What's the simplest, library-free code for implementing array intersections in javascript? I want to write
intersection([1,2,3], [2,3,4,5])
and get
[2, 3]
Use a combination of Array.prototype.filter and Array.prototype.includes:
const filteredArray = array1.filter(value => array2.includes(value));
For older browsers, with Array.prototype.indexOf and without an arrow function:
var filteredArray = array1.filter(function(n) {
return array2.indexOf(n) !== -1;
});
NB! Both .includes and .indexOf internally compares elements in the array by using ===, so if the array contains objects it will only compare object references (not their content). If you want to specify your own comparison logic, use Array.prototype.some instead.
Destructive seems simplest, especially if we can assume the input is sorted:
/* destructively finds the intersection of
* two arrays in a simple fashion.
*
* PARAMS
* a - first array, must already be sorted
* b - second array, must already be sorted
*
* NOTES
* State of input arrays is undefined when
* the function returns. They should be
* (prolly) be dumped.
*
* Should have O(n) operations, where n is
* n = MIN(a.length, b.length)
*/
function intersection_destructive(a, b)
{
var result = [];
while( a.length > 0 && b.length > 0 )
{
if (a[0] < b[0] ){ a.shift(); }
else if (a[0] > b[0] ){ b.shift(); }
else /* they're equal */
{
result.push(a.shift());
b.shift();
}
}
return result;
}
Non-destructive has to be a hair more complicated, since we’ve got to track indices:
/* finds the intersection of
* two arrays in a simple fashion.
*
* PARAMS
* a - first array, must already be sorted
* b - second array, must already be sorted
*
* NOTES
*
* Should have O(n) operations, where n is
* n = MIN(a.length(), b.length())
*/
function intersect_safe(a, b)
{
var ai=0, bi=0;
var result = [];
while( ai < a.length && bi < b.length )
{
if (a[ai] < b[bi] ){ ai++; }
else if (a[ai] > b[bi] ){ bi++; }
else /* they're equal */
{
result.push(a[ai]);
ai++;
bi++;
}
}
return result;
}
If your environment supports ECMAScript 6 Set, one simple and supposedly efficient (see specification link) way:
function intersect(a, b) {
var setA = new Set(a);
var setB = new Set(b);
var intersection = new Set([...setA].filter(x => setB.has(x)));
return Array.from(intersection);
}
Shorter, but less readable (also without creating the additional intersection Set):
function intersect(a, b) {
var setB = new Set(b);
return [...new Set(a)].filter(x => setB.has(x));
}
Note that when using sets you will only get distinct values, thus new Set([1, 2, 3, 3]).size evaluates to 3.
Using Underscore.js or lodash.js
_.intersection( [0,345,324] , [1,0,324] ) // gives [0,324]
// Return elements of array a that are also in b in linear time:
function intersect(a, b) {
return a.filter(Set.prototype.has, new Set(b));
}
// Example:
console.log(intersect([1,2,3], [2,3,4,5]));
I recommend above succinct solution which outperforms other implementations on large inputs. If performance on small inputs matters, check the alternatives below.
Alternatives and performance comparison:
See the following snippet for alternative implementations and check https://jsperf.com/array-intersection-comparison for performance comparisons.
function intersect_for(a, b) {
const result = [];
const alen = a.length;
const blen = b.length;
for (let i = 0; i < alen; ++i) {
const ai = a[i];
for (let j = 0; j < blen; ++j) {
if (ai === b[j]) {
result.push(ai);
break;
}
}
}
return result;
}
function intersect_filter_indexOf(a, b) {
return a.filter(el => b.indexOf(el) !== -1);
}
function intersect_filter_in(a, b) {
const map = b.reduce((map, el) => {map[el] = true; return map}, {});
return a.filter(el => el in map);
}
function intersect_for_in(a, b) {
const result = [];
const map = {};
for (let i = 0, length = b.length; i < length; ++i) {
map[b[i]] = true;
}
for (let i = 0, length = a.length; i < length; ++i) {
if (a[i] in map) result.push(a[i]);
}
return result;
}
function intersect_filter_includes(a, b) {
return a.filter(el => b.includes(el));
}
function intersect_filter_has_this(a, b) {
return a.filter(Set.prototype.has, new Set(b));
}
function intersect_filter_has_arrow(a, b) {
const set = new Set(b);
return a.filter(el => set.has(el));
}
function intersect_for_has(a, b) {
const result = [];
const set = new Set(b);
for (let i = 0, length = a.length; i < length; ++i) {
if (set.has(a[i])) result.push(a[i]);
}
return result;
}
Results in Firefox 53:
Ops/sec on large arrays (10,000 elements):
filter + has (this) 523 (this answer)
for + has 482
for-loop + in 279
filter + in 242
for-loops 24
filter + includes 14
filter + indexOf 10
Ops/sec on small arrays (100 elements):
for-loop + in 384,426
filter + in 192,066
for-loops 159,137
filter + includes 104,068
filter + indexOf 71,598
filter + has (this) 43,531 (this answer)
filter + has (arrow function) 35,588
My contribution in ES6 terms. In general it finds the intersection of an array with indefinite number of arrays provided as arguments.
Array.prototype.intersect = function(...a) {
return [this,...a].reduce((p,c) => p.filter(e => c.includes(e)));
}
var arrs = [[0,2,4,6,8],[4,5,6,7],[4,6]],
arr = [0,1,2,3,4,5,6,7,8,9];
document.write("<pre>" + JSON.stringify(arr.intersect(...arrs)) + "</pre>");
How about just using associative arrays?
function intersect(a, b) {
var d1 = {};
var d2 = {};
var results = [];
for (var i = 0; i < a.length; i++) {
d1[a[i]] = true;
}
for (var j = 0; j < b.length; j++) {
d2[b[j]] = true;
}
for (var k in d1) {
if (d2[k])
results.push(k);
}
return results;
}
edit:
// new version
function intersect(a, b) {
var d = {};
var results = [];
for (var i = 0; i < b.length; i++) {
d[b[i]] = true;
}
for (var j = 0; j < a.length; j++) {
if (d[a[j]])
results.push(a[j]);
}
return results;
}
The performance of #atk's implementation for sorted arrays of primitives can be improved by using .pop rather than .shift.
function intersect(array1, array2) {
var result = [];
// Don't destroy the original arrays
var a = array1.slice(0);
var b = array2.slice(0);
var aLast = a.length - 1;
var bLast = b.length - 1;
while (aLast >= 0 && bLast >= 0) {
if (a[aLast] > b[bLast] ) {
a.pop();
aLast--;
} else if (a[aLast] < b[bLast] ){
b.pop();
bLast--;
} else /* they're equal */ {
result.push(a.pop());
b.pop();
aLast--;
bLast--;
}
}
return result;
}
I created a benchmark using jsPerf. It's about three times faster to use .pop.
If you need to have it handle intersecting multiple arrays:
const intersect = (a1, a2, ...rest) => {
const a12 = a1.filter(value => a2.includes(value))
if (rest.length === 0) { return a12; }
return intersect(a12, ...rest);
};
console.log(intersect([1,2,3,4,5], [1,2], [1, 2, 3,4,5], [2, 10, 1]))
Sort it
check one by one from the index 0, create new array from that.
Something like this, Not tested well though.
function intersection(x,y){
x.sort();y.sort();
var i=j=0;ret=[];
while(i<x.length && j<y.length){
if(x[i]<y[j])i++;
else if(y[j]<x[i])j++;
else {
ret.push(x[i]);
i++,j++;
}
}
return ret;
}
alert(intersection([1,2,3], [2,3,4,5]));
PS:The algorithm only intended for Numbers and Normal Strings, intersection of arbitary object arrays may not work.
Using jQuery:
var a = [1,2,3];
var b = [2,3,4,5];
var c = $(b).not($(b).not(a));
alert(c);
A tiny tweak to the smallest one here (the filter/indexOf solution), namely creating an index of the values in one of the arrays using a JavaScript object, will reduce it from O(N*M) to "probably" linear time. source1 source2
function intersect(a, b) {
var aa = {};
a.forEach(function(v) { aa[v]=1; });
return b.filter(function(v) { return v in aa; });
}
This isn't the very simplest solution (it's more code than filter+indexOf), nor is it the very fastest (probably slower by a constant factor than intersect_safe()), but seems like a pretty good balance. It is on the very simple side, while providing good performance, and it doesn't require pre-sorted inputs.
For arrays containing only strings or numbers you can do something with sorting, as per some of the other answers. For the general case of arrays of arbitrary objects I don't think you can avoid doing it the long way. The following will give you the intersection of any number of arrays provided as parameters to arrayIntersection:
var arrayContains = Array.prototype.indexOf ?
function(arr, val) {
return arr.indexOf(val) > -1;
} :
function(arr, val) {
var i = arr.length;
while (i--) {
if (arr[i] === val) {
return true;
}
}
return false;
};
function arrayIntersection() {
var val, arrayCount, firstArray, i, j, intersection = [], missing;
var arrays = Array.prototype.slice.call(arguments); // Convert arguments into a real array
// Search for common values
firstArray = arrays.pop();
if (firstArray) {
j = firstArray.length;
arrayCount = arrays.length;
while (j--) {
val = firstArray[j];
missing = false;
// Check val is present in each remaining array
i = arrayCount;
while (!missing && i--) {
if ( !arrayContains(arrays[i], val) ) {
missing = true;
}
}
if (!missing) {
intersection.push(val);
}
}
}
return intersection;
}
arrayIntersection( [1, 2, 3, "a"], [1, "a", 2], ["a", 1] ); // Gives [1, "a"];
Simplest, fastest O(n) and shortest way:
function intersection (a, b) {
const setA = new Set(a);
return b.filter(value => setA.has(value));
}
console.log(intersection([1,2,3], [2,3,4,5]))
#nbarbosa has almost the same answer but he cast both arrays to Set and then back to array. There is no need for any extra casting.
Another indexed approach able to process any number of arrays at once:
// Calculate intersection of multiple array or object values.
function intersect (arrList) {
var arrLength = Object.keys(arrList).length;
// (Also accepts regular objects as input)
var index = {};
for (var i in arrList) {
for (var j in arrList[i]) {
var v = arrList[i][j];
if (index[v] === undefined) index[v] = 0;
index[v]++;
};
};
var retv = [];
for (var i in index) {
if (index[i] == arrLength) retv.push(i);
};
return retv;
};
It works only for values that can be evaluated as strings and you should pass them as an array like:
intersect ([arr1, arr2, arr3...]);
...but it transparently accepts objects as parameter or as any of the elements to be intersected (always returning array of common values). Examples:
intersect ({foo: [1, 2, 3, 4], bar: {a: 2, j:4}}); // [2, 4]
intersect ([{x: "hello", y: "world"}, ["hello", "user"]]); // ["hello"]
EDIT: I just noticed that this is, in a way, slightly buggy.
That is: I coded it thinking that input arrays cannot itself contain repetitions (as provided example doesn't).
But if input arrays happen to contain repetitions, that would produce wrong results. Example (using below implementation):
intersect ([[1, 3, 4, 6, 3], [1, 8, 99]]);
// Expected: [ '1' ]
// Actual: [ '1', '3' ]
Fortunately this is easy to fix by simply adding second level indexing. That is:
Change:
if (index[v] === undefined) index[v] = 0;
index[v]++;
by:
if (index[v] === undefined) index[v] = {};
index[v][i] = true; // Mark as present in i input.
...and:
if (index[i] == arrLength) retv.push(i);
by:
if (Object.keys(index[i]).length == arrLength) retv.push(i);
Complete example:
// Calculate intersection of multiple array or object values.
function intersect (arrList) {
var arrLength = Object.keys(arrList).length;
// (Also accepts regular objects as input)
var index = {};
for (var i in arrList) {
for (var j in arrList[i]) {
var v = arrList[i][j];
if (index[v] === undefined) index[v] = {};
index[v][i] = true; // Mark as present in i input.
};
};
var retv = [];
for (var i in index) {
if (Object.keys(index[i]).length == arrLength) retv.push(i);
};
return retv;
};
intersect ([[1, 3, 4, 6, 3], [1, 8, 99]]); // [ '1' ]
With some restrictions on your data, you can do it in linear time!
