Related
This is a slightly odd/unique request. I am trying to achieve a result where e.g "yes" becomes, "yyes", "yees", "yess", "yyees", "yyess", "yyeess".
I have looked at this: Find all lowercase and uppercase combinations of a string in Javascript which completes it for capitalisation, however my understanding is prohibiting me from manipulating this into character duplication (if this method is even possible to use in this way).
export function letterDuplication(level: number, input: string){
const houseLength = input.length;
if (level == 1){
var resultsArray: string[] = [];
const letters = input.split("");
const permCount = 1 << input.length;
for (let perm = 0; perm < permCount; perm++) {
// Update the capitalization depending on the current permutation
letters.reduce((perm, letter, i) => {
if (perm & 1) {
letters[i] = (letter.slice(0, perm) + letter.slice(perm-1, perm) + letter.slice(perm));
} else {
letters[i] = (letter.slice(0, perm - 1) + letter.slice(perm, perm) + letter.slice(perm))
}
return perm >> 1;
}, perm);
var result = letters.join("");
console.log(result);
resultsArray[perm] = result;
}
If I haven't explained this particularly well please let me know and I'll clarify. I'm finding it quite the challenge!
General idea
To get the list of all words we can get from ordered array of letters, we need to get all combinations of letters, passed 1 or 2 times into word, like:
word = 'sample'
array = 's'{1,2} + 'a'{1,2} + 'm'{1,2} + 'p'{1,2} + 'l'{1,2} + 'e'{1,2}
Amount of all possible words equal to 2 ^ word.length (8 for "yes"), so we can build binary table with 8 rows that represent all posible combinations just via convering numbers from 0 to 7 (from decimal to binary):
0 -> 000
1 -> 001
2 -> 010
3 -> 011
4 -> 100
5 -> 101
6 -> 110
7 -> 111
Each decimal we can use as pattern for new word, where 0 means letter must be used once, and 1 - letter must be used twice:
0 -> 000 -> yes
1 -> 001 -> yess
2 -> 010 -> yees
3 -> 011 -> yeess
4 -> 100 -> yyes
5 -> 101 -> yyess
6 -> 110 -> yyees
7 -> 111 -> yyeess
Code
So, your code representation may looks like this:
// Initial word
const word = 'yes';
// List of all possible words
const result = [];
// Iterating (2 ^ word.length) times
for (let i = 0; i < Math.pow(2, word.length); i++) {
// Get binary pattern for each word
const bin = decToBin(i, word.length);
// Make a new word from pattern ...
let newWord = '';
for (let i = 0; i < word.length; i++) {
// ... by adding letter 1 or 2 times based on bin char value
newWord += word[i].repeat(+bin[i] ? 2 : 1);
}
result.push(newWord);
}
// Print result (all possible words)
console.log(result);
// Method for decimal to binary convertion with leading zeroes
// (always returns string with length = len)
function decToBin(x, len) {
let rem, bin = 0, i = 1;
while (x != 0) {
rem = x % 2;
x = parseInt(x / 2);
bin += rem * i;
i = i * 10;
}
bin = bin.toString();
return '0'.repeat(len - bin.length) + bin;
}
Maybe this example can help you. It's a bit not neat and not optimal, but it seems to work:
function getCombinations(word = '') {
const allCombination = [];
const generate = (n, arr, i = 0) => {
if (i === n) {
return allCombination.push([...arr]);
} else {
arr[i] = 0;
generate(n, arr, i + 1);
arr[i] = 1;
generate(n, arr, i + 1);
}
}
generate(word.length, Array(word.length).fill(0));
return allCombination.map((el) => {
return el.map((isCopy, i) => isCopy ? word[i].repeat(2) : word[i]).join('')
});
}
console.log(getCombinations('yes'));
console.log(getCombinations('cat'));
I have to make a URL shortener for query strings. Have spent few days trying to compress array data into base64 strings. Thinking that the best approach may be to interpret something like "[[1,2,9,3],[1,0,2],[39,4]]" as base13 with numbers 0-9 and [], symbols.
how the current algorithm works:
convert the stringified arrays into an array of base13, where each element represents 1 unique character, convert this array to base10 number, convert this number to base 64 string.
but the problem is when converting the base13 array to base10 number, it makes large numbers like 5.304781188371057e+86 which cant be held in js.
I am open to alternative solutions of course, but please do not suggest something like creating a database of URLs as it won't work as I have up to 51!*51! unique URLs, better to just make a compact encodable and decodable query string and decode it as soon as the website is accessed.
