traxxx/src/scrapers/brazzers.js

'use strict';

/* eslint-disable newline-per-chained-call */
const qu = require('../utils/qu');
const slugify = require('../utils/slugify');
const { heightToCm, lbsToKg } = require('../utils/convert');

function scrapeAll(items, channel, upcoming) {
	return items.reduce((acc, { query }) => {
		const isUpcoming = query.exists('.icon-upcoming.active');

		if ((upcoming && !isUpcoming) || (!upcoming && isUpcoming)) {
			return acc;
		}

		const release = {};
		const pathname = query.url('a');

		release.url = `https://www.brazzers.com${pathname}`;
		release.entryId = pathname.match(/(\/view\/id\/|\/episode\/)(\d+)/)[2];

		release.title = query.q('a', 'title');
		release.date = query.date('time', 'MMMM DD, YYYY');

		release.actors = query.all('.model-names a', 'title');

		release.likes = query.number('.label-rating .like-amount');
		release.dislikes = query.number('.label-rating .dislike-amount');

		release.poster = query.img('.card-main-img');
		release.photos = query.imgs('.card-overlay .image-under');

		release.channel = slugify(query.q('.collection', 'title'), '');

		return acc.concat(release);
	}, []);
}

function getVideoData(html) {
	try {
		const videoScriptStart = html.indexOf('window.videoUiOptions');
		const videoScript = html.slice(videoScriptStart, html.indexOf('};', videoScriptStart));
		const videoString = videoScript.slice(videoScript.indexOf('{"stream_info"'), videoScript.lastIndexOf('},') + 1);

		return JSON.parse(videoString);
	} catch (error) {
		return null;
	}
}

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

	release.entryId = new URL(url).pathname.match(/(\/view\/id\/|\/episode\/)(\d+)/)[2];

	release.title = query.q('.scene-title[itemprop="name"]', true);
	release.description = query.text('#scene-description p[itemprop="description"]');

	release.date = query.date('.more-scene-info .scene-date', 'MMMM DD, YYYY');
	release.duration = Number(query.q('#trailer-player-container', 'data-duration')) // more accurate
		|| Number(query.q('.scene-length[itemprop="duration"]', 'content').slice(1, -1) * 60);

	// actor cards have avatar, but truncated name
	const actorImagesByActorId = query.imgs('.featured-model .card-image img').reduce((acc, img) => ({
		...acc,
		[img.match(/\/models\/(\d+)/)[1]]: [
			img.replace('medium', 'large'),
			img,
		],
	}), {});

	release.actors = query.all('.related-model a').map(actorEl => ({
		name: query.q(actorEl, null, 'title'),
		avatar: actorImagesByActorId[query.url(actorEl, null).match(/\/view\/id\/(\d+)/)?.[1]],
	}));

	release.likes = query.number('.label-rating .like');
	release.dislikes = query.number('.label-rating .dislike');

	const tags = query.all('.tag-card-container a', true);
	const categories = query.all('.timeline a[href*="/categories"]', 'title');

	release.tags = tags.concat(categories);
	release.channel = slugify(query.q('.scene-site .label-text', true) || query.q('.niche-site-logo', 'title'), '');

	const videoData = getVideoData(html);
	const poster = videoData?.poster || query.meta('itemprop="thumbnailUrl"') || query.q('#trailer-player-container', 'data-player-img');

	release.poster = qu.prefixUrl(poster);
	release.photos = query.urls('.carousel-thumb a');

	if (videoData) {
		release.trailer = Object.entries(videoData.stream_info.http.paths).map(([quality, path]) => ({
			src: qu.prefixUrl(path),
			quality: Number(quality.match(/\d{3,}/)[0]),
		}));
	}

	return release;
}

async function fetchActorReleases({ query }, accReleases = []) {
	const releases = scrapeAll(qu.initAll(query.all('.release-card.scene')));
	const next = query.url('.pagination .next a');

	if (next) {
		const url = `https://www.brazzers.com${next}`;
		const res = await qu.get(url);

		if (res.ok) {
			return fetchActorReleases(res.item, accReleases.concat(releases));
		}
	}

	return accReleases.concat(releases);
}

async function scrapeProfile({ query }, url, actorName, include) {
	const bioKeys = query.all('.profile-spec-list label', true).map(key => key.replace(/\n+|\s{2,}/g, '').trim());
	const bioValues = query.all('.profile-spec-list var', true).map(value => value.replace(/\n+|\s{2,}/g, '').trim());

	const bio = bioKeys.reduce((acc, key, index) => ({ ...acc, [key]: bioValues[index] }), {});

	const profile = {
		name: actorName,
	};

	profile.description = query.q('.model-profile-specs p', true);

	if (bio.Ethnicity) profile.ethnicity = bio.Ethnicity;
	if (bio.Measurements && bio.Measurements.match(/\d+[A-Z]+-\d+-\d+/)) [profile.bust, profile.waist, profile.hip] = bio.Measurements.split('-');
	if (bio['Date of Birth'] && bio['Date of Birth'] !== 'Unknown') profile.birthdate = qu.extractDate(bio['Date of Birth'], 'MMMM DD, YYYY');
	if (bio['Birth Location']) profile.birthPlace = bio['Birth Location'];
	if (bio['Pussy Type']) profile.pussy = bio['Pussy Type'].split(',').slice(-1)[0].toLowerCase();

	if (bio.Height) profile.height = heightToCm(bio.Height);
	if (bio.Weight) profile.weight = lbsToKg(bio.Weight.match(/\d+/)[0]);
	if (bio['Hair Color']) profile.hair = bio['Hair Color'].toLowerCase();

	if (bio['Tits Type'] && bio['Tits Type'].match('Natural')) profile.naturalBoobs = true;
	if (bio['Tits Type'] && bio['Tits Type'].match('Enhanced')) profile.naturalBoobs = false;

	if (bio['Body Art'] && bio['Body Art'].match('Tattoo')) profile.hasTattoos = true;
	if (bio['Body Art'] && bio['Body Art'].match('Piercing')) profile.hasPiercings = true;

	const avatarEl = query.q('.big-pic-model-container img');
	if (avatarEl) profile.avatar = `https:${avatarEl.src}`;

	if (include.releases) {
		profile.releases = await fetchActorReleases({ query });
	}

	return profile;
}

async function fetchLatest(channel, page = 1) {
	const res = await qu.getAll(`${channel.url}/page/${page}/`, '.release-card.scene');

	if (res.ok) {
		return scrapeAll(res.items, channel, false);
	}

	return res.status;
}

async function fetchUpcoming(channel) {
	const res = await qu.getAll(`${channel.url}/page/1`, '.release-card.scene');

	if (res.ok) {
		return scrapeAll(res.items, channel, true);
	}

	return res.status;
}

async function fetchScene(url, site) {
	const res = await qu.get(url);

	if (res.ok) {
		return scrapeScene(res.item, url, site);
	}

	return res.status;
}

async function fetchProfile(actorName, context, include) {
	const searchRes = await qu.get('https://brazzers.com/pornstars-search/', `a[title="${actorName}" i]`, {
		Cookie: `textSearch=${encodeURIComponent(actorName)};`,
	});

	const actorLink = searchRes.ok && searchRes.item.qu.url(null);

	if (actorLink) {
		const url = `https://brazzers.com${actorLink}`;
		const res = await qu.get(url);

		if (res.ok) {
			return scrapeProfile(res.item, url, actorName, include);
		}
	}

	return null;
}

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