Here is a trie made to work on alphabets/scripts from any language.
class TrieNode {
constructor(key) {
// the "key" value will be the character in sequence
this.key = key;
// we keep a reference to parent
this.parent = null;
// we have hash of children
this.children = {};
// check to see if the node is at the end
this.end = false;
}
getWord() {
let output = [];
let node = this;
while (node !== null) {
output.unshift(node.key)
node = node.parent
}
return output.join('')
}
}
class Trie {
constructor() {
this.base = new TrieNode(null)
}
insert(word) {
let node = this.base
const points = Array.from(word)
for (const i in points) {
const point = points[i]
if (!node.children[point]) {
const child = node.children[point] = new TrieNode(point)
child.parent = node
}
node = node.children[point]
if (i == word.length - 1) {
node.end = true
}
}
}
contains(word) {
let node = this.base
const points = Array.from(word)
for (const i in points) {
const point = points[i]
if (node.children[point]) {
node = node.children[point]
} else {
return false
}
}
return node.end;
}
find(prefix) {
let node = this.base
let output = []
const points = Array.from(prefix)
for (const i in points) {
const point = points[i]
// make sure prefix actually has words
if (node.children[point]) {
node = node.children[point]
} else {
// there's none. just return it.
return output
}
}
const stack = [node]
while (stack.length) {
node = stack.shift()
// base case, if node is at a word, push to output
if (node.end) {
output.unshift(node.getWord())
}
// iterate through each children, call recursive findAllWords
for (var child in node.children) {
stack.push(node.children[child])
}
}
return output
}
}
After asking How to efficiently store 1 million words and query them by starts_with, contains, or ends_with? and getting some answers, I am wondering how the "contains" and "unscrambles" parts can be implemented as a trie. The prefix ("starts with") search is easily handled by the trie.
const fs = require('fs')
const Trie = require('./Trie')
const words = fs.readFileSync('tmp/scrabble.csv', 'utf-8')
.trim()
.split(/\n+/)
.map(x => x.trim())
const trie = new Trie()
words.forEach(word => trie.insert(word))
console.log(trie.find('zy'))
[
'zygodactylous', 'zygomorphies', 'zygapophysis',
'zygapophyses', 'zygomorphic', 'zymologies',
'zygospores', 'zygosities', 'zygomorphy',
'zygodactyl', 'zymurgies', 'zymograms',
'zymogenes', 'zygotenes', 'zygospore',
'zygomatic', 'zyzzyvas', 'zymosans',
'zymology', 'zymogram', 'zymogens',
'zymogene', 'zygotene', 'zygosity',
'zygomata', 'zyzzyva', 'zymurgy',
'zymotic', 'zymosis', 'zymoses',
'zymosan', 'zymogen', 'zymases',
'zygotic', 'zygotes', 'zygosis',
'zygoses', 'zygomas', 'zydecos',
'zymase', 'zygote', 'zygose',
'zygoma', 'zygoid', 'zydeco',
'zymes', 'zyme'
]
Here is the tmp/scrabble.csv I used.
The question is, can a trie be used to find all the words which "contain" some input string, or find all the words which "unscramble" the input string? I am curious how to accomplish this with a Trie. If a trie cannot do this efficiently, then knowing why not, and potentially what I should be looking at instead, will be a great answer as well.
From my initial thought-attempts at solving "contains", it seems I would have to create a trie which maps all possible combinations of substrings to the final word, but that seems like it will be an explosion of memory, so not sure how to better reason about this.
For "unscrambles", where you put in "caldku" and it finds "duck", amongst other possible words, I think possibly, similar to the linked answer to the other question, maybe sorting "duck" into "cdku", and then storing that in the trie, and then sorting the input to "acdklu", and then searching for "contains" using the previous algorithm, but hmm, no that will break in a few cases. So maybe a trie is not the right approach for these two problems? If it is, roughly how would you do it (you don't need to provide a full implementation, unless you'd like and it's straightforward enough).
Here is an answer to unscrambles that ALSO handles pagination. It actually returns [words, startOfNextPage]. So to find the next page you call it starting at startOfNextPage.
The idea is to normalize words by sorting the letters, and then storing the word in the trie once. When searching we can dig into the tree both for the next letter we want, and any letter before it (because the letter we want may come after). So we also store a lookup to know what letters might be found, so that we can break out early without having to look at a lot of bad options. And, of course, we include counts so that we can calculate how much of the trie we have explored and/or skipped over.
class Trie {
constructor(depth=0) {
this.depth = depth;
this.children = {};
this.words = [];
this.count = 0;
this.isSorted = false;
this.hasChildWith = {};
}
insert (word) {
if (word != undefined) {
this._insert(Array.from(word).sort(), word);
}
}
_insert (key, word) {
this.count++;
this.isSorted = false;
if (key.length == this.depth) {
this.words.push(word);
}
else {
for (let i = this.depth+1; i < key.length; i++) {
this.hasChildWith[key[i]] = true;
}
if (! this.children[key[this.depth]] ) {
this.children[key[this.depth]] = new Trie(this.depth+1);
}
this.children[key[this.depth]]._insert(key, word);
}
}
sort () {
if (! this.isSorted) {
this.words.sort();
let letters = Object.keys(this.children);
letters.sort();
// Keys come out of a hash in insertion order.
let orderedChildren = {};
for (let i = 0; i < letters.length; i++) {
orderedChildren[letters[i]] = this.children[letters[i]];
}
this.children = orderedChildren;
}
}
find (letters, startAt=0, maxCount=null) {
return this._find(Array.from(letters).sort(), 0, startAt, maxCount || this.count, 0);
}
_find (key, keyPos, startAt, maxCount, curPos) {
if (curPos + this.count < startAt) {
return [[], curPos + this.count];
}
this.sort(); // Make sure we are sorted.
let answer = [];
if (keyPos < key.length) {
// We have not yet found all the letters.
