traxxx/src/scrapers/private.js

121 lines
4.1 KiB
JavaScript
Raw Normal View History

'use strict';
/* eslint-disable newline-per-chained-call */
const bhttp = require('bhttp');
const cheerio = require('cheerio');
const moment = require('moment');
async function getPhotos(entryId, site) {
const { hostname } = new URL(site.url);
const res = await bhttp.get(`https://${hostname}/gallery.php?type=highres&id=${entryId}`);
const html = res.body.toString();
const $ = cheerio.load(html, { normalizeWhitespace: true });
const photos = $('a.fakethumb').map((photoIndex, photoElement) => $(photoElement).attr('data-src') || $(photoElement).attr('href')).toArray();
return photos;
}
function scrapeLatest(html, site) {
const $ = cheerio.load(html, { normalizeWhitespace: true });
const sceneElements = $('.content-wrapper .scene').toArray();
return sceneElements.map((element) => {
const sceneLinkElement = $(element).find('h3 a');
const url = sceneLinkElement.attr('href');
const title = sceneLinkElement.text();
const entryId = url.split('/').slice(-1)[0];
const date = moment.utc($(element).find('.scene-date'), 'MM/DD/YYYY').toDate();
const actors = $(element).find('.scene-models a').map((actorIndex, actorElement) => $(actorElement).text()).toArray();
const likes = Number($(element).find('.scene-votes').text());
const thumbnailElement = $(element).find('img.img-responsive');
const photoCount = Number(thumbnailElement.attr('thumbs_num'));
const poster = thumbnailElement.attr('src');
const photos = Array.from({ length: photoCount }, (val, index) => thumbnailElement.attr(`src${index + 1}`));
const scene = {
url,
entryId,
title,
actors,
date,
poster,
photos,
rating: {
likes,
},
site,
};
return scene;
});
}
async function scrapeScene(html, url, site) {
const $ = cheerio.load(html, { normalizeWhitespace: true });
const release = {};
[release.entryId] = url.split('/').slice(-1);
release.title = $('.video-wrapper meta[itemprop="name"]').attr('content');
release.description = $('.video-wrapper meta[itemprop="description"]').attr('content');
release.date = moment.utc($('.video-wrapper meta[itemprop="uploadDate"]').attr('content'), 'MM/DD/YYYY').toDate();
release.actors = $('.content-wrapper .scene-models-list a').map((actorIndex, actorElement) => $(actorElement).text()).toArray();
const timestamp = $('.video-wrapper meta[itemprop="duration"]').attr('content');
if (timestamp) {
const [minutes, seconds] = timestamp.match(/\d+/g);
release.duration = Number(minutes) * 60 + Number(seconds);
}
release.tags = $('.content-desc .scene-tags a').map((tagIndex, tagElement) => $(tagElement).text()).toArray();
release.likes = Number($('.content-desc #social-actions #likes').text());
const posterScript = $('script:contains(poster)').html();
const posterLink = posterScript.slice(posterScript.indexOf('https://'), posterScript.indexOf('.jpg') + 4);
release.poster = $('meta[property="og:image"]').attr('content') || posterLink;
release.trailer = {
src: $('meta[property="og:video"]').attr('content') || $('#videojs-trailer source').attr('src'),
};
release.photos = await getPhotos(release.entryId, site);
release.movie = $('a[data-track="FULL MOVIE"]').attr('href');
const siteElement = $('.content-wrapper .logos-sites a');
if (siteElement) release.channel = siteElement.text().replace(/\s+/g, '').toLowerCase();
return release;
}
async function fetchLatest(site, page = 1) {
const { hostname } = new URL(site.url);
if (hostname.match('private.com')) {
const res = await bhttp.get(`${site.url}/${page}/`);
return scrapeLatest(res.body.toString(), site);
}
const res = await bhttp.get(`${site.url}/scenes/${page}/`);
return scrapeLatest(res.body.toString(), site);
}
async function fetchScene(url, site) {
const res = await bhttp.get(url);
return scrapeScene(res.body.toString(), url, site);
}
module.exports = {
fetchLatest,
fetchScene,
};