2019-04-12 01:12:53 +00:00
|
|
|
'use strict';
|
|
|
|
|
|
|
|
/* eslint-disable newline-per-chained-call */
|
|
|
|
const bhttp = require('bhttp');
|
|
|
|
const cheerio = require('cheerio');
|
|
|
|
const moment = require('moment');
|
|
|
|
|
|
|
|
const knex = require('../knex');
|
|
|
|
const { matchTags } = require('../tags');
|
|
|
|
|
|
|
|
function scrapeLatest(html, site) {
|
|
|
|
const $ = cheerio.load(html, { normalizeWhitespace: true });
|
|
|
|
const sceneElements = $('.echThumb').toArray();
|
|
|
|
|
|
|
|
return sceneElements.map((element) => {
|
|
|
|
const sceneLinkElement = $(element).find('.thmb_lnk');
|
|
|
|
const title = sceneLinkElement.attr('title');
|
|
|
|
const url = `https://bangbros.com${sceneLinkElement.attr('href')}`;
|
|
|
|
const shootId = sceneLinkElement.attr('id') && sceneLinkElement.attr('id').split('-')[1];
|
|
|
|
const entryId = url.split('/')[3].slice(5);
|
|
|
|
|
|
|
|
const date = moment.utc($(element).find('.thmb_mr_2 span.faTxt').text(), 'MMM D, YYYY').toDate();
|
|
|
|
const actors = $(element).find('.cast-wrapper a.cast').map((actorIndex, actorElement) => $(actorElement).text().trim()).toArray();
|
|
|
|
|
2019-10-31 00:53:26 +00:00
|
|
|
const photoElement = $(element).find('.rollover-image');
|
|
|
|
const poster = `https:${photoElement.attr('data-original')}`;
|
|
|
|
|
|
|
|
const photosUrl = photoElement.attr('data-rollover-url');
|
|
|
|
const photosMaxIndex = photoElement.attr('data-rollover-max-index');
|
|
|
|
const photos = Array.from({ length: photosMaxIndex }, (val, index) => `https:${photosUrl}big${index + 1}.jpg`);
|
|
|
|
|
2019-04-12 01:12:53 +00:00
|
|
|
const duration = moment.duration(`0:${$(element).find('.thmb_pic b.tTm').text()}`).asSeconds();
|
|
|
|
|
|
|
|
return {
|
|
|
|
url,
|
|
|
|
entryId,
|
|
|
|
shootId,
|
|
|
|
title,
|
|
|
|
actors,
|
|
|
|
date,
|
|
|
|
duration,
|
2019-10-31 00:53:26 +00:00
|
|
|
poster,
|
|
|
|
photos,
|
2019-04-12 01:12:53 +00:00
|
|
|
rating: null,
|
|
|
|
site,
|
|
|
|
};
|
|
|
|
});
|
|
|
|
}
|
|
|
|
|
|
|
|
async function scrapeScene(html, url, site) {
|
|
|
|
const $ = cheerio.load(html, { normalizeWhitespace: true });
|
|
|
|
const sceneElement = $('.playerSection');
|
|
|
|
|
|
|
|
const shootId = sceneElement.find('.vdoCast:contains("Release")').text().replace('Release: ', '');
|
|
|
|
const entryId = url.split('/')[3].slice(5);
|
|
|
|
const title = sceneElement.find('.ps-vdoHdd h1').text();
|
|
|
|
const description = sceneElement.find('.vdoDesc').text().trim();
|
|
|
|
|
|
|
|
const [siteName, ...actors] = sceneElement.find('.vdoCast a').map((actorIndex, actorElement) => $(actorElement).text()).toArray();
|
|
|
|
const siteId = siteName.replace(/[\s']+/g, '').toLowerCase();
|
|
|
|
|
2019-10-31 00:53:26 +00:00
|
|
|
const poster = `https:${$('img#player-overlay-image').attr('src')}`;
|
|
|
|
const trailer = `https:${$('source[type="video/mp4"]').attr('src')}`;
|
|
|
|
|
|
|
|
const firstPhotoUrl = `https:${$('img[data-slider-index="1"]').attr('src')}`;
|
|
|
|
// all scenes seem to have 12 album photos available, not always included on the page
|
|
|
|
const photos = Array.from({ length: 12 }, (val, index) => firstPhotoUrl.replace(/big\d+/, `big${index + 1}`));
|
|
|
|
|
2019-04-12 01:12:53 +00:00
|
|
|
const rawTags = $('.vdoTags a').map((tagIndex, tagElement) => $(tagElement).text()).toArray();
|
|
|
|
|
|
|
|
const [channelSite, tags] = await Promise.all([
|
2019-11-04 04:47:37 +00:00
|
|
|
site.isFallback
|
|
|
|
? knex('sites')
|
|
|
|
.where({ slug: siteId })
|
|
|
|
.orWhere({ name: siteName })
|
|
|
|
.first()
|
|
|
|
: site,
|
2019-04-12 01:12:53 +00:00
|
|
|
matchTags(rawTags),
|
|
|
|
]);
|
|
|
|
|
|
|
|
const stars = Number(sceneElement.find('.bVdPl_it_like .bVdPl_txt').text().replace('% like', '')) / 20;
|
|
|
|
|
|
|
|
return {
|
|
|
|
url,
|
|
|
|
shootId,
|
|
|
|
entryId,
|
|
|
|
title,
|
|
|
|
description,
|
|
|
|
actors,
|
|
|
|
tags,
|
2019-10-31 00:53:26 +00:00
|
|
|
poster,
|
|
|
|
photos,
|
|
|
|
trailer: {
|
|
|
|
src: trailer,
|
|
|
|
},
|
2019-04-12 01:12:53 +00:00
|
|
|
rating: {
|
|
|
|
stars,
|
|
|
|
},
|
|
|
|
site: channelSite || site,
|
|
|
|
};
|
|
|
|
}
|
|
|
|
|
|
|
|
async function fetchLatest(site, page = 1) {
|
2019-10-31 00:53:26 +00:00
|
|
|
const res = await bhttp.get(`https://bangbros.com/websites/${site.slug}/${page}`);
|
2019-04-12 01:12:53 +00:00
|
|
|
|
|
|
|
return scrapeLatest(res.body.toString(), site);
|
|
|
|
}
|
|
|
|
|
|
|
|
async function fetchScene(url, site) {
|
|
|
|
const { origin } = new URL(url);
|
|
|
|
const res = await bhttp.get(url);
|
|
|
|
|
|
|
|
if (origin !== 'https://bangbros.com') {
|
|
|
|
throw new Error('Cannot fetch from this URL. Please find the scene on https://bangbros.com and try again.');
|
|
|
|
}
|
|
|
|
|
|
|
|
return scrapeScene(res.body.toString(), url, site);
|
|
|
|
}
|
|
|
|
|
|
|
|
module.exports = {
|
|
|
|
fetchLatest,
|
|
|
|
fetchScene,
|
|
|
|
};
|