traxxx/src/scrapers/cherrypimps.js

'use strict';

const unprint = require('unprint');

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

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

		const url = query.url('.item-title a');
		const { pathname } = new URL(url);

		release.url = url;
		release.entryId = pathname.match(/\/trailers\/(.*).html/)[1];

		release.title = query.content('.item-title a');

		release.date = query.date('.item-date', 'MMMM D, YYYY');
		release.duration = query.duration('.item-date', /(\d{2}:)?\d{2}:\d{2}/);

		release.actors = query.all('.item-models a').map((actorEl) => ({
			name: unprint.query.content(actorEl),
			url: unprint.query.url(actorEl, null),
		}));

		release.poster = query.img('.item-video-thumb', { attribute: 'data-videoposter' }) || query.img('img.video_placeholder');
		release.teaser = query.video('.item-video-thumb', { attribute: 'data-videosrc' });

		release.photoCount = query.number('.item-date', { match: /(\d+) photos/i, matchIndex: 1 });
		release.channel = slugify(query.content('.item-sitename a'), '');

		return release;
	});
}

async function fetchLatest(channel, page = 1) {
	const slug = channel.parameters?.slug || new URL(channel.url).pathname.match(/\/series\/([\w-]+)/)[1];
	const res = await unprint.get(`https://cherrypimps.com/categories/${slug}_${page}_d.html`, { selectAll: '.item-updates .item-video' });

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

	return res.status;
}

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

	release.entryId = pathname.match(/\/trailers\/(.*).html/)[1];

	release.title = query.content('.item-title h1');
	release.description = query.content('.update-info-block p');

	release.date = query.date('.update-info-row:first-child', 'MMMM D, YYYY');
	release.duration = query.duration('.update-info-row:last-child');
	release.photoCount = query.number('.update-info-row:last-child', { match: /(\d+) photos/i, matchIndex: 1 });

	release.actors = query.all('.models-list-thumbs .model-list-item').map((actorEl) => ({
		name: unprint.query.content(actorEl, 'span'),
		url: unprint.query.url(actorEl, 'a'),
		avatar: [
			unprint.query.img(actorEl, 'img', { attribute: 'src0_3x' }),
			unprint.query.img(actorEl, 'img', { attribute: 'src0_2x' }),
			unprint.query.img(actorEl, 'img', { attribute: 'src0_1x' }),
		],
	}));

	release.tags = query.contents('.update-info-block a[href*="categories/"]');

	release.poster = [
		query.img('.update_thumb', { attribute: 'src0_3x' }),
		query.img('.update_thumb', { attribute: 'src0_2x' }),
		query.img('.update_thumb', { attribute: 'src0_1x' }), // usually only this one available
	].filter(Boolean);

	// faux video trailer player redirects to signup

	return release;
}

async function fetchScene(url, site, release) {
	const res = await unprint.get(url);

	if (res.ok) {
		return scrapeScene(res.context, url, site, release);
	}

	return res.status;
}

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

	const bio = Object.fromEntries(query.all('.model-stats li').map((bioEl) => [
		slugify(unprint.query.content(bioEl, 'strong'), '_'),
		unprint.query.text(bioEl),
	]));

	profile.height = Number(bio.height?.match(/\((\d+)\s*cm\)/)?.[1]) || null;
	profile.weight = Number(bio.weight?.match(/\((\d+)\s*kg\)/)?.[1]) || null;

	profile.age = parseInt(bio.age, 10) || null;
	profile.dateOfBirth = unprint.extractDate(bio.date_of_birth, 'MMMM D, YYYY');
	profile.birthPlace = bio.birthplace;
	profile.ethnicity = bio.race;

	profile.measurements = bio.measurements;

	profile.hair = bio.hair_color;
	profile.eyes = bio.eye_color;

	if (/various|several/i.test(bio.tattoos)) profile.hasTattoos = true;
	else if (/none/i.test(bio.tattoos)) profile.hasTattoos = false;
	else if (bio.tattoos) {
		profile.hasTattoos = true;
		profile.tattoos = bio.tattoos;
	}

	if (/various/i.test(bio.piercings)) profile.hasPiercings = true;
	else if (/none/i.test(bio.piercings)) profile.hasPiercings = false;
	else if (bio.piercings) {
		profile.hasPiercings = true;
		profile.piercings = bio.piercings;
	}

	profile.aliases = bio.aliases?.split(',').map((alias) => alias.trim());

	profile.avatar = [
		query.img('.model-img img, .model_bio_thumb', { attribute: 'src0_3x' }),
		query.img('.model-img img, .model_bio_thumb', { attribute: 'src0_2x' }),
		query.img('.model-img img, .model_bio_thumb', { attribute: 'src0_1x' }),
	];

	return profile;
}

async function fetchProfile({ name: actorName, url: actorUrl }, { channel, network }) {
	const origin = new URL(channel?.url || network.url).origin;

	const { res, url } = await tryUrls([
		actorUrl,
		`${origin}/models/${slugify(actorName, '')}.html`,
		`${origin}/models/${slugify(actorName)}.html`,
	]);

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

	return res.status;
}

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