//convert stringified array to array of base13(each element = each digit of base13 number)
function stringToArray(string)
{
let charSet = "[],1234567890";
let array = [];
for(let i = 0; i < string.length; i++)
{
array.push(charSet.indexOf(string[i]));
}
return array;
}
//convert base13 array to one large decimal number
function arrayToDecimal(array, base)
{
var decimal = 0;
for(let i = 0; i < array.length; i++)
{
decimal += array[i] * Math.pow(base, i)
}
return decimal;
}
//convert decimal number back to array
function decimalToArray(decimal, base)
{
var quotient = decimal;
var remainder = [];
while(quotient > base)
{
remainder.push(quotient % base)
quotient = Math.floor(quotient / base);
}
remainder.push(quotient % base)
return remainder;
}
const alphabet = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/';
// binary to string lookup table
const b2s = alphabet.split('');
// string to binary lookup table
// 123 == 'z'.charCodeAt(0) + 1
const s2b = new Array(123);
for(let i = 0; i < alphabet.length; i++)
{
s2b[alphabet.charCodeAt(i)] = i;
}
// number to base64
const ntob = (number) =>
{
if(number < 0) return `-${ntob(-number)}`;
let lo = number >>> 0;
let hi = (number / 4294967296) >>> 0;
let right = '';
while(hi > 0)
{
right = b2s[0x3f & lo] + right;
lo >>>= 6;
lo |= (0x3f & hi) << 26;
hi >>>= 6;
}
let left = '';
do {
left = b2s[0x3f & lo] + left;
lo >>>= 6;
} while(lo > 0);
return left + right;
};
// base64 to number
const bton = (base64) =>
{
let number = 0;
const sign = base64.charAt(0) === '-' ? 1 : 0;
for(let i = sign; i < base64.length; i++)
{
number = number * 64 + s2b[base64.charCodeAt(i)];
}
return sign ? -number : number;
};
console.log(decimalToArray(bton(ntob(arrayToDecimal([1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1], 13))), 13))
//encoded and decoded, works output:[1,1,1,1,1,1,1,1,1,1,1,1,1]
console.log(arrayToDecimal([1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1], 13))
//encoding doesnt work, array to decimal converts to 5.304781188371057e+86```
An interesting problem... The first thing you will need to assess is whether the base conversion compression you're seeking is worthwhile. Ie, how many base 64 characters are required to represent n characters of base 13? This involves solving...
13 ** n = 64 ** x
Solving for x, we get...
x = n * log(13) / log(64)
Ie, for every n digits of base 13, how many digits of base 64 are required. A sampling of a few values of n returns...
n = 6, x = 3.70
n = 7, x = 4.31
n = 8, x = 4.93
n = 9, x = 5.55
n = 10, x = 6.17
n = 11, x = 6.78
n = 12, x = 7.40
n = 13, x = 8.01
n = 14, x = 8.63
n = 15, x = 9.25
n = 16, x = 9.86
So how to interpret this? If you have 10 digits of base 13, you're going to need 7 digits (6.17 rounded up) of base 64. So the best ratio is when x is equal to, or just under, a whole number. So, 8 digits of base 13 requires 5 digits of base 64, achieving a best case of 5/8 or 62.5% compression ratio.
Assuming that's good enough to meet your requirement, then the following function converts the "base13" string to base 64.
const base13Chars = "0123456789[],";
const base64Chars = '0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz-_';
// see https://en.wikipedia.org/wiki/Query_string for URL parameter allowable characters.
function base13toBase64(x13) {
base13 = x13.split("").map( c => base13Chars.indexOf(c) );
// Make the array an even multiple of 8
for (i = base13.length; i % 8 !==0; i++) {
base13[i] = 0;
}
x64 = "";
for (i = 0; i < base13.length; i += 8) {
// Calculate base13 value of the next 8 characters.
let n = 0;
for (j = 0; j < 8; j++) {
n = n * 13 + base13[i + j];
}
// Now calculate the base64 of n.
for (j = 0; j < 5; j++) {
x64 = x64 + base64Chars.substr(n % 64,1);
n = Math.floor(n / 64);
}
}
return x64;
}
Running the above...
base13toBase64( "[[1,2,9,3],[1,0,2],[39,4]]" ) returns "ilYKerYlgEJ4PxAAjaJi"
Note that the original value is a length of 26 characters, and the base64 value is 20 characters, so the compression ratio is 77%, not quite the optimal 62.5%. This is because of the padding to bring the original array to 32 characters, an even multiple of 8. The longer the string to encode, though, the closer the ratio will be to 62.5%.
