forked from DebaucheryLibrarian/traxxx
284 lines
8.9 KiB
JavaScript
284 lines
8.9 KiB
JavaScript
'use strict';
|
|
|
|
const Promise = require('bluebird');
|
|
const bhttp = require('bhttp');
|
|
const cheerio = require('cheerio');
|
|
const { JSDOM } = require('jsdom');
|
|
const moment = require('moment');
|
|
|
|
const { heightToCm } = require('../utils/convert');
|
|
const { matchTags } = require('../tags');
|
|
|
|
async function fetchPhotos(url) {
|
|
const res = await bhttp.get(url);
|
|
|
|
return res.body.toString();
|
|
}
|
|
|
|
function scrapePhotos(html) {
|
|
const $ = cheerio.load(html, { normalizeWhitespace: true });
|
|
|
|
const photos = $('.photo_gallery_thumbnail_wrapper .thumbs')
|
|
.map((photoIndex, photoElement) => {
|
|
const src = $(photoElement).attr('src');
|
|
|
|
if (src.match(/dl\d+/)) {
|
|
// thumbnail URLs containing dl02/ or dl03/ don't appear to have
|
|
// a full photo available, fall back to thumbnail
|
|
return src;
|
|
}
|
|
|
|
return src.replace('thumbs/', 'photos/');
|
|
})
|
|
.toArray();
|
|
|
|
return photos;
|
|
}
|
|
|
|
async function getPhotos(entryId, site, page = 1) {
|
|
const albumUrl = `https://www.julesjordan.com/trial/gallery.php?id=${entryId}&type=highres&page=${page}`;
|
|
|
|
const html = await fetchPhotos(albumUrl);
|
|
const $ = cheerio.load(html, { normalizeWhitespace: true });
|
|
|
|
const photos = scrapePhotos(html);
|
|
const pagesString = $('.page_totals').text().trim();
|
|
const pages = pagesString.length > 0 ? Number($('.page_totals').text().trim().match(/\d+$/)[0]) : null;
|
|
|
|
const otherPhotos = pages
|
|
? await Promise.map(Array.from({ length: pages - 1 }), async (val, index) => {
|
|
const pageUrl = `https://www.julesjordan.com/trial/gallery.php?id=${entryId}&type=highres&page=${index + 2}`;
|
|
const pageHtml = await fetchPhotos(pageUrl);
|
|
|
|
return scrapePhotos(pageHtml);
|
|
}, {
|
|
concurrency: 2,
|
|
})
|
|
: [];
|
|
|
|
return photos.concat(otherPhotos.flat());
|
|
}
|
|
|
|
function scrapeLatest(html, site) {
|
|
const $ = cheerio.load(html, { normalizeWhitespace: true });
|
|
const scenesElements = $('.update_details').toArray();
|
|
|
|
return scenesElements.map((element) => {
|
|
const photoElement = $(element).find('a img.thumbs');
|
|
const photoCount = Number(photoElement.attr('cnt'));
|
|
const [poster, ...photos] = Array.from({ length: photoCount }, (value, index) => photoElement.attr(`src${index}_1x`)).filter(photoUrl => photoUrl !== undefined);
|
|
|
|
const sceneLinkElement = $(element).children('a').eq(1);
|
|
const url = sceneLinkElement.attr('href');
|
|
const title = sceneLinkElement.text();
|
|
|
|
const entryId = $(element).attr('data-setid');
|
|
|
|
const date = moment
|
|
.utc($(element).find('.update_date').text(), 'MM/DD/YYYY')
|
|
.toDate();
|
|
|
|
const actors = $(element).find('.update_models a')
|
|
.map((actorIndex, actorElement) => $(actorElement).text())
|
|
.toArray();
|
|
|
|
return {
|
|
url,
|
|
entryId,
|
|
title,
|
|
actors,
|
|
date,
|
|
poster,
|
|
photos,
|
|
site,
|
|
};
|
|
});
|
|
}
|
|
|
|
function scrapeUpcoming(html, site) {
|
|
const $ = cheerio.load(html, { normalizeWhitespace: true });
|
|
const scenesElements = $('#coming_soon_carousel').find('.table').toArray();
|
|
|
|
return scenesElements.map((element) => {
|
|
const entryId = $(element).find('.upcoming_updates_thumb').attr('id').match(/\d+/)[0];
|
|
|
|
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();
|
|
|
|
const photoElement = $(element).find('a img.thumbs');
|
|
const poster = photoElement.attr('src');
|
|
|
|
const videoClass = $(element).find('.update_thumbnail div').attr('class');
|
|
const videoScript = $(element).find(`script:contains(${videoClass})`).html();
|
|
const trailer = videoScript.slice(videoScript.indexOf('https://'), videoScript.indexOf('.mp4') + 4);
|
|
|
|
return {
|
|
url: null,
|
|
entryId,
|
|
title,
|
|
date,
|
|
actors,
|
|
poster,
|
|
trailer: {
|
|
src: trailer,
|
|
},
|
|
rating: null,
|
|
site,
|
|
};
|
|
});
|
|
}
|
|
|
|
async function scrapeScene(html, url, site) {
|
|
const $ = cheerio.load(html, { normalizeWhitespace: true });
|
|
|
|
const title = $('.title_bar_hilite').text().trim();
|
|
const entryId = $('.suggest_tags a').attr('href').match(/\d+/)[0];
|
|
