traxxx/src/scrapers/kellymadison.js

'use strict';

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

const siteMapByKey = {
	PF: 'pornfidelity',
	TF: 'teenfidelity',
	KM: 'kellymadison',
};

const siteMapBySlug = Object.entries(siteMapByKey).reduce((acc, [key, value]) => ({ ...acc, [value]: key }), {});

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

		release.shootId = query.q('.card-meta .text-right', true);

		const siteId = release.shootId.match(/\w{2}/)[0];
		const siteSlug = siteMapByKey[siteId];

		if (site.slug !== siteSlug) {
			// using generic network overview, scene is not from the site we want
			return null;
		}

		const { pathname } = new URL(query.url('h5 a'));
		[release.entryId] = pathname.match(/\d+$/);
		release.url = `${site.url}/episodes/${release.entryId}`;

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

		release.date = query.date('.card-meta .text-left', ['MMM D', 'MMM D, YYYY'], /\w+ \d+(, \w+)?/);
		release.actors = query.all('.models a', true);

		release.duration = query.dur('.content a');

		const duration = query.q('.content a', true).match(/(\d+) min/)[1];
		if (duration) release.duration = Number(duration) * 60;

		release.poster = query.img('.card-img-top');
		release.teaser = {
			src: query.video('video'),
		};

		return release;
	}).filter(scene => scene);
}

async function scrapeScene({ query, html }, url, baseRelease) {
	const { pathname, origin } = new URL(url);
	const release = {};

	[release.entryId] = pathname.match(/\d+$/);

	const titleString = query.q('.card-header.row h4', true);
	const episode = titleString.match(/#\d+$/)[0];

	release.title = titleString.match(/Trailer: ([\w\s]+) -/)[1];
	release.channel = slugify(titleString.match(/([\w\s]+) #\d+$/)[1], '');

	const siteKey = siteMapBySlug[release.channel];

	release.shootId = `${siteKey} ${episode}`;
	release.description = query.q('p.card-text', true);

	// order not reliable, get keys
	const detailElsByKey = query.all('.card-body h4.card-title').reduce((acc, rowEl) => ({
		...acc,
		[slugify(rowEl.textContent.match(/(\w+):/)?.[1])]: rowEl,
	}), {});

	release.date = query.date(detailElsByKey.published, null, 'YYYY-MM-DD');
	release.duration = query.dur(detailElsByKey.episode);

	release.actors = query.all(detailElsByKey.starring, 'a', true);

	const token = query.meta('name=_token');
	const trailerInfoUrl = `${origin}/episodes/trailer/sources/${release.entryId}?type=trailer`;
	const trailerInfoRes = await http.post(trailerInfoUrl, null, {
		'X-CSRF-Token': token,
		'X-Requested-With': 'XMLHttpRequest',
	});

	if (trailerInfoRes.ok && trailerInfoRes.body.sources.length > 0) {
		release.trailer = trailerInfoRes.body.sources.map(trailer => ({
			src: trailer.src,
			type: trailer.type,
			quality: trailer.res.replace(4000, 2160),
		}));
	}

	const posterPrefix = html.indexOf('poster:');
	const poster = html.slice(html.indexOf('http', posterPrefix), html.indexOf('.jpg', posterPrefix) + 4);

	if (poster) {
		if (baseRelease?.poster) {
			release.photos = [poster];
		} else {
			release.poster = poster;
		}
	}

	return release;
}

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

	const bioKeys = query.all('table.table td:nth-child(1)', true);
	const bioValues = query.all('table.table td:nth-child(2)', true);
	const bio = bioKeys.reduce((acc, key, index) => ({ ...acc, [key.slice(0, -1)]: bioValues[index] }), {});

	if (bio.Ethnicity) profile.ethnicity = bio.Ethnicity;
	if (bio.Measurements) [profile.bust, profile.waist, profile.hip] = bio.Measurements.split('-');
	if (bio.Birthplace) profile.birthPlace = bio.Birthplace;

	if (bio.Height) {
		const [feet, inches] = bio.Height.match(/\d+/g);
		profile.height = feetInchesToCm(feet, inches);
	}

	profile.avatar = query.img('img[src*="model"]');

	return profile;
}

async function fetchLatest(site, page = 1) {
	const url = `https://kellymadison.com/episodes/search?page=${page}`; // TLS issues with teenfidelity.com, same overview on all sites
	const res = await http.get(url, {
		'X-Requested-With': 'XMLHttpRequest',
	});

	if (res.ok && res.body.status === 'success') {
		return scrapeLatest(qu.extractAll(res.body.html, '.episode'), site);
	}

	return res.status;
}

async function fetchScene(url, channel, baseRelease) {
	const { pathname } = new URL(url);

	const res = await qu.get(`https://www.kellymadison.com${pathname}`, null, {
		'X-Requested-With': 'XMLHttpRequest',
	});

	return res.ok ? scrapeScene(res.item, url, baseRelease) : res.status;
}

async function fetchProfile(actorName) {
	const actorSlug = slugify(actorName);
	const res = await qu.get(`https://www.kellymadison.com/models/${actorSlug}`, null, {
		'X-Requested-With': 'XMLHttpRequest',
	});

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

	return res.status;
}

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