traxxx/src/scrapers/julesjordan.js

84 lines
2.1 KiB
JavaScript
Raw Normal View History

'use strict';
const bhttp = require('bhttp');
const cheerio = require('cheerio');
const moment = require('moment');
function scrapeLatest(html, site) {
const $ = cheerio.load(html, { normalizeWhitespace: true });
const scenesElements = $('.update_details').toArray();
return scenesElements.map((element) => {
const sceneLinkElement = $(element).children('a').eq(1);
const url = sceneLinkElement.attr('href');
const title = sceneLinkElement.text();
const date = moment
.utc($(element).find('.update_date').text(), 'MM/DD/YYYY')
.toDate();
const actors = $(element).find('.update_models a')
.map((actorIndex, actorElement) => $(actorElement).text())
.toArray();
return {
url,
title,
actors,
date,
rating: null,
site,
};
});
}
function scrapeUpcoming(html, site) {
const $ = cheerio.load(html, { normalizeWhitespace: true });
const scenesElements = $('#coming_soon_carousel').find('.table').toArray();
return scenesElements.map((element) => {
const details = $(element).find('.update_details_comingsoon')
.eq(1)
.children()
.remove();
const title = details
.end()
.text()
.trim();
const actors = details
.text()
.trim()
.split(', ');
const date = moment
.utc($(element).find('.update_date_comingsoon').text().slice(7), 'MM/DD/YYYY')
.toDate();
return {
url: null,
title,
actors,
date,
rating: null,
site,
};
});
}
async function fetchReleases(site) {
const [latestRes, upcomingRes] = await Promise.all([
bhttp.get(`${site.url}/categories/movies_1_d.html`),
bhttp.get(`${site.url}/index.php`),
]);
return [
...scrapeUpcoming(upcomingRes.body.toString(), site, true),
...scrapeLatest(latestRes.body.toString(), site),
];
}
module.exports = fetchReleases;