traxxx/src/scrapers/kink.js

'use strict';

const unprint = require('unprint');

const slugify = require('../utils/slugify');
const { stripQuery } = require('../utils/url');

function scrapeAll(scenes, entity) {
	return scenes.map(({ query }) => {
		const release = {};
		const networkUrl = entity.type === 'channel' ? entity.parent.url : entity.url;

		const href = query.url('a[href*="/shoot"]');

		release.url = `${networkUrl}${href}`;

		release.shootId = href.split('/').slice(-1)[0];
		release.entryId = release.shootId;

		release.title = query.content('.card-body a[href*="/shoot"]').trim();
		release.date = query.date('small > span', 'MMM D, YYYY');

		release.actors = query.all('a[href*="/model"]').map((actorEl) => ({
			name: unprint.query.content(actorEl),
			url: unprint.query.url(actorEl, null, { origin: networkUrl }),
		}));

		const poster = query.img('.ratio-thumbnail img');

		release.poster = [
			stripQuery(poster).replace('_thumb', '_full'),
			stripQuery(poster),
			poster,
		].filter(Boolean).map((src) => ({
			src,
			expectType: {
				PNG: 'image/png',
			},
		}));

		try {
			release.photos = JSON.parse(query.attribute('.ratio-thumbnail img', 'data-cycle'))
				.map((src) => Array.from(new Set([
					stripQuery(src).replace('_thumb', '_full'),
					stripQuery(src),
					src,
				])).filter(Boolean).map((source) => ({
					src: source,
					expectType: {
						PNG: 'image/png',
					},
				})));
		} catch (error) {
			// no photos
		}

		release.trailer = `https://cdnp.kink.com/imagedb/${release.entryId}/trailer/${release.entryId}_trailer_high.mp4`;

		release.channel = slugify(query.content('.shoot-thumbnail-footer a[href*="/channel"]'), '');
		release.rating = query.number('.thumb-up') / 10;

		return release;
	});
}

async function fetchLatest(channel, page = 1) {
	const url = `${channel.parent.url}/search?type=shoots&channelIds=${channel.parameters?.slug || channel.slug}&sort=published&page=${page}`;

	const res = await unprint.browserRequest(url, {
		selectAll: '.container .card',
	});

	if (res.status === 200) {
		// const items = unprint.initAll(html, '.container .card');

		const scenes = scrapeAll(res.context, channel);

		return scenes;
	}

	return res.status;
}

function scrapeScene({ query }, url, entity) {
	const release = { url };
	const data = query.json('div[data-setup]', { attribute: 'data-setup' });

	release.shootId = data?.id || new URL(url).pathname.split('/')[2];
	release.entryId = data?.id || release.shootId;

	release.title = data?.title || query.attribute('#shootPage #favoriteShootButton', 'data-title') || query.content('#shootPage h1');
	release.description = query.content('//*[contains(text(), \'Description\')]/following-sibling::span/p');

	release.date = query.date('.shoot-detail-legend', 'MMM D, YYYY');

	release.duration = data?.duration
		? data.duration / 1000
		: query.duration('#shootPage .clock');

	release.actors = query.elements('#shootPage h1 + span a[href*="/model"]').map((actorEl) => ({
		name: unprint.query.content(actorEl).replace(/,\s*/, ''),
		url: unprint.query.url(actorEl, null, { origin: entity.type === 'channel' ? entity.parent.url : entity.url }),
	}));

	release.director = query.content('.director-name')?.trim();

	const poster = data?.posterUrl || query.poster();

	release.poster = [
		stripQuery(poster),
		poster,
	].filter(Boolean).map((src) => ({
		src,
		expectType: {
			PNG: 'image/png',
		},
	}));

	release.photos = query.json('#galleryImagesContainer', { attribute: 'data-images' })?.map((src) => [
		src.fullPath,
		src.thumbFullPath,
	].filter(Boolean).map((source) => ({
		src: source,
		expectType: {
			PNG: 'image/png',
		},
	})));