Then, on the server side you'll need the constants above plus the following function to "uncompress" the base64 to the base13 stringified URL...
function base64toBase13(x64) {
base64 = x64.split("").map( c => base64Chars.indexOf(c) );
x13 = "";
for (i = 0; i < base64.length; i += 5) {
// Calculate base64 value of the next 5 characters.
let n = 0;
for (j = 5 - 1; 0 <= j; j--) {
n = n * 64 + base64[i + j];
}
// Now calculate the base13 of n.
let x = "";
for (j = 0; j < 8; j++) {
x = base13Chars.substr(n % 13,1) + x;
n = Math.floor(n / 13);
}
x13 = x13 + x;
}
// Removed the trailing 0's as a result of the buffering in
// base13toBase64 to make the array an even multiple of 8.
while (x13.substr(-1,1) === "0") {
x13 = x13.substr(0, x13.length - 1);
}
return x13;
}
Running the above...
base64toBase13 ( "ilYKerYlgEJ4PxAAjaJi" ) returns "[[1,2,9,3],[1,0,2],[39,4]]"
Hope this helps...
The best compression is when you can leave stuff out.
assuming your data structure is Array<Array<int>> given by the one sample, we can leave out pretty much everything that doesn't contribute to the data itself.
I'm not compressing the string, but the data itself with 1 b64Character / 5 bits needed to represent a number. as for the structure we only store the number of sub-arrays and their respective lengths; so more or less an additional character per Array in your data.
boils down to:
function encode(data) {
const alphabet = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/';
let str = "";
function encode(nr, hasMoreDigits) {
if (nr > 31) {
// I need more bits/characters to encode this number.
//encode the more significant bits with the 0b100000 flag
encode(nr >>> 5, 32);
}
// 0b011111 payload | 0b100000 flag
const index = nr & 31 | hasMoreDigits;
str += alphabet[index];
}
encode(data.length);
data.forEach(arr => {
encode(arr.length);
arr.forEach(v => encode(v >>> 0 /* int32 -> uint32 */));
});
return str;
}
function decode(str) {
const alphabet = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/';
let i = 0;
function parse() {
let nr = 0,
hasMoreDigits;
do {
const index = alphabet.indexOf(str.charAt(i++));
nr = nr << 5 | index & 31; // 0b011111 payload
hasMoreDigits = index & 32; // 0b100000 flag
} while (hasMoreDigits);
return nr; // int32 due to the bit operations above
}
let data = Array(parse());
for (let j = 0; j < data.length; ++j) {
let arr = data[j] = Array(parse());
for (let k = 0; k < arr.length; ++k) {
arr[k] = parse();
}
}
return data;
}
let data = [
[1, 2, 9, 3],
[1, 0, 2],
[39, 4]
];
let text = encode(data);
let data2 = decode(text);
console.log("input:", data);
console.log("encoded:", text, "length:", text.length);
console.log("output:", data2);
console.log("equal:", JSON.stringify(data) === JSON.stringify(data2));
.as-console-wrapper{top:0;max-height:100%!important}
The encoding of the numbers. Ideally you would encode a number as binary with a static size, but this means 32bit/int which would be 6characters/number, so multibytes.
We split the number into chunks of 'n' bits, ignore the leading zeroes and encode the rest. ideally we can encode small number with very few characters, downside: we loose 1bit/chunk if n is too small and the average numbers are big. It's a tradeoff; that's why I left this configurable.
the current format is 6bits/number. 1 for the Structure, 5 bits as payload. In the format (1.....)*0.....
I would suggest you directly encode the Base13 string into Base64.
Although that might not result in better compression than your solution, it removes the heavy multiplications you are performing. Moreover how do you guarantee that no collisions happen when converting through arrayToDecimal ?
the purpose of this code is to to write a function which increments a string, to create a new string. If the string already ends with a number, the number should be incremented by 1. If the string does not end with a number the number 1 should be appended to the new string. e.g. "foo123" --> "foo124" or "foo" --> "foo1".
