traxxx/src/scrapers/realitykings.js

'use strict';

/* eslint-disable newline-per-chained-call */
const Promise = require('bluebird');
const bhttp = require('bhttp');
const { CookieJar } = Promise.promisifyAll(require('tough-cookie'));
const { JSDOM } = require('jsdom');

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

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

    const scriptString = document.querySelector('script').textContent;
    const prefix = 'window.__JUAN.initialState = {';

    const dataStart = scriptString.slice(scriptString.indexOf(prefix) + prefix.length - 1);
    const dataString = dataStart.slice(0, dataStart.indexOf('};') + 1);
    const data = JSON.parse(dataString);

    const actorsMap = data.entities.actors;
    const tagsMap = data.entities.tags;
    const scenes = Object.values(data.entities.releases);

    return Promise.all(scenes.map(async (scene) => {
        const {
            id: entryId,
            title,
            description,
        } = scene;

        const url = `https://www.realitykings.com/scene/${entryId}`;
        const date = new Date(scene.dateReleased);
        const actors = scene.actors.map(actorId => actorsMap[actorId].name);

        const rawTags = scene.tags.map(tagId => tagsMap[tagId].name);
        const tags = await matchTags(rawTags);

        if (!scene.images.poster) {
            console.log(site.name, site.id);
            console.log(scene);
            console.log(title, url, scene.images);
        }

        const [poster, ...photos] = scene.images.poster.map(image => image.xl.url);

        const duration = scene.videos.mediabook.length;
        const trailer720p = scene.videos.mediabook.files['720p'] && scene.videos.mediabook.files['720p'].urls.view;
        const trailer360p = scene.videos.mediabook.files['360p'] && scene.videos.mediabook.files['360p'].urls.view;

        const { likes, dislikes } = scene.stats;

        return {
            url,
            entryId,
            title,
            description,
            actors,
            date,
            tags,
            duration,
            poster,
            photos,
            trailer: {
                src: trailer720p || trailer360p,
                quality: trailer720p ? 720 : 360,
            },
            rating: { likes, dislikes },
            site,
        };
    }));
}

async function scrapeScene(data, url, site) {
    const {
        id: entryId,
        title,
        description,
    } = data;

    const date = new Date(data.dateReleased);
    const actors = data.actors.map(actor => actor.name);

    const { likes, dislikes } = data.stats;

    const rawTags = data.tags.map(tag => tag.name);
    const tags = await matchTags(rawTags);

    const [poster, ...photos] = data.images.poster.map(image => image.xl.url);

    const duration = data.videos.mediabook.length;
    const trailer720p = data.videos.mediabook.files['720p'] && data.videos.mediabook.files['720p'].urls.view;
    const trailer360p = data.videos.mediabook.files['360p'] && data.videos.mediabook.files['360p'].urls.view;

    return {
        url,
        entryId,
        title,
        description,
        actors,
        date,
        duration,
        tags,
        poster,
        photos,
        trailer: {
            src: trailer720p || trailer360p,
            quality: trailer720p ? 720 : 360,
        },
        rating: {
            likes,
            dislikes,
        },
        site,
    };
}

async function fetchLatest(site, page = 1) {
    const { hostname, search } = new URL(site.url);

    if (hostname.match(/(www\.)?realitykings\.com/) && search.match(/\?site=\d+/)) {
        const res = await bhttp.get(`${site.url}&page=${page}`);

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

    if (site.parameters && site.parameters.siteId) {
        const res = await bhttp.get(`https://www.realitykings.com/scenes?site=${site.parameters.siteId}&page=${page}`);

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

    throw new Error(`Reality Kings site '${site.name}' (${site.url}) not supported`);
}

async function fetchScene(url, site) {
    const entryId = url.match(/\d+/)[0];

    const cookieJar = new CookieJar();
    const session = bhttp.session({ cookieJar });

    await session.get(url);

    const cookies = await cookieJar.getCookieStringAsync(url);
    const instanceToken = cookies.split(';')[0].split('=')[1];

    const res = await session.get(`https://site-api.project1service.com/v2/releases/${entryId}`, {
        headers: {
            Instance: instanceToken,
        },
    });

    return scrapeScene(res.body.result.parent || res.body.result, url, site);
}

module.exports = {
    fetchLatest,
    fetchScene,
};