size()];
for(int index = 0; index < listB.size(); index ++){
int position = listB.get(index); //this may have to be cast as an int
orderedArray[position] = listA.get(index);
}
//if you receive UnsupportedOperationException when running listA.clear()
//you should replace the line with listA = new List<Object>()
//using your actual implementation of the List interface
listA.clear();
listA.addAll(orderedArray);
Just encountered the same problem.
I have a list of ordered keys, and I need to order the objects in a list according to the order of the keys.
My lists are long enough to make the solutions with time complexity of N^2 unusable.
My solution:
<K, T> List<T> sortByOrder(List<K> orderedKeys, List<T> objectsToOrder, Function<T, K> keyExtractor) {
AtomicInteger ind = new AtomicInteger(0);
Map<K, Integer> keyToIndex = orderedKeys.stream().collect(Collectors.toMap(k -> k, k -> ind.getAndIncrement(), (oldK, newK) -> oldK));
SortedMap<Integer, T> indexToObj = new TreeMap<>();
objectsToOrder.forEach(obj -> indexToObj.put(keyToIndex.get(keyExtractor.apply(obj)), obj));
return new ArrayList<>(indexToObj.values());
}
The time complexity is O(N * Log(N)).
The solution assumes that all the objects in the list to sort have distinct keys. If not then just replace SortedMap<Integer, T> indexToObj by SortedMap<Integer, List<T>> indexToObjList.
To avoid having a very inefficient look up, you should index the items in listB and then sort listA based on it.
Map<Item, Integer> index = IntStream.range(0, listB.size()).boxed()
.collect(Collectors.toMap(listB::get, x -> x));
listA.sort((e1, e2) -> Integer.compare(index.get(c1), index.get(c2));
So for me the requirement was to sort originalList with orderedList. originalList always contains all element from orderedList, but not vice versa. No new elements.
fun <T> List<T>.sort(orderedList: List<T>): List<T> {
return if (size == orderedList.size) {
orderedList
} else {
var keepIndexCount = 0
mapIndexed { index, item ->
if (orderedList.contains(item)) {
orderedList[index - keepIndexCount]
} else {
keepIndexCount++
item
}
}
}}
P.S. my case was that I have list that user can sort by drag and drop, but some items might be filtered out, so we preserve hidden items position.
If you want to do it manually. Solution based on bubble sort (same length required):
public void sortAbasedOnB(String[] listA, double[] listB) {
for (int i = 0; i < listB.length - 1; i++) {
for (int j = listB.length - 1; j > i; j--) {
if (listB[j] < listB[j - 1]){
double tempD = listB[j - 1];
listB[j - 1] = listB[j];
listB[j] = tempD;
String tempS = listA[j - 1];
listA[j - 1] = listA[j];
listA[j] = tempS;
}
}
}
}
If the object references should be the same, you can initialize listA new.
listA = new ArrayList(listB)
In Java there are set of classes which can be useful to sort lists or arrays. Most of the following examples will use lists but the same concept can be applied for arrays. A example will show this.
We can use this by creating a list of Integers and sort these using the Collections.sort(). The Collections (Java Doc) class (part of the Java Collection Framework) provides a list of static methods which we can use when working with collections such as list, set and the like. So in a nutshell, we can sort a list by simply calling: java.util.Collections.sort(the list) as shown in the following example:
import java.util.ArrayList;
import java.util.Collections;
import java.util.List;
public class example {
public static void main(String[] args) {
List<Integer> ints = new ArrayList<Integer>();
ints.add(4);
ints.add(3);
ints.add(7);
ints.add(5);
Collections.sort(ints);
System.out.println(ints);
}
}
The above class creates a list of four integers and, using the collection sort method, sorts this list (in one line of code) without us having to worry about the sorting algorithm.
I am trying to find the third largest element of an array by sorting it through treeset in a decreasing order, but some of the test cases fail for certain input values and most of it pass for certain input values.