With my code below, pretty much all my test cases are passed except a corner case for "foo999" did not print out "foo1000". I know that there should be a way to do with regex to fix my problem, but I am not too familiar with it. Can anyone please help?
function incrementString (input) {
var reg = /[0-9]/;
var result = "";
if(reg.test(input[input.length - 1]) === true){
input = input.split("");
for(var i = 0; i < input.length; i++){
if(parseInt(input[i]) === NaN){
result += input[i];
}
else if(i === input.length - 1){
result += (parseInt(input[i]) + 1).toString();
}
else{
result += input[i];
}
}
return result;
}
else if (reg.test(input[input.length - 1]) === false){
return input += 1;
}
}
You can use replace with a callback:
'foo'.replace(/(\d*)$/, function($0, $1) { return $1*1+1; });
//=> "foo1"
'foo999'.replace(/(\d*)$/, function($0, $1) { return $1*1+1; });
//=> "foo1000"
'foo123'.replace(/(\d*)$/, function($0, $1) { return $1*1+1; });
//=> "foo124"
Explanation:
/(\d*)$/ # match 0 or more digits at the end of string
function($0, $1) {...} # callback function with 2nd parameter as matched group #1
return $1*1+1; # return captured number+1. $1*1 is a trick to convert
# string to number
The most concise way that also accounts for leading zeros, non-numeric endings, and empty strings I've seen is:
''.replace(/[0-8]?9*$/, w => ++w)
//=> 1
'foo'.replace(/[0-8]?9*$/, w => ++w)
//=> foo1
'foo099'.replace(/[0-8]?9*$/, w => ++w)
//=> foo100
'foo999'.replace(/[0-8]?9*$/, w => ++w)
//=> foo1000
function pad(number, length, filler) {
number = number + "";
if (number.length < length) {
for (var i = number.length; i < length; i += 1) {
number = filler + number;
}
}
return number;
}
function incrementString (input) {
var orig = input.match(/\d+$/);
if (orig.length === 1) {
orig = pad(parseInt(orig[0]) + 1, orig[0].length, '0');
input = input.replace(/\d+$/, orig);
return input;
}
return input + "1";
}
What does it do?
It first checks if there is a trailing number. If yes, increment it and pad left it with zeros (with the function "pad" which you'd be able to sort it out yourself).
string.replace is a function which works with argument 1 the substring to search (string, regex), argument 2 the element to replace with (string, function).
In this case I've used a regex as first argument and the incremented, padded number.
The regex is pretty simple: \d means "integer" and + means "one or more of the preceeding", which means one or more digits. $ means the end of the string.
More info about regular expressions (in JavaScript): https://developer.mozilla.org/en/docs/Web/JavaScript/Reference/Global_Objects/RegExp (thanks #Casimir et Hippolyte for the link)
I think you can simplify your code greatly:
function incrementString(input) {
var splits = input.split(/(\d+)$/),
num = 1;
if (splits[1] !== undefined) num = parseInt(splits[1]) + 1;
return splits[0] + num;
}
This checks for any number of digits at the end of your string.
function incrementString (strng) {
//separateing number from string
let x = (strng).replace( /^\D+/g, '');
//getting the length of original number from the string
let len = x.length;
//getting the string part from strng
str = strng.split(x);
//incrementing number by 1
let number = Number(x) + 1 + '';
//padding the number with 0 to make it's length exactly to orignal number
while(number.length < len){
number = '0' + number;
}
//new string by joining string and the incremented number
str = (str + number).split(',').join('');
//return new string
return str;
}
My quick answer will be :
let newStr = string.split('');
let word = [];
let num = [];
for (let i = 0 ; i<string.length ;i++){
isNaN(string[i])? word.push(string[i]) : num.push(string[i])
}
let l=num.length-1;
let pureNum=0;
for (let i = 0 ; i<num.length ;i++){
pureNum += num[i] * Math.pow(10,l);
l--;
}
let wordNum = (pureNum+1).toString().split('');
for (let i = wordNum.length ; i<num.length ;i++){
wordNum.unshift("0");
}
return word.join("")+wordNum.join("");
}
I'm creating some client side functions for a mappable spreadsheet export feature.
I'm using jQuery to manage the sort order of the columns, but each column is ordered like an Excel spreadsheet i.e. a b c d e......x y z aa ab ac ad etc etc
How can I generate a number as a letter? Should I define a fixed array of values? Or is there a dynamic way to generate this?
