traxxx/src/scrapers/teamskeet.js

'use strict';

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

function extractTitle(pathname) {
    return pathname
        .split('/')
        .slice(-2)[0]
        .split('_')
        .map(seg => `${seg.charAt(0).toUpperCase()}${seg.slice(1)}`)
        .join(' ');
}

function extractActors(str) {
    return str
        .split(/,|\band\b/ig)
        .filter(actor => !/\.{3}/.test(actor))
        .map(actor => actor.trim())
        .filter(actor => actor.length > 0);
}

function scrapeLatest(html, site) {
    const { document } = new JSDOM(html).window;

    const scenes = Array.from(document.querySelectorAll('#updatesList li.grey, #updatesList li.white'));

    return scenes.map((scene) => {
        const release = { site };

        const link = scene.querySelector('.info a');
        const poster = scene.querySelector('img');
        const { pathname } = new URL(link);

        [release.entryId] = poster.id.match(/\d+/);

        release.url = `https://www.teamskeet.com${pathname}`;
        release.title = extractTitle(pathname);

        release.date = moment.utc(scene.querySelector('strong').textContent, 'MM/DD/YYYY').toDate();

        const photos = Array.from({ length: 5 }, (_value, index) => poster.dataset.original.replace(/\d+.jpg/, `${String(index + 1).padStart(2, '0')}.jpg`));
        [release.poster] = photos;
        release.photos = photos.slice(1);

        const actors = scene.querySelector('div span[rel="test"]').textContent;
        release.actors = extractActors(actors);

        return release;
    });
}

function scrapeScene(html, site, url) {
    const { document } = new JSDOM(html).window;
    const release = { site };

    release.entryId = document.querySelector('#story-and-tags .scene_rater').attributes.rel.value;
    release.description = document.querySelector('#story-and-tags td:nth-child(2) div').textContent;
    const [actors, title, channel] = document.querySelector('title').textContent.split('|').map(item => item.trim());

    release.url = url;
    release.title = title;
    release.actors = extractActors(actors);
    release.channel = channel.toLowerCase();
    release.tags = Array.from(document.querySelectorAll('#story-and-tags tr:nth-child(2) a'), el => el.rel);

    const date = document.querySelector('h3 ~ div:nth-child(4), h3 ~ div div.gray:not(.scene_rater)').textContent.split(':')[1].trim();
    release.date = moment.utc(date, 'MMMM Do, YYYY').toDate();

    const { poster } = document.querySelector('video');
    if (poster && !/gen/.test(poster)) release.poster = [poster.replace('low', 'hi'), poster];

    const siteId = document.querySelector('#story-and-tags img').src.match(/\w+.jpg/)[0].replace('.jpg', '');
    const actorsSlug = document.querySelector('h3 a').href.split('/').slice(-2)[0];

    release.photos = Array.from({ length: 5 }, (value, index) => `https://images.psmcdn.net/teamskeet/${siteId}/${actorsSlug}/shared/scenes/new/${String(index + 1).padStart(2, '0')}.jpg`);

    const trailer = document.querySelector('div.right.gray a').href;
    if (trailer) release.trailer = { src: trailer };

    return release;
}

function scrapeSceneA(html, site, sceneX, url) {
    const scene = sceneX || new JSDOM(html).window.document;
    const release = { site };

    release.description = scene.querySelector('.scene-story').textContent.replace('...read more', '...').trim();

    release.date = moment.utc(scene.querySelector('.scene-date').textContent, 'MM/DD/YYYY').toDate();
    release.actors = Array.from(scene.querySelectorAll('.starring span'), el => extractActors(el.textContent)).flat();

    const durationString = scene.querySelector('.time').textContent.trim();
    const duration = ['00'].concat(durationString.split(':')).slice(-3).join(':'); // ensure hh:mm:ss
    release.duration = moment.duration(duration).asSeconds();

    if (sceneX) {
        const titleEl = scene.querySelector(':scope > a');

        release.url = titleEl.href;
        release.entryId = titleEl.id;
        release.title = titleEl.title;

        const [poster, ...photos] = Array.from(scene.querySelectorAll('.scene img'), el => el.src);
        release.poster = [poster.replace('bio_big', 'video'), poster];
        release.photos = photos;
    }

    if (!sceneX) {
        release.title = scene.querySelector('.title span').textContent;
        release.url = url;

        release.poster = scene.querySelector('video').poster;
        release.photos = [release.poster.replace('video', 'bio_small'), release.poster.replace('video', 'bio_small2')];
    }

    const [, entryIdA, entryIdB] = new URL(release.url).pathname.split('/');
    release.entryId = entryIdA === 'scenes' ? entryIdB : entryIdA;

    return release;
}

function scrapeLatestA(html, site) {
    const { document } = new JSDOM(html).window;

    const scenes = Array.from(document.querySelectorAll('.scenewrapper'));

    return scenes.map(scene => scrapeSceneA(null, site, scene));
}

async function fetchLatestTeamSkeet(site, page = 1) {
    const url = `https://www.teamskeet.com/t1/updates/load?fltrs[site]=${site.parameters.id}&page=${page}&view=newest&fltrs[time]=ALL&order=DESC`;
    const res = await bhttp.get(url);

    if (res.statusCode === 200) {
        return scrapeLatest(res.body.toString(), site);
    }

    return null;
}

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

    if (res.statusCode === 200) {
        return scrapeLatestA(res.body.toString(), site);
    }

    return null;
}

async function fetchLatest(site, page = 1) {
    if (site.parameters.id) {
        return fetchLatestTeamSkeet(site, page);
    }

    if (site.parameters.scraper === 'A') {
        return fetchLatestA(site, page);
    }

    return null;
}

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

    if (site.parameters?.scraper === 'A') {
        return scrapeSceneA(res.body.toString(), site, null, url);
    }

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

module.exports = {
    fetchLatest,
    fetchScene,
};