	release.trailer = [
		...(data?.trailer?.sources?.map((source) => ({
			src: source.url,
			quality: source.resolution,
		})) || []),
		`https://cdnp.kink.com/imagedb/${release.entryId}/trailer/${release.entryId}_trailer_high.mp4`,
	];

	release.tags = query.contents('#shootPage a[href*="/tag"]').map((tag) => tag.trim());
	release.channel = data?.channelName?.name || slugify(query.url('.shoot-detail-legend a[href*="/channel"]')?.split('/').slice(-1)[0], '');

	release.qualities = data?.resolutions
		? Object.entries(data.resolutions).filter(([, enabled]) => enabled).map(([res]) => parseInt(res, 10))
		: null;

	return release;
}

async function fetchScene(url, channel) {
	const res = await unprint.browserRequest(url);

	if (res.status === 200) {
		const scene = scrapeScene(res.context, url, channel);

		return scene;
	}

	return res.status;
}

async function scrapeProfile({ query }, actorUrl) {
	const profile = { url: actorUrl };

	profile.entryId = actorUrl.match(/\/model\/(\d+)\//)?.[1] || query.attribute('h1 + button[data-id]', 'data-id');
	profile.description = query.content('.content-container #expand-text')?.trim();

	const tags = query.contents('.content-container a[href*="/tag"]').map((tag) => tag.toLowerCase().trim());

	if (tags.includes('brunette') || tags.includes('brunet')) profile.hairColor = 'brown';
	if (tags.includes('blonde') || tags.includes('blond')) profile.hairColor = 'blonde';
	if (tags.includes('black hair')) profile.hairColor = 'black';
	if (tags.includes('redhead')) profile.hairColor = 'red';

	if (tags.includes('natural boobs')) profile.naturalBoobs = true;
	if (tags.includes('fake boobs')) profile.naturalBoobs = false;

	if (tags.includes('white')) profile.ethnicity = 'white';
	if (tags.includes('latin')) profile.ethnicity = 'latin';
	if (tags.includes('Black')) profile.ethnicity = 'black';

	if (tags.includes('pierced nipples')) profile.hasPiercings = true;
	if (tags.includes('tattoo')) profile.hasTattoos = true;

	if (tags.includes('foreskin')) profile.hasForeskin = true;

	if ((tags.includes('big dick') || tags.includes('foreskin'))
		&& (tags.includes('fake boobs') || tags.includes('big tits'))) profile.gender = 'transsexual';

	[profile.avatar, ...profile.photos] = query.imgs('.kink-slider-images img:not([data-src*="missing"])', { attribute: 'data-src' });
	profile.social = query.urls('.content-container a[href*="twitter.com"], .content-container a[href*="x.com"]');

	return profile;
}

async function getActorUrl({ name: actorName, url }, networkUrl) {
	if (url) {
		return url;
	}

	// const searchRes = await tab.goto(`${networkUrl}/search?type=performers&q=${actorName}`);
	const searchApiRes = await unprint.browserRequest(`https://www.kink.com/api/v2/search/suggestions/performers?term=${actorName}`);

	if (searchApiRes.status === 200) {
		const data = searchApiRes.context.query.json('body pre');
		const actorId = data.find((actor) => actor.label === actorName)?.id;

		if (actorId) {
			const actorUrl = `${networkUrl}/model/${actorId}/${slugify(actorName)}`;

			return actorUrl;
		}
	}

	return null;
}

async function fetchProfile(actor, entity) {
	const networkUrl = entity.type === 'channel' ? entity.parent.url : entity.url;
	const actorUrl = await getActorUrl(actor, networkUrl);

	if (actorUrl) {
		const actorRes = await unprint.browserRequest(actorUrl);

		if (actorRes.status === 200) {
			return scrapeProfile(actorRes.context, actorUrl);
		}

		return actorRes.status;
	}

	return null;
}

module.exports = {
	// beforeNetwork,
	fetchLatest,
	fetchScene,
	fetchProfile,
};