For positive integers: use an array mapping the values to a "seen/not seen" boolean.
function intersectIntegers(array1,array2) {
var seen=[],
result=[];
for (var i = 0; i < array1.length; i++) {
seen[array1[i]] = true;
}
for (var i = 0; i < array2.length; i++) {
if ( seen[array2[i]])
result.push(array2[i]);
}
return result;
}
There is a similar technique for objects: take a dummy key, set it to "true" for each element in array1, then look for this key in elements of array2. Clean up when you're done.
function intersectObjects(array1,array2) {
var result=[];
var key="tmpKey_intersect"
for (var i = 0; i < array1.length; i++) {
array1[i][key] = true;
}
for (var i = 0; i < array2.length; i++) {
if (array2[i][key])
result.push(array2[i]);
}
for (var i = 0; i < array1.length; i++) {
delete array1[i][key];
}
return result;
}
Of course you need to be sure the key didn't appear before, otherwise you'll be destroying your data...
function intersection(A,B){
var result = new Array();
for (i=0; i<A.length; i++) {
for (j=0; j<B.length; j++) {
if (A[i] == B[j] && $.inArray(A[i],result) == -1) {
result.push(A[i]);
}
}
}
return result;
}
For simplicity:
// Usage
const intersection = allLists
.reduce(intersect, allValues)
.reduce(removeDuplicates, []);
// Implementation
const intersect = (intersection, list) =>
intersection.filter(item =>
list.some(x => x === item));
const removeDuplicates = (uniques, item) =>
uniques.includes(item) ? uniques : uniques.concat(item);
// Example Data
const somePeople = [bob, doug, jill];
const otherPeople = [sarah, bob, jill];
const morePeople = [jack, jill];
const allPeople = [...somePeople, ...otherPeople, ...morePeople];
const allGroups = [somePeople, otherPeople, morePeople];
// Example Usage
const intersection = allGroups
.reduce(intersect, allPeople)
.reduce(removeDuplicates, []);
intersection; // [jill]
Benefits:
dirt simple
data-centric
works for arbitrary number of lists
works for arbitrary lengths of lists
works for arbitrary types of values
works for arbitrary sort order
retains shape (order of first appearance in any array)
exits early where possible
memory safe, short of tampering with Function / Array prototypes
Drawbacks:
higher memory usage
higher CPU usage
requires an understanding of reduce
requires understanding of data flow
You wouldn't want to use this for 3D engine or kernel work, but if you have problems getting this to run in an event-based app, your design has bigger problems.
I'll contribute with what has been working out best for me:
if (!Array.prototype.intersect){
Array.prototype.intersect = function (arr1) {
var r = [], o = {}, l = this.length, i, v;
for (i = 0; i < l; i++) {
o[this[i]] = true;
}
l = arr1.length;
for (i = 0; i < l; i++) {
v = arr1[i];
if (v in o) {
r.push(v);
}
}
return r;
};
}
A functional approach with ES2015
A functional approach must consider using only pure functions without side effects, each of which is only concerned with a single job.
These restrictions enhance the composability and reusability of the functions involved.
// small, reusable auxiliary functions
const createSet = xs => new Set(xs);
const filter = f => xs => xs.filter(apply(f));
const apply = f => x => f(x);
// intersection
const intersect = xs => ys => {
const zs = createSet(ys);
return filter(x => zs.has(x)
? true
: false
) (xs);
};
// mock data
const xs = [1,2,2,3,4,5];
const ys = [0,1,2,3,3,3,6,7,8,9];
// run it
console.log( intersect(xs) (ys) );
Please note that the native Set type is used, which has an advantageous
lookup performance.
Avoid duplicates
Obviously repeatedly occurring items from the first Array are preserved, while the second Array is de-duplicated. This may be or may be not the desired behavior. If you need a unique result just apply dedupe to the first argument:
// auxiliary functions
const apply = f => x => f(x);
const comp = f => g => x => f(g(x));
const afrom = apply(Array.from);
const createSet = xs => new Set(xs);
const filter = f => xs => xs.filter(apply(f));
// intersection
const intersect = xs => ys => {
const zs = createSet(ys);
return filter(x => zs.has(x)
? true
: false
) (xs);
};
// de-duplication
const dedupe = comp(afrom) (createSet);
// mock data
const xs = [1,2,2,3,4,5];
const ys = [0,1,2,3,3,3,6,7,8,9];
// unique result
console.log( intersect(dedupe(xs)) (ys) );
Compute the intersection of any number of Arrays
If you want to compute the intersection of an arbitrarily number of Arrays just compose intersect with foldl. Here is a convenience function:
// auxiliary functions
const apply = f => x => f(x);
const uncurry = f => (x, y) => f(x) (y);
const createSet = xs => new Set(xs);
const filter = f => xs => xs.filter(apply(f));
const foldl = f => acc => xs => xs.reduce(uncurry(f), acc);
// intersection
const intersect = xs => ys => {
const zs = createSet(ys);
return filter(x => zs.has(x)
? true
: false
) (xs);
};
// intersection of an arbitrarily number of Arrays
const intersectn = (head, ...tail) => foldl(intersect) (head) (tail);
// mock data
const xs = [1,2,2,3,4,5];
const ys = [0,1,2,3,3,3,6,7,8,9];
const zs = [0,1,2,3,4,5,6];
// run
console.log( intersectn(xs, ys, zs) );
.reduce to build a map, and .filter to find the intersection. delete within the .filter allows us to treat the second array as though it's a unique set.
function intersection (a, b) {
var seen = a.reduce(function (h, k) {
h[k] = true;
return h;
}, {});
return b.filter(function (k) {
var exists = seen[k];
delete seen[k];
return exists;
});
}
I find this approach pretty easy to reason about. It performs in constant time.
I have written an intesection function which can even detect intersection of array of objects based on particular property of those objects.
For instance,
if arr1 = [{id: 10}, {id: 20}]
and arr2 = [{id: 20}, {id: 25}]
and we want intersection based on the id property, then the output should be :
[{id: 20}]
As such, the function for the same (note: ES6 code) is :
const intersect = (arr1, arr2, accessors = [v => v, v => v]) => {
const [fn1, fn2] = accessors;
const set = new Set(arr2.map(v => fn2(v)));
return arr1.filter(value => set.has(fn1(value)));
};
and you can call the function as:
intersect(arr1, arr2, [elem => elem.id, elem => elem.id])
Also note: this function finds intersection considering the first array is the primary array and thus the intersection result will be that of the primary array.
This function avoids the N^2 problem, taking advantage of the power of dictionaries. Loops through each array only once, and a third and shorter loop to return the final result.
It also supports numbers, strings, and objects.
function array_intersect(array1, array2)
{
var mergedElems = {},
result = [];
// Returns a unique reference string for the type and value of the element
function generateStrKey(elem) {
var typeOfElem = typeof elem;
if (typeOfElem === 'object') {
typeOfElem += Object.prototype.toString.call(elem);
}
return [typeOfElem, elem.toString(), JSON.stringify(elem)].join('__');
}
array1.forEach(function(elem) {
var key = generateStrKey(elem);
if (!(key in mergedElems)) {
mergedElems[key] = {elem: elem, inArray2: false};
}
});
array2.forEach(function(elem) {
var key = generateStrKey(elem);
if (key in mergedElems) {
mergedElems[key].inArray2 = true;
}
});
Object.values(mergedElems).forEach(function(elem) {
if (elem.inArray2) {
result.push(elem.elem);
}
});
return result;
}
If there is a special case that cannot be solved, just by modifying the generateStrKey function, it could surely be solved. The trick of this function is that it uniquely represents each different data according to type and value.
This variant has some performance improvements. Avoid loops in case any array is empty. It also starts by walking through the shorter array first, so if it finds all the values of the first array in the second array, exits the loop.
function array_intersect(array1, array2)
{
var mergedElems = {},
result = [],
firstArray, secondArray,
firstN = 0,
secondN = 0;
function generateStrKey(elem) {
var typeOfElem = typeof elem;
if (typeOfElem === 'object') {
typeOfElem += Object.prototype.toString.call(elem);
}
return [typeOfElem, elem.toString(), JSON.stringify(elem)].join('__');
}
// Executes the loops only if both arrays have values
if (array1.length && array2.length)
{
// Begins with the shortest array to optimize the algorithm
if (array1.length < array2.length) {
firstArray = array1;
secondArray = array2;
} else {
firstArray = array2;
secondArray = array1;
}
firstArray.forEach(function(elem) {
var key = generateStrKey(elem);
if (!(key in mergedElems)) {
mergedElems[key] = {elem: elem, inArray2: false};
// Increases the counter of unique values in the first array
firstN++;
}
});
secondArray.some(function(elem) {
var key = generateStrKey(elem);
if (key in mergedElems) {
if (!mergedElems[key].inArray2) {
mergedElems[key].inArray2 = true;
// Increases the counter of matches
secondN++;
// If all elements of first array have coincidence, then exits the loop
return (secondN === firstN);
}
}
});
Object.values(mergedElems).forEach(function(elem) {
if (elem.inArray2) {
result.push(elem.elem);
}
});
}
return result;
}
Here is underscore.js implementation:
_.intersection = function(array) {
if (array == null) return [];
var result = [];
var argsLength = arguments.length;
for (var i = 0, length = array.length; i < length; i++) {
var item = array[i];
if (_.contains(result, item)) continue;
for (var j = 1; j < argsLength; j++) {
if (!_.contains(arguments[j], item)) break;
}
if (j === argsLength) result.push(item);
}
return result;
};
Source: http://underscorejs.org/docs/underscore.html#section-62
Create an Object using one array and loop through the second array to check if the value exists as key.
function intersection(arr1, arr2) {
var myObj = {};
var myArr = [];
for (var i = 0, len = arr1.length; i < len; i += 1) {
if(myObj[arr1[i]]) {
myObj[arr1[i]] += 1;
} else {
myObj[arr1[i]] = 1;
}
}
for (var j = 0, len = arr2.length; j < len; j += 1) {
if(myObj[arr2[j]] && myArr.indexOf(arr2[j]) === -1) {
myArr.push(arr2[j]);
}
}
return myArr;
}
I think using an object internally can help with computations and could be performant too.
// Approach maintains a count of each element and works for negative elements too
function intersect(a,b){
const A = {};
a.forEach((v)=>{A[v] ? ++A[v] : A[v] = 1});
const B = {};
b.forEach((v)=>{B[v] ? ++B[v] : B[v] = 1});
const C = {};
Object.entries(A).map((x)=>C[x[0]] = Math.min(x[1],B[x[0]]))
return Object.entries(C).map((x)=>Array(x[1]).fill(Number(x[0]))).flat();
}
const x = [1,1,-1,-1,0,0,2,2];
const y = [2,0,1,1,1,1,0,-1,-1,-1];
const result = intersect(x,y);
console.log(result); // (7) [0, 0, 1, 1, 2, -1, -1]
I am using map even object could be used.
//find intersection of 2 arrs
const intersections = (arr1,arr2) => {
let arrf = arr1.concat(arr2)
let map = new Map();
let union = [];
for(let i=0; i<arrf.length; i++){
if(map.get(arrf[i])){
map.set(arrf[i],false);
}else{
map.set(arrf[i],true);
}
}
map.forEach((v,k)=>{if(!v){union.push(k);}})
return union;
}
This is a proposed standard: With the currently stage 2 proposal https://github.com/tc39/proposal-set-methods, you could use
mySet.intersection(mySet2);
Until then, you could use Immutable.js's Set, which inspired that proposal
Immutable.Set(mySet).intersect(mySet2)
I extended tarulen's answer to work with any number of arrays. It also should work with non-integer values.
function intersect() {
const last = arguments.length - 1;
var seen={};
var result=[];
for (var i = 0; i < last; i++) {
for (var j = 0; j < arguments[i].length; j++) {
if (seen[arguments[i][j]]) {
seen[arguments[i][j]] += 1;
}
else if (!i) {
seen[arguments[i][j]] = 1;
}
}
}
for (var i = 0; i < arguments[last].length; i++) {
if ( seen[arguments[last][i]] === last)
result.push(arguments[last][i]);
}
return result;
}
If your arrays are sorted, this should run in O(n), where n is min( a.length, b.length )
function intersect_1d( a, b ){
var out=[], ai=0, bi=0, acurr, bcurr, last=Number.MIN_SAFE_INTEGER;
while( ( acurr=a[ai] )!==undefined && ( bcurr=b[bi] )!==undefined ){
if( acurr < bcurr){
if( last===acurr ){
out.push( acurr );
}
last=acurr;
ai++;
}
else if( acurr > bcurr){
if( last===bcurr ){
out.push( bcurr );
}
last=bcurr;
bi++;
}
else {
out.push( acurr );
last=acurr;
ai++;
bi++;
}
}
return out;
}
Related
Imagine two arrays:
const array1 = [1,2,3];
const array2 = [2,3,4];
Now, I want to get all the differences of these two arrays and put them in two new arrays.