I think you're looking for something like this
function colName(n) {
var ordA = 'a'.charCodeAt(0);
var ordZ = 'z'.charCodeAt(0);
var len = ordZ - ordA + 1;
var s = "";
while(n >= 0) {
s = String.fromCharCode(n % len + ordA) + s;
n = Math.floor(n / len) - 1;
}
return s;
}
// Example:
for(n = 0; n < 125; n++)
document.write(n + ":" + colName(n) + "<br>");
This is a very easy way:
function numberToLetters(num) {
let letters = ''
while (num >= 0) {
letters = 'ABCDEFGHIJKLMNOPQRSTUVWXYZ'[num % 26] + letters
num = Math.floor(num / 26) - 1
}
return letters
}
function getColumnDescription(i) {
const m = i % 26;
const c = String.fromCharCode(65 + m);
const r = i - m;
return r > 0
? `${getColumnDescription((r - 1) / 26)}${c}`
: `Column ${c}`
}
Usage:
getColumnDescription(15)
"Column P"
getColumnDescription(26)
"Column AA"
getColumnDescription(4460)
"Column FOO"
If you have your data in a two-dimensional array, e.g.
var data = [
['Day', 'score],
['Monday', 99],
];
you can map the rows/columns to spreadsheet cell numbers as follows (building on the code examples above):
function getSpreadSheetCellNumber(row, column) {
let result = '';
// Get spreadsheet column letter
let n = column;
while (n >= 0) {
result = String.fromCharCode(n % 26 + 65) + result;
n = Math.floor(n / 26) - 1;
}
// Get spreadsheet row number
result += `${row + 1}`;
return result;
};
E.g. the 'Day' value from data[0][0] would go in spreadsheet cell A1.
> getSpreadSheetCellNumber(0, 0)
> "A1"
This also works when you have 26+ columns:
> getSpreadSheetCellNumber(0, 26)
> "AA1"
You can use code like this, assuming that numbers contains the numbers of your columns. So after this code you'll get the string names for your columns:
var letters = ['a', 'b', 'c', ..., 'z'];
var numbers = [1, 2, 3, ...];
var columnNames = [];
for(var i=0;i<numbers.length;i++) {
var firstLetter = parseInt(i/letters.length) == 0 ? '' : letters[parseInt(i/letters.length)];
var secondLetter = letters[i%letters.length-1];
columnNames.push(firstLetter + secondLetter);
}
Simple recursive solution:
function numberToColumn(n) {
const res = 'ABCDEFGHIJKLMNOPQRSTUVWXYZ'[n % 26];
return n >= 26 ? numberToColumn(Math.floor(n / 26) - 1) + res : res;
}
Here is an alternative approach that relies on .toString(26). It uses conversion to base-26 and then translates the characters so they are in the a..z range:
const conv = ((base, alpha) => { // Closure for preparing the function
const map = Object.fromEntries(Array.from(alpha, (c, i) => [c, alpha[i + 10]]));
return n => (n + base).toString(26).replace(/o*p/, "").replace(/./g, m => map[m]);
})(parseInt("ooooooooop0", 26), "0123456789abcdefghijklmnopqrstuvwxyz");
// Example:
for (let n = 0; n < 29; n++) console.log(n, conv(n));
console.log("...");
for (let n = 690; n < 705; n++) console.log(n, conv(n));
About the magical number
The magical value "ooooooooop0" is derived as follows:
It is a number expressed in radix 26, in the standard way, i.e. where the ten digits also play a role, and then the first letters of the alphabet.
The greatest "digit" in this radix 26 is "p" (the 16th letter of the Latin alphabet), and "o" is the second greatest.
The magical value is formed by a long enough series of the one-but-greatest digit, followed by the greatest digit and ended by a 0.
As JavaScript integer numbers max out around Number.MAX_SAFE_INTEGER (greater integers numbers would suffer from rounding errors), there is no need to have a longer series of "o" than was selected. We can see that Number.MAX_SAFE_INTEGER.toString(26) has 12 digits, so precision is ensured up to 11 digits in radix 26, meaning we need 9 "o".
This magical number ensures that if we add units to it (in radix 26), we will always have a representation which starts with a series of "o" and then a "p". That is because at some point the last digit will wrap around to 0 again, and the "p" will also wrap around to 0, bringing the preceding "o" to "p". And so we have this invariant that the number always starts with zero or more "o" and then a "p".
More generic
The above magic number could be derived via code, and we could make it more generic by providing the target alphabet. The length of that target alphabet then also directly determines the radix (i.e. the number of characters in that string).