// tmpPos will track how much we looked at in detail
let tmpPos = curPos;
tmpPos += this.words.length;
for (const [k, v] of Object.entries(this.children)) {
if (k < key[keyPos]) {
// The next letter we want can be deeper in the trie?
if (v.hasChildWith[key[keyPos]]) {
// It is! Let's find it.
let result = v._find(key, keyPos, startAt, maxCount - answer.length, tmpPos);
answer = answer.concat(result[0]);
if (maxCount <= answer.length) {
// We finished our answer, return it and what we found.
return [answer, result[1]];
}
}
tmpPos += v.count; // Didn't find it, but track that we've seen this.
}
else if (k == key[keyPos]) {
// We found our next letter! Search deeper.
let result = v._find(key, keyPos+1, startAt, maxCount - answer.length, tmpPos);
answer = answer.concat(result[0]);
if (maxCount <= answer.length) {
// We finished the search.
return [answer, result[1]];
}
else {
// No other letter can match.
break;
}
}
else {
// Neither this or any later letter can match.
break;
}
}
// Return our partial answer and mark that we went
// through this whole node.
return [answer, curPos + this.count];
}
else {
// We have our letters and are recursively finding words.
if (startAt <= curPos + this.words.length) {
answer = this.words.slice(startAt - curPos);
if (maxCount <= answer.length) {
// No need to search deeper, we're done.
answer = answer.slice(0, maxCount);
return [answer, curPos + answer.length];
}
curPos += answer.length;
}
for (const child of Object.values(this.children)) {
let result = child._find(key, keyPos, startAt, maxCount - answer.length, curPos);
answer = answer.concat(result[0])
if (maxCount <= answer.length) {
return [answer, result[1]];
}
else {
curPos += child.count;
}
}
return [answer, curPos];
}
}
}
exports.Trie = Trie
And here is some code to test it with.
const fs = require('fs')
const AnagramTrie = require('./AnagramTrie')
const words = fs.readFileSync('tmp/scrabble.csv', 'utf-8')
.trim()
.split(/\n+/)
.map(x => x.trim())
const trie = new AnagramTrie.Trie()
words.forEach(word => trie.insert(word))
console.log(trie.find('b', 0, 12))
console.log(trie.find('b', 0, 6))
console.log(trie.find('b', 11, 6))
console.log(trie.find('zx', 0, 12))
console.log(trie.find('zx', 0, 6))
console.log(trie.find('zx', 60875, 6))
It will return words in the order of their sorted anagram, followed by the word itself. So if you search for admired you'll find unadmired first because the n in un comes before the r in admired. And you'll find that disembarrassed comes before either because it has 2 a's in it.
And here is an answer to contains. Note how, despite using a Trie both times, it needs a lot of customization to the problem we are solving.
Sample code to use it is like the other one. You can see the deduplication at work if you search for a. The answer aa (correctly) only comes up once. Also it is a little slower to start up because you have to put words into the data structure multiple times.
class Trie {
constructor(depth=0, key=[]) {
this.depth = depth;
this.children = {};
this.words = [];
this.count = 0;
this.isSorted = false;
this.path = key.slice(0, depth).join("");
this.char = key[this.depth-1];
}
insert (word) {
if (word != undefined) {
const key = Array.from(word);
for (let i = 0; i < key.length; i++) {
this._insert(key.slice(i), word);
}
}
}
_insert (key, word) {
this.count++;
if (this.depth == key.length) {
this.words.push(word);
}
else {
if (! this.children[key[this.depth]] ) {
this.children[key[this.depth]] = new Trie(this.depth+1, key);
}
this.children[key[this.depth]]._insert(key, word);
}
}
sort () {
if (! this.isSorted) {
this.words.sort();
let letters = Object.keys(this.children);
letters.sort();
// Keys come out of a hash in insertion order.
let orderedChildren = {};
for (let i = 0; i < letters.length; i++) {
orderedChildren[letters[i]] = this.children[letters[i]];
}
this.children = orderedChildren;
this.isSorted
}
}
find (letters, startAt=0, maxCount=null) {
// Defaults, special cases, etc.
if (this.count <= startAt) {
return [[], startAt];
}
if (maxCount == null) {
maxCount = this.count;
}
if (letters == "") {
// Special case.
this.sort();
answer = this.words.slice(startAt, startAt + maxCount);
return [answer, startAt + answer.length];
}
// We will do the recursive search.
const key = Array.from(letters);
// The challenge is that each word is stored multiple times.
// We want to only find them once. So we will stop searching as soon
// as we match our search string twice.
//
// This requires keeping track of where we are in trying to match our
// search string again. And if we fail, backtracking in our attempt
// to match.
//
// Here is the complication. Partial matches can overlap.
//
// Consider the following search string:
//
// search: a b a a b a c a b
// 0 1 2 3 4 5 6 7 8
//
// Now suppose that I'm looking for "c" at 6 next and it isn't an
// "c". Well I know I matched "aba" and should next check if I got
// character 3, "a". We can encode this logic in an array.
//
// a b a a b a c a b
// backtrackTo: [-1, 0,-1, 1, 0,-1, 3,-1, 0]
//
// So if I've matched through 5, I check "c", then "a", then "b"
// before deciding that I'm not still partway through a match.
//
// Let's calculate that backtrackTo. We will also calculate a
// variable, matchAtAtEnd. Which is how long our longest
// running match is at a full match.