One Array will be for all the items that where missing in the first.
The other will be for the items missing in the second.
The result would look something like this:
const newArray1 = [1];
const newArray2 = [4];
How would I go about this and what is the most efficient way?
const array1 = [2,1,3,5,2,1,3,5];
const array2 = [4,3,2,6,7,4,3,2,6,7];
function diff(arr1, arr2) {
const dontAddDuplicates = true;
arr1.sort();
arr2.sort();
let a1 = [];
let a2 = [];
let i = 0;
let j = 0;
while (i < array1.length || j < array2.length) {
if (i >= arr1.length) {
if (!dontAddDuplicates || (a2.length == 0 || a2[a2.length - 1] != arr2[j])) {
a2.push(arr2[j]);
}
j++;
} else if (j >= array2.length) {
if (!dontAddDuplicates || (a1.length == 0 || a1[a1.length - 1] != arr1[i])) {
a1.push(arr1[i]);
}
i++;
} else if (arr1[i] < arr2[j]) {
if (!dontAddDuplicates || (a1.length == 0 || a1[a1.length - 1] != arr1[i])) {
a1.push(arr1[i]);
}
i++;
} else if (arr2[j] < arr1[i]) {
if (!dontAddDuplicates || (a2.length == 0 || a2[a2.length - 1] != arr2[j])) {
a2.push(arr2[j]);
}
j++;
} else {
// Same value, do nothing
i++;
j++;
}
}
return [a1, a2];
}
console.log(diff(array1, array2));
// OUTPUT: [[1, 5], [4, 6, 7]]
Here's another potential implementation using sorting, but it has the side effect of leaving array1 and array2 in a sorted fashion. Sorting allows you to avoid needing to rescan the other array every time. If they are already sorted then great you can skip this step. If the side effect is a problem, then use a deep copy of array1 and array2 before calling sort.
Flip dontAddDuplicates if you want duplicates or not. I notice the other implementations don't account for that, but easy enough to add.
Run time should be: SORT N + SORT M + N + M = SORT N = N LOG N depending on your input sizes and distributions SORT will be the significant O Notation https://www.bigocheatsheet.com/
https://jsfiddle.net/buscgtL2/1/
If you want to do it in N + M + N + M = N time you can use this implementation which uses a hash map instead of sorting. This has a disadvantage on memory space.
const array1 = [2,1,3,5,2,1,3,5];
const array2 = [4,3,2,6,7,4,3,2,6,7];
function diff(arr1, arr2) {
let dontAddDuplicates = true;
let a1 = [];
let a2 = [];
let a1hash = {};
let a2hash = {};
for (let i = 0; i < arr1.length; i++) {
a1hash[arr1[i]] = 0;
}
for (let i = 0; i < arr2.length; i++) {
a2hash[arr2[i]] = 0;
}
for (let i = 0; i < arr1.length; i++) {
if (!a2hash.hasOwnProperty(arr1[i])) {
if (!dontAddDuplicates || a1hash[arr1[i]] == 0) {
a1hash[arr1[i]] = 1;
a1.push(arr1[i]);
}
}
}
for (let i = 0; i < arr2.length; i++) {
if (!a1hash.hasOwnProperty(arr2[i])) {
if (!dontAddDuplicates || a2hash[arr2[i]] == 0) {
a2hash[arr2[i]] = 1;
a2.push(arr2[i]);
}
}
}
return [a1, a2];
}
console.log(diff(array1, array2));
//OUTPUT: [[1, 5], [4, 6, 7]]
https://jsfiddle.net/2945y3an/1/
The worse performance would be any algorithm where for every N element you scan array M searching for a match. This would be N * M = N^2
You can achieve it in very simple way with minimum line of codes by using Array.filter() along with Array.includes() methods of JavaScript.
Working Demo :
const array1 = [1,2,3];
const array2 = [2,3,4];
const updatedArray1 = array1.filter(item => !array2.includes(item));
const updatedArray2 = array2.filter(item => !array1.includes(item));
console.log(updatedArray1); // [1]
console.log(updatedArray2); // [4]
Something like this would be relatively efficient, you're only iterating through two arrays.
Iterate through the first array and add any missing items to the first newArray. Repeat the same process for the second array.
If you need to only loop through the array once, you would need to do something similar to what is done below, but only loop through the longest array.
const array1 = [1,2,3];
const array2 = [2,3];
function diff (arr1, arr2) {
const newA1 = [];
const newA2 = [];
arr1.forEach((item, i) => {
let index = arr2.findIndex(it => it === item);
if(index < 0) newA1.push(item);
});
arr2.forEach((item, i) => {
let index = arr1.findIndex(it => it === item);
if(index < 0) newA2.push(item);
});
return [newA1, newA2];
}
console.log(diff(array1, array2));
More efficent method (only loop through 1 array). This way you choose the longest array, loop through it, and check for duplicates of the current item of the long array and if an item in the secondary array exists at the same position, check for duplicates of this item in the longest array as well. This method is similar to the method above, but there is only 1 for each loop.
const array1 = [1,2,3];
const array2 = [3,4,5];
function diff (arr1, arr2) {
const newA1 = [];
const newA2 = [];
let baseArr, secondaryArr;
if(arr1.length > arr2.length) {
baseArr = arr1;
secondaryArr = arr2;
} else {
baseArr = arr2;
secondaryArr = arr1;
}
baseArr.forEach((item, i) => {
const secondaryArrI = secondaryArr.findIndex(it => it === item);
if(secondaryArrI < 0) newA1.push(item)
if(typeof secondaryArr[i] !== "undefined") {
const removeI = baseArr.findIndex(it => it === secondaryArr[i]);
if(removeI < 0) newA2.push(secondaryArr[i]);
}
})
return [newA1, newA2];
}
console.log(diff(array1, array2));
Lodash difference if you dot't mind
const array1 = [1,2,3];
const array2 = [2,3,4];
console.log(_.difference(array1, array2));
console.log(_.difference(array2, array1));
.as-console-wrapper{min-height: 100%!important; top: 0}
<script src="https://cdn.jsdelivr.net/npm/lodash#4.17.21/lodash.min.js"></script>
What's the simplest, library-free code for implementing array intersections in javascript? I want to write
intersection([1,2,3], [2,3,4,5])
and get
[2, 3]
Use a combination of Array.prototype.filter and Array.prototype.includes:
const filteredArray = array1.filter(value => array2.includes(value));
For older browsers, with Array.prototype.indexOf and without an arrow function:
var filteredArray = array1.filter(function(n) {
return array2.indexOf(n) !== -1;
});
NB! Both .includes and .indexOf internally compares elements in the array by using ===, so if the array contains objects it will only compare object references (not their content). If you want to specify your own comparison logic, use Array.prototype.some instead.
Destructive seems simplest, especially if we can assume the input is sorted:
/* destructively finds the intersection of
* two arrays in a simple fashion.
*
* PARAMS
* a - first array, must already be sorted
* b - second array, must already be sorted
*
* NOTES
* State of input arrays is undefined when
* the function returns. They should be
* (prolly) be dumped.
*
* Should have O(n) operations, where n is
* n = MIN(a.length, b.length)
*/
function intersection_destructive(a, b)
{
var result = [];
while( a.length > 0 && b.length > 0 )
{
if (a[0] < b[0] ){ a.shift(); }
else if (a[0] > b[0] ){ b.shift(); }
else /* they're equal */
{
result.push(a.shift());
b.shift();
}
}
return result;
}
Non-destructive has to be a hair more complicated, since we’ve got to track indices:
/* finds the intersection of
* two arrays in a simple fashion.
*
* PARAMS
* a - first array, must already be sorted
* b - second array, must already be sorted
*
* NOTES
*
* Should have O(n) operations, where n is
* n = MIN(a.length(), b.length())
*/
function intersect_safe(a, b)
{
var ai=0, bi=0;
var result = [];
while( ai < a.length && bi < b.length )
{
if (a[ai] < b[bi] ){ ai++; }
else if (a[ai] > b[bi] ){ bi++; }
else /* they're equal */
{
result.push(a[ai]);
ai++;
bi++;
}
}
return result;
}
If your environment supports ECMAScript 6 Set, one simple and supposedly efficient (see specification link) way:
function intersect(a, b) {
var setA = new Set(a);
var setB = new Set(b);
var intersection = new Set([...setA].filter(x => setB.has(x)));
return Array.from(intersection);
}
Shorter, but less readable (also without creating the additional intersection Set):
function intersect(a, b) {
var setB = new Set(b);
return [...new Set(a)].filter(x => setB.has(x));
}
Note that when using sets you will only get distinct values, thus new Set([1, 2, 3, 3]).size evaluates to 3.
Using Underscore.js or lodash.js
_.intersection( [0,345,324] , [1,0,324] ) // gives [0,324]
// Return elements of array a that are also in b in linear time:
function intersect(a, b) {
return a.filter(Set.prototype.has, new Set(b));
}
// Example:
console.log(intersect([1,2,3], [2,3,4,5]));
I recommend above succinct solution which outperforms other implementations on large inputs. If performance on small inputs matters, check the alternatives below.
Alternatives and performance comparison:
See the following snippet for alternative implementations and check https://jsperf.com/array-intersection-comparison for performance comparisons.
function intersect_for(a, b) {
const result = [];
const alen = a.length;
const blen = b.length;
for (let i = 0; i < alen; ++i) {
const ai = a[i];
for (let j = 0; j < blen; ++j) {
if (ai === b[j]) {
result.push(ai);
break;
}
}
}
return result;
}
function intersect_filter_indexOf(a, b) {
return a.filter(el => b.indexOf(el) !== -1);
}
function intersect_filter_in(a, b) {
const map = b.reduce((map, el) => {map[el] = true; return map}, {});
return a.filter(el => el in map);
}
function intersect_for_in(a, b) {
const result = [];
const map = {};
for (let i = 0, length = b.length; i < length; ++i) {
map[b[i]] = true;
}
for (let i = 0, length = a.length; i < length; ++i) {
if (a[i] in map) result.push(a[i]);
}
return result;
}
function intersect_filter_includes(a, b) {
return a.filter(el => b.includes(el));
}
function intersect_filter_has_this(a, b) {
return a.filter(Set.prototype.has, new Set(b));
}
function intersect_filter_has_arrow(a, b) {
const set = new Set(b);
return a.filter(el => set.has(el));
}
function intersect_for_has(a, b) {
const result = [];
const set = new Set(b);
for (let i = 0, length = a.length; i < length; ++i) {
if (set.has(a[i])) result.push(a[i]);
}
return result;
}
Results in Firefox 53:
Ops/sec on large arrays (10,000 elements):
filter + has (this) 523 (this answer)
for + has 482
for-loop + in 279
filter + in 242
for-loops 24
filter + includes 14
filter + indexOf 10
Ops/sec on small arrays (100 elements):
for-loop + in 384,426
filter + in 192,066
for-loops 159,137
filter + includes 104,068
filter + indexOf 71,598
filter + has (this) 43,531 (this answer)
filter + has (arrow function) 35,588
My contribution in ES6 terms. In general it finds the intersection of an array with indefinite number of arrays provided as arguments.
Array.prototype.intersect = function(...a) {
return [this,...a].reduce((p,c) => p.filter(e => c.includes(e)));
}
var arrs = [[0,2,4,6,8],[4,5,6,7],[4,6]],
arr = [0,1,2,3,4,5,6,7,8,9];
document.write("<pre>" + JSON.stringify(arr.intersect(...arrs)) + "</pre>");
How about just using associative arrays?
function intersect(a, b) {
var d1 = {};
var d2 = {};
var results = [];
for (var i = 0; i < a.length; i++) {
d1[a[i]] = true;
}
for (var j = 0; j < b.length; j++) {
d2[b[j]] = true;
}
for (var k in d1) {
if (d2[k])
results.push(k);
}
return results;
}
edit:
// new version
function intersect(a, b) {
var d = {};
var results = [];
for (var i = 0; i < b.length; i++) {
d[b[i]] = true;
}
for (var j = 0; j < a.length; j++) {
if (d[a[j]])
results.push(a[j]);
}
return results;
}
The performance of #atk's implementation for sorted arrays of primitives can be improved by using .pop rather than .shift.
function intersect(array1, array2) {
var result = [];
// Don't destroy the original arrays
var a = array1.slice(0);
var b = array2.slice(0);
var aLast = a.length - 1;
var bLast = b.length - 1;
while (aLast >= 0 && bLast >= 0) {
if (a[aLast] > b[bLast] ) {
a.pop();
aLast--;
} else if (a[aLast] < b[bLast] ){
b.pop();
bLast--;
} else /* they're equal */ {
result.push(a.pop());
b.pop();
aLast--;
bLast--;
}
}
return result;
}
I created a benchmark using jsPerf. It's about three times faster to use .pop.