My code:
// { Driver Code Starts
import java.util.Scanner;
import java.util.*;
import java.io.*;
class ThirdLargestElement
{
public static void main(String args[])
{
Scanner sc = new Scanner(System.in);
int t = sc.nextInt();
while(t>0)
{
long n =sc.nextLong();
Long arr[] = new Long[(int)n];
for(long i=0;i<n;i++)
arr[(int)i] = sc.nextLong();
GfG g = new GfG();
System.out.println(g.thirdLargest(arr));
t--;
}
}
}// } Driver Code Ends
class GfG
{
long thirdLargest(Long a[])
{
// Your code here
if(a.length<3)
return -1;
else{
TreeSet<Long> ts=new TreeSet<Long>(new myComparator());
for(long i:a)
ts.add(i);
ArrayList<Long> al=new ArrayList<Long>(ts);
return al.get(2);
}
}
}
class myComparator implements Comparator{
public int compare(Object obj1,Object obj2){
Long a=(Long) obj1;
Long b=(Long) obj2;
if(a<b)
return 1;
else if(a>b)
return -1;
else
return 0;
}
}
The testcase it failed:
Link to the question where you can run the code
Please explain why this code failed to pass the given test case.
Try This one
class GfG {
long thirdLargest(Long a[]) {
Arrays.sort(a);
List<Long> numbers = Arrays.asList(a);
Collections.reverse(numbers);
return numbers.size() >= 3 ? numbers.get(2) : -1;
}
}
You can directly add elements to TreeSet and get third largest number using stream,
while(t>0)
{
long n =sc.nextLong();
TreeSet<Long> ts=new TreeSet<>(Comparator.comparingLong(Long::longValue).reversed());
for(long i=0;i<n;i++)
ts.add(sc.nextLong());
long thirdLast = ts.stream()
.limit(3)
.skip(2)
.mapToLong(e->e)
.findAny().orElse(0l);
System.out.println(thirdLast);
t--;
}
The question states:
the function thirdLargest ... takes two argument. The first argument is the array a[] and the second argument is the size of the array (n).
Although the question states that the array is...
an array of distinct elements
the test cases indicate that we are dealing with an array of integers.
Personally, I don't see the need for the second method parameter, because in java an array is an object and has a length member. So my implementation below only takes one parameter, namely an array of int. Maybe the people at geeksforgeeks.org simply converted a question that was originally for the C language to java, since, in C, it is difficult to determine the size of any array.
Each element in a TreeSet must be an object, so we need to convert the elements in the int array to Integer objects. Autoboxing will do this automatically, nonetheless my code below contains an explicit conversion. So in the method I create a TreeSet. Since class Integer implements interface Comparable, the default TreeSet constructor is sufficient. I add all the elements of the int array to the TreeSet, then obtain a descending iterator and then iterate to the third element returned by the iterator, which is the value that the method needs to return.
int thirdLargest(int[] arr) {
int third = -1;
if (arr != null && arr.length > 2) {
TreeSet<Integer> set = new TreeSet<Integer>();
for (int elem : arr) {
set.add(Integer.valueOf(elem));
}
Iterator<Integer> iter = set.descendingIterator();
if (iter.hasNext()) {
iter.next();
if (iter.hasNext()) {
iter.next();
if (iter.hasNext()) {
third = iter.next().intValue();
}
}
}
}
return third;
}
Of-course, if you want to ignore the conditions imposed by the original question, you could get the third largest element using the stream API
IntStream.of(2, 4, 1, 3, 5)
.boxed()
.sorted(Collections.reverseOrder())
.collect(Collectors.toList()).get(2)
This is my situation: I have list A of values. I also have list B which contains a hierarchy of ranks. The first being of the highest, last being of the lowest. List A will contain one, some, or all of the values from list B. I want to see which value from list A is of the highest degree (or lowest index) on list B. How would I do this best?