// We start by assuming that we go back to the beginning.
let backtrackTo = [-1];
let rematches = [];
let matchAtEnd = 0;
for (let i = 1; i < key.length; i++) {
if (key[i] == key[0]) {
backtrackTo[i] = -1;
rematches.push(i);
}
else {
backtrackTo[i] = 0;
}
}
// In our example we have:
//
// backtrackTo = [-1, 0, -1, -1, 0, -1, 0, -1, 0]
// rematches = [2, 3, 5, 7]
//
// Now let `k` be the length of the current rematch.
let k = 1;
while (0 < rematches.length) {
let nextRematches = [];
for (let i = 0; i < rematches.length; i++) {
if (i + k == key.length) {
matchAtEnd = k-1;
}
else {
if (key[k] == key[i+k]) {
nextRematches.push(i);
}
else {
backtrackTo[i+k] = k;
}
}
}
rematches = nextRematches;
k++;
// In our example we get:
//
// k = 1
// backtrackTo = [-1, 0, -1, -1, 0, -1, 0, -1, 0]
// rematches = [2, 3, 5, 7]
// matchAtEnd = -1
//
// k = 2
// backtrackTo = [-1, 0, -1, 1, 0, -1, 0, -1, 0]
// rematches = [3, 7]
// matchAtEnd = -1
//
// k = 3
// backtrackTo = [-1, 0, -1, 1, 0, -1, 3, -1, 0]
// rematches = []
// matchAtEnd = 2
//
// and we see that we got the expected backtrackTo AND
// we have recorded the fact that at matching
// abaabacab we are currently also matching the ab at the
// start.
//
// Now let's find the first match.
let node = this;
for (let i = 0; i < key.length; i++) {
node = node.children[key[i]];
if (!node) {
return [[], startAt];
}
}
return node._find(key, startAt, maxCount, 0, backtrackTo, matchAtEnd)
}
_find (key, startAt, maxCount, curPos, backtrackTo, nextMatch) {
// console.log([key, startAt, maxCount, curPos, backtrackTo, nextMatch, [this.path, this.count]]);
// Skip me?
if ((curPos + this.count <= startAt) || (key.length <= nextMatch)) {
return [[], curPos + this.count];
}
this.sort();
let answer = [];
if (curPos < startAt) {
if (startAt < curPos + this.words.length) {
answer = this.words.slice(startAt-curPos, startAt-curPos+maxCount);
}
else {
curPos += this.words.length; // Count the words we are skipping.
}
}
else {
answer = this.words.slice(0, maxCount);
}
curPos += answer.length;
if (maxCount <= answer.length) {
return [answer, curPos];
}
for (const [k, v] of Object.entries(this.children)) {
let thisMatch = nextMatch;
while ((-1 < thisMatch) && (key[thisMatch] != k)) {
thisMatch = backtrackTo[thisMatch];
}
thisMatch++;
let partialAnswer = null;
[partialAnswer, curPos] = v._find(key, startAt, maxCount - answer.length, curPos, backtrackTo, thisMatch);
answer = answer.concat(partialAnswer);
if (maxCount <= answer.length) {
break; // We are done.
}
}
return [answer, curPos];
}
}
exports.Trie = Trie
I'm looking for a good JavaScript equivalent of the C/PHP printf() or for C#/Java programmers, String.Format() (IFormatProvider for .NET).
My basic requirement is a thousand separator format for numbers for now, but something that handles lots of combinations (including dates) would be good.
I realize Microsoft's Ajax library provides a version of String.Format(), but we don't want the entire overhead of that framework.
Current JavaScript
From ES6 on you could use template strings:
let soMany = 10;
console.log(`This is ${soMany} times easier!`);
// "This is 10 times easier!"
See Kim's answer below for details.
Older answer
Try sprintf() for JavaScript.
If you really want to do a simple format method on your own, don’t do the replacements successively but do them simultaneously.
Because most of the other proposals that are mentioned fail when a replace string of previous replacement does also contain a format sequence like this:
"{0}{1}".format("{1}", "{0}")
Normally you would expect the output to be {1}{0} but the actual output is {1}{1}. So do a simultaneous replacement instead like in fearphage’s suggestion.
Building on the previously suggested solutions:
// First, checks if it isn't implemented yet.
if (!String.prototype.format) {
String.prototype.format = function() {
var args = arguments;
return this.replace(/{(\d+)}/g, function(match, number) {
return typeof args[number] != 'undefined'
? args[number]
: match
;
});
};
}
"{0} is dead, but {1} is alive! {0} {2}".format("ASP", "ASP.NET")
outputs
ASP is dead, but ASP.NET is alive! ASP {2}
If you prefer not to modify String's prototype:
if (!String.format) {
String.format = function(format) {
var args = Array.prototype.slice.call(arguments, 1);
return format.replace(/{(\d+)}/g, function(match, number) {
return typeof args[number] != 'undefined'
? args[number]
: match
;
});
};
}
Gives you the much more familiar:
String.format('{0} is dead, but {1} is alive! {0} {2}', 'ASP', 'ASP.NET');
with the same result:
ASP is dead, but ASP.NET is alive! ASP {2}
It's funny because Stack Overflow actually has their own formatting function for the String prototype called formatUnicorn. Try it! Go into the console and type something like:
"Hello, {name}, are you feeling {adjective}?".formatUnicorn({name:"Gabriel", adjective: "OK"});
You get this output:
Hello, Gabriel, are you feeling OK?
You can use objects, arrays, and strings as arguments! I got its code and reworked it to produce a new version of String.prototype.format:
String.prototype.formatUnicorn = String.prototype.formatUnicorn ||
function () {
"use strict";
var str = this.toString();
if (arguments.length) {
var t = typeof arguments[0];
var key;
var args = ("string" === t || "number" === t) ?
Array.prototype.slice.call(arguments)
: arguments[0];
for (key in args) {
str = str.replace(new RegExp("\\{" + key + "\\}", "gi"), args[key]);
}
}
return str;
};
Note the clever Array.prototype.slice.call(arguments) call -- that means if you throw in arguments that are strings or numbers, not a single JSON-style object, you get C#'s String.Format behavior almost exactly.
