traxxx/src/scrapers/pierrewoodman.js

'use strict';

const unprint = require('unprint');

const slugify = require('../utils/slugify');
const capitalize = require('../utils/capitalize');
const tryUrls = require('../utils/try-urls');

function scrapeAll(scenes, channel, discard = true) {
	return scenes.reduce((acc, { query, element }) => {
		const release = {};

		release.url = unprint.query.url(element, null, { attribute: 'href', origin: channel.url });

		const { hostname, pathname } = new URL(release.url);

		release.entryId = pathname.match(/_(\d+)/)?.[1];
		release.channel = hostname.match(/(\w+)\.com/)?.[1];

		if (discard && release.channel !== channel.slug) {
			acc.unextracted.concat(release);

			return acc;
		}

		release.title = query.content('.title, .informations h3');
		release.duration = query.duration('.duration, .timer, .infos');

		release.actors = query.content('.sub')?.split(/,\s*/);

		release.poster = query.img('.thumb, picture img');

		acc.scenes = acc.scenes.concat(release);

		return acc;
	}, {
		scenes: [],
		unextracted: [],
	});
}

async function fetchLatest(channel, page) {
	const res = await unprint.get(channel.parameters?.latest
		? `${channel.parameters.latest}?page=${page}`
		: `${channel.url}/videos?page=${page}`, { selectAll: '.items .scene' });

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

	return res.status;
}

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

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

	const title = query.content('.page_title h1, h2');
	const wunfTitle = title.match(/wunf \d+/i)?.[0];

	release.title = wunfTitle ? wunfTitle.toUpperCase() : title;
	release.description = query.content('.info_container .description');

	release.date = query.date('.info_container .info_line:nth-child(1)', 'YYYY-MM-DD')
		|| query.date('.description', 'D MMMM YYYY', { match: /\d{1,2} \w+ \d{4}/ });

	release.actors = query.all('.girl_item, .starring .item').map((actorEl) => {
		const avatar = unprint.query.img(actorEl);

		return {
			name: capitalize(unprint.query.content(actorEl, '.name, .informations p'), { uncapitalize: true }),
			url: unprint.query.url(actorEl, null, { origin: entity.url }),
			avatar,
		};
	});

	release.duration = query.duration('.infos .description');

	if (!release.duration) {
		const duration = query.content('.info_container .info_line:nth-child(2)');

		release.duration = (duration.match(/(\d+) hour/)?.[1] || 0) * 3600
			+ (duration.match(/(\d+) minutes/)?.[1] || 0) * 60;
	}

	release.tags = query.contents('.tags a:not(.more_tag)');
	release.poster = html.match(/image: "(.*?)"/)?.[1];

	release.trailer = html.match(/url: "(.*mp4.*)"/g)?.map((src) => ({
		src: src.match(/"(.*)"/)?.[1],
		quality: Number(src.match(/[-/](\d+)p/)?.[1]),
	}));

	if (query.exists('.download-icon-4k')) {
		release.qualities = [2160];
	}

	return release;
}

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

	profile.avatar = query.img('.actor img, .avatar img');
	profile.nationality = query.content(['.nationality, .nationnality', '//strong[contains(text(), "Nationnality")]'])?.replace(/nationn?ality\s*:/i, '').trim(); // sic

	profile.scenes = scrapeAll(unprint.initAll(query.all('.videos .item, .list .scene.item')), entity, false);

	return profile;
}

async function getActorUrl(actor) {
	if (actor.url) {
		return [actor.url];
	}

	const res = await unprint.get('https://www.woodmancastingx.com');

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

	const searchUrl = unprint.prefixUrl(res.context.html.match(/"(.*searchCompletion\.js)"/)?.[1], 'https://www.woodmancastingx.com');

	if (!searchUrl) {
		return null;
	}

	const searchRes = await unprint.get(searchUrl);

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

	const [actorId] = searchRes.data.actors.find(([_actorId, actorName]) => slugify(actorName) === actor.slug) || [];

	if (!actorId) {
		return null;
	}

	// WUNF has the same avatars at higher quality, but not all performers
	return [
		`https://www.wakeupnfuck.com/actor/${actor.slug}_${actorId}`,
		`https://www.woodmancastingx.com/girl/${actor.slug}_${actorId}`,
	];
}

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

	if (!Array.isArray(actorUrls)) {
		return actorUrls;
	}

	const { res, url } = await tryUrls(actorUrls);

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

	return res.status;
}

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