scrape-meta/dist/rules/author.js

A library to easily scrape metadata from an article on the web using Open Graph metadata, regular HTML metadata, and series of fallbacks.

'use strict';

var isUrl = require('is-url');
var toTitle = require('to-title-case'

/**
 * Wrap a rule with validation and formatting logic.
 *
 * @param {Function} rule
 * @return {Function} wrapped
 */

);function wrap(rule) {
  return function ($) {
    var value = rule($);
    if (typeof value != 'string') return;
    if (isUrl(value)) return;
    if (value.indexOf('www.') === 0) return;
    if (value.includes('|')) return;

    // trim extra whitespace
    value = value.replace(/\s+/g, ' ');
    value = value.trim

    // remove any extra "by" in the start of the string
    ();value = value.replace(/^[\s\n]*by[\s\n]*/im, ''

    // make it title case, since some sites have it in weird casing
    );value = toTitle(value);

    return value;
  };
}

/**
 * Enforce stricter matching for a `rule`.
 *
 * @param {Function} rule
 * @return {Function} stricter
 */

function strict(rule) {
  return function ($) {
    var value = rule($);
    var regexp = /^\S+\s+\S+/;
    if (!regexp.test(value)) return;
    return value;
  };
}

/**
 * Rules.
 */

module.exports = [wrap(function ($) {
  return $('meta[property="article:author"]').attr('content');
}), wrap(function ($) {
  return $('meta[name="author"]').attr('content');
}), wrap(function ($) {
  return $('meta[name="sailthru.author"]').attr('content');
}), wrap(function ($) {
  return $('[rel="author"]').first().text();
}), wrap(function ($) {
  return $('[itemprop*="author"] [itemprop="name"]').first().text();
}), wrap(function ($) {
  return $('[itemprop*="author"]').first().text();
}), wrap(function ($) {
  return $('meta[property="book:author"]').attr('content');
}), strict(wrap(function ($) {
  return $('a[class*="author"]').first().text();
})), strict(wrap(function ($) {
  return $('[class*="author"] a').first().text();
})), strict(wrap(function ($) {
  return $('[class*="author"]').first().text();
})), strict(wrap(function ($) {
  return $('[class*="byline"]').text();
})), strict(wrap(function ($) {
  return $('a[href*="/author/"]').text();
}))];