If you need to have it handle intersecting multiple arrays:
const intersect = (a1, a2, ...rest) => {
const a12 = a1.filter(value => a2.includes(value))
if (rest.length === 0) { return a12; }
return intersect(a12, ...rest);
};
console.log(intersect([1,2,3,4,5], [1,2], [1, 2, 3,4,5], [2, 10, 1]))
Sort it
check one by one from the index 0, create new array from that.
Something like this, Not tested well though.
function intersection(x,y){
x.sort();y.sort();
var i=j=0;ret=[];
while(i<x.length && j<y.length){
if(x[i]<y[j])i++;
else if(y[j]<x[i])j++;
else {
ret.push(x[i]);
i++,j++;
}
}
return ret;
}
alert(intersection([1,2,3], [2,3,4,5]));
PS:The algorithm only intended for Numbers and Normal Strings, intersection of arbitary object arrays may not work.
Using jQuery:
var a = [1,2,3];
var b = [2,3,4,5];
var c = $(b).not($(b).not(a));
alert(c);
A tiny tweak to the smallest one here (the filter/indexOf solution), namely creating an index of the values in one of the arrays using a JavaScript object, will reduce it from O(N*M) to "probably" linear time. source1 source2
function intersect(a, b) {
var aa = {};
a.forEach(function(v) { aa[v]=1; });
return b.filter(function(v) { return v in aa; });
}
This isn't the very simplest solution (it's more code than filter+indexOf), nor is it the very fastest (probably slower by a constant factor than intersect_safe()), but seems like a pretty good balance. It is on the very simple side, while providing good performance, and it doesn't require pre-sorted inputs.
For arrays containing only strings or numbers you can do something with sorting, as per some of the other answers. For the general case of arrays of arbitrary objects I don't think you can avoid doing it the long way. The following will give you the intersection of any number of arrays provided as parameters to arrayIntersection:
var arrayContains = Array.prototype.indexOf ?
function(arr, val) {
return arr.indexOf(val) > -1;
} :
function(arr, val) {
var i = arr.length;
while (i--) {
if (arr[i] === val) {
return true;
}
}
return false;
};
function arrayIntersection() {
var val, arrayCount, firstArray, i, j, intersection = [], missing;
var arrays = Array.prototype.slice.call(arguments); // Convert arguments into a real array
// Search for common values
firstArray = arrays.pop();
if (firstArray) {
j = firstArray.length;
arrayCount = arrays.length;
while (j--) {
val = firstArray[j];
missing = false;
// Check val is present in each remaining array
i = arrayCount;
while (!missing && i--) {
if ( !arrayContains(arrays[i], val) ) {
missing = true;
}
}
if (!missing) {
intersection.push(val);
}
}
}
return intersection;
}
arrayIntersection( [1, 2, 3, "a"], [1, "a", 2], ["a", 1] ); // Gives [1, "a"];
Simplest, fastest O(n) and shortest way:
function intersection (a, b) {
const setA = new Set(a);
return b.filter(value => setA.has(value));
}
console.log(intersection([1,2,3], [2,3,4,5]))
#nbarbosa has almost the same answer but he cast both arrays to Set and then back to array. There is no need for any extra casting.
Another indexed approach able to process any number of arrays at once:
// Calculate intersection of multiple array or object values.
function intersect (arrList) {
var arrLength = Object.keys(arrList).length;
// (Also accepts regular objects as input)
var index = {};
for (var i in arrList) {
for (var j in arrList[i]) {
var v = arrList[i][j];
if (index[v] === undefined) index[v] = 0;
index[v]++;
};
};
var retv = [];
for (var i in index) {
if (index[i] == arrLength) retv.push(i);
};
return retv;
};
It works only for values that can be evaluated as strings and you should pass them as an array like:
intersect ([arr1, arr2, arr3...]);
...but it transparently accepts objects as parameter or as any of the elements to be intersected (always returning array of common values). Examples:
intersect ({foo: [1, 2, 3, 4], bar: {a: 2, j:4}}); // [2, 4]
intersect ([{x: "hello", y: "world"}, ["hello", "user"]]); // ["hello"]
EDIT: I just noticed that this is, in a way, slightly buggy.
That is: I coded it thinking that input arrays cannot itself contain repetitions (as provided example doesn't).
But if input arrays happen to contain repetitions, that would produce wrong results. Example (using below implementation):
intersect ([[1, 3, 4, 6, 3], [1, 8, 99]]);
// Expected: [ '1' ]
// Actual: [ '1', '3' ]
Fortunately this is easy to fix by simply adding second level indexing. That is:
Change:
if (index[v] === undefined) index[v] = 0;
index[v]++;
by:
if (index[v] === undefined) index[v] = {};
index[v][i] = true; // Mark as present in i input.
...and:
if (index[i] == arrLength) retv.push(i);
by:
if (Object.keys(index[i]).length == arrLength) retv.push(i);
Complete example:
// Calculate intersection of multiple array or object values.
function intersect (arrList) {
var arrLength = Object.keys(arrList).length;
// (Also accepts regular objects as input)
var index = {};
for (var i in arrList) {
for (var j in arrList[i]) {
var v = arrList[i][j];
if (index[v] === undefined) index[v] = {};
index[v][i] = true; // Mark as present in i input.
};
};
var retv = [];
for (var i in index) {
if (Object.keys(index[i]).length == arrLength) retv.push(i);
};
return retv;
};
intersect ([[1, 3, 4, 6, 3], [1, 8, 99]]); // [ '1' ]
With some restrictions on your data, you can do it in linear time!
For positive integers: use an array mapping the values to a "seen/not seen" boolean.
function intersectIntegers(array1,array2) {
var seen=[],
result=[];
for (var i = 0; i < array1.length; i++) {
seen[array1[i]] = true;
}
for (var i = 0; i < array2.length; i++) {
if ( seen[array2[i]])
result.push(array2[i]);
}
return result;
}
There is a similar technique for objects: take a dummy key, set it to "true" for each element in array1, then look for this key in elements of array2. Clean up when you're done.
function intersectObjects(array1,array2) {
var result=[];
var key="tmpKey_intersect"
for (var i = 0; i < array1.length; i++) {
array1[i][key] = true;
}
for (var i = 0; i < array2.length; i++) {
if (array2[i][key])
result.push(array2[i]);
}
for (var i = 0; i < array1.length; i++) {
delete array1[i][key];
}
return result;
}
Of course you need to be sure the key didn't appear before, otherwise you'll be destroying your data...
function intersection(A,B){
var result = new Array();
for (i=0; i<A.length; i++) {
for (j=0; j<B.length; j++) {
if (A[i] == B[j] && $.inArray(A[i],result) == -1) {
result.push(A[i]);
}
}
}
return result;
}
For simplicity:
// Usage
const intersection = allLists
.reduce(intersect, allValues)
.reduce(removeDuplicates, []);
// Implementation
const intersect = (intersection, list) =>
intersection.filter(item =>
list.some(x => x === item));
const removeDuplicates = (uniques, item) =>
uniques.includes(item) ? uniques : uniques.concat(item);
// Example Data
const somePeople = [bob, doug, jill];
const otherPeople = [sarah, bob, jill];
const morePeople = [jack, jill];
const allPeople = [...somePeople, ...otherPeople, ...morePeople];
const allGroups = [somePeople, otherPeople, morePeople];
// Example Usage
const intersection = allGroups
.reduce(intersect, allPeople)
.reduce(removeDuplicates, []);
intersection; // [jill]
Benefits:
dirt simple
data-centric
works for arbitrary number of lists
works for arbitrary lengths of lists
works for arbitrary types of values
works for arbitrary sort order
retains shape (order of first appearance in any array)
exits early where possible
memory safe, short of tampering with Function / Array prototypes
Drawbacks:
higher memory usage
higher CPU usage
requires an understanding of reduce
requires understanding of data flow
You wouldn't want to use this for 3D engine or kernel work, but if you have problems getting this to run in an event-based app, your design has bigger problems.
I'll contribute with what has been working out best for me:
if (!Array.prototype.intersect){
Array.prototype.intersect = function (arr1) {
var r = [], o = {}, l = this.length, i, v;
for (i = 0; i < l; i++) {
o[this[i]] = true;
}
l = arr1.length;
for (i = 0; i < l; i++) {
v = arr1[i];
if (v in o) {
r.push(v);
}
}
return r;
};
}
A functional approach with ES2015
A functional approach must consider using only pure functions without side effects, each of which is only concerned with a single job.
These restrictions enhance the composability and reusability of the functions involved.
// small, reusable auxiliary functions
const createSet = xs => new Set(xs);
const filter = f => xs => xs.filter(apply(f));
const apply = f => x => f(x);
// intersection
const intersect = xs => ys => {
const zs = createSet(ys);
return filter(x => zs.has(x)
? true
: false
) (xs);
};
// mock data
const xs = [1,2,2,3,4,5];
const ys = [0,1,2,3,3,3,6,7,8,9];
// run it
console.log( intersect(xs) (ys) );
Please note that the native Set type is used, which has an advantageous
lookup performance.
Avoid duplicates
Obviously repeatedly occurring items from the first Array are preserved, while the second Array is de-duplicated. This may be or may be not the desired behavior. If you need a unique result just apply dedupe to the first argument:
// auxiliary functions
const apply = f => x => f(x);
const comp = f => g => x => f(g(x));
const afrom = apply(Array.from);
const createSet = xs => new Set(xs);
const filter = f => xs => xs.filter(apply(f));
// intersection
const intersect = xs => ys => {
const zs = createSet(ys);
return filter(x => zs.has(x)
? true
: false
) (xs);
};
// de-duplication
const dedupe = comp(afrom) (createSet);
// mock data
const xs = [1,2,2,3,4,5];
const ys = [0,1,2,3,3,3,6,7,8,9];
// unique result
console.log( intersect(dedupe(xs)) (ys) );
Compute the intersection of any number of Arrays
If you want to compute the intersection of an arbitrarily number of Arrays just compose intersect with foldl. Here is a convenience function:
// auxiliary functions
const apply = f => x => f(x);
const uncurry = f => (x, y) => f(x) (y);
const createSet = xs => new Set(xs);
const filter = f => xs => xs.filter(apply(f));
const foldl = f => acc => xs => xs.reduce(uncurry(f), acc);
// intersection
const intersect = xs => ys => {
const zs = createSet(ys);
return filter(x => zs.has(x)
? true
: false
) (xs);
};
// intersection of an arbitrarily number of Arrays
const intersectn = (head, ...tail) => foldl(intersect) (head) (tail);
// mock data
const xs = [1,2,2,3,4,5];
const ys = [0,1,2,3,3,3,6,7,8,9];
const zs = [0,1,2,3,4,5,6];
// run
console.log( intersectn(xs, ys, zs) );
.reduce to build a map, and .filter to find the intersection. delete within the .filter allows us to treat the second array as though it's a unique set.
function intersection (a, b) {
var seen = a.reduce(function (h, k) {
h[k] = true;
return h;
}, {});
return b.filter(function (k) {
var exists = seen[k];
delete seen[k];
return exists;
});
}
I find this approach pretty easy to reason about. It performs in constant time.
I have written an intesection function which can even detect intersection of array of objects based on particular property of those objects.
For instance,
if arr1 = [{id: 10}, {id: 20}]
and arr2 = [{id: 20}, {id: 25}]
and we want intersection based on the id property, then the output should be :
[{id: 20}]
As such, the function for the same (note: ES6 code) is :
const intersect = (arr1, arr2, accessors = [v => v, v => v]) => {
const [fn1, fn2] = accessors;
const set = new Set(arr2.map(v => fn2(v)));
return arr1.filter(value => set.has(fn1(value)));
};
and you can call the function as:
intersect(arr1, arr2, [elem => elem.id, elem => elem.id])
Also note: this function finds intersection considering the first array is the primary array and thus the intersection result will be that of the primary array.
This function avoids the N^2 problem, taking advantage of the power of dictionaries. Loops through each array only once, and a third and shorter loop to return the final result.
