rrl-parser.js

const date = require('date.js');
const cheerio = require('cheerio');
const { readFileSync } = require('fs');

const html = readFileSync('active-popular.html', 'utf8');

const parseFiction = ($, el) => ({
  image: $(el).find('img').attr('src'),
  title: $(el).find('.fiction-title').children('a').text(),
  id: $(el).find('.fiction-title').children('a').attr('href').split('/')[2],
  type: $(el).find('span.label.bg-blue-hoki').text(),
  tags: $(el).find('span.label.bg-blue-dark')
    .map((i, el) => $(el).text()).get(),
});

const parseLatest = (html) => {
  const $ = cheerio.load(html);

  const fictions = $('.fiction-list-item').map((i, el) => Object.assign({
    recent: $(el).find('li.list-item').map((i, el) => ({
      name: $(el).find('span.col-xs-8').text(),
      created: date($(el).find('time').text()),
    })).get(),
  }, parseFiction($, el))).get();

  return fictions;
};

const parsePopular = (html) => {
  const parseStats = ($, el) => {
    let stats = {};

    stats.rating = $(el).find('.star').attr('title');
    stats.latest = date($(el).find('time').attr('datetime'));

    $(el).find('span').each((i, el) => {
      const text = $(el).text().toLowerCase();
      const key = text.split(' ')[1];
      const value = parseInt(text.split(' ')[0], 10);

      if (!key || !value) { return; }

      stats[key] = value;
    });

    return stats;
  }

  const $ = cheerio.load(html);

  const fictions = $('.fiction-list-item').map((i, el) => {
    return Object.assign(parseStats($, el), parseFiction($, el));
  }).get();

  return fictions;
}

console.log(JSON.stringify(parsePopular(html), 0, 2));