traxxx/src/scrapers/private.js

'use strict';

const unprint = require('unprint');

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

const headers = {
	'Accept-Language': 'en-US,en', // will translate to server language if not specified
};

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

		release.url = query.url('h3 a');
		release.entryId = new URL(release.url).pathname.split('/').at(-1);

		release.title = query.content('h3 a')
			|| query.attribute('a img', 'alt')?.split(/:(.*)/)[1]?.trim()
			|| query.attribute('a img', 'alt');

		release.date = query.date('.scene-date', ['MM/DD/YYYY', 'YYYY-MM-DD']);

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

		release.poster = query.img('a img');
		release.teaser = query.video('.mini_video_player source');

		return release;
	});
}

async function fetchLatest(site, page = 1) {
	const { hostname } = new URL(site.url);

	if (hostname.match('private.com')) {
		const res = await unprint.get(`${site.url}/${page}/`, { selectAll: '.content-wrapper .scene', headers });

		return scrapeAll(res.context, site);
	}

	const res = await unprint.get(`${site.url}/scenes/${page}/`, { selectAll: '.content-wrapper .scene', headers });

	return scrapeAll(res.context, site);
}

async function getPhotos(entryId, site) {
	const { hostname } = new URL(site.url);

	const res = await unprint.get(`https://${hostname}/gallery.php?type=highres&id=${entryId}`);

	if (res.ok) {
		return res.context.query.urls('#lightgallery a.item');
	}

	return [];
}

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

	release.entryId = new URL(url).pathname.split('/').at(-1);
	release.channel = slugify(query.content('.title-site'), '');

	release.title = query.content(['.title-zone h1', '.content-desc h1']);
	release.description = query.content('meta[itemprop="description"]', { attribute: 'content' }) || query.content('#description-section');

	release.date = query.date('meta[itemprop="uploadDate"]', 'YYYY-MM-DD', { attribute: 'content' });
	release.duration = query.duration('meta[itemprop="duration"]', { attribute: 'content' });

	release.actors = query.all('.title-zone .tag-models a, .scene-models-list a').map((actorEl) => ({
		name: unprint.query.content(actorEl),
		url: unprint.query.url(actorEl, null),
	}));

	release.tags = query.contents('.title-zone .tag-tags, .scene-tags a');

	release.poster = query.img('meta[itemprop="thumbnailUrl"]', { attribute: 'content' }) || query.img('#trailer_player_finished img');
	release.photos = await getPhotos(release.entryId, entity);
	release.trailer = query.url('meta[itemprop="contentURL"]', { attribute: 'content' }) || query.video('#trailer_player source');

	const movieUrl = query.url('.content-desc a[href*="movie/"]');

	if (movieUrl) {
		release.movie = {
			url: movieUrl,
			entryId: new URL(movieUrl).pathname.match(/movie\/(\d+)/)?.[1],
		};
	}

	return release;
}

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

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

	release.title = query.content('.dvds-wrapper h1');
	release.description = query.content('meta[itemprop="description"]', { attribute: 'content' }) || query.content('p.sinopsys');

	release.date = query.date('.date-duration [itemprop="datePublished"]', 'YYYY-MM-DD')
		|| query.date('.date-duration em + span', 'DD/MM/YYYY')
		|| query.date('meta[itemprop="uploadDate"]', 'YYYY-MM-DD', { attribute: 'content' }); // upload date does not match release date

	release.duration = query.number('.date-duration', { match: /(\d+) min/, matchIndex: 1 }) * 60 || null;

	release.actors = query.all('.dvd-performers a').map((actorEl) => ({
		name: unprint.query.content(actorEl),
		url: unprint.query.url(actorEl, null),
	}));

	release.poster = query.img('meta[itemprop="thumbnailUrl"]', { attribute: 'content' }) || query.img('#trailer_player_finished img');
	release.covers = query.imgs('.dvds-picture img');

	release.trailer = query.url('meta[itemprop="contentURL"]', { attribute: 'content' }) || query.video('#trailer_player source');

	release.scenes = scrapeAll(unprint.initAll(query.all('.scenes .card')));

	return release;
}

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

	const bioKeys = query.contents('.bio-pornstar .model-facts:last-child em');
	const bioValues = query.texts('.bio-pornstar .model-facts:last-child li');

	const bio = Object.fromEntries(bioKeys.map((key, index) => [slugify(key, '_'), bioValues[index]]).filter(([, value]) => value !== '-'));

	profile.description = query.content('.model-facts-long');

	profile.birthPlace = bio.birth_place;
	profile.nationality = bio.nationality;

	profile.measurements = bio.measurements;

	profile.height = Number(bio.height?.match(/(\d+)cm/)?.[1]) || null;
	profile.weight = Number(bio.weight?.match(/(\d+)kg/i)?.[1]) || null;

	profile.hairColor = bio.hair_color;
	profile.eye = bio.eye_color;

	if (bio.tattoos) {
		profile.hasTattoos = true;
		profile.tattoos = bio.tattoos;
	}

	if (bio.tattoos) {
		profile.hasTattoos = true;
		profile.tattoos = bio.tattoos;
	}

	if (bio.piercings) {
		profile.hasPiercings = true;
		profile.piercings = bio.piercings;
	}

	profile.avatar = query.img('meta[property="og:image"]', { attribute: 'content' }); // other images are not suitable as avatar

	// profile.releases = scrapeAll(unprint.initAll(query.all('.scenes .card'))); // works for good measure, but no longer in use

	return profile;
}

async function searchProfile(actorName) {
	const actorSearchSlug = slugify(actorName, '+');
	const url = `https://www.private.com/search.php?query=${actorSearchSlug}`;

	const res = await unprint.get(url, { headers });

	return res.context.query.url(`.model a[title="${actorName}"]`);
}

async function fetchProfile({ name: actorName, url }) {
	const actorUrl = url || await searchProfile(actorName);

	if (!actorUrl) {
		return null;
	}

	const res = await unprint.get(actorUrl, { headers });

	if (res.ok) {
		return scrapeProfile(res.context);
	}

	return res.status;
}

module.exports = {
	fetchLatest,
	fetchProfile,
	scrapeScene: {
		scraper: scrapeScene,
		headers,
	},
	scrapeMovie: {
		scraper: scrapeMovie,
		headers,
	},
};