It also supports numbers, strings, and objects.
function array_intersect(array1, array2)
{
var mergedElems = {},
result = [];
// Returns a unique reference string for the type and value of the element
function generateStrKey(elem) {
var typeOfElem = typeof elem;
if (typeOfElem === 'object') {
typeOfElem += Object.prototype.toString.call(elem);
}
return [typeOfElem, elem.toString(), JSON.stringify(elem)].join('__');
}
array1.forEach(function(elem) {
var key = generateStrKey(elem);
if (!(key in mergedElems)) {
mergedElems[key] = {elem: elem, inArray2: false};
}
});
array2.forEach(function(elem) {
var key = generateStrKey(elem);
if (key in mergedElems) {
mergedElems[key].inArray2 = true;
}
});
Object.values(mergedElems).forEach(function(elem) {
if (elem.inArray2) {
result.push(elem.elem);
}
});
return result;
}
If there is a special case that cannot be solved, just by modifying the generateStrKey function, it could surely be solved. The trick of this function is that it uniquely represents each different data according to type and value.
This variant has some performance improvements. Avoid loops in case any array is empty. It also starts by walking through the shorter array first, so if it finds all the values of the first array in the second array, exits the loop.
function array_intersect(array1, array2)
{
var mergedElems = {},
result = [],
firstArray, secondArray,
firstN = 0,
secondN = 0;
function generateStrKey(elem) {
var typeOfElem = typeof elem;
if (typeOfElem === 'object') {
typeOfElem += Object.prototype.toString.call(elem);
}
return [typeOfElem, elem.toString(), JSON.stringify(elem)].join('__');
}
// Executes the loops only if both arrays have values
if (array1.length && array2.length)
{
// Begins with the shortest array to optimize the algorithm
if (array1.length < array2.length) {
firstArray = array1;
secondArray = array2;
} else {
firstArray = array2;
secondArray = array1;
}
firstArray.forEach(function(elem) {
var key = generateStrKey(elem);
if (!(key in mergedElems)) {
mergedElems[key] = {elem: elem, inArray2: false};
// Increases the counter of unique values in the first array
firstN++;
}
});
secondArray.some(function(elem) {
var key = generateStrKey(elem);
if (key in mergedElems) {
if (!mergedElems[key].inArray2) {
mergedElems[key].inArray2 = true;
// Increases the counter of matches
secondN++;
// If all elements of first array have coincidence, then exits the loop
return (secondN === firstN);
}
}
});
Object.values(mergedElems).forEach(function(elem) {
if (elem.inArray2) {
result.push(elem.elem);
}
});
}
return result;
}
Here is underscore.js implementation:
_.intersection = function(array) {
if (array == null) return [];
var result = [];
var argsLength = arguments.length;
for (var i = 0, length = array.length; i < length; i++) {
var item = array[i];
if (_.contains(result, item)) continue;
for (var j = 1; j < argsLength; j++) {
if (!_.contains(arguments[j], item)) break;
}
if (j === argsLength) result.push(item);
}
return result;
};
Source: http://underscorejs.org/docs/underscore.html#section-62
Create an Object using one array and loop through the second array to check if the value exists as key.
function intersection(arr1, arr2) {
var myObj = {};
var myArr = [];
for (var i = 0, len = arr1.length; i < len; i += 1) {
if(myObj[arr1[i]]) {
myObj[arr1[i]] += 1;
} else {
myObj[arr1[i]] = 1;
}
}
for (var j = 0, len = arr2.length; j < len; j += 1) {
if(myObj[arr2[j]] && myArr.indexOf(arr2[j]) === -1) {
myArr.push(arr2[j]);
}
}
return myArr;
}
I think using an object internally can help with computations and could be performant too.
// Approach maintains a count of each element and works for negative elements too
function intersect(a,b){
const A = {};
a.forEach((v)=>{A[v] ? ++A[v] : A[v] = 1});
const B = {};
b.forEach((v)=>{B[v] ? ++B[v] : B[v] = 1});
const C = {};
Object.entries(A).map((x)=>C[x[0]] = Math.min(x[1],B[x[0]]))
return Object.entries(C).map((x)=>Array(x[1]).fill(Number(x[0]))).flat();
}
const x = [1,1,-1,-1,0,0,2,2];
const y = [2,0,1,1,1,1,0,-1,-1,-1];
const result = intersect(x,y);
console.log(result); // (7) [0, 0, 1, 1, 2, -1, -1]
I am using map even object could be used.
//find intersection of 2 arrs
const intersections = (arr1,arr2) => {
let arrf = arr1.concat(arr2)
let map = new Map();
let union = [];
for(let i=0; i<arrf.length; i++){
if(map.get(arrf[i])){
map.set(arrf[i],false);
}else{
map.set(arrf[i],true);
}
}
map.forEach((v,k)=>{if(!v){union.push(k);}})
return union;
}
This is a proposed standard: With the currently stage 2 proposal https://github.com/tc39/proposal-set-methods, you could use
mySet.intersection(mySet2);
Until then, you could use Immutable.js's Set, which inspired that proposal
Immutable.Set(mySet).intersect(mySet2)
I extended tarulen's answer to work with any number of arrays. It also should work with non-integer values.
function intersect() {
const last = arguments.length - 1;
var seen={};
var result=[];
for (var i = 0; i < last; i++) {
for (var j = 0; j < arguments[i].length; j++) {
if (seen[arguments[i][j]]) {
seen[arguments[i][j]] += 1;
}
else if (!i) {
seen[arguments[i][j]] = 1;
}
}
}
for (var i = 0; i < arguments[last].length; i++) {
if ( seen[arguments[last][i]] === last)
result.push(arguments[last][i]);
}
return result;
}
If your arrays are sorted, this should run in O(n), where n is min( a.length, b.length )
function intersect_1d( a, b ){
var out=[], ai=0, bi=0, acurr, bcurr, last=Number.MIN_SAFE_INTEGER;
while( ( acurr=a[ai] )!==undefined && ( bcurr=b[bi] )!==undefined ){
if( acurr < bcurr){
if( last===acurr ){
out.push( acurr );
}
last=acurr;
ai++;
}
else if( acurr > bcurr){
if( last===bcurr ){
out.push( bcurr );
}
last=bcurr;
bi++;
}
else {
out.push( acurr );
last=acurr;
ai++;
bi++;
}
}
return out;
}
This question already has answers here:
In Javascript, how do I check if an array has duplicate values?
(9 answers)
Closed 10 months ago.
I wanted to write a javascript function which checks if array contains duplicate values or not.
I have written the following code but its giving answer as "true" always.
Can anybody please tell me what am I missing.
function checkIfArrayIsUnique(myArray)
{
for (var i = 0; i < myArray.length; i++)
{
for (var j = 0; j < myArray.length; j++)
{
if (i != j)
{
if (myArray[i] == myArray[j])
{
return true; // means there are duplicate values
}
}
}
}
return false; // means there are no duplicate values.
}
An easy solution, if you've got ES6, uses Set:
function checkIfArrayIsUnique(myArray) {
return myArray.length === new Set(myArray).size;
}
let uniqueArray = [1, 2, 3, 4, 5];
console.log(`${uniqueArray} is unique : ${checkIfArrayIsUnique(uniqueArray)}`);
let nonUniqueArray = [1, 1, 2, 3, 4, 5];
console.log(`${nonUniqueArray} is unique : ${checkIfArrayIsUnique(nonUniqueArray)}`);
let arr = [11,22,11,22];
let hasDuplicate = arr.some((val, i) => arr.indexOf(val) !== i);
// hasDuplicate = true
True -> array has duplicates
False -> uniqe array
This should work with only one loop:
function checkIfArrayIsUnique(arr) {
var map = {}, i, size;
for (i = 0, size = arr.length; i < size; i++){
if (map[arr[i]]){
return false;
}
map[arr[i]] = true;
}
return true;
}
You got the return values the wrong way round:
As soon as you find two values that are equal, you can conclude that the array is not unique and return false.
At the very end, after you've checked all the pairs, you can return true.
If you do this a lot, and the arrays are large, you might want to investigate the possibility of sorting the array and then only comparing adjacent elements. This will have better asymptotic complexity than your current method.
Assuming you're targeting browsers that aren't IE8,
this would work as well:
function checkIfArrayIsUnique(myArray)
{
for (var i = 0; i < myArray.length; i++)
{
if (myArray.indexOf(myArray[i]) !== myArray.lastIndexOf(myArray[i])) {
return false;
}
}
return true; // this means not unique
}
Here's an O(n) solution:
function hasDupes(arr) {
/* temporary object */
var uniqOb = {};
/* create object attribute with name=value in array, this will not keep dupes*/
for (var i in arr)
uniqOb[arr[i]] = "";
/* if object's attributes match array, then no dupes! */
if (arr.length == Object.keys(uniqOb).length)
alert('NO dupes');
else
alert('HAS dupes');
}
var arr = ["1/1/2016", "1/1/2016", "2/1/2016"];
hasDupes(arr);
https://jsfiddle.net/7kkgy1j3/
Another solution:
Array.prototype.checkIfArrayIsUnique = function() {
this.sort();
for ( var i = 1; i < this.length; i++ ){
if(this[i-1] == this[i])
return false;
}
return true;
}
function hasNoDuplicates(arr) {
return arr.every(num => arr.indexOf(num) === arr.lastIndexOf(num));
}
hasNoDuplicates accepts an array and returns true if there are no duplicate values. If there are any duplicates, the function returns false.
Without a for loop, only using Map().
You can also return the duplicates.
(function(a){
let map = new Map();
a.forEach(e => {
if(map.has(e)) {
let count = map.get(e);
console.log(count)
map.set(e, count + 1);
} else {
map.set(e, 1);
}
});
let hasDup = false;
let dups = [];
map.forEach((value, key) => {
if(value > 1) {
hasDup = true;
dups.push(key);
}
});
console.log(dups);
return hasDup;
})([2,4,6,2,1,4]);
Late answer but can be helpful
function areThereDuplicates(args) {
let count = {};
for(let i = 0; i < args.length; i++){
count[args[i]] = 1 + (count[args[i]] || 0);
}
let found = Object.keys(count).filter(function(key) {
return count[key] > 1;
});
return found.length ? true : false;
}
areThereDuplicates([1,2,5]);
The code given in the question can be better written as follows
function checkIfArrayIsUnique(myArray)
{
for (var i = 0; i < myArray.length; i++)
{
for (var j = i+1; j < myArray.length; j++)
{
if (myArray[i] == myArray[j])
{
return true; // means there are duplicate values
}
}
}
return false; // means there are no duplicate values.
}
Returns the duplicate item in array and creates a new array with no duplicates:
var a = ["hello", "hi", "hi", "juice", "juice", "test"];
var b = ["ding", "dong", "hi", "juice", "juice", "test"];
var c = a.concat(b);
var dupClearArr = [];
function dupArray(arr) {
for (i = 0; i < arr.length; i++) {
if (arr.indexOf(arr[i]) != i && arr.indexOf(arr[i]) != -1) {
console.log('duplicate item ' + arr[i]);
} else {
dupClearArr.push(arr[i])
}
}
console.log('actual array \n' + arr + ' \nno duplicate items array \n' + dupClearArr)
}
dupArray(c);
const containsMatches = (a1, a2) => a1.some((v) => a2.includes(v));
If your array nests other arrays/objects, using the Set approach may not be what you want since comparing two objects compares their references. If you want to check that their contained values are equal, something else is needed. Here are a couple different approaches.
Approach 1: Map using JSON.stringify for keys
If you want to consider objects with the same contained values as equal, here's one simple way to do it using a Map object. It uses JSON.stringify to make a unique id for each element in the array.
I believe the runtime of this would be O(n * m) on arrays, assuming JSON.stringify serializes in linear time. n is the length of the outer array, m is size of the arrays. If the objects get very large, however, this may slow down since the keys will be very long. Not a very space-efficient implementation, but it is simple and works for many data types.
function checkArrayDupeFree(myArray, idFunc) {
const dupeMap = new Map();
for (const el of myArray) {
const id = idFunc(el);
if (dupeMap.has(id))
return false;
dupeMap.set(id, el);
}
return true;
}
const notUnique = [ [1, 2], [1, 3], [1, 2] ];
console.log(`${JSON.stringify(notUnique)} has no duplicates? ${checkArrayDupeFree(notUnique, JSON.stringify)}`);
const unique = [ [2, 1], [1, 3], [1, 2] ];
console.log(`${JSON.stringify(unique)} has no duplicates? ${checkArrayDupeFree(unique, JSON.stringify)}`);
Of course, you could also write your own id-generator function, though I'm not sure you can do much better than JSON.stringify.
