compromise/src/terms/tagger/contraction/01-irregulars.js

natural language processing in the browser

'use strict';
const fixContraction = require('./fix');

const irregulars = {
  'wanna': ['want', 'to'],
  'gonna': ['going', 'to'],
  'im': ['i', 'am'],
  'alot': ['a', 'lot'],

  'dont': ['do', 'not'],
  'dun': ['do', 'not'],

  'won\'t': ['will', 'not'],
  'wont': ['will', 'not'],

  'can\'t': ['can', 'not'],
  'cant': ['can', 'not'],
  'cannot': ['can', 'not'],

  'aint': ['is', 'not'], //or 'are'
  'ain\'t': ['is', 'not'],
  'shan\'t': ['should', 'not'],

  'where\'d': ['where', 'did'],
  'whered': ['where', 'did'],
  'when\'d': ['when', 'did'],
  'whend': ['when', 'did'],
  'how\'d': ['how', 'did'],
  'howd': ['how', 'did'],
  'what\'d': ['what', 'did'],
  'whatd': ['what', 'did'],
  'let\'s': ['let', 'us'],
// 'dunno': ['do', 'not', 'know'],
// 'brb': ['be', 'right', 'back']
};

//check irregulars
const checkIrregulars = (ts) => {
  let irreg = Object.keys(irregulars);
  for(let i = 0; i < irreg.length; i++) {
    for(let t = 0; t < ts.terms.length; t++) {
      if (ts.terms[t].normal === irreg[i]) {
        let fix = irregulars[irreg[i]];
        ts = fixContraction(ts, fix, t);
        break;
      }
    }
  }
  return ts;
};
module.exports = checkIrregulars;