traxxx/src/scrapers/wankzvr.js

'use strict';

const unprint = require('unprint');

const slugify = require('../utils/slugify');

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

		release.url = query.url('a', { origin: channel.origin });
		release.entryId = new URL(release.url).pathname.match(/(\d+)\/?$/)?.[1];

		release.title = query.content('.card__h');
		release.date = query.date('.card__date', 'D MMMM, YYYY', { match: null });

		release.actors = query.all('.card__links a').map((actorEl) => ({
			name: unprint.query.content(actorEl),
			url: unprint.query.url(actorEl, null, { origin: channel.url }),
		}));

		const poster = query.sourceSet('picture source[type="image/jpeg"]', 'data-srcset')
			|| query.sourceSet('picture source[type="image/jpeg"]', 'srcset')
			|| query.sourceSet('.video__cover', 'srcset');

		if (poster?.[0]) {
			release.poster = [
				poster[0].replace(/small|tiny/, 'large'),
				...poster,
			].map((src) => ({
				src,
				referer: url,
			}));

			const teaser = poster[0].replace(/\b(cover|hero|\d+)\/[a-z0-9_]+\.[a-z]+$/i, 'roll.webm'); // actually how site generates teaser URL

			release.teaser = {
				src: teaser,
				referer: url,
			};
		}

		release.channel = channel.slug; // avoid being assigned to WankzVR network

		return release;
	});
}

async function fetchLatest(channel, page) {
	const url = `${channel.url}/videos?o=d&p=${page}`;
	const res = await unprint.get(url, { selectAll: '.layout__content > .cards-list .card' }); // .cards-list is also used for hidden upcoming scenes

	if (res.ok) {
		return scrapeAll(res.context, channel, url);
	}

	return res.status;
}

async function getTrailerUrl(release, channel, cookies, referer) {
	const csrfToken = cookies.csrfst;

	if (!csrfToken) {
		return null;
	}

	const res = await unprint.post(`${channel.url}/ajax/player-config.json`, {
		item_id: release.entryId,
	}, {
		form: true,
		headers: {
			'X-Requested-With': 'XMLHttpRequest',
			'X-CSRF-Token': csrfToken,
		},
		cookies,
	});

	if (res.ok) {
		const trailers = res.data.streams.map((trailer) => ({
			src: trailer.url,
			quality: Number(trailer.id?.match(/\d+/)?.[0] || trailer?.name.match(/\d+/)?.[0]),
			vr: true,
			referer,
		}));

		const poster = unprint.prefixUrl(res.data.poster, res.data.thumbCDN);

		return {
			trailers,
			poster: poster && {
				src: poster,
				referer,
			},
		};
	}

	return null;
}

async function scrapeScene({ query }, { url, entity, include, cookies }) {
	const release = {};
	const data = query.json('script[type="application/ld+json"]');

	release.entryId = new URL(url).pathname.match(/(\d+)\/?$/)?.[1];

	release.title = query.content('.detail__title');
	release.description = query.content('.detail__txt');

	release.date = query.date('.detail__date', 'D MMMM, YYYY', { match: null });
	release.duration = query.number('.time') * 60;

	release.actors = (query.all('.detail__header-lg .detail__models a') || query.all('.detail__header-sm .detail__models a')).map((el) => ({
		name: unprint.query.content(el),
		url: unprint.query.url(el, null, { origin: entity.origin }),
	}));

	release.tags = query.contents('.tag-list .tag').concat(query.contents('.detail__specs-list .detail__specs-item'));

	release.photos = query.all('.photo-strip__slide').map((el) => ([
		unprint.query.img(el, null, 'data-src'),
		unprint.query.img(el, 'img'),
	].map((src) => ({ src, referer: url }))));

	if (data?.thumbnailUrl) {
		release.poster = [
			data.thumbnailUrl.replace(/small|tiny/, 'large'),
			data.thumbnailUrl,
		].map((src) => ({ src, referer: url }));
	}

	if (include.trailers || (!release.poster && include.poster)) {
		const { trailers, poster } = await getTrailerUrl(release, entity, cookies, url) || {};

		release.trailer = trailers;
		release.poster = poster;
	}

	return release;
}

async function fetchActorScenes({ query }, url, entity, page = 1, accScenes = []) {
	const scenes = scrapeAll(unprint.initAll(query.all('.cards-list .card')), entity);
	const hasNextPage = !query.exists('.pagenav__link.inactive');

	if (hasNextPage) {
		const { origin, pathname, searchParams } = new URL(url);
		searchParams.set('p', page + 1);

		const res = await unprint.get(`${origin}${pathname}?${searchParams}`);

		if (res.ok) {
			return fetchActorScenes(res.context, url, entity, page + 1, accScenes.concat(scenes));
		}
	}

	return accScenes.concat(scenes);
}

async function scrapeProfile({ query }, url, entity, options) {
	const profile = {};

	const bio = query.all('.person__meta__item').reduce((acc, el) => ({
		...acc,
		[slugify(unprint.query.content(el, '.person__meta__label'))]: unprint.query.text(el),
	}), {});

	profile.description = query.content('.person__content');

	profile.gender = entity.slug === 'tranzvr' ? 'transsexual' : 'female';
	profile.age = Number(bio.age) || null;

	profile.birthPlace = bio.birthplace;

	// height shown in imperial with cm between brackets when requested from North American IP, but only in cm for European IPs
	profile.height = unprint.extractNumber(bio.height, { match: /(\d+)cm/, matchIndex: 1 });
	profile.measurements = bio.measurements;

	profile.avatar = query.sourceSet('.person__avatar img').map((src) => ({
		src,
		referer: url,
	}));

	if (options.includeActorScenes) {
		profile.scenes = await fetchActorScenes({ query }, url, entity);
	}

	return profile;
}

async function fetchProfile(baseActor, { entity }, options) {
	const url = `${entity.url}/${baseActor.slug}`;
	const res = await unprint.get(url);

	if (res.ok) {
		return scrapeProfile(res.context, url, entity, options);
	}

	return res.status;
}

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