Approach 2: Custom HashMap, Hashcode, and Equality implementations
If you have a lot of big arrays, it may be better performance-wise to implement your own hash/equality functions and use a Map as a HashMap.
In the following implementation, we hash the array. If there is a collision, map a key to an array of collided values, and check to see if any of the array values match according to the equality function.
The downside of this approach is that you may have to consider a wide range of types for which to make hashcode/equality functions, depending on what's in the array.
function checkArrayDupeFreeWHashes(myArray, hashFunc, eqFunc) {
const hashMap = new Map();
for (const el of myArray) {
const hash = hashFunc(el);
const hit = hashMap.get(hash);
if (hit == null)
hashMap.set(hash, [el]);
else if (hit.some(v => eqFunc(v, el)))
return false;
else
hit.push(el);
}
return true;
}
Here's a demo of the custom HashMap in action. I implemented a hashing function and an equality function for arrays of arrays.
function checkArrayDupeFreeWHashes(myArray, hashFunc, eqFunc) {
const hashMap = new Map();
for (const el of myArray) {
const hash = hashFunc(el);
const hit = hashMap.get(hash);
if (hit == null)
hashMap.set(hash, [el]);
else if (hit.some(v => eqFunc(v, el)))
return false;
else
hit.push(el);
}
return true;
}
function arrayHasher(arr) {
let hash = 19;
for (let i = 0; i < arr.length; i++) {
const el = arr[i];
const toHash = Array.isArray(el)
? arrayHasher(el)
: el * 23;
hash = hash * 31 + toHash;
}
return hash;
}
function arrayEq(a, b) {
if (a.length != b.length)
return false;
for (let i = 0; i < a.length; i++) {
if ((Array.isArray(a) || Array.isArray(b)) && !arrayEq(a[i], b[i]))
return false;
else if (a[i] !== b[i])
return false;
}
return true;
}
const notUnique = [ [1, 2], [1, 3], [1, 2] ];
const unique = [ [2, 1], [1, 3], [1, 2] ];
console.log(`${JSON.stringify(notUnique)} has no duplicates? ${checkArrayDupeFreeWHashes(notUnique, arrayHasher, arrayEq)}`);
console.log(`${JSON.stringify(unique)} has no duplicates? ${checkArrayDupeFreeWHashes(unique, arrayHasher, arrayEq)}`);
function checkIfArrayIsUnique(myArray)
{
isUnique=true
for (var i = 0; i < myArray.length; i++)
{
for (var j = 0; j < myArray.length; j++)
{
if (i != j)
{
if (myArray[i] == myArray[j])
{
isUnique=false
}
}
}
}
return isUnique;
}
This assume that the array is unique at the start.
If find two equals values, then change to false
i think this is the simple way
$(document).ready(function() {
var arr = [1,2,3,9,6,5,6];
console.log( "result =>"+ if_duplicate_value (arr));
});
function if_duplicate_value (arr){
for(i=0;i<arr.length-1;i++){
for(j=i+1;j<arr.length;j++){
if(arr[i]==arr[j]){
return true;
}
}
}
return false;
}
<script src="https://cdnjs.cloudflare.com/ajax/libs/jquery/3.3.1/jquery.min.js"></script>
var c=[2,2,3,3,5,5,4,4,8,8];
for(var i=0; i<b.length; i++){
for(var j=i+1; j<b.length; j++){
if(c[i]==c[j]){
console.log(c[j]);
}
}
}
This question already has answers here:
Cartesian product of multiple arrays in JavaScript
(35 answers)
Closed 1 year ago.
I'm having trouble coming up with code to generate combinations from n number of arrays with m number of elements in them, in JavaScript. I've seen similar questions about this for other languages, but the answers incorporate syntactic or library magic that I'm unsure how to translate.
Consider this data:
[[0,1], [0,1,2,3], [0,1,2]]
3 arrays, with a different number of elements in them. What I want to do is get all combinations by combining an item from each array.
For example:
0,0,0 // item 0 from array 0, item 0 from array 1, item 0 from array 2
0,0,1
0,0,2
0,1,0
0,1,1
0,1,2
0,2,0
0,2,1
0,2,2
And so on.
If the number of arrays were fixed, it would be easy to make a hard coded implementation. But the number of arrays may vary:
[[0,1], [0,1]]
[[0,1,3,4], [0,1], [0], [0,1]]
Any help would be much appreciated.
Here is a quite simple and short one using a recursive helper function:
function cartesian(...args) {
var r = [], max = args.length-1;
function helper(arr, i) {
for (var j=0, l=args[i].length; j<l; j++) {
var a = arr.slice(0); // clone arr
a.push(args[i][j]);
if (i==max)
r.push(a);
else
helper(a, i+1);
}
}
helper([], 0);
return r;
}
Usage:
cartesian([0,1], [0,1,2,3], [0,1,2]);
To make the function take an array of arrays, just change the signature to function cartesian(args) instead of using rest parameter syntax.
I suggest a simple recursive generator function:
// JS
function* cartesianIterator(head, ...tail) {
const remainder = tail.length ? cartesianIterator(...tail) : [[]];
for (let r of remainder) for (let h of head) yield [h, ...r];
}
// get values:
const cartesian = items => [...cartesianIterator(items)];
console.log(cartesian(input));
// TS
function* cartesianIterator<T>(items: T[][]): Generator<T[]> {
const remainder = items.length > 1 ? cartesianIterator(items.slice(1)) : [[]];
for (let r of remainder) for (let h of items.at(0)!) yield [h, ...r];
}
// get values:
const cartesian = <T>(items: T[][]) => [...cartesianIterator(items)];
console.log(cartesian(input));
You could take an iterative approach by building sub arrays.
var parts = [[0, 1], [0, 1, 2, 3], [0, 1, 2]],
result = parts.reduce((a, b) => a.reduce((r, v) => r.concat(b.map(w => [].concat(v, w))), []));
console.log(result.map(a => a.join(', ')));
.as-console-wrapper { max-height: 100% !important; top: 0; }
After doing a little research I discovered a previous related question:
Finding All Combinations of JavaScript array values
I've adapted some of the code from there so that it returns an array of arrays containing all of the permutations:
function(arraysToCombine) {
var divisors = [];
for (var i = arraysToCombine.length - 1; i >= 0; i--) {
divisors[i] = divisors[i + 1] ? divisors[i + 1] * arraysToCombine[i + 1].length : 1;
}
function getPermutation(n, arraysToCombine) {
var result = [],
curArray;
for (var i = 0; i < arraysToCombine.length; i++) {
curArray = arraysToCombine[i];
result.push(curArray[Math.floor(n / divisors[i]) % curArray.length]);
}
return result;
}
var numPerms = arraysToCombine[0].length;
for(var i = 1; i < arraysToCombine.length; i++) {
numPerms *= arraysToCombine[i].length;
}
var combinations = [];
for(var i = 0; i < numPerms; i++) {
combinations.push(getPermutation(i, arraysToCombine));
}
return combinations;
}
I've put a working copy at http://jsfiddle.net/7EakX/ that takes the array you gave earlier ([[0,1], [0,1,2,3], [0,1,2]]) and outputs the result to the browser console.
const charSet = [["A", "B"],["C", "D", "E"],["F", "G", "H", "I"]];
console.log(charSet.reduce((a,b)=>a.flatMap(x=>b.map(y=>x+y)),['']))
Just for fun, here's a more functional variant of the solution in my first answer:
function cartesian() {
var r = [], args = Array.from(arguments);
args.reduceRight(function(cont, factor, i) {
return function(arr) {
for (var j=0, l=factor.length; j<l; j++) {
var a = arr.slice(); // clone arr
a[i] = factor[j];
cont(a);
}
};
}, Array.prototype.push.bind(r))(new Array(args.length));
return r;
}
Alternative, for full speed we can dynamically compile our own loops:
function cartesian() {
return (cartesian.cache[arguments.length] || cartesian.compile(arguments.length)).apply(null, arguments);
}
cartesian.cache = [];
cartesian.compile = function compile(n) {
var args = [],
indent = "",
up = "",
down = "";
for (var i=0; i<n; i++) {
var arr = "$"+String.fromCharCode(97+i),
ind = String.fromCharCode(105+i);
args.push(arr);
up += indent+"for (var "+ind+"=0, l"+arr+"="+arr+".length; "+ind+"<l"+arr+"; "+ind+"++) {\n";
down = indent+"}\n"+down;
indent += " ";
up += indent+"arr["+i+"] = "+arr+"["+ind+"];\n";
}
var body = "var res=[],\n arr=[];\n"+up+indent+"res.push(arr.slice());\n"+down+"return res;";
return cartesian.cache[n] = new Function(args, body);
}
var f = function(arr){
if(typeof arr !== 'object'){
return false;
}
arr = arr.filter(function(elem){ return (elem !== null); }); // remove empty elements - make sure length is correct
var len = arr.length;
var nextPerm = function(){ // increase the counter(s)
var i = 0;
while(i < len)
{
arr[i].counter++;
if(arr[i].counter >= arr[i].length){
arr[i].counter = 0;
i++;
}else{
return false;
}
}
return true;
};
var getPerm = function(){ // get the current permutation
var perm_arr = [];
for(var i = 0; i < len; i++)
{
perm_arr.push(arr[i][arr[i].counter]);
}
return perm_arr;
};
var new_arr = [];
for(var i = 0; i < len; i++) // set up a counter property inside the arrays
{
arr[i].counter = 0;
}
while(true)
{
new_arr.push(getPerm()); // add current permutation to the new array
if(nextPerm() === true){ // get next permutation, if returns true, we got them all
break;
}
}
return new_arr;
};
Here's another way of doing it. I treat the indices of all of the arrays like a number whose digits are all different bases (like time and dates), using the length of the array as the radix.
So, using your first set of data, the first digit is base 2, the second is base 4, and the third is base 3. The counter starts 000, then goes 001, 002, then 010. The digits correspond to indices in the arrays, and since order is preserved, this is no problem.
I have a fiddle with it working here: http://jsfiddle.net/Rykus0/DS9Ea/1/
and here is the code:
// Arbitrary base x number class
var BaseX = function(initRadix){
this.radix = initRadix ? initRadix : 1;
this.value = 0;
this.increment = function(){
return( (this.value = (this.value + 1) % this.radix) === 0);
}
}
function combinations(input){
var output = [], // Array containing the resulting combinations
counters = [], // Array of counters corresponding to our input arrays
remainder = false, // Did adding one cause the previous digit to rollover?
temp; // Holds one combination to be pushed into the output array
// Initialize the counters
for( var i = input.length-1; i >= 0; i-- ){
counters.unshift(new BaseX(input[i].length));
}
// Get all possible combinations
// Loop through until the first counter rolls over
while( !remainder ){
temp = []; // Reset the temporary value collection array
remainder = true; // Always increment the last array counter
// Process each of the arrays
for( i = input.length-1; i >= 0; i-- ){
temp.unshift(input[i][counters[i].value]); // Add this array's value to the result
// If the counter to the right rolled over, increment this one.
if( remainder ){
remainder = counters[i].increment();
}
}
output.push(temp); // Collect the results.
}
return output;
}
// Input is an array of arrays
console.log(combinations([[0,1], [0,1,2,3], [0,1,2]]));
You can use a recursive function to get all combinations
const charSet = [["A", "B"],["C", "D", "E"],["F", "G", "H", "I"]];
let loopOver = (arr, str = '', final = []) => {
if (arr.length > 1) {
arr[0].forEach(v => loopOver(arr.slice(1), str + v, final))
} else {
arr[0].forEach(v => final.push(str + v))
}
return final
}
console.log(loopOver(charSet))
This code can still be shorten using ternary but i prefer the first version for readability 😊
const charSet = [["A", "B"],["C", "D", "E"],["F", "G", "H", "I"]];
let loopOver = (arr, str = '') => arr[0].map(v => arr.length > 1 ? loopOver(arr.slice(1), str + v) : str + v).flat()
console.log(loopOver(charSet))
Another implementation with ES6 recursive style
Array.prototype.cartesian = function(a,...as){
return a ? this.reduce((p,c) => (p.push(...a.cartesian(...as).map(e => as.length ? [c,...e] : [c,e])),p),[])
: this;
};
console.log(JSON.stringify([0,1].cartesian([0,1,2,3], [[0],[1],[2]])));
What's the simplest, library-free code for implementing array intersections in javascript? I want to write
intersection([1,2,3], [2,3,4,5])
and get
[2, 3]
Use a combination of Array.prototype.filter and Array.prototype.includes:
const filteredArray = array1.filter(value => array2.includes(value));
For older browsers, with Array.prototype.indexOf and without an arrow function:
var filteredArray = array1.filter(function(n) {
return array2.indexOf(n) !== -1;
});
NB! Both .includes and .indexOf internally compares elements in the array by using ===, so if the array contains objects it will only compare object references (not their content). If you want to specify your own comparison logic, use Array.prototype.some instead.