"a{0}bcd{1}ef".formatUnicorn("FOO", "BAR"); // yields "aFOObcdBARef"
That's because Array's slice will force whatever's in arguments into an Array, whether it was originally or not, and the key will be the index (0, 1, 2...) of each array element coerced into a string (eg, "0", so "\\{0\\}" for your first regexp pattern).
Neat.
Number Formatting in JavaScript
I got to this question page hoping to find how to format numbers in JavaScript, without introducing yet another library. Here's what I've found:
Rounding floating-point numbers
The equivalent of sprintf("%.2f", num) in JavaScript seems to be num.toFixed(2), which formats num to 2 decimal places, with rounding (but see #ars265's comment about Math.round below).
(12.345).toFixed(2); // returns "12.35" (rounding!)
(12.3).toFixed(2); // returns "12.30" (zero padding)
Exponential form
The equivalent of sprintf("%.2e", num) is num.toExponential(2).
(33333).toExponential(2); // "3.33e+4"
Hexadecimal and other bases
To print numbers in base B, try num.toString(B). JavaScript supports automatic conversion to and from bases 2 through 36 (in addition, some browsers have limited support for base64 encoding).
(3735928559).toString(16); // to base 16: "deadbeef"
parseInt("deadbeef", 16); // from base 16: 3735928559
Reference Pages
Quick tutorial on JS number formatting
Mozilla reference page for toFixed() (with links to toPrecision(), toExponential(), toLocaleString(), ...)
From ES6 on you could use template strings:
let soMany = 10;
console.log(`This is ${soMany} times easier!`);
// "This is 10 times easier!"
Be aware that template strings are surrounded by backticks ` instead of (single) quotes.
For further information:
https://developers.google.com/web/updates/2015/01/ES6-Template-Strings
https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/template_strings
Note:
Check the mozilla-site to find a list of supported browsers.
jsxt, Zippo
This option fits better.
String.prototype.format = function() {
var formatted = this;
for (var i = 0; i < arguments.length; i++) {
var regexp = new RegExp('\\{'+i+'\\}', 'gi');
formatted = formatted.replace(regexp, arguments[i]);
}
return formatted;
};
With this option I can replace strings like these:
'The {0} is dead. Don\'t code {0}. Code {1} that is open source!'.format('ASP', 'PHP');
With your code the second {0} wouldn't be replaced. ;)
I use this simple function:
String.prototype.format = function() {
var formatted = this;
for( var arg in arguments ) {
formatted = formatted.replace("{" + arg + "}", arguments[arg]);
}
return formatted;
};
That's very similar to string.format:
"{0} is dead, but {1} is alive!".format("ASP", "ASP.NET")
For Node.js users there is util.format which has printf-like functionality:
util.format("%s world", "Hello")
I'm surprised no one used reduce, this is a native concise and powerful JavaScript function.
ES6 (EcmaScript2015)
String.prototype.format = function() {
return [...arguments].reduce((p,c) => p.replace(/%s/,c), this);
};
console.log('Is that a %s or a %s?... No, it\'s %s!'.format('plane', 'bird', 'SOman'));
< ES6
function interpolate(theString, argumentArray) {
var regex = /%s/;
var _r=function(p,c){return p.replace(regex,c);}
return argumentArray.reduce(_r, theString);
}
interpolate("%s, %s and %s", ["Me", "myself", "I"]); // "Me, myself and I"
How it works:
reduce applies a function against an accumulator and each element in the array (from left to right) to reduce it to a single value.
var _r= function(p,c){return p.replace(/%s/,c)};
console.log(
["a", "b", "c"].reduce(_r, "[%s], [%s] and [%s]") + '\n',
[1, 2, 3].reduce(_r, "%s+%s=%s") + '\n',
["cool", 1337, "stuff"].reduce(_r, "%s %s %s")
);
Here's a minimal implementation of sprintf in JavaScript: it only does "%s" and "%d", but I have left space for it to be extended. It is useless to the OP, but other people who stumble across this thread coming from Google might benefit from it.
function sprintf() {
var args = arguments,
string = args[0],
i = 1;
return string.replace(/%((%)|s|d)/g, function (m) {
// m is the matched format, e.g. %s, %d
var val = null;
if (m[2]) {
val = m[2];
} else {
val = args[i];
// A switch statement so that the formatter can be extended. Default is %s
switch (m) {
case '%d':
val = parseFloat(val);
if (isNaN(val)) {
val = 0;
}
break;
}
i++;
}
return val;
});
}
Example:
alert(sprintf('Latitude: %s, Longitude: %s, Count: %d', 41.847, -87.661, 'two'));
// Expected output: Latitude: 41.847, Longitude: -87.661, Count: 0
In contrast with similar solutions in previous replies, this one does all substitutions in one go, so it will not replace parts of previously replaced values.
3 different ways to format javascript string
There are 3 different ways to format a string by replacing placeholders with the variable value.
Using template literal (backticks ``)
let name = 'John';
let age = 30;
// using backticks
console.log(`${name} is ${age} years old.`);
// John is 30 years old.
Using concatenation
let name = 'John';
let age = 30;
// using concatenation
console.log(name + ' is ' + age + ' years old.');
// John is 30 years old.
Creating own format function
String.prototype.format = function () {
var args = arguments;
return this.replace(/{([0-9]+)}/g, function (match, index) {
// check if the argument is there
return typeof args[index] == 'undefined' ? match : args[index];
});
};
console.log('{0} is {1} years old.'.format('John', 30));
JavaScript programmers can use String.prototype.sprintf at https://github.com/ildar-shaimordanov/jsxt/blob/master/js/String.js. Below is example:
var d = new Date();
var dateStr = '%02d:%02d:%02d'.sprintf(
d.getHours(),
d.getMinutes(),
d.getSeconds());
I want to share my solution for the 'problem'. I haven't re-invented the wheel but tries to find a solution based on what JavaScript already does. The advantage is, that you get all implicit conversions for free. Setting the prototype property $ of String gives a very nice and compact syntax (see examples below). It is maybe not the most efficient way, but in most cases dealing with output it does not have to be super optimized.
