forked from DebaucheryLibrarian/traxxx
139 lines
4.4 KiB
JavaScript
139 lines
4.4 KiB
JavaScript
'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 scrape(html, site, upcoming) {
|
|
const $ = cheerio.load(html, { normalizeWhitespace: true });
|
|
const sceneElements = $('.release-card.scene').toArray();
|
|
|
|
return sceneElements.reduce((acc, element) => {
|
|
const isUpcoming = $(element).find('.icon-upcoming.active').length === 1;
|
|
|
|
if ((upcoming && !isUpcoming) || (!upcoming && isUpcoming)) {
|
|
return acc;
|
|
}
|
|
|
|
const sceneLinkElement = $(element).find('a');
|
|
|
|
const url = `https://www.brazzers.com${sceneLinkElement.attr('href')}`;
|
|
const title = sceneLinkElement.attr('title');
|
|
const shootId = url.split('/').slice(-3, -2)[0];
|
|
|
|
const date = moment.utc($(element).find('time').text(), 'MMMM DD, YYYY').toDate();
|
|
const actors = $(element).find('.model-names a').map((actorIndex, actorElement) => $(actorElement).attr('title')).toArray();
|
|
|
|
const likes = Number($(element).find('.label-rating .like-amount').text());
|
|
const dislikes = Number($(element).find('.label-rating .dislike-amount').text());
|
|
|
|
const poster = `https:${$(element).find('.card-main-img').attr('data-src')}`;
|
|
const photos = $(element).find('.card-overlay .image-under').map((photoIndex, photoElement) => `https:${$(photoElement).attr('data-src')}`).toArray();
|
|
|
|
return acc.concat({
|
|
url,
|
|
shootId,
|
|
title,
|
|
actors,
|
|
date,
|
|
poster,
|
|
photos,
|
|
rating: {
|
|
likes,
|
|
dislikes,
|
|
},
|
|
site,
|
|
});
|
|
}, []);
|
|
}
|
|
|
|
async function scrapeScene(html, url, site) {
|
|
const $ = cheerio.load(html, { normalizeWhitespace: true });
|
|
const videoJson = $('script:contains("window.videoUiOptions")').html();
|
|
const videoData = JSON.parse(videoJson.slice(videoJson.indexOf('{"stream_info":'), videoJson.lastIndexOf('"},') + 2));
|
|
|
|
const shootId = url.split('/').slice(-3, -2)[0];
|
|
const title = $('.scene-title[itemprop="name"]').text();
|
|
|
|
const description = $('#scene-description p[itemprop="description"]')
|
|
.contents()
|
|
.first()
|
|
.text()
|
|
.trim();
|
|
|
|
const date = moment.utc($('.more-scene-info .scene-date').text(), 'MMMM DD, YYYY').toDate();
|
|
const actors = $('.related-model a').map((actorIndex, actorElement) => $(actorElement).text()).toArray();
|
|
const duration = Number($('.scene-length[itemprop="duration"]').attr('content').slice(1, -1)) * 60;
|
|
|
|
const likes = Number($('.label-rating .like').text());
|
|
const dislikes = Number($('.label-rating .dislike').text());
|
|
|
|
const siteElement = $('.niche-site-logo');
|
|
const siteUrl = `https://www.brazzers.com${siteElement.attr('href').slice(0, -1)}`;
|
|
const siteName = siteElement.attr('title');
|
|
|
|
const rawTags = $('.tag-card-container a').map((tagIndex, tagElement) => $(tagElement).text()).toArray();
|
|
|
|
const poster = `https:${videoData.poster}`;
|
|
const trailer = `https:${videoData.stream_info.http.paths.mp4_480_1500}`;
|
|
const photos = $('.carousel-thumb a').map((photoIndex, photoElement) => `https:${$(photoElement).attr('href')}`).toArray();
|
|
|
|
const [tags, channelSite] = await Promise.all([
|
|
matchTags(rawTags),
|
|
knex('sites')
|
|
.where({ url: siteUrl })
|
|
.orWhere({ name: siteName })
|
|
.first(),
|
|
]);
|
|
|
|
return {
|
|
url,
|
|
shootId,
|
|
title,
|
|
description,
|
|
actors,
|
|
date,
|
|
poster,
|
|
photos,
|
|
trailer: {
|
|
src: trailer,
|
|
quality: 480,
|
|
},
|
|
duration,
|
|
rating: {
|
|
likes,
|
|
dislikes,
|
|
},
|
|
tags,
|
|
site: channelSite || site,
|
|
};
|
|
}
|
|
|
|
async function fetchLatest(site, page = 1) {
|
|
const res = await bhttp.get(`${site.url}/page/${page}/`);
|
|
|
|
return scrape(res.body.toString(), site, false);
|
|
}
|
|
|
|
async function fetchUpcoming(site) {
|
|
const res = await bhttp.get(`${site.url}/`);
|
|
|
|
return scrape(res.body.toString(), site, true);
|
|
}
|
|
|
|
async function fetchScene(url, site) {
|
|
const res = await bhttp.get(url);
|
|
|
|
return scrapeScene(res.body.toString(), url, site);
|
|
}
|
|
|
|
module.exports = {
|
|
fetchLatest,
|
|
fetchUpcoming,
|
|
fetchScene,
|
|
};
|