Just in case its still unclear, this is an example:
List A: Merchant, Peasant, Queen
List B: King, Queen, Knight, Merchant, Peasant
I'd want the method to spit out Queen in this case
Assuming List B is already sorted from Top Rank -> Bottom rank, one arbitary way you could solve it is with
public static void main (String[] args) throws Exception {
String[] firstList = { "Merchant", "Peasant", "Queen" };
String[] secondList = { "King", "Queen", "Knight", "Merchant", "Peasant" };
for (String highRank : secondList) {
for (String lowRank : firstList) {
if (highRank.equalsIgnoreCase(lowRank)) {
System.out.println(highRank);
return;
}
}
}
}
What you are describing is called a "partial ordering", and the proper way to implement the behavior you're looking for in Java is with a Comparator that defines the ordering; something like:
public class PartialOrdering<T> implements Comparator<T> {
private final Map<T, Integer> listPositions = new HashMap<>();
public PartialOrdering(List<T> elements) {
for (int i = 0; i < elements.size(); i++) {
listPositions.put(elements.get(i), i);
}
}
public int compare(T a, T b) {
Integer aPos = listPositions.get(a);
Integer bPos = listPositions.get(b);
if (aPos == null || bPos == null) {
throw new IllegalArgumentException(
"PartialOrdering can only compare elements it's aware of.");
}
return Integer.compare(aPos, bPos);
}
}
You can then simply call Collections.max() to find the largest value in your first list.
This is much more efficient than either of the other answers, which are both O(n^2) and don't handle unknown elements coherently (they assume we have a total ordering).
Even better than implementing your own PartialOrdering, however, is to use Guava's Ordering class, which provides an efficient partial ordering and a number of other useful tools. With Guava all you need to do is:
// Or store the result of Ordering.explicit() if you need to reuse it
Ordering.explicit(listB).max(listA);
I think this might work give it a Try:
function int getHighest(List<String> listA, List<String> listB)
{
int index = 0;
int max = 100;
int tmpMax = 0;
for(String test:lista)
{
for(int i =0;i<listb.size();++i)
{
if(list.get(i).equals(test))
{
tmpMax = index;
}
}
if(tmpMax < max) max = tmpMax;
++index;
}
return max;
}
I am building a data structure to learn more about java. I understand this program might be useless.
Here's what I want. I want to create a data structure that store smallest 3 values. if value is high, then ignore it. When storing values than I also want to put them in correct place so I don't have to sort them later. I can enter values by calling the add method.
so let's say I want to add 20, 10, 40, 30 than the result will be [10,20,30]. note I can only hold 3 smallest values and it store them as I place them.
I also understand that there are a lot of better ways for doing this but again this is just for learning purposes.
Question: I need help creating add method. I wrote some code but I am getting stuck with add method. Please help.
My Thinking: we might have to use a Iterator in add method?
public class MyJavaApp {
public static void main(String[] args){
MyClass<Integer> m = new MyClass<Integer>(3);
m.add(10);
m.add(20);
m.add(30);
m.add(40);
}
}
public class MyClass<V extends Comparable<V>> {
private V v[];
public MyClass(int s){
this.v = (V[])new Object[s];
}
public void add(V a){
}
}
Here is a rough sketch of the add method you have to implement.
You have to use the appropriate implementation of the compareTo method when comparing elements.
public void add(V a){
V temp = null;
if(a.compareTo( v[0]) == -1 ){
/*
keeping the v[0] in a temp variable since, v[0] could be the second
smallest value or the third smallest value.
Therefore call add method again to assign it to the correct
position.
*/
temp = v[0];
v[0] = a;
add(temp);
}else if(a.compareTo(v[0]) == 1 && a.compareTo(v[1]) == -1){
temp = v[1];
v[1] = a;
add(temp);
}else if(a.compareTo(v[1]) == 1 && a.compareTo(v[2]) == -1){
temp = v[2];
v[2] = a;
add(temp);
}
}
Therefore the v array will contain the lowerest elements.
Hope this helps.
A naive, inefficient approach would be (as you suggest) to iterate through the values and add / remove based on what you find:
public void add(Integer a)
{
// If fewer than 3 elements in the list, add and we're done.
if (m.size() < 3)
{
m.add(a);
return;
}
// If there's 3 elements, find the maximum.
int max = Integer.MIN_VALUE;
int index = -1;
for (int i=0; i<3; i++) {
int v = m.get(i);
if (v > max) {
max = v;
index = i;
}
}
// If a is less than the max, we need to add it and remove the existing max.
if (a < max) {
m.remove(index);
m.add(a);
}
}
Note: this has been written for Integer, not a generic type V. You'll need to generalise. It also doesn't keep the list sorted - another of your requirements.