String.form = function(str, arr) {
var i = -1;
function callback(exp, p0, p1, p2, p3, p4) {
if (exp=='%%') return '%';
if (arr[++i]===undefined) return undefined;
exp = p2 ? parseInt(p2.substr(1)) : undefined;
var base = p3 ? parseInt(p3.substr(1)) : undefined;
var val;
switch (p4) {
case 's': val = arr[i]; break;
case 'c': val = arr[i][0]; break;
case 'f': val = parseFloat(arr[i]).toFixed(exp); break;
case 'p': val = parseFloat(arr[i]).toPrecision(exp); break;
case 'e': val = parseFloat(arr[i]).toExponential(exp); break;
case 'x': val = parseInt(arr[i]).toString(base?base:16); break;
case 'd': val = parseFloat(parseInt(arr[i], base?base:10).toPrecision(exp)).toFixed(0); break;
}
val = typeof(val)=='object' ? JSON.stringify(val) : val.toString(base);
var sz = parseInt(p1); /* padding size */
var ch = p1 && p1[0]=='0' ? '0' : ' '; /* isnull? */
while (val.length<sz) val = p0 !== undefined ? val+ch : ch+val; /* isminus? */
return val;
}
var regex = /%(-)?(0?[0-9]+)?([.][0-9]+)?([#][0-9]+)?([scfpexd%])/g;
return str.replace(regex, callback);
}
String.prototype.$ = function() {
return String.form(this, Array.prototype.slice.call(arguments));
}
Here are a few examples:
String.format("%s %s", [ "This is a string", 11 ])
console.log("%s %s".$("This is a string", 11))
var arr = [ "12.3", 13.6 ]; console.log("Array: %s".$(arr));
var obj = { test:"test", id:12 }; console.log("Object: %s".$(obj));
console.log("%c", "Test");
console.log("%5d".$(12)); // ' 12'
console.log("%05d".$(12)); // '00012'
console.log("%-5d".$(12)); // '12 '
console.log("%5.2d".$(123)); // ' 120'
console.log("%5.2f".$(1.1)); // ' 1.10'
console.log("%10.2e".$(1.1)); // ' 1.10e+0'
console.log("%5.3p".$(1.12345)); // ' 1.12'
console.log("%5x".$(45054)); // ' affe'
console.log("%20#2x".$("45054")); // ' 1010111111111110'
console.log("%6#2d".$("111")); // ' 7'
console.log("%6#16d".$("affe")); // ' 45054'
Adding to zippoxer's answer, I use this function:
String.prototype.format = function () {
var a = this, b;
for (b in arguments) {
a = a.replace(/%[a-z]/, arguments[b]);
}
return a; // Make chainable
};
var s = 'Hello %s The magic number is %d.';
s.format('world!', 12); // Hello World! The magic number is 12.
I also have a non-prototype version which I use more often for its Java-like syntax:
function format() {
var a, b, c;
a = arguments[0];
b = [];
for(c = 1; c < arguments.length; c++){
b.push(arguments[c]);
}
for (c in b) {
a = a.replace(/%[a-z]/, b[c]);
}
return a;
}
format('%d ducks, 55 %s', 12, 'cats'); // 12 ducks, 55 cats
ES 2015 update
All the cool new stuff in ES 2015 makes this a lot easier:
function format(fmt, ...args){
return fmt
.split("%%")
.reduce((aggregate, chunk, i) =>
aggregate + chunk + (args[i] || ""), "");
}
format("Hello %%! I ate %% apples today.", "World", 44);
// "Hello World, I ate 44 apples today."
I figured that since this, like the older ones, doesn't actually parse the letters, it might as well just use a single token %%. This has the benefit of being obvious and not making it difficult to use a single %. However, if you need %% for some reason, you would need to replace it with itself:
format("I love percentage signs! %%", "%%");
// "I love percentage signs! %%"
+1 Zippo with the exception that the function body needs to be as below or otherwise it appends the current string on every iteration:
String.prototype.format = function() {
var formatted = this;
for (var arg in arguments) {
formatted = formatted.replace("{" + arg + "}", arguments[arg]);
}
return formatted;
};
I use a small library called String.format for JavaScript which supports most of the format string capabilities (including format of numbers and dates), and uses the .NET syntax. The script itself is smaller than 4 kB, so it doesn't create much of overhead.
I'll add my own discoveries which I've found since I asked:
number_format (for thousand separator/currency formatting)
sprintf (same author as above)
Sadly it seems sprintf doesn't handle thousand separator formatting like .NET's string format.
If you are looking to handle the thousands separator, you should really use toLocaleString() from the JavaScript Number class since it will format the string for the user's region.
The JavaScript Date class can format localized dates and times.
Very elegant:
String.prototype.format = function (){
var args = arguments;
return this.replace(/\{\{|\}\}|\{(\d+)\}/g, function (curlyBrack, index) {
return ((curlyBrack == "{{") ? "{" : ((curlyBrack == "}}") ? "}" : args[index]));
});
};
// Usage:
"{0}{1}".format("{1}", "{0}")
Credit goes to (broken link) https://gist.github.com/0i0/1519811
There is "sprintf" for JavaScript which you can find at http://www.webtoolkit.info/javascript-sprintf.html.
The PHPJS project has written JavaScript implementations for many of PHP's functions. Since PHP's sprintf() function is basically the same as C's printf(), their JavaScript implementation of it should satisfy your needs.