Destructive seems simplest, especially if we can assume the input is sorted:
/* destructively finds the intersection of
* two arrays in a simple fashion.
*
* PARAMS
* a - first array, must already be sorted
* b - second array, must already be sorted
*
* NOTES
* State of input arrays is undefined when
* the function returns. They should be
* (prolly) be dumped.
*
* Should have O(n) operations, where n is
* n = MIN(a.length, b.length)
*/
function intersection_destructive(a, b)
{
var result = [];
while( a.length > 0 && b.length > 0 )
{
if (a[0] < b[0] ){ a.shift(); }
else if (a[0] > b[0] ){ b.shift(); }
else /* they're equal */
{
result.push(a.shift());
b.shift();
}
}
return result;
}
Non-destructive has to be a hair more complicated, since we’ve got to track indices:
/* finds the intersection of
* two arrays in a simple fashion.
*
* PARAMS
* a - first array, must already be sorted
* b - second array, must already be sorted
*
* NOTES
*
* Should have O(n) operations, where n is
* n = MIN(a.length(), b.length())
*/
function intersect_safe(a, b)
{
var ai=0, bi=0;
var result = [];
while( ai < a.length && bi < b.length )
{
if (a[ai] < b[bi] ){ ai++; }
else if (a[ai] > b[bi] ){ bi++; }
else /* they're equal */
{
result.push(a[ai]);
ai++;
bi++;
}
}
return result;
}
If your environment supports ECMAScript 6 Set, one simple and supposedly efficient (see specification link) way:
function intersect(a, b) {
var setA = new Set(a);
var setB = new Set(b);
var intersection = new Set([...setA].filter(x => setB.has(x)));
return Array.from(intersection);
}
Shorter, but less readable (also without creating the additional intersection Set):
function intersect(a, b) {
var setB = new Set(b);
return [...new Set(a)].filter(x => setB.has(x));
}
Note that when using sets you will only get distinct values, thus new Set([1, 2, 3, 3]).size evaluates to 3.
Using Underscore.js or lodash.js
_.intersection( [0,345,324] , [1,0,324] ) // gives [0,324]
// Return elements of array a that are also in b in linear time:
function intersect(a, b) {
return a.filter(Set.prototype.has, new Set(b));
}
// Example:
console.log(intersect([1,2,3], [2,3,4,5]));
I recommend above succinct solution which outperforms other implementations on large inputs. If performance on small inputs matters, check the alternatives below.
Alternatives and performance comparison:
See the following snippet for alternative implementations and check https://jsperf.com/array-intersection-comparison for performance comparisons.
function intersect_for(a, b) {
const result = [];
const alen = a.length;
const blen = b.length;
for (let i = 0; i < alen; ++i) {
const ai = a[i];
for (let j = 0; j < blen; ++j) {
if (ai === b[j]) {
result.push(ai);
break;
}
}
}
return result;
}
function intersect_filter_indexOf(a, b) {
return a.filter(el => b.indexOf(el) !== -1);
}
function intersect_filter_in(a, b) {
const map = b.reduce((map, el) => {map[el] = true; return map}, {});
return a.filter(el => el in map);
}
function intersect_for_in(a, b) {
const result = [];
const map = {};
for (let i = 0, length = b.length; i < length; ++i) {
map[b[i]] = true;
}
for (let i = 0, length = a.length; i < length; ++i) {
if (a[i] in map) result.push(a[i]);
}
return result;
}
function intersect_filter_includes(a, b) {
return a.filter(el => b.includes(el));
}
function intersect_filter_has_this(a, b) {
return a.filter(Set.prototype.has, new Set(b));
}
function intersect_filter_has_arrow(a, b) {
const set = new Set(b);
return a.filter(el => set.has(el));
}
function intersect_for_has(a, b) {
const result = [];
const set = new Set(b);
for (let i = 0, length = a.length; i < length; ++i) {
if (set.has(a[i])) result.push(a[i]);
}
return result;
}
Results in Firefox 53:
Ops/sec on large arrays (10,000 elements):
filter + has (this) 523 (this answer)
for + has 482
for-loop + in 279
filter + in 242
for-loops 24
filter + includes 14
filter + indexOf 10
Ops/sec on small arrays (100 elements):
for-loop + in 384,426
filter + in 192,066
for-loops 159,137
filter + includes 104,068
filter + indexOf 71,598
filter + has (this) 43,531 (this answer)
filter + has (arrow function) 35,588
My contribution in ES6 terms. In general it finds the intersection of an array with indefinite number of arrays provided as arguments.
Array.prototype.intersect = function(...a) {
return [this,...a].reduce((p,c) => p.filter(e => c.includes(e)));
}
var arrs = [[0,2,4,6,8],[4,5,6,7],[4,6]],
arr = [0,1,2,3,4,5,6,7,8,9];
document.write("<pre>" + JSON.stringify(arr.intersect(...arrs)) + "</pre>");
How about just using associative arrays?
function intersect(a, b) {
var d1 = {};
var d2 = {};
var results = [];
for (var i = 0; i < a.length; i++) {
d1[a[i]] = true;
}
for (var j = 0; j < b.length; j++) {
d2[b[j]] = true;
}
for (var k in d1) {
if (d2[k])
results.push(k);
}
return results;
}
edit:
// new version
function intersect(a, b) {
var d = {};
var results = [];
for (var i = 0; i < b.length; i++) {
d[b[i]] = true;
}
for (var j = 0; j < a.length; j++) {
if (d[a[j]])
results.push(a[j]);
}
return results;
}
The performance of #atk's implementation for sorted arrays of primitives can be improved by using .pop rather than .shift.
function intersect(array1, array2) {
var result = [];
// Don't destroy the original arrays
var a = array1.slice(0);
var b = array2.slice(0);
var aLast = a.length - 1;
var bLast = b.length - 1;
while (aLast >= 0 && bLast >= 0) {
if (a[aLast] > b[bLast] ) {
a.pop();
aLast--;
} else if (a[aLast] < b[bLast] ){
b.pop();
bLast--;
} else /* they're equal */ {
result.push(a.pop());
b.pop();
aLast--;
bLast--;
}
}
return result;
}
I created a benchmark using jsPerf. It's about three times faster to use .pop.
If you need to have it handle intersecting multiple arrays:
const intersect = (a1, a2, ...rest) => {
const a12 = a1.filter(value => a2.includes(value))
if (rest.length === 0) { return a12; }
return intersect(a12, ...rest);
};
console.log(intersect([1,2,3,4,5], [1,2], [1, 2, 3,4,5], [2, 10, 1]))
Sort it
check one by one from the index 0, create new array from that.
Something like this, Not tested well though.
function intersection(x,y){
x.sort();y.sort();
var i=j=0;ret=[];
while(i<x.length && j<y.length){
if(x[i]<y[j])i++;
else if(y[j]<x[i])j++;
else {
ret.push(x[i]);
i++,j++;
}
}
return ret;
}
alert(intersection([1,2,3], [2,3,4,5]));
PS:The algorithm only intended for Numbers and Normal Strings, intersection of arbitary object arrays may not work.
Using jQuery:
var a = [1,2,3];
var b = [2,3,4,5];
var c = $(b).not($(b).not(a));
alert(c);
A tiny tweak to the smallest one here (the filter/indexOf solution), namely creating an index of the values in one of the arrays using a JavaScript object, will reduce it from O(N*M) to "probably" linear time. source1 source2
function intersect(a, b) {
var aa = {};
a.forEach(function(v) { aa[v]=1; });
return b.filter(function(v) { return v in aa; });
}
This isn't the very simplest solution (it's more code than filter+indexOf), nor is it the very fastest (probably slower by a constant factor than intersect_safe()), but seems like a pretty good balance. It is on the very simple side, while providing good performance, and it doesn't require pre-sorted inputs.
For arrays containing only strings or numbers you can do something with sorting, as per some of the other answers. For the general case of arrays of arbitrary objects I don't think you can avoid doing it the long way. The following will give you the intersection of any number of arrays provided as parameters to arrayIntersection:
var arrayContains = Array.prototype.indexOf ?
function(arr, val) {
return arr.indexOf(val) > -1;
} :
function(arr, val) {
var i = arr.length;
while (i--) {
if (arr[i] === val) {
return true;
}
}
return false;
};
function arrayIntersection() {
var val, arrayCount, firstArray, i, j, intersection = [], missing;
var arrays = Array.prototype.slice.call(arguments); // Convert arguments into a real array
// Search for common values
firstArray = arrays.pop();
if (firstArray) {
j = firstArray.length;
arrayCount = arrays.length;
while (j--) {
val = firstArray[j];
missing = false;
// Check val is present in each remaining array
i = arrayCount;
while (!missing && i--) {
if ( !arrayContains(arrays[i], val) ) {
missing = true;
}
}
if (!missing) {
intersection.push(val);
}
}
}
return intersection;
}
arrayIntersection( [1, 2, 3, "a"], [1, "a", 2], ["a", 1] ); // Gives [1, "a"];
Simplest, fastest O(n) and shortest way:
function intersection (a, b) {
const setA = new Set(a);
return b.filter(value => setA.has(value));
}
console.log(intersection([1,2,3], [2,3,4,5]))
#nbarbosa has almost the same answer but he cast both arrays to Set and then back to array. There is no need for any extra casting.
Another indexed approach able to process any number of arrays at once:
// Calculate intersection of multiple array or object values.
function intersect (arrList) {
var arrLength = Object.keys(arrList).length;
// (Also accepts regular objects as input)
var index = {};
for (var i in arrList) {
for (var j in arrList[i]) {
var v = arrList[i][j];
if (index[v] === undefined) index[v] = 0;
index[v]++;
};
};
var retv = [];
for (var i in index) {
if (index[i] == arrLength) retv.push(i);
};
return retv;
};
It works only for values that can be evaluated as strings and you should pass them as an array like:
intersect ([arr1, arr2, arr3...]);
...but it transparently accepts objects as parameter or as any of the elements to be intersected (always returning array of common values). Examples:
intersect ({foo: [1, 2, 3, 4], bar: {a: 2, j:4}}); // [2, 4]
intersect ([{x: "hello", y: "world"}, ["hello", "user"]]); // ["hello"]
EDIT: I just noticed that this is, in a way, slightly buggy.
That is: I coded it thinking that input arrays cannot itself contain repetitions (as provided example doesn't).
But if input arrays happen to contain repetitions, that would produce wrong results. Example (using below implementation):
intersect ([[1, 3, 4, 6, 3], [1, 8, 99]]);
// Expected: [ '1' ]
// Actual: [ '1', '3' ]
Fortunately this is easy to fix by simply adding second level indexing. That is:
Change:
if (index[v] === undefined) index[v] = 0;
index[v]++;
by:
if (index[v] === undefined) index[v] = {};
index[v][i] = true; // Mark as present in i input.
...and:
if (index[i] == arrLength) retv.push(i);
by:
if (Object.keys(index[i]).length == arrLength) retv.push(i);
Complete example:
// Calculate intersection of multiple array or object values.
function intersect (arrList) {
var arrLength = Object.keys(arrList).length;
// (Also accepts regular objects as input)
var index = {};
for (var i in arrList) {
for (var j in arrList[i]) {
var v = arrList[i][j];
if (index[v] === undefined) index[v] = {};
index[v][i] = true; // Mark as present in i input.
};
};
var retv = [];
for (var i in index) {
if (Object.keys(index[i]).length == arrLength) retv.push(i);
};
return retv;
};
intersect ([[1, 3, 4, 6, 3], [1, 8, 99]]); // [ '1' ]
With some restrictions on your data, you can do it in linear time!
