traxxx/src/scrapers/naughtyamerica.js

138 lines
4.5 KiB
JavaScript
Executable File

'use strict';
const unprint = require('unprint');
const slugify = require('../utils/slugify');
// Naughty America network
function scrapeLatest(scenes, channel) {
return scenes.map(({ query }) => {
const release = {};
const url = query.url('a');
release.url = url;
release.entryId = query.attribute('a', 'data-scene-id') || (url && new URL(url).pathname.match(/-(\d+)$/)?.[1]) || null;
release.date = query.date('.entry-date, .scene-date', 'MMM D, YYYY');
release.duration = query.duration('.scene-runtime');
release.actors = query.all('.contain-actors a, .scene-actors a').map((actorEl) => ({
name: unprint.query.content(actorEl),
url: channel.parameters?.useActorUrl === false
? null // actor URL is scene link in native layout
: unprint.query.url(actorEl, null),
}));
release.poster = [
...(query.sourceSet('source[data-srcset*="scenes/"][type="image/jpeg"]', 'data-srcset') || []),
query.img('.main-scene-img', { attribute: 'srcset' }),
query.img('.scene-thumb'),
].filter(Boolean);
release.tags = query.contents('.flag-bg');
release.qualities = [
query.exists('//a[contains(@class, "label-four-k") and contains(text(), "4K")] | //span[contains(@class, "bug-4k")]') && 2160, // label-four-k is also used for non-4K tags
query.exists('//a[contains(@class, "label-hd") and contains(text(), "HD")]') && 720,
].filter(Boolean);
release.channel = slugify(query.content('.site-title'), '');
return release;
});
}
function scrapeScene({ query }, { url }) {
const release = {};
release.entryId = new URL(url).pathname.match(/-(\d+)$/)?.[1];
release.title = query.content('.breadcrumb-item.active') || query.content('.scene-title, .grey-title'); // main title has performer name instead of scene title in live scenes
release.description = query.text('.synopsis, .scene-description');
release.date = query.date('.entry-date, .released-date', ['MMM D, YYYY', 'MM/DD/YY']);
release.duration = query.duration('.duration');
release.actors = query.exists('.performer-list') || query.exists('.scene-info a[href*="/pornstar"].scene-title') // title links to performer in live scenes
? query.all('.performer-list a, .grey-performers a, .scene-info a[href*="/pornstar"].scene-title').map((actorEl) => ({
name: unprint.query.content(actorEl),
url: unprint.query.url(actorEl, null),
}))
: query.content('.grey-performers')?.split(',').map((actorName) => actorName.trim());
release.poster = [
...(query.sourceSet('.play-trailer source[data-srcset*="scenes/"][type="image/jpeg"]', 'data-srcset') || []),
...(query.sourceSet('.scenepage-video source[srcset*="scenes/"][type="image/jpeg"]', 'srcset') || []),
query.img('.play-trailer img[data-srcset*="scenes/"]', { attribute: 'data-srcset' }),
query.img('.scenepage-video .playcard'),
query.img('.scene-page .start-card'),
].filter(Boolean);
release.photos = query.els('.contain-scene-images.desktop-only .scene-image').map((imgEl) => [
unprint.query.url(imgEl, null),
unprint.query.img(imgEl, 'img', { attribute: 'srcset' }),
]);
const trailer = query.video('video source');
if (trailer) {
release.trailer = [
{
source: trailer.replace(/_\d+\.mp4/, '_1080.mp4'),
quality: 1080,
},
trailer,
];
}
release.channel = slugify(query.content('.site-title'), '');
release.tags = query.contents('.categories a, .category a');
release.qualities = [
query.exists('//a[contains(@class, "label-four-k") and contains(text(), "4K")]') && 2160, // label-four-k is also used for non-4K tags
query.exists('img.icon-1080') && 1080,
query.exists('//a[contains(@class, "label-hd") and contains(text(), "HD")]') && 720,
].filter(Boolean);
return release;
}
async function scrapeProfile({ query }) {
const profile = {};
profile.description = query.content('.bio_about_text');
profile.avatar = query.img('img.performer-pic, img.performer-img, img.peformer-img'); // sic peformer
return profile;
}
async function fetchLatest(site, page = 1) {
const res = await unprint.get(`${site.url}${site.parameters?.scenes || ''}?page=${page}`, { selectAll: '.site-list .scene-item, .panel-body' });
if (res.ok) {
return scrapeLatest(res.context, site);
}
return res.status;
}
async function fetchProfile({ slug }, { channel }) {
const res = await unprint.get(`${channel.url}/pornstar/${slug}`, { select: '.bio-info, .performer-details' });
if (res.ok) {
return scrapeProfile(res.context);
}
return res.status;
}
module.exports = {
fetchLatest,
fetchProfile,
scrapeScene: {
scraper: scrapeScene,
unprint: true,
},
};