I use this one:
String.prototype.format = function() {
var newStr = this, i = 0;
while (/%s/.test(newStr))
newStr = newStr.replace("%s", arguments[i++])
return newStr;
}
Then I call it:
"<h1>%s</h1><p>%s</p>".format("Header", "Just a test!");
I have a solution very close to Peter's, but it deals with number and object case.
if (!String.prototype.format) {
String.prototype.format = function() {
var args;
args = arguments;
if (args.length === 1 && args[0] !== null && typeof args[0] === 'object') {
args = args[0];
}
return this.replace(/{([^}]*)}/g, function(match, key) {
return (typeof args[key] !== "undefined" ? args[key] : match);
});
};
}
Maybe it could be even better to deal with the all deeps cases, but for my needs this is just fine.
"This is an example from {name}".format({name:"Blaine"});
"This is an example from {0}".format("Blaine");
PS: This function is very cool if you are using translations in templates frameworks like AngularJS:
<h1> {{('hello-message'|translate).format(user)}} <h1>
<h1> {{('hello-by-name'|translate).format( user ? user.name : 'You' )}} <h1>
Where the en.json is something like
{
"hello-message": "Hello {name}, welcome.",
"hello-by-name": "Hello {0}, welcome."
}
One very slightly different version, the one I prefer (this one uses {xxx} tokens rather than {0} numbered arguments, this is much more self-documenting and suits localization much better):
String.prototype.format = function(tokens) {
var formatted = this;
for (var token in tokens)
if (tokens.hasOwnProperty(token))
formatted = formatted.replace(RegExp("{" + token + "}", "g"), tokens[token]);
return formatted;
};
A variation would be:
var formatted = l(this);
that calls an l() localization function first.
For basic formatting:
var template = jQuery.validator.format("{0} is not a valid value");
var result = template("abc");
We can use a simple lightweight String.Format string operation library for Typescript.
String.Format():
var id = image.GetId()
String.Format("image_{0}.jpg", id)
output: "image_2db5da20-1c5d-4f1a-8fd4-b41e34c8c5b5.jpg";
String Format for specifiers:
var value = String.Format("{0:L}", "APPLE"); //output "apple"
value = String.Format("{0:U}", "apple"); // output "APPLE"
value = String.Format("{0:d}", "2017-01-23 00:00"); //output "23.01.2017"
value = String.Format("{0:s}", "21.03.2017 22:15:01") //output "2017-03-21T22:15:01"
value = String.Format("{0:n}", 1000000);
//output "1.000.000"
value = String.Format("{0:00}", 1);
//output "01"
String Format for Objects including specifiers:
var fruit = new Fruit();
fruit.type = "apple";
fruit.color = "RED";
fruit.shippingDate = new Date(2018, 1, 1);
fruit.amount = 10000;
String.Format("the {type:U} is {color:L} shipped on {shippingDate:s} with an amount of {amount:n}", fruit);
// output: the APPLE is red shipped on 2018-01-01 with an amount of 10.000
Just in case someone needs a function to prevent polluting global scope, here is the function that does the same:
function _format (str, arr) {
return str.replace(/{(\d+)}/g, function (match, number) {
return typeof arr[number] != 'undefined' ? arr[number] : match;
});
};
For those who like Node.JS and its util.format feature, I've just extracted it out into its vanilla JavaScript form (with only functions that util.format uses):
exports = {};
function isString(arg) {
return typeof arg === 'string';
}
function isNull(arg) {
return arg === null;
}
function isObject(arg) {
return typeof arg === 'object' && arg !== null;
}
function isBoolean(arg) {
return typeof arg === 'boolean';
}
function isUndefined(arg) {
return arg === void 0;
}
function stylizeNoColor(str, styleType) {
return str;
}
function stylizeWithColor(str, styleType) {
var style = inspect.styles[styleType];
if (style) {
return '\u001b[' + inspect.colors[style][0] + 'm' + str +
'\u001b[' + inspect.colors[style][3] + 'm';
} else {
return str;
}
}
function isFunction(arg) {
return typeof arg === 'function';
}
function isNumber(arg) {
return typeof arg === 'number';
}
function isSymbol(arg) {
return typeof arg === 'symbol';
}
function formatPrimitive(ctx, value) {
if (isUndefined(value))
return ctx.stylize('undefined', 'undefined');
if (isString(value)) {
var simple = '\'' + JSON.stringify(value).replace(/^"|"$/g, '')
.replace(/'/g, "\\'")
.replace(/\\"/g, '"') + '\'';
return ctx.stylize(simple, 'string');
}
if (isNumber(value)) {
// Format -0 as '-0'. Strict equality won't distinguish 0 from -0,
// so instead we use the fact that 1 / -0 < 0 whereas 1 / 0 > 0 .
if (value === 0 && 1 / value < 0)
return ctx.stylize('-0', 'number');
return ctx.stylize('' + value, 'number');
}
if (isBoolean(value))
return ctx.stylize('' + value, 'boolean');
// For some reason typeof null is "object", so special case here.
