fiction-word/dist/word-length.js

A random word generator.

"use strict";
/*
 * This file contains functions for generating random word lengths based on a predefined distribution.
 * The distribution is based on the frequency of word lengths in default linux English dictionary.
 */
Object.defineProperty(exports, "__esModule", { value: true });
exports.getRandomWordLength = exports.generateDistribution = void 0;
var dictionaryDistribution = [
    [1, 52],
    [2, 488],
    [3, 1385],
    [4, 3688],
    [5, 6717],
    [6, 10268],
    [7, 13451],
    [8, 13869],
    [9, 12363],
    [10, 9823],
    [11, 6922],
    [12, 4454],
    [13, 2549],
    [14, 1284],
    [15, 629],
    [16, 260],
    [17, 124],
    [18, 38],
    [19, 18],
    [20, 13], // 20+
];
var corpusDistribution = [
    [1, 0.03],
    [2, 0.06],
    [3, 0.18],
    [4, 0.19],
    [5, 0.16],
    [6, 0.13],
    [7, 0.1],
    [8, 0.07],
    [9, 0.04],
    [10, 0.02],
    [11, 0.02], // 11+
];
/**
 * @description Computes the cumulative distribution of word lengths.
 * @returns An array of cumulative probabilities for each word length.
 */
function generateDistribution(mode) {
    if (mode === void 0) { mode = "dictionary"; }
    // Distribution of word lengths
    var distribution = mode === "dictionary" ? dictionaryDistribution : corpusDistribution;
    var total = distribution.reduce(function (sum, _a) {
        var count = _a[1];
        return sum + count;
    }, 0);
    // Precompute cumulative probabilities
    var cumulative = [];
    var sum = 0;
    for (var _i = 0, distribution_1 = distribution; _i < distribution_1.length; _i++) {
        var _a = distribution_1[_i], length = _a[0], count = _a[1];
        // Don't include 20+ as it's a tail distribution
        if (length === 20)
            break;
        sum += count;
        cumulative.push([length, sum / total]);
    }
    return cumulative;
}
exports.generateDistribution = generateDistribution;
/**
 * Sample a word length from the long tail distribution.
 * @param min The minimum word length.
 * @param max The maximum word length.
 * @param decay The decay rate.
 * @returns A random word length.
 */
function sampleLongTail(min, max, decay) {
    if (min === void 0) { min = 20; }
    if (max === void 0) { max = 30; }
    if (decay === void 0) { decay = 0.6; }
    // A geometric decay for 20+ length
    var r = Math.random();
    var cumulative = 0;
    for (var i = 0; i <= max - min; i++) {
        var prob = Math.pow(decay, i) * (1 - decay);
        cumulative += prob;
        if (r < cumulative)
            return min + i;
    }
    return max; // fallback
}
/**
 * Sample a word length from the distribution.
 * @returns A random word length.
 */
function getRandomWordLength(distribution) {
    distribution = distribution || generateDistribution();
    var rand = Math.random();
    for (var _i = 0, distribution_2 = distribution; _i < distribution_2.length; _i++) {
        var _a = distribution_2[_i], length = _a[0], prob = _a[1];
        if (rand < prob)
            return length;
    }
    // If it falls into the 20+ tail
    return sampleLongTail();
}
exports.getRandomWordLength = getRandomWordLength;