traxxx/src/scrapers/brazzers.js

'use strict';

/* eslint-disable newline-per-chained-call */
const bhttp = require('bhttp');
const cheerio = require('cheerio');
const { JSDOM } = require('jsdom');
const moment = require('moment');

const { heightToCm, lbsToKg } = require('../utils/convert');
const { fetchSites } = require('../sites');
const { matchTags } = require('../tags');

const hairMap = {
    Blonde: 'blonde',
    Brunette: 'brown',
    'Black Hair': 'black',
    Redhead: 'red',
};

function scrape(html, site, upcoming) {
    const $ = cheerio.load(html, { normalizeWhitespace: true });
    const sceneElements = $('.release-card.scene').toArray();

    return sceneElements.reduce((acc, element) => {
        const isUpcoming = $(element).find('.icon-upcoming.active').length === 1;

        if ((upcoming && !isUpcoming) || (!upcoming && isUpcoming)) {
            return acc;
        }

        const sceneLinkElement = $(element).find('a');

        const url = `https://www.brazzers.com${sceneLinkElement.attr('href')}`;
        const title = sceneLinkElement.attr('title');
        const entryId = url.split('/').slice(-3, -2)[0];

        const date = moment.utc($(element).find('time').text(), 'MMMM DD, YYYY').toDate();
        const actors = $(element).find('.model-names a').map((actorIndex, actorElement) => $(actorElement).attr('title')).toArray();

        const likes = Number($(element).find('.label-rating .like-amount').text());
        const dislikes = Number($(element).find('.label-rating .dislike-amount').text());

        const poster = `https:${$(element).find('.card-main-img').attr('data-src')}`;
        const photos = $(element).find('.card-overlay .image-under').map((photoIndex, photoElement) => `https:${$(photoElement).attr('data-src')}`).toArray();

        return acc.concat({
            url,
            entryId,
            title,
            actors,
            date,
            poster,
            photos,
            rating: {
                likes,
                dislikes,
            },
            site,
        });
    }, []);
}

async function scrapeScene(html, url, site) {
    const $ = cheerio.load(html, { normalizeWhitespace: true });
    const videoJson = $('script:contains("window.videoUiOptions")').html();
    const videoData = JSON.parse(videoJson.slice(videoJson.indexOf('{"stream_info":'), videoJson.lastIndexOf('"},') + 2));

    const entryId = url.split('/').slice(-3, -2)[0];
    const title = $('.scene-title[itemprop="name"]').text();

    const description = $('#scene-description p[itemprop="description"]')
        .contents()
        .first()
        .text()
        .trim();

    const date = moment.utc($('.more-scene-info .scene-date').text(), 'MMMM DD, YYYY').toDate();
    const actors = $('.related-model a').map((actorIndex, actorElement) => $(actorElement).text()).toArray();
    const duration = Number($('.scene-length[itemprop="duration"]').attr('content').slice(1, -1)) * 60;

    const likes = Number($('.label-rating .like').text());
    const dislikes = Number($('.label-rating .dislike').text());

    const siteElement = $('.niche-site-logo');
    const siteUrl = `https://www.brazzers.com${siteElement.attr('href').slice(0, -1)}`;
    const siteName = siteElement.attr('title');
    const siteSlug = siteName.replace(/\s+/g, '').toLowerCase();

    const rawTags = $('.tag-card-container a').map((tagIndex, tagElement) => $(tagElement).text()).toArray();

    const poster = `https:${videoData.poster}`;
    const trailer = `https:${videoData.stream_info.http.paths.mp4_480_1500}`;
    const photos = $('.carousel-thumb a').map((photoIndex, photoElement) => `https:${$(photoElement).attr('href')}`).toArray();

    const [tags, [channelSite]] = await Promise.all([
        matchTags(rawTags),
        site.isFallback
            ? fetchSites({
                slug: siteSlug,
                name: siteName,
                url: siteUrl,
            })
            : [site],
    ]);

    return {
        url,
        entryId,
        title,
        description,
        actors,
        date,
        poster,
        photos,
        trailer: {
            src: trailer,
            quality: 480,
        },
        duration,
        rating: {
            likes,
            dislikes,
        },
        tags,
        site: channelSite || site,
    };
}

function scrapeActorSearch(html, url, actorName) {
    const { document } = new JSDOM(html).window;
    const actorLink = document.querySelector(`a[title="${actorName}"]`);

    return actorLink;
}

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

    const avatarEl = document.querySelector('.big-pic-model-container img');
    const descriptionEl = document.querySelector('.model-profile-specs p');
    const bioKeys = Array.from(document.querySelectorAll('.profile-spec-list label'), el => el.textContent.replace(/\n+|\s{2,}/g, '').trim());
    const bioValues = Array.from(document.querySelectorAll('.profile-spec-list var'), el => el.textContent.replace(/\n+|\s{2,}/g, '').trim());

    const bio = bioKeys.reduce((acc, key, index) => ({ ...acc, [key]: bioValues[index] }), {});

    const profile = {
        name: actorName,
    };

    if (bio.Ethnicity) profile.ethnicity = bio.Ethnicity;
    if (bio.Measurements) [profile.bust, profile.waist, profile.hip] = bio.Measurements.split('-');
    if (bio['Date of Birth'] && bio['Date of Birth'] !== 'Unknown') profile.birthdate = moment.utc(bio['Date of Birth'], 'MMMM DD, YYYY').toDate();
    if (bio['Birth Location']) profile.birthPlace = bio['Birth Location'];
    if (bio['Pussy Type']) profile.pussy = bio['Pussy Type'].split(',').slice(-1)[0].toLowerCase();

    if (bio.Height) profile.height = heightToCm(bio.Height);
    if (bio.Weight) profile.weight = lbsToKg(bio.Weight.match(/\d+/)[0]);
    if (bio['Hair Color']) profile.hair = hairMap[bio['Hair Color']] || bio['Hair Color'].toLowerCase();

    if (bio['Body Art']) {
        profile.hasTattoo = !!bio['Body Art'].match('Tattoo');
        profile.hasPiercing = !!bio['Body Art'].match('Piercing');
    }

    if (descriptionEl) profile.description = descriptionEl.textContent.trim();
    if (avatarEl) profile.avatar = `https:${avatarEl.src}`;

    profile.releases = Array.from(document.querySelectorAll('.release-card-container .scene-card-title a'), el => `https://brazzers.com${el.href}`);

    return profile;
}

async function fetchLatest(site, page = 1) {
    const res = await bhttp.get(`${site.url}/page/${page}/`);

    return scrape(res.body.toString(), site, false);
}

async function fetchUpcoming(site) {
    const res = await bhttp.get(`${site.url}/`);

    return scrape(res.body.toString(), site, true);
}

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

    return scrapeScene(res.body.toString(), url, site);
}

async function fetchProfile(actorName) {
    const searchUrl = 'https://brazzers.com/pornstars-search/';
    const searchRes = await bhttp.get(searchUrl, {
        headers: {
            Cookie: `textSearch=${encodeURIComponent(actorName)};`,
        },
    });

    const actorLink = scrapeActorSearch(searchRes.body.toString(), searchUrl, actorName);

    if (actorLink) {
        const url = `https://brazzers.com${actorLink}`;
        const res = await bhttp.get(url);

        return scrapeProfile(res.body.toString(), url, actorName);
    }

    return null;
}

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