traxxx/src/scrapers/inthecrack.js

'use strict';

const moment = require('moment');

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

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

		release.url = query.url('a', 'href', { origin: channel.url });
		release.entryId = new URL(release.url).pathname.match(/\/Collection\/(\d+)/)[1];

		release.shootId = query.cnt('a span:nth-of-type(1)').match(/^\d+/)?.[0];
		release.date = query.date('a span:nth-of-type(2)', 'YYYY-MM-DD');

		release.actors = (query.q('a img', 'alt') || query.cnt('a span:nth-of-type(1)'))?.match(/[a-zA-Z]+(\s[A-Za-z]+)*/g);

		release.poster = release.shootId
			? `https://inthecrack.com/assets/images/posters/collections/${release.shootId}.jpg`
			: query.img('a img', 'src', { origin: channel.url });

		return release;
	});
}

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

	release.entryId = new URL(url).pathname.match(/\/Collection\/(\d+)/)[1];
	release.shootId = query.cnt('h2 span').match(/^\d+/)?.[0];

	release.actors = query.cnt('h2 span')?.match(/[a-zA-Z]+(\s[A-Za-z]+)*/g);

	release.description = query.cnt('p#CollectionDescription');
	release.productionLocation = query.cnt('.modelCollectionHeader p')?.match(/Shoot Location: (.*)/)?.[1];

	release.poster = qu.prefixUrl(html.match(/background-image: url\('(.*)'\)/)?.[1], channel.url);

	release.chapters = query.all('.ClipOuter').map((el) => {
		const chapter = {};

		chapter.title = query.text(el, 'h4');
		chapter.description = query.cnt(el, 'p');
		chapter.duration = query.dur(el, '.InlineDuration');

		const posterStyle = query.style(el, '.clipImage', 'background-image');
		const poster = qu.prefixUrl(posterStyle.match(/url\((.*)\)/)?.[1], channel.url);

		if (poster) {
			const { origin, pathname } = new URL(poster);

			chapter.poster = [
				`${origin}${pathname}`, // full size
				poster,
			];
		}

		if (query.exists(el, '.ThreeDInfo')) {
			chapter.tags = ['3d'];
		}

		return chapter;
	});

	return release;
}

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

	profile.description = query.cnt('.bio-text');
	profile.birthPlace = query.cnt('.birth-place span');

	profile.avatar = query.img('.actor-photo img');

	if (include.releases) {
		return scrapeAll(qu.initAll(el, '.scene'));
	}

	console.log(profile);
	return profile;
}

async function fetchLatest(channel, page = 1) {
	const year = moment().subtract(page - 1, ' year').year();

	const url = `${channel.url}/Collections/Date/${year}`;
	const res = await qu.getAll(url, '.collectionGridLayout li');

	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 }, entity, include) {
	const url = `${entity.url}/actors/${slugify(actorName, '_')}`;
	const res = await qu.get(url);

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

	return res.status;
}

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