JavaScript random word generator

wordGen.js

//  DO WHAT THE FUCK YOU WANT TO PUBLIC LICENSE
// Version 2, December 2004
//
// Copyright (C) 2015 Swen Wenzel <swenzel@uos.de>
//
// Everyone is permitted to copy and distribute verbatim or modified
// copies of this license document, and changing it is allowed as long
// as the name is changed.
//
// DO WHAT THE FUCK YOU WANT TO PUBLIC LICENSE
// TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
//
// 0. You just DO WHAT THE FUCK YOU WANT TO.

// This generator yields everything from ~2% real english words over ~75% english sounding words and ~10% strange words to ~13% unreadable crap.
// Try it out. It was created as a tradeoff between size and good results, if you need something better, well do it yourself.
// If you drop me a line I might also be able to dig out the initial python source for this thing which was more powerful and worked
// with more than just 1grams and 10 followers.
// If you think you have an improvement, write a comment.

// list of characters with their followers' corresponding accumulated propabilities
// accumulated probabilities are usable as keys and work great with Math.random()
var grams = {
    " ": {0.6294: "s", 1.0: "t", 0.5166: "i", 0.2635: "c", 0.1255: "h", 0.7882: "a", 0.4297: "w", 0.1915: "b", 0.3464: "o", 0.0603: "f"},
    "a": {1.0: "n", 0.507: "l", 0.1379: "i", 0.3973: " ", 0.0446: "y", 0.6368: "r", 0.1865: "c", 0.0895: "m", 0.2906: "s", 0.7853: "t"},
    "c": {1.0: "o", 0.078: "r", 0.7882: "h", 0.3303: "t", 0.6334: "e", 0.181: "i", 0.0365: " ", 0.4808: "a", 0.2455: "k", 0.1214: "l"},
    "b": {1.0: "e", 0.0168: "s", 0.5666: "o", 0.3311: "l", 0.7064: "a", 0.042: " ", 0.2285: "i", 0.1634: "r", 0.1023: "y", 0.445: "u"},
    "e": {1.0: " ", 0.3763: "s", 0.2035: "a", 0.2859: "d", 0.0326: "c", 0.6218: "r", 0.1016: "e", 0.0666: "t", 0.14: "l", 0.4732: "n"},
    "d": {0.4397: "e", 1.0: " ", 0.0217: "d", 0.0671: "u", 0.044: "r", 0.0109: "y", 0.2946: "i", 0.0974: "s", 0.1495: "a", 0.2076: "o"},
    "g": {0.4683: "h", 1.0: " ", 0.0461: "s", 0.3648: "o", 0.0214: "n", 0.1991: "r", 0.2811: "a", 0.0805: "u", 0.1315: "i", 0.6221: "e"},
    "f": {0.1547: "a", 1.0: " ", 0.0923: "t", 0.0555: "u", 0.3867: "e", 0.2186: "f", 0.2981: "r", 0.0235: "l", 0.5046: "i", 0.7035: "o"},
    "i": {0.125: "e", 1.0: "n", 0.5324: "s", 0.1826: "d", 0.4017: "c", 0.3255: "o", 0.6677: "t", 0.2528: "l", 0.0382: "g", 0.081: "r"},
    "h": {0.0636: "t", 1.0: "e", 0.0099: "y", 0.3911: "i", 0.0382: "r", 0.0043: "n", 0.1563: "o", 0.2699: " ", 0.0219: "u", 0.5663: "a"},
    "k": {1.0: "e", 0.0184: "y", 0.2099: "s", 0.0336: "o", 0.1346: "n", 0.0828: "a", 0.664: " ", 0.0066: "u", 0.0526: "l", 0.3508: "i"},
    "j": {0.3455: "a", 1.0: "u", 0.0009: "s", 0.0314: " ", 0.2002: "e", 0.0618: "i", 0.0051: "p", 0.0021: "c", 0.6376: "o", 0.0132: "r"},
    "m": {1.0: "e", 0.7376: "a", 0.0544: "b", 0.2917: "i", 0.1154: "u", 0.4078: "o", 0.0265: "s", 0.1847: "p", 0.5582: " ", 0.0833: "m"},
    "l": {1.0: " ", 0.8256: "e", 0.0557: "s", 0.5064: "i", 0.2714: "o", 0.0221: "u", 0.1913: "y", 0.6551: "l", 0.3799: "a", 0.1173: "d"},
    "o": {0.4793: "u", 1.0: "n", 0.8009: " ", 0.0373: "o", 0.3502: "f", 0.1282: "w", 0.0817: "l", 0.1832: "t", 0.6375: "r", 0.2536: "m"},
    "n": {1.0: " ", 0.5597: "g", 0.4228: "t", 0.1167: "i", 0.3088: "e", 0.7117: "d", 0.2199: "s", 0.0764: "c", 0.0377: "a", 0.1673: "o"},
    "q": {1.0: "u", 0.0059: "l", 0.0136: "b", 0.0017: "s", 0.0348: "i", 0.0218: "a", 0.003: "w", 0.0043: "w", 0.0006: "v", 0.0888: " "},
    "p": {1.0: "e", 0.18: "i", 0.5315: "a", 0.4013: "l", 0.8133: "r", 0.124: "p", 0.2815: " ", 0.0687: "u", 0.6631: "o", 0.0274: "t"},
    "s": {1.0: " ", 0.2935: "i", 0.1824: "a", 0.3978: "e", 0.5411: "t", 0.0947: "s", 0.1379: "h", 0.053: "u", 0.0224: "p", 0.2328: "o"},
    "r": {1.0: "e", 0.1439: "t", 0.0554: "d", 0.2183: "s", 0.0245: "n", 0.5049: "o", 0.7461: " ", 0.0896: "y", 0.4002: "i", 0.3042: "a"},
    "u": {1.0: "r", 0.5156: "s", 0.2004: " ", 0.3647: "l", 0.259: "p", 0.1475: "c", 0.6674: "n", 0.0476: "e", 0.8267: "t", 0.0963: "g"},
    "t": {0.1525: "a", 0.0368: "u", 1.0: "h", 0.1056: "r", 0.4593: "o", 0.071: "s", 0.3457: "e", 0.7133: " ", 0.0183: "t", 0.243: "i"},
    "w": {0.0373: "s", 0.0059: "l", 0.8036: "a", 1.0: "i", 0.6122: "e", 0.3096: "o", 0.1913: " ", 0.0183: "r", 0.4543: "h", 0.0822: "n"},
    "v": {1.0: "e", 0.007: "s", 0.0014: "d", 0.0737: "o", 0.0234: " ", 0.0029: "r", 0.1484: "a", 0.0112: "y", 0.339: "i", 0.0045: "u"},
    "y": {0.204: "e", 1.0: " ", 0.0681: "i", 0.0136: "l", 0.0314: "t", 0.0223: "m", 0.303: "o", 0.0475: "a", 0.0056: "p", 0.1217: "s"},
    "x": {1.0: "p", 0.7464: " ", 0.0368: "u", 0.2706: "i", 0.1089: "c", 0.0215: "h", 0.01: "o", 0.5265: "t", 0.1895: "e", 0.3517: "i"},
    "z": {1.0: "e", 0.4678: " ", 0.2055: "o", 0.3289: "i", 0.0684: "y", 0.043: "u", 0.0075: "h", 0.024: "l", 0.6314: "a", 0.1182: "z"}
};
// create a sorted list for all keys
for(var key in grams){
    var sorted = [];
    for(var p in grams[key])
        sorted.push(p);
    grams[key].sorted = sorted.sort();
}

