traxxx/src/scrapers/karups.js

'use strict';

const unprint = require('unprint');

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

const channelSlugs = {
	kpc: 'karupsprivatecollection',
	kha: 'karupshometownamateurs',
	kow: 'karupsolderwomen',
};

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

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

		release.title = query.content('.title');
		release.date = query.date('.date', ['MMM Do, YYYY', 'DD MMM YYYY'], { match: null });

		release.channel = channelSlugs[query.content('.site')];

		const poster = query.img('.thumb img');

		if (poster && /\.\w{3,4}$/.test(poster)) { // missing poster points to https://media.karups.com/thumbs_pg/
			release.poster = Array.from(new Set([
				poster.replace('.jpg', '-feat_lg.jpg'),
				poster,
			]));
		}

		return release;
	});
}

async function fetchLatest(channel, page) {
	const res = await unprint.get(new URL(`./videos/page${page}.html`, channel.url).href, { // some sites require a trailing slash, join paths properly; don't use origin in case channel path is used
		selectAll: '.listing-videos .item',
		cookies: {
			warningHidden: 'hide',
		},
	});

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

	return res.status;
}

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

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

	release.title = query.content('.title');
	release.description = query.content('.content-information-description p');

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

	release.actors = query.all('.models .content a').map((modelEl) => ({
		name: unprint.query.content(modelEl),
		url: unprint.query.url(modelEl, null),
	}));

	// videos and photos seem to be removed, query educated guess just in case
	const poster = query.poster('.video-player video') || query.img('.video-poster img');

	if (poster && /\.\w{3,4}$/.test(poster)) { // missing poster points to https://media.karups.com/thumbs_pg/
		release.poster = Array.from(new Set([
			poster,
			poster.replace('-feat_lg', ''),
		]));
	}

	release.photos = query.imgs('.video-thumbs img').slice(1);
	release.trailer = query.video('.video-player source');

	return release;
}

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

	const bio = Object.fromEntries(query.all('.model-table .item').map((bioEl) => [
		slugify(unprint.query.content(bioEl, '.label'), '_'),
		unprint.query.content(bioEl, '.value'),
	]));

	profile.age = unprint.extractNumber(bio.date_of_birth); // seemingly only used on Boyfun and always age
	profile.height = unprint.extractNumber(bio.height);
	profile.weight = unprint.extractNumber(bio.height);

	profile.penisLength = unprint.extractNumber(bio.dick_size);

	if (bio.cut_uncut?.toLowerCase() === 'cut') profile.isCircumcised = true;
	if (bio.cut_uncut?.toLowerCase() === 'uncut') profile.isCircumcised = false;

	profile.avatar = query.img('.model-thumb img[src*=".jpg"]');
	profile.scenes = scrapeAll(unprint.initAll(query.all('.listing-videos .item')), entity);

	return profile;
}

async function getActorUrl(actor, entity) {
	if (actor.url) {
		return actor.url;
	}

	const res = await unprint.get(`${entity.origin}/models/search/${actor.slug}/`, {
		selectAll: '.listing-models .item',
		cookies: {
			warningHidden: 'hide',
		},
	});

	if (!res.ok) {
		return res.status;
	}

	const actorUrl = res.context.find((item) => slugify(item.query.content('.title')) === actor.slug)?.query.url('a');

	return actorUrl;
}

async function fetchProfile(actor, entity) {
	const actorUrl = await getActorUrl(actor, entity);

	if (!actorUrl) {
		return null;
	}

	const actorRes = await unprint.get(actorUrl, {
		cookies: {
			warningHidden: 'hide',
		},
	});

	if (actorRes.ok) {
		return scrapeProfile(actorRes.context, entity);
	}

	return actorRes.status;
}

module.exports = {
	fetchLatest,
	fetchProfile,
	scrapeScene: {
		scraper: scrapeScene,
		cookies: {
			warningHidden: 'hide',
		},
	},
};