traxxx/src/scrapers/xempire.js

55 lines
1.6 KiB
JavaScript

'use strict';
const bhttp = require('bhttp');
const cheerio = require('cheerio');
const moment = require('moment');
function scrape(html, site) {
const $ = cheerio.load(html, { normalizeWhitespace: true });
const scenesElements = $('.sceneInfo').toArray();
return scenesElements.map((element) => {
const sceneLinkElement = $(element).find('.sceneTitle a');
const url = `${site.url}${sceneLinkElement.attr('href')}`;
const title = sceneLinkElement.attr('title');
const [likes, dislikes] = $(element).find('.value')
.toArray()
.map(value => Number($(value).text()));
const stars = Math.floor(((likes * 5 + dislikes) / (likes + dislikes)) * 100) / 100;
const actors = $(element).find('.sceneActors a')
.map((actorIndex, actorElement) => $(actorElement).attr('title'))
.toArray();
const date = moment.utc($(element).find('.sceneDate').text(), 'MM-DD-YYYY').toDate();
return {
url,
title,
actors,
date,
rating: {
likes,
dislikes,
stars,
},
siteId: site.id,
};
});
}
async function fetchReleases(site) {
const [latestRes, upcomingRes] = await Promise.all([
bhttp.get(`${site.url}/en/videos`),
bhttp.get(`${site.url}/en/videos/AllCategories/0/1/upcoming`),
]);
return {
latest: scrape(latestRes.body.toString(), site),
upcoming: scrape(upcomingRes.body.toString(), site),
};
}
module.exports = fetchReleases;