For positive integers: use an array mapping the values to a "seen/not seen" boolean.
function intersectIntegers(array1,array2) {
var seen=[],
result=[];
for (var i = 0; i < array1.length; i++) {
seen[array1[i]] = true;
}
for (var i = 0; i < array2.length; i++) {
if ( seen[array2[i]])
result.push(array2[i]);
}
return result;
}
There is a similar technique for objects: take a dummy key, set it to "true" for each element in array1, then look for this key in elements of array2. Clean up when you're done.
function intersectObjects(array1,array2) {
var result=[];
var key="tmpKey_intersect"
for (var i = 0; i < array1.length; i++) {
array1[i][key] = true;
}
for (var i = 0; i < array2.length; i++) {
if (array2[i][key])
result.push(array2[i]);
}
for (var i = 0; i < array1.length; i++) {
delete array1[i][key];
}
return result;
}
Of course you need to be sure the key didn't appear before, otherwise you'll be destroying your data...
function intersection(A,B){
var result = new Array();
for (i=0; i<A.length; i++) {
for (j=0; j<B.length; j++) {
if (A[i] == B[j] && $.inArray(A[i],result) == -1) {
result.push(A[i]);
}
}
}
return result;
}
For simplicity:
// Usage
const intersection = allLists
.reduce(intersect, allValues)
.reduce(removeDuplicates, []);
// Implementation
const intersect = (intersection, list) =>
intersection.filter(item =>
list.some(x => x === item));
const removeDuplicates = (uniques, item) =>
uniques.includes(item) ? uniques : uniques.concat(item);
// Example Data
const somePeople = [bob, doug, jill];
const otherPeople = [sarah, bob, jill];
const morePeople = [jack, jill];
const allPeople = [...somePeople, ...otherPeople, ...morePeople];
const allGroups = [somePeople, otherPeople, morePeople];
// Example Usage
const intersection = allGroups
.reduce(intersect, allPeople)
.reduce(removeDuplicates, []);
intersection; // [jill]
Benefits:
dirt simple
data-centric
works for arbitrary number of lists
works for arbitrary lengths of lists
works for arbitrary types of values
works for arbitrary sort order
retains shape (order of first appearance in any array)
exits early where possible
memory safe, short of tampering with Function / Array prototypes
Drawbacks:
higher memory usage
higher CPU usage
requires an understanding of reduce
requires understanding of data flow
You wouldn't want to use this for 3D engine or kernel work, but if you have problems getting this to run in an event-based app, your design has bigger problems.
I'll contribute with what has been working out best for me:
if (!Array.prototype.intersect){
Array.prototype.intersect = function (arr1) {
var r = [], o = {}, l = this.length, i, v;
for (i = 0; i < l; i++) {
o[this[i]] = true;
}
l = arr1.length;
for (i = 0; i < l; i++) {
v = arr1[i];
if (v in o) {
r.push(v);
}
}
return r;
};
}
A functional approach with ES2015
A functional approach must consider using only pure functions without side effects, each of which is only concerned with a single job.
These restrictions enhance the composability and reusability of the functions involved.
// small, reusable auxiliary functions
const createSet = xs => new Set(xs);
const filter = f => xs => xs.filter(apply(f));
const apply = f => x => f(x);
// intersection
const intersect = xs => ys => {
const zs = createSet(ys);
return filter(x => zs.has(x)
? true
: false
) (xs);
};
// mock data
const xs = [1,2,2,3,4,5];
const ys = [0,1,2,3,3,3,6,7,8,9];
// run it
console.log( intersect(xs) (ys) );
Please note that the native Set type is used, which has an advantageous
lookup performance.
Avoid duplicates
Obviously repeatedly occurring items from the first Array are preserved, while the second Array is de-duplicated. This may be or may be not the desired behavior. If you need a unique result just apply dedupe to the first argument:
// auxiliary functions
const apply = f => x => f(x);
const comp = f => g => x => f(g(x));
const afrom = apply(Array.from);
const createSet = xs => new Set(xs);
const filter = f => xs => xs.filter(apply(f));
// intersection
const intersect = xs => ys => {
const zs = createSet(ys);
return filter(x => zs.has(x)
? true
: false
) (xs);
};
// de-duplication
const dedupe = comp(afrom) (createSet);
// mock data
const xs = [1,2,2,3,4,5];
const ys = [0,1,2,3,3,3,6,7,8,9];
// unique result
console.log( intersect(dedupe(xs)) (ys) );
Compute the intersection of any number of Arrays
If you want to compute the intersection of an arbitrarily number of Arrays just compose intersect with foldl. Here is a convenience function:
// auxiliary functions
const apply = f => x => f(x);
const uncurry = f => (x, y) => f(x) (y);
const createSet = xs => new Set(xs);
const filter = f => xs => xs.filter(apply(f));
const foldl = f => acc => xs => xs.reduce(uncurry(f), acc);
// intersection
const intersect = xs => ys => {
const zs = createSet(ys);
return filter(x => zs.has(x)
? true
: false
) (xs);
};
// intersection of an arbitrarily number of Arrays
const intersectn = (head, ...tail) => foldl(intersect) (head) (tail);
// mock data
const xs = [1,2,2,3,4,5];
const ys = [0,1,2,3,3,3,6,7,8,9];
const zs = [0,1,2,3,4,5,6];
// run
console.log( intersectn(xs, ys, zs) );
.reduce to build a map, and .filter to find the intersection. delete within the .filter allows us to treat the second array as though it's a unique set.
function intersection (a, b) {
var seen = a.reduce(function (h, k) {
h[k] = true;
return h;
}, {});
return b.filter(function (k) {
var exists = seen[k];
delete seen[k];
return exists;
});
}
I find this approach pretty easy to reason about. It performs in constant time.
I have written an intesection function which can even detect intersection of array of objects based on particular property of those objects.
For instance,
if arr1 = [{id: 10}, {id: 20}]
and arr2 = [{id: 20}, {id: 25}]
and we want intersection based on the id property, then the output should be :
[{id: 20}]
As such, the function for the same (note: ES6 code) is :
const intersect = (arr1, arr2, accessors = [v => v, v => v]) => {
const [fn1, fn2] = accessors;
const set = new Set(arr2.map(v => fn2(v)));
return arr1.filter(value => set.has(fn1(value)));
};
and you can call the function as:
intersect(arr1, arr2, [elem => elem.id, elem => elem.id])
Also note: this function finds intersection considering the first array is the primary array and thus the intersection result will be that of the primary array.
This function avoids the N^2 problem, taking advantage of the power of dictionaries. Loops through each array only once, and a third and shorter loop to return the final result.
It also supports numbers, strings, and objects.
function array_intersect(array1, array2)
{
var mergedElems = {},
result = [];
// Returns a unique reference string for the type and value of the element
function generateStrKey(elem) {
var typeOfElem = typeof elem;
if (typeOfElem === 'object') {
typeOfElem += Object.prototype.toString.call(elem);
}
return [typeOfElem, elem.toString(), JSON.stringify(elem)].join('__');
}
array1.forEach(function(elem) {
var key = generateStrKey(elem);
if (!(key in mergedElems)) {
mergedElems[key] = {elem: elem, inArray2: false};
}
});
array2.forEach(function(elem) {
var key = generateStrKey(elem);
if (key in mergedElems) {
mergedElems[key].inArray2 = true;
}
});
Object.values(mergedElems).forEach(function(elem) {
if (elem.inArray2) {
result.push(elem.elem);
}
});
return result;
}
If there is a special case that cannot be solved, just by modifying the generateStrKey function, it could surely be solved. The trick of this function is that it uniquely represents each different data according to type and value.
This variant has some performance improvements. Avoid loops in case any array is empty. It also starts by walking through the shorter array first, so if it finds all the values of the first array in the second array, exits the loop.
function array_intersect(array1, array2)
{
var mergedElems = {},
result = [],
firstArray, secondArray,
firstN = 0,
secondN = 0;
function generateStrKey(elem) {
var typeOfElem = typeof elem;
if (typeOfElem === 'object') {
typeOfElem += Object.prototype.toString.call(elem);
}
return [typeOfElem, elem.toString(), JSON.stringify(elem)].join('__');
}
// Executes the loops only if both arrays have values
if (array1.length && array2.length)
{
// Begins with the shortest array to optimize the algorithm
if (array1.length < array2.length) {
firstArray = array1;
secondArray = array2;
} else {
firstArray = array2;
secondArray = array1;
}
firstArray.forEach(function(elem) {
var key = generateStrKey(elem);
if (!(key in mergedElems)) {
mergedElems[key] = {elem: elem, inArray2: false};
// Increases the counter of unique values in the first array
firstN++;
}
});
secondArray.some(function(elem) {
var key = generateStrKey(elem);
if (key in mergedElems) {
if (!mergedElems[key].inArray2) {
mergedElems[key].inArray2 = true;
// Increases the counter of matches
secondN++;
// If all elements of first array have coincidence, then exits the loop
return (secondN === firstN);
}
}
});
Object.values(mergedElems).forEach(function(elem) {
if (elem.inArray2) {
result.push(elem.elem);
}
});
}
return result;
}
Here is underscore.js implementation:
_.intersection = function(array) {
if (array == null) return [];
var result = [];
var argsLength = arguments.length;
for (var i = 0, length = array.length; i < length; i++) {
var item = array[i];
if (_.contains(result, item)) continue;
for (var j = 1; j < argsLength; j++) {
if (!_.contains(arguments[j], item)) break;
}
if (j === argsLength) result.push(item);
}
return result;
};
Source: http://underscorejs.org/docs/underscore.html#section-62
Create an Object using one array and loop through the second array to check if the value exists as key.
function intersection(arr1, arr2) {
var myObj = {};
var myArr = [];
for (var i = 0, len = arr1.length; i < len; i += 1) {
if(myObj[arr1[i]]) {
myObj[arr1[i]] += 1;
} else {
myObj[arr1[i]] = 1;
}
}
for (var j = 0, len = arr2.length; j < len; j += 1) {
if(myObj[arr2[j]] && myArr.indexOf(arr2[j]) === -1) {
myArr.push(arr2[j]);
}
}
return myArr;
}
I think using an object internally can help with computations and could be performant too.
// Approach maintains a count of each element and works for negative elements too
function intersect(a,b){
const A = {};
a.forEach((v)=>{A[v] ? ++A[v] : A[v] = 1});
const B = {};
b.forEach((v)=>{B[v] ? ++B[v] : B[v] = 1});
const C = {};
Object.entries(A).map((x)=>C[x[0]] = Math.min(x[1],B[x[0]]))
return Object.entries(C).map((x)=>Array(x[1]).fill(Number(x[0]))).flat();
}
const x = [1,1,-1,-1,0,0,2,2];
const y = [2,0,1,1,1,1,0,-1,-1,-1];
const result = intersect(x,y);
console.log(result); // (7) [0, 0, 1, 1, 2, -1, -1]
I am using map even object could be used.
//find intersection of 2 arrs
const intersections = (arr1,arr2) => {
let arrf = arr1.concat(arr2)
let map = new Map();
let union = [];
for(let i=0; i<arrf.length; i++){
if(map.get(arrf[i])){
map.set(arrf[i],false);
}else{
map.set(arrf[i],true);
}
}
map.forEach((v,k)=>{if(!v){union.push(k);}})
return union;
}
This is a proposed standard: With the currently stage 2 proposal https://github.com/tc39/proposal-set-methods, you could use
mySet.intersection(mySet2);
Until then, you could use Immutable.js's Set, which inspired that proposal
Immutable.Set(mySet).intersect(mySet2)
I extended tarulen's answer to work with any number of arrays. It also should work with non-integer values.
function intersect() {
const last = arguments.length - 1;
var seen={};
var result=[];
for (var i = 0; i < last; i++) {
for (var j = 0; j < arguments[i].length; j++) {
if (seen[arguments[i][j]]) {
seen[arguments[i][j]] += 1;
}
else if (!i) {
seen[arguments[i][j]] = 1;
}
}
}
for (var i = 0; i < arguments[last].length; i++) {
if ( seen[arguments[last][i]] === last)
result.push(arguments[last][i]);
}
return result;
}
If your arrays are sorted, this should run in O(n), where n is min( a.length, b.length )
function intersect_1d( a, b ){
var out=[], ai=0, bi=0, acurr, bcurr, last=Number.MIN_SAFE_INTEGER;
while( ( acurr=a[ai] )!==undefined && ( bcurr=b[bi] )!==undefined ){
if( acurr < bcurr){
if( last===acurr ){
out.push( acurr );
}
last=acurr;
ai++;
}
else if( acurr > bcurr){
if( last===bcurr ){
out.push( bcurr );
}
last=bcurr;
bi++;
}
else {
out.push( acurr );
last=acurr;
ai++;
bi++;
}
}
return out;
}