traxxx/src/scrapers/pervcity.js

'use strict';

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

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

		release.url = query.url('.videoPic a');
		release.entryId = query.q('.videoPic img', 'id').match(/set-target-(\d+)/)[1];

		release.title = query.q('h3 a', true);
		release.description = query.q('.runtime + p', true);

		release.date = query.date('.date', 'MM-DD-YYYY');
		release.duration = query.dur('.runtime');

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

		release.poster = query.img('.videoPic img');
		release.entity = entity;

		return release;
	});
}

function scrapeScene({ query }) {
	const release = {};

	release.entryId = query.q('.trailerLeft img', 'id').match(/set-target-(\d+)/)[1];

	release.title = query.q('.infoHeader h1', true);
	release.description = query.q('.infoBox p', true);

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

	release.poster = query.img('.posterimg');
	release.photos = query.imgs('.trailerSnaps img').slice(1); // first photo is poster in lower quality

	const trailer = query.q('script')?.textContent.match(/\/trailers\/.+\.mp4/)?.[0];

	if (trailer) {
		release.trailer = {
			src: `https://pervcity.com${trailer}`,
		};
	}

	return release;
}

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

	const bio = query.all('.moreInfo li').reduce((acc, el) => ({
		...acc,
		[slugify(query.q(el, 'span', true), '_')]: query.text(el),
	}), {});

	profile.description = query.q('.aboutModel p', true);
	profile.dateOfBirth = qu.extractDate(bio.date_of_birth, ['MMMM D, YYYY', 'DD-MMM-YY']);

	profile.birthPlace = bio.birth_location;
	profile.ethnicity = bio.ethnicity;

	profile.height = feetInchesToCm(bio.height);
	profile.weight = lbsToKg(bio.weight);

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

	profile.avatar = query.img('.starPic img');
	profile.releases = scrapeAll(qu.initAll(query.all('.aboutScenes .videoBlock')));

	return profile;
}

async function fetchLatest(channel, page = 1) {
	if (channel.parameters?.siteId) {
		const url = `https://pervcity.com/search.php?site[]=${channel.parameters.siteId}&page=${page}`;
		const res = await qu.getAll(url, '.videoBlock');

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

	return null;
}

async function fetchUpcoming(channel) {
	const url = 'https://pervcity.com';
	const res = await qu.getAll(url, '.upcoming .videoBlock');

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

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

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

async function fetchProfile(actorName) {
	const url = `https://pervcity.com/models/${slugify(actorName)}.html`;
	const res = await qu.get(url);

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

	const url2 = `https://pervcity.com/models/${slugify(actorName, '')}.html`;
	const res2 = await qu.get(url2);

	if (res2.ok) {
		return scrapeProfile(res2.item);
	}

	return res2.status;
}

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