traxxx/src/scrapers/xempire.js

'use strict';

const bhttp = require('bhttp');
const cheerio = require('cheerio');
const moment = require('moment');

const { matchTags } = require('../tags');

function scrape(html, site) {
    const $ = cheerio.load(html, { normalizeWhitespace: true });
    const scenesElements = $('li[data-itemtype=scene]').toArray();

    return scenesElements.map((element) => {
        const sceneLinkElement = $(element).find('.sceneTitle a');
        const url = `${site.url}${sceneLinkElement.attr('href')}`;
        const title = sceneLinkElement.attr('title');

        const shootId = $(element).attr('data-itemid');

        const date = moment
            .utc($(element).find('.sceneDate').text(), 'MM-DD-YYYY')
            .toDate();

        const actors = $(element).find('.sceneActors a')
            .map((actorIndex, actorElement) => $(actorElement).attr('title'))
            .toArray();

        const [likes, dislikes] = $(element).find('.value')
            .toArray()
            .map(value => Number($(value).text()));

        return {
            url,
            shootId,
            title,
            actors,
            director: 'Mason',
            date,
            rating: {
                likes,
                dislikes,
            },
            site,
        };
    });
}

async function scrapeSceneFallback($, url, site) {
    const shootId = new URL(url).pathname.split('/').slice(-1)[0];
    const title = $('h1.title').text();
    const date = moment.utc($('.updatedDate').text(), 'MM-DD-YYYY').toDate();
    const actors = $('.sceneColActors a').map((actorIndex, actorElement) => $(actorElement).text()).toArray();

    const description = ($('.sceneDesc').text() || '').replace(/Video Description:/g, ' ').trim();
    const stars = $('.currentRating').text().split('/')[0] / 2;

    const rawTags = $('.sceneColCategories > a').map((tagIndex, tagElement) => $(tagElement).text()).toArray();
    const tags = await matchTags(rawTags);

    return {
        url,
        shootId,
        title,
        date,
        actors,
        director: 'Mason',
        description,
        tags,
        rating: {
            stars,
        },
        site,
    };
}

async function scrapeScene(html, url, site) {
    const $ = cheerio.load(html, { normalizeWhitespace: true });
    const json = $('script[type="application/ld+json"]').html();

    if (!json) {
        return scrapeSceneFallback($, url, site);
    }

    const data = JSON.parse(json)[0];
    const shootId = new URL(url).pathname.split('/').slice(-1)[0];

    const title = data.isPartOf.name;
    const date = moment.utc(data.dateCreated, 'YYYY-MM-DD').toDate();

    const actors = data.actor
        .sort(({ genderA }, { genderB }) => {
            if (genderA === 'female' && genderB === 'male') return 1;
            if (genderA === 'male' && genderB === 'female') return -1;

            return 0;
        })
        .map(actor => actor.name);

    const description = data.description || undefined;
    const stars = (data.aggregateRating.ratingValue / data.aggregateRating.bestRating) * 5;

    const duration = moment.duration(data.duration.slice(2).split(':')).asSeconds();

    const rawTags = data.keywords.split(', ');
    const tags = await matchTags(rawTags);

    return {
        url,
        shootId,
        title,
        date,
        actors,
        director: 'Mason',
        description,
        duration,
        tags,
        rating: {
            stars,
        },
        site,
    };
}

async function fetchLatest(site, page = 1) {
    const res = await bhttp.get(`${site.url}/en/videos/AllCategories/0/${page}`);

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

async function fetchUpcoming(site) {
    const res = await bhttp.get(`${site.url}/en/videos/AllCategories/0/1/upcoming`);

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

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

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

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