if (isNull(value))
return ctx.stylize('null', 'null');
// es6 symbol primitive
if (isSymbol(value))
return ctx.stylize(value.toString(), 'symbol');
}
function arrayToHash(array) {
var hash = {};
array.forEach(function (val, idx) {
hash[val] = true;
});
return hash;
}
function objectToString(o) {
return Object.prototype.toString.call(o);
}
function isDate(d) {
return isObject(d) && objectToString(d) === '[object Date]';
}
function isError(e) {
return isObject(e) &&
(objectToString(e) === '[object Error]' || e instanceof Error);
}
function isRegExp(re) {
return isObject(re) && objectToString(re) === '[object RegExp]';
}
function formatError(value) {
return '[' + Error.prototype.toString.call(value) + ']';
}
function formatPrimitiveNoColor(ctx, value) {
var stylize = ctx.stylize;
ctx.stylize = stylizeNoColor;
var str = formatPrimitive(ctx, value);
ctx.stylize = stylize;
return str;
}
function isArray(ar) {
return Array.isArray(ar);
}
function hasOwnProperty(obj, prop) {
return Object.prototype.hasOwnProperty.call(obj, prop);
}
function formatProperty(ctx, value, recurseTimes, visibleKeys, key, array) {
var name, str, desc;
desc = Object.getOwnPropertyDescriptor(value, key) || {value: value[key]};
if (desc.get) {
if (desc.set) {
str = ctx.stylize('[Getter/Setter]', 'special');
} else {
str = ctx.stylize('[Getter]', 'special');
}
} else {
if (desc.set) {
str = ctx.stylize('[Setter]', 'special');
}
}
if (!hasOwnProperty(visibleKeys, key)) {
name = '[' + key + ']';
}
if (!str) {
if (ctx.seen.indexOf(desc.value) < 0) {
if (isNull(recurseTimes)) {
str = formatValue(ctx, desc.value, null);
} else {
str = formatValue(ctx, desc.value, recurseTimes - 1);
}
if (str.indexOf('\n') > -1) {
if (array) {
str = str.split('\n').map(function (line) {
return ' ' + line;
}).join('\n').substr(2);
} else {
str = '\n' + str.split('\n').map(function (line) {
return ' ' + line;
}).join('\n');
}
}
} else {
str = ctx.stylize('[Circular]', 'special');
}
}
if (isUndefined(name)) {
if (array && key.match(/^\d+$/)) {
return str;
}
name = JSON.stringify('' + key);
if (name.match(/^"([a-zA-Z_][a-zA-Z_0-9]*)"$/)) {
name = name.substr(1, name.length - 2);
name = ctx.stylize(name, 'name');
} else {
name = name.replace(/'/g, "\\'")
.replace(/\\"/g, '"')
.replace(/(^"|"$)/g, "'")
.replace(/\\\\/g, '\\');
name = ctx.stylize(name, 'string');
}
}
return name + ': ' + str;
}
function formatArray(ctx, value, recurseTimes, visibleKeys, keys) {
var output = [];
for (var i = 0, l = value.length; i < l; ++i) {
if (hasOwnProperty(value, String(i))) {
output.push(formatProperty(ctx, value, recurseTimes, visibleKeys,
String(i), true));
} else {
output.push('');
}
}
keys.forEach(function (key) {
if (!key.match(/^\d+$/)) {
output.push(formatProperty(ctx, value, recurseTimes, visibleKeys,
key, true));
}
});
return output;
}
function reduceToSingleString(output, base, braces) {
var length = output.reduce(function (prev, cur) {
return prev + cur.replace(/\u001b\[\d\d?m/g, '').length + 1;
}, 0);
if (length > 60) {
return braces[0] +
(base === '' ? '' : base + '\n ') +
' ' +
output.join(',\n ') +
' ' +
braces[1];
}
return braces[0] + base + ' ' + output.join(', ') + ' ' + braces[1];
}
function formatValue(ctx, value, recurseTimes) {
// Provide a hook for user-specified inspect functions.
// Check that value is an object with an inspect function on it
if (ctx.customInspect &&
value &&
isFunction(value.inspect) &&
// Filter out the util module, it's inspect function is special
value.inspect !== exports.inspect &&
// Also filter out any prototype objects using the circular check.
!(value.constructor && value.constructor.prototype === value)) {
var ret = value.inspect(recurseTimes, ctx);
if (!isString(ret)) {
ret = formatValue(ctx, ret, recurseTimes);
}
return ret;
}
// Primitive types cannot have properties
var primitive = formatPrimitive(ctx, value);
if (primitive) {
return primitive;
}
// Look up the keys of the object.
var keys = Object.keys(value);
var visibleKeys = arrayToHash(keys);
if (ctx.showHidden) {
keys = Object.getOwnPropertyNames(value);
}
// This could be a boxed primitive (new String(), etc.), check valueOf()
// NOTE: Avoid calling `valueOf` on `Date` instance because it will return
// a number which, when object has some additional user-stored `keys`,
// will be printed out.
var formatted;
var raw = value;
try {
// the .valueOf() call can fail for a multitude of reasons
if (!isDate(value))
raw = value.valueOf();
} catch (e) {
// ignore...
}
if (isString(raw)) {
// for boxed Strings, we have to remove the 0-n indexed entries,
// since they just noisey up the output and are redundant
keys = keys.filter(function (key) {
return !(key >= 0 && key < raw.length);
});
}
// Some type of object without properties can be shortcutted.