const date = moment
|
|
.utc($('.update_date').text(), 'MM/DD/YYYY')
|
|
.toDate();
|
|
|
|
const actors = $('.backgroundcolor_info > .update_models a')
|
|
.map((_actorIndex, actorElement) => $(actorElement).text())
|
|
.toArray();
|
|
|
|
const description = $('.update_description').text().trim();
|
|
|
|
const stars = Number($('.avg_rating').text().trim().replace(/[\s|Avg Rating:]/g, ''));
|
|
|
|
const infoLines = $('script:contains("useimage")')
|
|
.html()
|
|
.split('\n');
|
|
|
|
const poster = infoLines.find(line => line.match('useimage')).replace('useimage = "', '').slice(0, -2);
|
|
|
|
const trailerLine = infoLines.find(line => line.match('movie["Trailer_720"]'));
|
|
const trailer = trailerLine.slice(trailerLine.indexOf('path:"') + 6, trailerLine.indexOf('",movie'));
|
|
|
|
const photos = await getPhotos(entryId, site);
|
|
|
|
const rawTags = $('.update_tags a').map((tagIndex, tagElement) => $(tagElement).text()).toArray();
|
|
const tags = await matchTags(rawTags);
|
|
|
|
return {
|
|
url,
|
|
entryId,
|
|
title,
|
|
date,
|
|
actors,
|
|
description,
|
|
poster,
|
|
photos,
|
|
trailer: {
|
|
src: trailer,
|
|
quality: 720,
|
|
},
|
|
rating: {
|
|
stars,
|
|
},
|
|
tags,
|
|
site,
|
|
};
|
|
}
|
|
|
|
function scrapeProfile(html, url, actorName) {
|
|
const { document } = new JSDOM(html).window;
|
|
|
|
const bio = document.querySelector('.model_bio').textContent;
|
|
const avatarEl = document.querySelector('.model_bio_pic');
|
|
|
|
const profile = {
|
|
name: actorName,
|
|
};
|
|
|
|
const heightString = bio.match(/\d+ feet \d+ inches/);
|
|
const ageString = bio.match(/Age:\s*\d{2}/);
|
|
const measurementsString = bio.match(/\w+-\d+-\d+/);
|
|
|
|
if (heightString) profile.height = heightToCm(heightString[0]);
|
|
if (ageString) profile.age = Number(ageString[0].match(/\d{2}/)[0]);
|
|
if (measurementsString) [profile.bust, profile.waist, profile.hip] = measurementsString[0].split('-');
|
|
|
|
if (avatarEl) {
|
|
const src = avatarEl.innerHTML.slice(avatarEl.innerHTML.indexOf('src') + 5, avatarEl.innerHTML.indexOf('set.jpg') + 7).trim();
|
|
const src0 = avatarEl.innerHTML.slice(avatarEl.innerHTML.indexOf('src0') + 6, avatarEl.innerHTML.indexOf('set.jpg') + 7).trim();
|
|
const src1 = avatarEl.innerHTML.slice(avatarEl.innerHTML.indexOf('src0_1x') + 9, avatarEl.innerHTML.indexOf('1x.jpg') + 6).trim();
|
|
const src2 = avatarEl.innerHTML.slice(avatarEl.innerHTML.indexOf('src0_2x') + 9, avatarEl.innerHTML.indexOf('2x.jpg') + 6).trim();
|
|
const src3 = avatarEl.innerHTML.slice(avatarEl.innerHTML.indexOf('src0_3x') + 9, avatarEl.innerHTML.indexOf('3x.jpg') + 6).trim();
|
|
|
|
const avatar = src3 || src2 || src1 || src0 || src;
|
|
|
|
if (avatar) profile.avatar = avatar;
|
|
}
|
|
|
|
profile.releases = Array.from(document.querySelectorAll('.category_listing_block .update_details > a:first-child'), el => el.href);
|
|
|
|
return profile;
|
|
}
|
|
|
|
async function fetchLatest(site, page = 1) {
|
|
const res = await bhttp.get(`${site.url}/trial/categories/movies_${page}_d.html`);
|
|
|
|
return scrapeLatest(res.body.toString(), site);
|
|
}
|
|
|
|
async function fetchUpcoming(site) {
|
|
const res = await bhttp.get(`${site.url}/trial/index.php`);
|
|
|
|
return scrapeUpcoming(res.body.toString(), site);
|
|
}
|
|
|
|
async function fetchScene(url, site) {
|
|
const res = await bhttp.get(url);
|
|
|
|
return scrapeScene(res.body.toString(), url, site);
|
|
}
|
|
|
|
async function fetchProfile(actorName) {
|
|
const actorSlugA = actorName.toLowerCase().replace(/\s+/g, '-');
|
|
const actorSlugB = actorName.toLowerCase().replace(/\s+/g, '');
|
|
|
|
const urlA = `https://julesjordan.com/trial/models/${actorSlugA}.html`;
|
|
const urlB = `https://julesjordan.com/trial/models/${actorSlugB}.html`;
|
|
|
|
const resA = await bhttp.get(urlA);
|
|
|
|
if (resA.statusCode === 200) {
|
|
const profile = scrapeProfile(resA.body.toString(), urlA, actorName);
|
|
|
|
return profile;
|
|
}
|
|
|
|
const resB = await bhttp.get(urlB);
|
|
|
|
if (resB.statusCode === 200) {
|
|
const profile = scrapeProfile(resB.body.toString(), urlB, actorName);
|
|
|
|
return profile;
|
|
}
|
|
|
|
return null;
|
|
}
|
|
|
|
module.exports = {
|
|
fetchLatest,
|
|
fetchProfile,
|
|
fetchUpcoming,
|
|
fetchScene,
|
|
};
|