yoastseo/build/languageProcessing/researches/getProminentWordsForInsights.js

Yoast client-side content analysis

"use strict";

Object.defineProperty(exports, "__esModule", {
  value: true
});
exports.default = void 0;
var _lodash = require("lodash");
var _determineProminentWords = require("../helpers/prominentWords/determineProminentWords");
var _removeURLs = _interopRequireDefault(require("../helpers/sanitize/removeURLs.js"));
var _removeEmailAddresses = _interopRequireDefault(require("../helpers/sanitize/removeEmailAddresses"));
function _interopRequireDefault(e) { return e && e.__esModule ? e : { default: e }; }
/**
 * Retrieves the prominent words from the given paper.
 *
 * @param {Paper} paper             The paper to determine the prominent words of.
 * @param {Researcher} researcher   The researcher to use for analysis.
 *
 * @returns {WordCombination[]} Prominent words for this paper, filtered and sorted.
 */
function getProminentWordsForInsights(paper, researcher) {
  const functionWords = researcher.getConfig("functionWords");
  // An optional custom helper to return custom function to return the stem of a word.
  const customStemmer = researcher.getHelper("customGetStemmer");
  const stemmer = customStemmer ? customStemmer(researcher) : researcher.getHelper("getStemmer")(researcher);
  // An optional custom helper to get words from the text.
  const getWordsCustomHelper = researcher.getHelper("getWordsCustomHelper");
  let text = paper.getText();
  // We don't want to include URLs or email addresses in prominent words.
  text = (0, _removeURLs.default)(text);
  text = (0, _removeEmailAddresses.default)(text);

  // If the language has a custom helper to get words from the text, we don't retrieve the abbreviation.
  const abbreviations = getWordsCustomHelper ? [] : (0, _determineProminentWords.retrieveAbbreviations)(text);
  const prominentWordsFromText = (0, _determineProminentWords.getProminentWords)(text, abbreviations, stemmer, functionWords, getWordsCustomHelper);
  const collapsedWords = (0, _determineProminentWords.collapseProminentWordsOnStem)(prominentWordsFromText);
  (0, _determineProminentWords.sortProminentWords)(collapsedWords);

  /*
   * Collapse the list of prominent words on stems, sort it, filter out all words that occur less than
   * 5 times in the text. Return the 20 top items from this list.
   */
  return (0, _lodash.take)((0, _determineProminentWords.filterProminentWords)(collapsedWords, 5), 20);
}
var _default = exports.default = getProminentWordsForInsights;
//# sourceMappingURL=getProminentWordsForInsights.js.map