esthesaurus/src/thesaurus.js

Tesauro de castellano

const fs = require('fs');

/**
 * @module Thesaurus loads partialThesaurus based on the given word.
 * Exposes some methods, basically includes.
 */

/** @todo Class */
let Thesaurus = { words: {} };

/**
 * Normalizes the word and gets the possible words to check. If the partials dicts haven't been
 * loaded, loads it into Thesaurus.
 * @param {String} word 
 * @returns Normalized word and beggining.
 */
function _initWord (word) {
  return word.normalize('NFD').replace(/[\u0300-\u036f]/g, "").toLowerCase().trim();
}

/** Loads the file with the given word beggining */
function _getPartialDict (word) {
  let beggining = word.length > 1
    ? word.slice(0,2)
    : word.slice(0,1);
  
  const path = `./words/${beggining}.js`;

  if (!Thesaurus.words[beggining]) {
    try {
      Thesaurus.words[beggining] = require(path);
    } catch (err) {
      return false;
    }
  }
  return Thesaurus.words[beggining];
}

/**
 * Checks if a word is included in the Thesaurus.
 */
Thesaurus.includes = (w) => {
  let word = _initWord(w);
  let toCheck = _getWordsToCheck(word);
  
  if (!toCheck.length) {
    throw "Unexpected error. Not word to check";
  } 
  
  return Boolean(toCheck.some(w => {
    let partial = _getPartialDict(w);
    return partial && partial.includes(w);
  }));
}


/**
 * Chechs if the word is plural, femenin or has a prefix.
 * @param {string} word 
 * @returns array of words to check
 */
function _getWordsToCheck (word) {
  let toCheck = [..._removePrefix(word), word];
  toCheck.unshift(..._getSingulars(toCheck));
  /** May induce false positives */
  toCheck.unshift(..._getMasculin(toCheck));
  return toCheck;
}
/**
 * Gets an array with all the prefixes.
 * @param update Boolean; reads all files in Thesaurus and updates the prefixes file.
 * This is so not to be checjing all words every time this function is called.
 */
Thesaurus.getPrefixes = (update = false) => {
  let prefixesPath = './src/prefixes.js';
  if (update) {
    let all = Thesaurus.all();
    Thesaurus.prefixes = all.filter(el => /.*[-]$/.test(el)).map(el => el.slice(0, -1));
    let data = 'module.exports = ' + JSON.stringify(Thesaurus.prefixes);
    fs.writeFileSync(prefixesPath, data);
  } else if (!Thesaurus.prefixes) {
    Thesaurus.prefixes = require('./prefixes.js');
  }
  return Thesaurus.prefixes;
}

/** Loads all the data in the Thesaurus */
Thesaurus.loadAll = () => {
  fs.readdirSync('src/words').forEach(f => _getPartialDict(f.replace('.js', '')));
}

/** Returns all the words in the Thesaurs as a single array */
Thesaurus.all = () => {
  Thesaurus.loadAll();
  let all = [];
  Object.values(Thesaurus.words).forEach(w => all.push(...w));
  return all;
}

function _removePrefix(w) {
  let result = [];
  let reg = (p) => `^${p}(.*)`;
  Thesaurus.getPrefixes().forEach(prefix => {
    let match = w.match(reg(prefix));
    // console.log('testing', prefix, 'against', w)
    match && result.push(match[1]);
  });
  return result;
}


/**
 * If the word it's a plural, checks its singular form.
 * @param {string} word 
 * @param {object} partialDict 
 * @example 'bueyes' -> first checks buey; 
 * if not included checks bueye
 * and if not included then bueyes
 */
function _getSingulars (words){
  let sing = [];
  words.forEach(word => {
    let letters = word.split('');
    letters.pop() === 's' && sing.unshift(letters.join('')); 
    letters.pop() === 'e' && sing.unshift(letters.join(''));
  });
  return sing;
}

function _getMasculin(words) {
  let ret = [];
  words.forEach(word => {
    let letters = word.split('');
    if (letters.pop() === 'a') {
      letters.push('o');
      ret.push( letters.join(''))
    }
  });
  
  return ret;
}

module.exports = Thesaurus;