Wordish

A generator of word like phrases. Useful for passwords and publishing. Try it out: cmroanirgo.github.io/wordish/.

Installation (for commandline use)

npm install -g wordish
wordish

Installation (for api use)

npm install -s wordish

Installation (for use in browser)

In your script, add the following and you're good to go!

<script type="text/javascript" src='https://unpkg.com/wordish@latest/dist/wordish.min.js'></script>
<script type="text/javascript">
	function generate(){
		var dict = new Wordish(10); // 10=high accuracy/readbility of words
		dict.learn(sourceText, { });
		var words = dict.createWords({
		});
		$('#result').value  = words.join(' ');
	}
</script>

Wordish API

The main class that you interact with is Wordish and is always instantiated by using the following:

var accuracy = 5; // from 2 to 20..ish. 10 is very accurate, 2 is very random/ gibberish
var dict = new Wordish(accuracy);
or
var dict = new Wordish(); // the default accuracy is 5

Note about accuracy

The accuracy controls how many consecutive letters are remembered. For instance, if the accuracy is 5, when a word longer than that appears (eg 'welcome', then the tree would stop remembering after 'welco' and the 'me' gets munged into the rest of the tree)

learn

This is the first function to be called in order to populate the dictionary. It may be called multiple times to load different source texts. It has the form:

dict.learn(sourceText[,options]);

where options may contain:

• minWordLen. Any words shorter than this will be removed
• maxWordLen. Any words longer than this will be removed
• validator. This is a callback function which simply sanitises the input sourceText. Any characters (whether english letters or not), will be added to the dictionary.

The default validator is good for English words and looks like:

function defaultValidator(text) {
	return text.toLowerCase() // ignore case
		.replace(/[^a-z]/gi, ' ') // make invalid chars a SPACE
		.replace(/\b(?:and|an|or|the)\b/gi, ' ')// trim *very* common words
		.replace(/ {2,}/g, ' ').trim(); // remove excess whitespace
};

A Klingon validator would be different, because it has uppercase characters and apostrophes inside a word:

function klingonValidator(text) {
	var reInvalidChars = /[^a-z\']/gi; // uses upper/lower AND '
	return text 
		.replace(reInvalidChars, ' ') // make invalid chars a SPACE
		.replace(/ {2,}/g, ' ').trim(); // remove excess whitespace
}

The default set of options for learn are:

var _defaultLearnOptions = {
	minWordLen: 0,
	maxWordLen: 25,
	validator: defaultValidator
};

createWords

Once learning is complete, you may begin generating words. Usage:

var words = dict.createWords([options]);

The returned data is an array of words that match the constraints set by options.

The default set of options for createWords are:

var _defaultCreateOptions = {
	numWords: 4,
	randomizeNumWords: 1,
	minWordLen: 5,
	maxWordLen: 10
};

To generate a password, separated by a -:

var words = dict.createWords();
var password = words.join('-')

To generate a phrase (with a leading capital letter and ending with a period):

var options = { numWords:5, randomizeNumWords:12 }; // between 5 and 17 words will be generated.
var words = dict.createWords(options);
var phrase = words.join(' ').replace(/^(.)/, function(m, m1)  { return m1.toUpperCase(); }) + '.';

reset

Resets any 'learnt' state back to original, but has little real world use. This is effectively the same as recreating a new Wordish object. ie:

dict.reset();
...
dict.learn(...);
var words = dict.createWords(...)

API Examples

Simple

const fs = require('fs')
const Wordish = require('Wordish');
const filename = "somefile.txt"; 

// Load up a text document that we'll use for source words
var accuracy = 5; // from 2 to 20..ish. 10 is very accurate, 2 is very random/ gibberish. Words longer than 5 will get 'munged' into the dictionary.
var dict = new Wordish(accuracy);
dict.learn(fs.readFileSync(filename, {encoding:'utf8'}))

// generate a list of words
var options = { // these options are all optional
	minWordLen: 5,
	maxWordLen: 10,
	numWords: 10,
	randomizeNumWords: 2 // between 10 and 12 words will be made
}
console.log(dict.createWords(options).join('-'));

A correct battery horse staple password generator.

Note how the learn option, minWordLen here matches the createWord option & means that shorter words cannot be used to 'make up' a word. You need to use a long section of source text for this to work securely, however. (eg War and Peace, The Bible, or a list of the most common words):

const fs = require('fs')
const Wordish = require('wordish');
const filename = "somefile.txt"; 

// Load up a text document that we'll use for source words
var accuracy = 10; 
var dict = new Wordish(accuracy);
dict.learn(fs.readFileSync(filename, {encoding:'utf8'}), {minWordLen: 5}) 

// generate a list of words
var options = { // these options are all optional
	minWordLen: 5,
	maxWordLen: 10,
	numWords: 4,
	randomizeNumWords: 2 // betwen 4 and 6 words made
}
for (var i=0; i<5; i++)
	console.log(dict.createWords(options).join(' '));

Support Klingon!:

const fs = require('fs')
const Wordish = require('wordish');
const filename = "klingon.txt"; 

// Load up a text document that we'll use for source words
var accuracy = 10; 
var dict = new Wordish(accuracy);
dict.learn(fs.readFileSync(filename, {encoding:'utf8'}), klingonValidator) // the learn function can also take just a validatorFunction as an options argument


var options = {}; // use defaults
for (var i=0; i<5; i++) // create a series of phrases
	console.log(dict.createWords(options).join(' '));

function klingonValidator(text) {
	var reInvalidChars = /[^a-z\']/gi; // uses upper/lower AND '
	return text 
		.replace(reInvalidChars, ' ') // make invalid chars a SPACE
		.replace(/ {2,}/g, ' ').trim(); // remove excess whitespace
}

Demo klingon output (*):

neH puql loD afflichen ghaH womb loD
Earthly chenmo tlhuH Hev tlhIH vo' vaj SoS
palaces 'a' Hegh 'ej Hoch 'ej

(*) Not actually klingon!