traxxx/src/scrapers/pornhub.js

'use strict';

const { JSDOM } = require('jsdom');
const moment = require('moment');

const http = require('../utils/http');

const ethnicityMap = {
	White: 'Caucasian',
};

const hairMap = {
	Brunette: 'brown',
};

async function scrapeProfile(html, _url, actorName) {
	const { document } = new JSDOM(html).window;

	const entries = Array.from(document.querySelectorAll('.infoPiece'), el => el.textContent.replace(/\n|\t/g, '').split(':'));
	const bio = entries.reduce((acc, [key, value]) => (key ? { ...acc, [key.trim()]: value.trim() } : acc), {});

	const profile = {
		name: actorName,
	};

	const descriptionString = document.querySelector('div[itemprop="description"]') || document.querySelector('.longBio');
	const avatarEl = document.querySelector('#getAvatar') || document.querySelector('.thumbImage img');

	if (bio.Gender) profile.gender = bio.Gender.toLowerCase();
	if (bio.ethnicity) profile.ethnicity = ethnicityMap[bio.Ethnicity] || bio.Ethnicity;

	if (descriptionString) profile.description = descriptionString.textContent;

	if (bio.Birthday && !/-0001/.test(bio.Birthday)) profile.birthdate = moment.utc(bio.Birthday, 'MMM D, YYYY').toDate(); // birthyear sometimes -0001, see Spencer Bradley as of january 2020
	if (bio.Born) profile.birthdate = moment.utc(bio.Born, 'YYYY-MM-DD').toDate();

	profile.birthPlace = bio['Birth Place'] || bio.Birthplace;
	profile.residencePlace = bio['City and Country'];

	if (bio.Measurements && bio.Measurements !== '--') [profile.bust, profile.waist, profile.hip] = bio.Measurements.split('-');
	if (bio['Fake Boobs']) profile.naturalBoobs = bio['Fake Boobs'] === 'No';

	if (bio.Height) profile.height = Number(bio.Height.match(/\(\d+/)[0].slice(1));
	if (bio.Weight) profile.weight = Number(bio.Weight.match(/\(\d+/)[0].slice(1));
	if (bio['Hair Color']) profile.hair = hairMap[bio['Hair Color']] || bio['Hair Color'].toLowerCase();
	if (bio.Piercings) profile.hasPiercings = bio.Piercings === 'Yes';
	if (bio.Tattoos) profile.hasTattoos = bio.Tattoos === 'Yes';

	if (avatarEl && !/default\//.test(avatarEl.src)) profile.avatar = avatarEl.src;
	profile.social = Array.from(document.querySelectorAll('.socialList a'), el => el.href).filter(link => link !== 'https://www.twitter.com/'); // PH links to Twitter itself for some reason

	return profile;
}

async function fetchProfile({ name: actorName }) {
	const actorSlug = actorName.toLowerCase().replace(/\s+/g, '-');

	/* Model pages are not reliably associated with actual porn stars
    const modelUrl = `https://pornhub.com/model/${actorSlug}`;
    const pornstarUrl = `https://pornhub.com/pornstar/${actorSlug}`;

    const [modelRes, pornstarRes] = await Promise.all([
        http.get(modelUrl),
        http.get(pornstarUrl),
    ]);

    const model = modelRes.statusCode === 200 && await scrapeProfile(modelRes.body.toString(), modelUrl, actorName);
    const pornstar = pornstarRes.statusCode === 200 && await scrapeProfile(pornstarRes.body.toString(), pornstarUrl, actorName);

    if (model && pornstar) {
        return {
            ...model,
            ...pornstar,
        };
    }
    */

	const pornstarUrl = `https://pornhub.com/pornstar/${actorSlug}`;
	const pornstarRes = await http.get(pornstarUrl);

	return scrapeProfile(pornstarRes.body.toString(), pornstarUrl, actorName);
}

module.exports = {
	fetchProfile,
};