Here's an implementation of that algorithm. It consists of looking for the right place to insert. Then it can be optimized for your requirements:
Don't bother looking past the size you want
Don't add more items than necessary
Here's the code. I added the toString() method for convenience. Only the add() method is interesting. Also this implementation is a bit more flexible as it respects the size you give to the constructor and doesn't assume 3.
I used a List rather than an array because it makes dealing with generics a lot easier. You'll find that using an array of generics makes using your class a bit more ugly (i.e. you have to deal with type erasure by providing a Class<V>).
import java.util.*;
public class MyClass<V extends Comparable<V>> {
private int s;
private List<V> v;
public MyClass(int s) {
this.s = s;
this.v = new ArrayList<V>(s);
}
public void add(V a) {
int i=0;
int l = v.size();
// Find the right index
while(i<l && v.get(i).compareTo(a) < 0) i++;
if(i<s) {
v.add(i, a);
// Truncate the list to make sure we don't store more values than needed
if(v.size() > s) v.remove(v.size()-1);
}
}
public String toString() {
StringBuilder result = new StringBuilder();
for(V item : v) {
result.append(item).append(',');
}
return result.toString();
}
}
I have defined my own compare function for a priority queue, however the compare function needs information of an array. The problem is that when the values of the array changed, it did not affect the compare function. How do I deal with this?
Code example:
import java.util.Arrays;
import java.util.Comparator;
import java.util.PriorityQueue;
import java.util.Scanner;
public class Main {
public static final int INF = 100;
public static int[] F = new int[201];
public static void main(String[] args){
PriorityQueue<Integer> Q = new PriorityQueue<Integer>(201,
new Comparator<Integer>(){
public int compare(Integer a, Integer b){
if (F[a] > F[b]) return 1;
if (F[a] == F[b]) return 0;
return -1;
}
});
Arrays.fill(F, INF);
F[0] = 0; F[1] = 1; F[2] = 2;
for (int i = 0; i < 201; i ++) Q.add(i);
System.out.println(Q.peek()); // Prints 0, because F[0] is the smallest
F[0] = 10;
System.out.println(Q.peek()); // Still prints 0 ... OMG
}
}
So, essentially, you are changing your comparison criteria on the fly, and that's just not the functionality that priority queue contracts offer. Note that this might seem to work on some cases (e.g. a heap might sort some of the items when removing or inserting another item) but since you have no guarantees, it's just not a valid approach.
What you could do is, every time you change your arrays, you get all the elements out, and put them back in. This is of course very expensive ( O(n*log(n))) so you should probably try to work around your design to avoid changing the array values at all.
Your comparator is only getting called when you modify the queue (that is, when you add your items). After that, the queue has no idea something caused the order to change, which is why it remains the same.
It is quite confusing to have a comparator like this. If you have two values, A and B, and A>B at some point, everybody would expect A to stay bigger than B. I think your usage of a priority queue for this problem is wrong.
Use custom implementation of PriorityQueue that uses comparator on peek, not on add:
public class VolatilePriorityQueue <T> extends AbstractQueue <T>
{
private final Comparator <? super T> comparator;
private final List <T> elements = new ArrayList <T> ();
public VolatilePriorityQueue (Comparator <? super T> comparator)
{
this.comparator = comparator;
}
#Override
public boolean offer (T e)
{
return elements.add (e);
}
#Override
public T poll ()
{
if (elements.isEmpty ()) return null;
else return elements.remove (getMinimumIndex ());
}
#Override
public T peek ()
{
if (elements.isEmpty ()) return null;
else return elements.get (getMinimumIndex ());
}
#Override
public Iterator <T> iterator ()
{
return elements.iterator ();
}
#Override
public int size ()
{
return elements.size ();
}
private int getMinimumIndex ()
{
T e = elements.get (0);
int index = 0;
for (int count = elements.size (), i = 1; i < count; i++)
{
T ee = elements.get (i);
if (comparator.compare (e, ee) > 0)
{
e = ee;
index = i;
}
}
return index;
}
}