function generateWords(n){
    if(!n){
        n = 1;
    }
    var words = [];
    for(var i=0; i<n; i++){
        var w = ' ';
        // last will be our 1gram used to find a proper follower
        var last = w;
        while(true){

            var rand = Math.random();
            var p_list = grams[last].sorted;
            // find the follower corresponding to the random number
            // Note: p_list containes the accumulated probabilities of
            // the followers.
            for(var k=0; k+1 < p_list.length && p_list[k] < rand; k++);

            char = grams[last][p_list[k]];
            if(char === ' '){
                // space indicates end, but we do not want words shorter
                // than 5 so make sure we have minimum length
                if(w.length > 5)
                    break;
            }
            else{
                w += char;
            }
            last = char;
        }
        // strip the leading space before appending to the list
        words.push(w.substring(1));
    }
    return words;
}

Thanks for sharing Swen. This has been very helpful for a project I'm working on. Do you by chance know of a JS or Node lib that outputs a grams object with that same structure given a text input?

Hey @gragland
I'm glad you find it helpful. TBH I completely forgot I even created and published this...
Unfortunately I don't know of any such library and I'm afraid I cannot really uphold my promise "to dig out the initial python source for this thing" the HDD I assume this might be on broke down a few years ago.

But it's not very complicated actually. Should be something like this:

1. replace punctuation with space (or apply an arbitrarily complex stemming mechanism you like)
2. build pairs of adjacent characters
3. group by first character
4. count occurrences of second character and divide by size of group
5. build cumulative sum of those relative counts

IIRC I even experimented with triplets but the issue was that this led to too many real words.

Hey, I appreciate the reply! I actually ended up figuring out and doing basically what you describe.