traxxx/src/scrapers/fcuk.js

'use strict';

const qu = require('../utils/qu');
const slugify = require('../utils/slugify');
const { feetInchesToCm } = require('../utils/convert');

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

		release.url = query.url('a.more:not([href*="/join.php"])', 'href', { origin: channel.url });

		if (release.url) {
			release.entryId = new URL(release.url).pathname.match(/\/scene\/(\d+)\/(\d+)/).slice(1, 3).join('-');
		} else {
			release.entryId = query.img('.bigthumb').match(/\/scenes\/(\w+)/)?.[1];
		}

		release.title = query.q('h5 strong', true)?.match(/. - (.+)$/)[1] || query.text('.videos h3');
		release.description = query.text('p');
		release.date = query.date('h5 strong, .videos h3', 'MMM. DD, YYYY', /\w+. \d{2}, \d{4}/);

		// remove common patterns so only the name is left
		const curatedTitle = release.title.replace(/\b(part \d|\banal|bts)\b/gi, '').trim();

		if (!/\band\b/.test(curatedTitle) && new RegExp(curatedTitle).test(release.description)) {
			// scene title is probably the actor name
			release.actors = [release.title];
		}

		release.poster = query.img('.bigthumb', null, { origin: channel.url });
		release.photos = query.imgs('.smallthumb', null, { origin: channel.url });

		release.tags = query.all('a[href*="/keywords"]', true);

		return release;
	});
}

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

		release.url = query.url('.updateInfo h5 a:not([href*="content/"]):not([href*="#coming"])');
		release.entryId = query.url('.updateThumb img', 'alt');

		release.title = query.q('.updateInfo h5 a', true);

		release.actors = query.all('.tour_update_models a', true);
		release.date = query.date('.availdate, .updateInfo p span:nth-child(2)', 'MM/DD/YYYY');

		release.poster = query.img('.updateThumb img');

		const trailer = query.q('.updateInfo h5 a', 'onclick')?.match(/'(.+)'/)?.[1];

		if (trailer) {
			release.trailer = {
				src: `${channel.url}${trailer}`,
			};
		}

		return release;
	});
}

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

	release.entryId = new URL(url).pathname.match(/\/scene\/(\d+)\/(\d+)/).slice(1, 3).join('-');

	release.title = query.text('h4 strong, .videos h3');
	release.description = query.q('#about p, .videos p', true);

	const actors = query.urls('a[href*="/girl/"]').map(actorUrl => actorUrl.match(/video-([\w\s]+)/)?.[1]).filter(Boolean);

	if (actors.length > 0) {
		release.actors = actors;
	} else {
		// release.actors = [query.q('.previewmed h5 strong', true)?.match(/^([\w\s]+),/)?.[0] || query.q('.videos h3', true)].filter(Boolean);
		release.actors = [release.title];
	}

	release.tags = query.all('.info a[href*="/keywords"], .buttons a[href*="/keywords"]', true);

	release.poster = query.img('#info .main-preview, .bigthumb', null, { origin: channel.url });
	release.photos = [query.img('.previewmed img', null, { origin: channel.url })].concat(query.imgs('.hd-clip img, .smallthumb', null, { origin: channel.url })).filter(photo => photo);

	return release;
}

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

	release.title = query.q('.updatesBlock h2', true);
	release.poster = query.meta('property="og:image"');
	release.entryId = release.poster.match(/\/content\/(.*)\//)?.[1];

	const trailer = html.match(/src="(.+\.mp4)"/)?.[1];

	if (trailer) {
		release.trailer = {
			src: `${channel.url}${trailer}`,
		};
	}

	return release;
}

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

	const bio = query.cnts('.info p').reduce((acc, info) => {
		const [key, value] = info.match(/(\w+):\s*(.*)/).slice(1);

		return { ...acc, [slugify(key, '_')]: value };
	}, {});

	profile.age = Number(bio.age);
	profile.height = feetInchesToCm(bio.height);
	profile.eyes = bio.eyes || bio.eyecolor;

	if (bio.figure || bio.measurements) {
		const [bust, cup, waist, hip] = (bio.figure || bio.measurements)?.match(/(\d+)(\w+)-(\d+)-(\d+)/).slice(1);

		profile.bust = Number(bust);
		profile.cup = cup;
		profile.waist = Number(waist);
		profile.hip = Number(hip);
	}

	profile.avatar = query.img('img.main-preview', 'src', { origin: entity.url });

	return profile;
}

async function fetchLatestBlog(channel, page) {
	const url = `${channel.url}/free/updates/videos/${(page - 1) * 10}`;
	const res = await qu.getAll(url, '.videos');

	return res.ok ? scrapeLatestBlog(res.items, channel) : res.status;
}

async function fetchLatest(channel, page = 1) {
	if (channel.parameters?.blog) {
		return fetchLatestBlog(channel, page);
	}

	const url = `${channel.url}/categories/Movies_${page}_d.html`;
	const res = await qu.getAll(url, '.bodyArea .updateItem');

	return res.ok ? scrapeAll(res.items, channel) : res.status;
}

async function fetchUpcoming(channel) {
	if (channel.parameters?.blog) {
		return [];
	}

	const res = await qu.getAll(channel.url, '#owl-upcomingScenes .updateItem');

	return res.ok ? scrapeAll(res.items, channel) : res.status;
}

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

	if (res.ok) {
		if (channel.parameters?.blog) {
			return scrapeSceneBlog(res.item, url, channel);
		}

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

	return res.status;
}

async function fetchProfile(baseActor, entity) {
	const modelsRes = await qu.getAll(`${entity.url}/free/girls.php?alpha=${baseActor.name.slice(0, 1)}`, '.model');

	console.log(baseActor);

	if (modelsRes.ok) {
		const models = modelsRes.items.filter(({ query }) => query.cnt('strong') === baseActor.name);

		return Promise.all(models.map(async (model) => {
			const modelUrl = model.query.url('a', 'href', { origin: entity.url });
			const modelRes = await qu.get(modelUrl);

			if (modelRes.ok) {
				return scrapeProfile(modelRes.item, entity);
			}

			return modelRes.status;
		}));
	}

	return modelsRes.status;
}

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