Here is the same output generated as above, but with a more generic function:
function createConverter(targetDigits) {
const radix = targetDigits.length,
alpha = "0123456789abcdefghijklmnopqrstuvwxyz",
map = Object.fromEntries(Array.from(alpha,
(src, i) => [src, targetDigits[i]]
)),
base = parseInt((alpha[radix-1]+'0').padStart(
Number.MAX_SAFE_INTEGER.toString(radix).length - 1, alpha[radix-2]
), radix),
trimmer = RegExp("^" + alpha[radix-2] + "*" + alpha[radix-1]);
return n => (n + base).toString(radix)
.replace(trimmer, "")
.replace(/./g, m => map[m]);
}
// Example:
const conv = createConverter("abcdefghijklmnopqrstuvwxyz");
for (let n = 0; n < 29; n++) console.log(n, conv(n));
console.log("...");
for (let n = 690; n < 705; n++) console.log(n, conv(n));
This can now easily be adapted to use a more reduced target alphabet (like without the letters "l" and "o"), giving a radix of 24 instead of 26:
function createConverter(targetDigits) {
const radix = targetDigits.length,
alpha = "0123456789abcdefghijklmnopqrstuvwxyz",
map = Object.fromEntries(Array.from(alpha,
(src, i) => [src, targetDigits[i]]
)),
base = parseInt((alpha[radix-1]+'0').padStart(
Number.MAX_SAFE_INTEGER.toString(radix).length - 1, alpha[radix-2]
), radix),
trimmer = RegExp("^" + alpha[radix-2] + "*" + alpha[radix-1]);
return n => (n + base).toString(radix)
.replace(trimmer, "")
.replace(/./g, m => map[m]);
}
// Example without "l" and "o" in target alphabet:
const conv = createConverter("abcdefghijkmnpqrstuvwxyz");
for (let n = 0; n < 29; n++) console.log(n, conv(n));
console.log("...");
for (let n = 690; n < 705; n++) console.log(n, conv(n));
This covers the range from 1 to 1000. Beyond that I haven't checked.
function colToletters(num) {
let a = " ABCDEFGHIJKLMNOPQRSTUVWXYZ";
if (num < 27) return a[num % a.length];
if (num > 26) {
num--;
let letters = ''
while (num >= 0) {
letters = 'ABCDEFGHIJKLMNOPQRSTUVWXYZ'[num % 26] + letters
num = Math.floor(num / 26) - 1
}
return letters;
}
}
I could be wrong but I've checked the other functions in this answer and they seem to fail at 26 which should be Z. Remember there are 26 letters in the alphabet not 25.
In JavaScript, how can I convert a sequence of numbers in an array to a range of numbers? In other words, I want to express consecutive occurring integers (no gaps) as hyphenated ranges.
[2,3,4,5,10,18,19,20] would become [2-5,10,18-20]
[1,6,7,9,10,12] would become [1,6-7,9-10,12]
[3,5,99] would remain [3,5,99]
[5,6,7,8,9,10,11] would become [5-11]
Here is an algorithm that I made some time ago, originally written for C#, now I ported it to JavaScript:
function getRanges(array) {
var ranges = [], rstart, rend;
for (var i = 0; i < array.length; i++) {
rstart = array[i];
rend = rstart;
while (array[i + 1] - array[i] == 1) {
rend = array[i + 1]; // increment the index if the numbers sequential
i++;
}
ranges.push(rstart == rend ? rstart+'' : rstart + '-' + rend);
}
return ranges;
}
getRanges([2,3,4,5,10,18,19,20]);
// returns ["2-5", "10", "18-20"]
getRanges([1,2,3,5,7,9,10,11,12,14 ]);
// returns ["1-3", "5", "7", "9-12", "14"]
getRanges([1,2,3,4,5,6,7,8,9,10])
// returns ["1-10"]
Just having fun with solution from CMS :
function getRanges (array) {
for (var ranges = [], rend, i = 0; i < array.length;) {
ranges.push ((rend = array[i]) + ((function (rstart) {
while (++rend === array[++i]);
return --rend === rstart;
})(rend) ? '' : '-' + rend));
}
return ranges;
}
Very nice question: here's my attempt:
function ranges(numbers){
var sorted = numbers.sort(function(a,b){return a-b;});
var first = sorted.shift();
return sorted.reduce(function(ranges, num){
if(num - ranges[0][1] <= 1){
ranges[0][1] = num;
} else {
ranges.unshift([num,num]);
}
return ranges;
},[[first,first]]).map(function(ranges){
return ranges[0] === ranges[1] ?
ranges[0].toString() : ranges.join('-');
}).reverse();
}
Demo on JSFiddler
I needed TypeScript code today to solve this very problem -- many years after the OP -- and decided to try a version written in a style more functional than the other answers here. Of course, only the parameter and return type annotations distinguish this code from standard ES6 JavaScript.
function toRanges(values: number[],
separator = '\u2013'): string[] {
return values
.slice()
.sort((p, q) => p - q)
.reduce((acc, cur, idx, src) => {
if ((idx > 0) && ((cur - src[idx - 1]) === 1))
acc[acc.length - 1][1] = cur;
else acc.push([cur]);
return acc;
}, [])
.map(range => range.join(separator));
}
Note that slice is necessary because sort sorts in place and we can't change the original array.
