traxxx/src/scrapers/nubiles.js

'use strict';

const qu = require('../utils/qu');
const slugify = require('../utils/slugify');
const { heightToCm } = require('../utils/convert');

const slugUrlMap = {
	nubiles: 'https://www.nubiles.net',
	nubilesporn: 'https://www.nubiles-porn.com',
};

async function getPhotos(albumUrl) {
	const res = await qu.getAll(albumUrl, '.photo-thumb');

	return res.ok
		? res.items.map(({ query }) => query.q('source').srcset)
		: [];
}

function scrapeAll(scenes, site, origin) {
	return scenes.map(({ query }) => {
		const release = {};

		release.title = query.q('.title a', true);

		const url = query.url('.title a').split('?')[0];
		const channelUrl = query.url('.site-link');

		if (/^http/.test(url)) {
			const { pathname } = new URL(url);
			// release.entryId = pathname.split('/')[3];

			if (channelUrl) release.url = `${channelUrl}${pathname}`;
			else release.url = url;
		} else if (!/\/join/.test(url)) {
			// release.entryId = url.split('/')[3];

			if (channelUrl) release.url = `${channelUrl}${url}`;
			else if (site?.url) release.url = `${site.url}${url}`;
			else if (origin) release.url = `${origin}${url}`;
		} else {
			// release.entryId = qu.q('a img', 'tube_tour_thumb_id');
		}

		release.date = query.date('.date', 'MMM D, YYYY');
		release.actors = query.all('.models a.model', true);

		// no reliable entry ID between upcoming and released scenes
		release.entryId = `${qu.formatDate(release.date, 'YYYY-MM-DD')}-${slugify(release.title)}`;

		const poster = query.sourceSet('img', 'data-srcset')?.[0];

		release.poster = poster && [
			poster.replace('_640', '_1280'),
			poster,
		];

		release.stars = query.number('.rating');
		release.likes = query.number('.likes');

		return release;
	});
}

async function scrapeScene({ query }, url, site) {
	const release = {};

	const { origin, pathname } = new URL(url);
	release.url = `${origin}${pathname}`;

	release.entryId = new URL(url).pathname.split('/')[3];
	release.title = query.q('.content-pane-title h2', true);
	release.description = query.q('.content-pane-column div', true);

	release.date = query.date('.date', 'MMM D, YYYY');

	release.actors = query.all('.content-pane-performers .model', true);
	release.tags = query.all('.categories a', true);

	release.poster = query.poster() || query.img('.fake-video-player img');
	release.trailer = query.all('source').map((source) => ({
		src: source.src,
		quality: Number(source.getAttribute('res')),
	}));

	release.stars = Number(query.q('.score', true));
	release.likes = Number(query.q('#likecount', true));

	const albumLink = query.url('.content-pane-related-links a[href*="gallery"]');
	if (albumLink) release.photos = await getPhotos(`${site.url}${albumLink}`);

	return release;
}

function scrapeProfile({ query }, _actorName, origin) {
	const profile = {};

	const keys = query.all('.model-profile h5', true);
	const values = query.all('.model-profile h5 + p', true);

	const bio = keys.reduce((acc, key, index) => ({ ...acc, [slugify(key, '_')]: values[index] }), {});

	profile.age = Number(bio.age);
	profile.description = query.q('.model-bio', true);

	profile.residencePlace = bio.location;

	profile.height = heightToCm(bio.height);
	[profile.bust, profile.waist, profile.hip] = bio.figure.split('-').map((v) => Number(v) || v);

	profile.avatar = query.img('.model-profile img');

	const releases = query.all('.content-grid-item').filter((el) => /video\//.test(query.url(el, '.img-wrapper a'))); // filter out photos
	profile.releases = scrapeAll(query.initAll(releases), null, origin);

	return profile;
}

async function fetchLatest(site, page = 1) {
	const url = `${site.url}/video/gallery/${(page - 1) * 12}`;
	const res = await qu.getAll(url, '.content-grid-item');

	return res.ok ? scrapeAll(res.items, site) : res.status;
}

async function fetchUpcoming(site) {
	if (site.parameters?.upcoming) {
		const url = `${site.url}/video/upcoming`;
		const res = await qu.getAll(url, '.content-grid-item');

		return res.ok ? scrapeAll(res.items, site) : res.status;
	}

	return [];
}

async function fetchProfile({ name: actorName }, { site }) {
	const firstLetter = actorName.charAt(0).toLowerCase();
	const origin = slugUrlMap[site.slug] || site.url;

	const url = `${origin}/model/alpha/${firstLetter}`;
	const resModels = await qu.get(url);

	if (!resModels.ok) return resModels.status;

	const modelPath = resModels.item.qu.all('.content-grid-item a.title').find((el) => slugify(el.textContent) === slugify(actorName));

	if (modelPath) {
		const modelUrl = `${origin}${modelPath}`;
		const resModel = await qu.get(modelUrl);

		return resModel.ok ? scrapeProfile(resModel.item, actorName, origin) : resModel.status;
	}

	return null;
}

module.exports = {
	fetchLatest,
	fetchUpcoming,
	fetchProfile,
	scrapeScene,
};