traxxx/src/scrapers/dorcel.js

'use strict';

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

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

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

		release.title = query.cnt('.title a');

		release.date = query.date('.date', 'MMMM DD, YYYY', /\w+ \d{1,2}, \d{4}/);
		release.duration = query.number('.length') * 60;

		release.actors = query.all('.actors a').map(actorEl => ({
			name: query.cnt(actorEl),
			url: query.url(actorEl, null),
		}));

		release.poster = query.img('.poster noscript img');
		release.stars = query.count('.rating .star1');

		release.tags = [query.cnt('.collection a')];

		return release;
	});
}

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

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

	release.title = query.cnt('.infos .title h1');
	release.description = query.cnt('#description p:nth-child(2)');

	release.date = query.date('.infos .date', 'MMMM DD, YYYY', /\w+ \d{1,2}, \d{4}/);
	release.duration = query.number('.infos .length') * 60;

	release.actors = query.all('.infos .actors a').map(actorEl => ({
		name: query.cnt(actorEl),
		url: query.url(actorEl, null),
	}));

	release.poster = query.img('.poster noscript img');
	release.stars = query.count('.infos .rating .star1');

	if (query.exists('.movie')) {
		release.movie = {
			name: query.cnt('.movie a'),
			url: query.url('.movie a'),
		};

		release.movie.entryId = new URL(release.movie.url).pathname.split('/').slice(-1)[0];
	}

	return release;
}

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

	profile.birthdate = qu.parseDate(query.text('.birthdate'), 'MMMM DD, YYYY');
	profile.nationality = query.text('.nationality');
	profile.hairColor = query.text('.hair');

	profile.description = query.cnt('.bio_results p');

	if (avatar) {
		profile.avatar = [
			avatar.replace('_crop', ''),
			avatar,
		];
	}

	// TODO: add pagination
	profile.releases = scrapeAll(qu.initAll(el, '.scene'));

	return profile;
}

// TODO: add movies

async function fetchLatest(channel, page = 1) {
	const url = `${channel.url}/en/news-videos-x-marc-dorcel-ajax?page=${page}&sorting=publish_date`;
	const res = await qu.getAll(url, '.scene', {
		'X-Requested-With': 'XMLHttpRequest',
	});

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

	return res.status;
}

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

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

	return res.status;
}

async function fetchProfile({ name: actorName, url: actorUrl }, entity, include) {
	const searchRes = await qu.getAll(`${entity.url}/en/pornstars?search=${slugify(actorName, '+')}`, '.actor');

	const actorItem = searchRes.ok && searchRes.items.find(actor => slugify(actor.query.cnt('h2')) === slugify(actorName));
	const actorItemUrl = actorItem?.query.url();
	const actorItemAvatar = actorItem?.query.img();

	const url = actorUrl || actorItemUrl || `${entity.url}/en/pornstar/${slugify(actorName, '-')}`;
	const res = await qu.get(url);

	if (res.ok) {
		return scrapeProfile(res.item, actorItemAvatar, entity, include);
	}

	return res.status;
}

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