Here's my take on this...
function getRanges(input) {
//setup the return value
var ret = [], ary, first, last;
//copy and sort
var ary = input.concat([]);
ary.sort(function(a,b){
return Number(a) - Number(b);
});
//iterate through the array
for (var i=0; i<ary.length; i++) {
//set the first and last value, to the current iteration
first = last = ary[i];
//while within the range, increment
while (ary[i+1] == last+1) {
last++;
i++;
}
//push the current set into the return value
ret.push(first == last ? first : first + "-" + last);
}
//return the response array.
return ret;
}
Using ES6, a solution is:
function display ( vector ) { // assume vector sorted in increasing order
// display e.g.vector [ 2,4,5,6,9,11,12,13,15 ] as "2;4-6;9;11-13;15"
const l = vector.length - 1; // last valid index of vector array
// map [ 2,4,5,6,9,11,12,13,15 ] into array of strings (quote ommitted)
// --> [ "2;", "4-", "-", "6;", "9;", "11-", "-", "13;", "15;" ]
vector = vector.map ( ( n, i, v ) => // n is current number at index i of vector v
i < l && v [ i + 1 ] - n === 1 ? // next number is adjacent ?
`${ i > 0 && n - v [ i - 1 ] === 1 ? "" : n }-` :
`${ n };`
);
return vector.join ( "" ). // concatenate all strings in vector array
replace ( /-+/g, "-" ). // replace multiple dashes by single dash
slice ( 0, -1 ); // remove trailing ;
}
If you want to add extra spaces for readability, just add extra calls to string.prototype.replace().
If the input vector is not sorted, you can add the following line right after the opening brace of the display() function:
vector.sort ( ( a, b ) => a - b ); // sort vector in place, in increasing order.
Note that this could be improved to avoid testing twice for integer adjacentness (adjacenthood? I'm not a native English speaker;-).
And of course, if you don't want a single string as output, split it with ";".
Rough outline of the process is as follows:
Create an empty array called ranges
For each value in sorted input array
If ranges is empty then insert the item {min: value, max: value}
Else if max of last item in ranges and the current value are consecutive then set max of last item in ranges = value
Else insert the item {min: value, max: value}
Format the ranges array as desired e.g. by combining min and max if same
The following code uses Array.reduce and simplifies the logic by combining step 2.1 and 2.3.
function arrayToRange(array) {
return array
.slice()
.sort(function(a, b) {
return a - b;
})
.reduce(function(ranges, value) {
var lastIndex = ranges.length - 1;
if (lastIndex === -1 || ranges[lastIndex].max !== value - 1) {
ranges.push({ min: value, max: value });
} else {
ranges[lastIndex].max = value;
}
return ranges;
}, [])
.map(function(range) {
return range.min !== range.max ? range.min + "-" + range.max : range.min.toString();
});
}
console.log(arrayToRange([2, 3, 4, 5, 10, 18, 19, 20]));
If you simply want a string that represents a range, then you'd find the mid-point of your sequence, and that becomes your middle value (10 in your example). You'd then grab the first item in the sequence, and the item that immediately preceded your mid-point, and build your first-sequence representation. You'd follow the same procedure to get your last item, and the item that immediately follows your mid-point, and build your last-sequence representation.
// Provide initial sequence
var sequence = [1,2,3,4,5,6,7,8,9,10];
// Find midpoint
var midpoint = Math.ceil(sequence.length/2);
// Build first sequence from midpoint
var firstSequence = sequence[0] + "-" + sequence[midpoint-2];
// Build second sequence from midpoint
var lastSequence = sequence[midpoint] + "-" + sequence[sequence.length-1];
// Place all new in array
var newArray = [firstSequence,midpoint,lastSequence];
alert(newArray.join(",")); // 1-4,5,6-10
Demo Online: http://jsbin.com/uvahi/edit
; For all cells of the array
;if current cell = prev cell + 1 -> range continues
;if current cell != prev cell + 1 -> range ended
int[] x = [2,3,4,5,10,18,19,20]
string output = '['+x[0]
bool range = false; --current range
for (int i = 1; i > x[].length; i++) {
if (x[i+1] = [x]+1) {
range = true;
} else { //not sequential
if range = true
output = output || '-'
else
output = output || ','
output.append(x[i]','||x[i+1])
range = false;
}
}
Something like that.