if (keys.length === 0) {
if (isFunction(value)) {
var name = value.name ? ': ' + value.name : '';
return ctx.stylize('[Function' + name + ']', 'special');
}
if (isRegExp(value)) {
return ctx.stylize(RegExp.prototype.toString.call(value), 'regexp');
}
if (isDate(value)) {
return ctx.stylize(Date.prototype.toString.call(value), 'date');
}
if (isError(value)) {
return formatError(value);
}
// now check the `raw` value to handle boxed primitives
if (isString(raw)) {
formatted = formatPrimitiveNoColor(ctx, raw);
return ctx.stylize('[String: ' + formatted + ']', 'string');
}
if (isNumber(raw)) {
formatted = formatPrimitiveNoColor(ctx, raw);
return ctx.stylize('[Number: ' + formatted + ']', 'number');
}
if (isBoolean(raw)) {
formatted = formatPrimitiveNoColor(ctx, raw);
return ctx.stylize('[Boolean: ' + formatted + ']', 'boolean');
}
}
var base = '', array = false, braces = ['{', '}'];
// Make Array say that they are Array
if (isArray(value)) {
array = true;
braces = ['[', ']'];
}
// Make functions say that they are functions
if (isFunction(value)) {
var n = value.name ? ': ' + value.name : '';
base = ' [Function' + n + ']';
}
// Make RegExps say that they are RegExps
if (isRegExp(value)) {
base = ' ' + RegExp.prototype.toString.call(value);
}
// Make dates with properties first say the date
if (isDate(value)) {
base = ' ' + Date.prototype.toUTCString.call(value);
}
// Make error with message first say the error
if (isError(value)) {
base = ' ' + formatError(value);
}
// Make boxed primitive Strings look like such
if (isString(raw)) {
formatted = formatPrimitiveNoColor(ctx, raw);
base = ' ' + '[String: ' + formatted + ']';
}
// Make boxed primitive Numbers look like such
if (isNumber(raw)) {
formatted = formatPrimitiveNoColor(ctx, raw);
base = ' ' + '[Number: ' + formatted + ']';
}
// Make boxed primitive Booleans look like such
if (isBoolean(raw)) {
formatted = formatPrimitiveNoColor(ctx, raw);
base = ' ' + '[Boolean: ' + formatted + ']';
}
if (keys.length === 0 && (!array || value.length === 0)) {
return braces[0] + base + braces[1];
}
if (recurseTimes < 0) {
if (isRegExp(value)) {
return ctx.stylize(RegExp.prototype.toString.call(value), 'regexp');
} else {
return ctx.stylize('[Object]', 'special');
}
}
ctx.seen.push(value);
var output;
if (array) {
output = formatArray(ctx, value, recurseTimes, visibleKeys, keys);
} else {
output = keys.map(function (key) {
return formatProperty(ctx, value, recurseTimes, visibleKeys, key, array);
});
}
ctx.seen.pop();
return reduceToSingleString(output, base, braces);
}
function inspect(obj, opts) {
// default options
var ctx = {
seen: [],
stylize: stylizeNoColor
};
// legacy...
if (arguments.length >= 3) ctx.depth = arguments[2];
if (arguments.length >= 4) ctx.colors = arguments[3];
if (isBoolean(opts)) {
// legacy...
ctx.showHidden = opts;
} else if (opts) {
// got an "options" object
exports._extend(ctx, opts);
}
// set default options
if (isUndefined(ctx.showHidden)) ctx.showHidden = false;
if (isUndefined(ctx.depth)) ctx.depth = 2;
if (isUndefined(ctx.colors)) ctx.colors = false;
if (isUndefined(ctx.customInspect)) ctx.customInspect = true;
if (ctx.colors) ctx.stylize = stylizeWithColor;
return formatValue(ctx, obj, ctx.depth);
}
exports.inspect = inspect;
// http://en.wikipedia.org/wiki/ANSI_escape_code#graphics
inspect.colors = {
'bold': [1, 22],
'italic': [3, 23],
'underline': [4, 24],
'inverse': [7, 27],
'white': [37, 39],
'grey': [90, 39],
'black': [30, 39],
'blue': [34, 39],
'cyan': [36, 39],
'green': [32, 39],
'magenta': [35, 39],
'red': [31, 39],
'yellow': [33, 39]
};
// Don't use 'blue' not visible on cmd.exe
inspect.styles = {
'special': 'cyan',
'number': 'yellow',
'boolean': 'yellow',
'undefined': 'grey',
'null': 'bold',
'string': 'green',
'symbol': 'green',
'date': 'magenta',
// "name": intentionally not styling
'regexp': 'red'
};
var formatRegExp = /%[sdj%]/g;
exports.format = function (f) {
if (!isString(f)) {
var objects = [];
for (var j = 0; j < arguments.length; j++) {
objects.push(inspect(arguments[j]));
}
return objects.join(' ');
}
var i = 1;
var args = arguments;
var len = args.length;
var str = String(f).replace(formatRegExp, function (x) {
if (x === '%%') return '%';
if (i >= len) return x;
switch (x) {
case '%s':
return String(args[i++]);
case '%d':
return Number(args[i++]);
case '%j':
try {
return JSON.stringify(args[i++]);
} catch (_) {
return '[Circular]';
}
default:
return x;
}
});
for (var x = args[i]; i < len; x = args[++i]) {
if (isNull(x) || !isObject(x)) {
str += ' ' + x;
} else {
str += ' ' + inspect(x);
}
}
return str;
};
Harvested from: https://github.com/joyent/node/blob/master/lib/util.js
I have a slightly longer formatter for JavaScript here...
You can do formatting several ways:
String.format(input, args0, arg1, ...)
String.format(input, obj)
"literal".format(arg0, arg1, ...)
"literal".format(obj)
Also, if you have say a ObjectBase.prototype.format (such as with DateJS) it will use that.
Examples...
var input = "numbered args ({0}-{1}-{2}-{3})";
console.log(String.format(input, "first", 2, new Date()));
//Outputs "numbered args (first-2-Thu May 31 2012...Time)-{3})"
console.log(input.format("first", 2, new Date()));
//Outputs "numbered args(first-2-Thu May 31 2012...Time)-{3})"
console.log(input.format(
"object properties ({first}-{second}-{third:yyyy-MM-dd}-{fourth})"
,{
'first':'first'
,'second':2
,'third':new Date() //assumes Date.prototype.format method
}
));
//Outputs "object properties (first-2-2012-05-31-{3})"
I've also aliased with .asFormat and have some detection in place in case there's already a string.format (such as with MS Ajax Toolkit (I hate that library).
Using Lodash you can get template functionality:
Use the ES template literal delimiter as an "interpolate" delimiter.
Disable support by replacing the "interpolate" delimiter.
var compiled = _.template('hello ${ user }!');
compiled({ 'user': 'pebbles' });
// => 'hello pebbles!