An adaptation of CMS's javascript solution for Cold Fusion
It does sort the list first so that 1,3,2,4,5,8,9,10 (or similar) properly converts to 1-5,8-10.
<cfscript>
function getRanges(nArr) {
arguments.nArr = listToArray(listSort(arguments.nArr,"numeric"));
var ranges = [];
var rstart = "";
var rend = "";
for (local.i = 1; i <= ArrayLen(arguments.nArr); i++) {
rstart = arguments.nArr[i];
rend = rstart;
while (i < ArrayLen(arguments.nArr) and (val(arguments.nArr[i + 1]) - val(arguments.nArr[i])) == 1) {
rend = val(arguments.nArr[i + 1]); // increment the index if the numbers sequential
i++;
}
ArrayAppend(ranges,rstart == rend ? rstart : rstart & '-' & rend);
}
return arraytolist(ranges);
}
</cfscript>
Tiny ES6 module for you guys. It accepts a function to determine when we must break the sequence (breakDetectorFunc param - default is the simple thing for integer sequence input).
NOTICE: since input is abstract - there's no auto-sorting before processing, so if your sequence isn't sorted - do it prior to calling this module
function defaultIntDetector(a, b){
return Math.abs(b - a) > 1;
}
/**
* #param {Array} valuesArray
* #param {Boolean} [allArraysResult=false] if true - [1,2,3,7] will return [[1,3], [7,7]]. Otherwise [[1.3], 7]
* #param {SequenceToIntervalsBreakDetector} [breakDetectorFunc] must return true if value1 and value2 can't be in one sequence (if we need a gap here)
* #return {Array}
*/
const sequenceToIntervals = function (valuesArray, allArraysResult, breakDetectorFunc) {
if (!breakDetectorFunc){
breakDetectorFunc = defaultIntDetector;
}
if (typeof(allArraysResult) === 'undefined'){
allArraysResult = false;
}
const intervals = [];
let from = 0, to;
if (valuesArray instanceof Array) {
const cnt = valuesArray.length;
for (let i = 0; i < cnt; i++) {
to = i;
if (i < cnt - 1) { // i is not last (to compare to next)
if (breakDetectorFunc(valuesArray[i], valuesArray[i + 1])) {
// break
appendLastResult();
}
}
}
appendLastResult();
} else {
throw new Error("input is not an Array");
}
function appendLastResult(){
if (isFinite(from) && isFinite(to)) {
const vFrom = valuesArray[from];
const vTo = valuesArray[to];
if (from === to) {
intervals.push(
allArraysResult
? [vFrom, vTo] // same values array item
: vFrom // just a value, no array
);
} else if (Math.abs(from - to) === 1) { // sibling items
if (allArraysResult) {
intervals.push([vFrom, vFrom]);
intervals.push([vTo, vTo]);
} else {
intervals.push(vFrom, vTo);
}
} else {
intervals.push([vFrom, vTo]); // true interval
}
from = to + 1;
}
}
return intervals;
};
module.exports = sequenceToIntervals;
/** #callback SequenceToIntervalsBreakDetector
#param value1
#param value2
#return bool
*/
first argument is the input sequence sorted array, second is a boolean flag controlling the output mode: if true - single item (outside the intervals) will be returned as arrays anyway: [1,7],[9,9],[10,10],[12,20], otherwise single items returned as they appear in the input array
for your sample input
[2,3,4,5,10,18,19,20]
it will return:
sequenceToIntervals([2,3,4,5,10,18,19,20], true) // [[2,5], [10,10], [18,20]]
sequenceToIntervals([2,3,4,5,10,18,19,20], false) // [[2,5], 10, [18,20]]
sequenceToIntervals([2,3,4,5,10,18,19,20]) // [[2,5], 10, [18,20]]
Here's a version in Coffeescript
getRanges = (array) ->
ranges = []
rstart
rend
i = 0
while i < array.length
rstart = array[i]
rend = rstart
while array[i + 1] - array[i] is 1
rend = array[i + 1] # increment the index if the numbers sequential
i = i + 1
if rstart == rend
ranges.push rstart + ''
else
ranges.push rstart + '-' + rend
i = i + 1
return ranges
I've written my own method that's dependent on Lo-Dash, but doesn't just give you back an array of ranges, rather, it just returns an array of range groups.
[1,2,3,4,6,8,10] becomes:
[[1,2,3,4],[6,8,10]]
http://jsfiddle.net/mberkom/ufVey/