traxxx/src/scrapers/evilangel.js

'use strict';

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

const { getPhotos } = require('./gamma');

async function scrape(json, site) {
    return Promise.all(json.map(async (scene) => {
        const {
            title,
            description,
            length,
            master_categories: tags,
            ratings_up: likes,
            ratings_down: dislikes,
        } = scene;

        const entryId = scene.clip_id;
        const url = `https://evilangel.com/en/video/${scene.url_title}/${entryId}`;
        const date = moment(scene.release_date, 'YYYY-MM-DD').toDate();
        const actors = scene.actors.map(({ name }) => name);
        const director = scene.directors[0].name;

        const poster = `https://images-evilangel.gammacdn.com/movies${scene.pictures.resized}`;
        const movie = `https://evilangel.com/en/movie/${scene.url_movie_title}/${scene.movie_id}`;

        return {
            url,
            entryId,
            title,
            description,
            length,
            actors,
            director,
            date,
            tags,
            poster,
            rating: {
                likes,
                dislikes,
            },
            movie,
            site,
        };
    }));
}

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

    const [data, data2] = JSON.parse(json);
    const videoData = JSON.parse(videoJson.slice(videoJson.indexOf('{'), videoJson.indexOf('};') + 1));
    const entryId = new URL(url).pathname.split('/').slice(-1)[0];

    const {
        name: title,
        description,
    } = data;
    // date in data object is not the release date of the scene, but the date the entry was added
    const date = moment.utc($('.updatedDate').first().text(), 'MM-DD-YYYY').toDate();

    const actors = data.actor.map(actor => actor.name);
    const hasTrans = data.actor.some(actor => actor.gender === 'shemale');

    const director = (data.director && data.director[0].name) || (data2.director && data2.director[0].name) || null;
    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 = hasTrans ? [...rawTags, 'transsexual'] : rawTags;

    const poster = videoData.picPreview;
    const trailer = `${videoData.playerOptions.host}${videoData.url}`;

    const photos = await getPhotos($('.picturesItem a').attr('href'), 'evilangel.com', site);

    return {
        url,
        entryId,
        title,
        date,
        actors,
        director,
        description,
        duration,
        tags,
        poster,
        photos,
        trailer: {
            src: trailer,
            quality: parseInt(videoData.sizeOnLoad, 10),
        },
        rating: {
            stars,
        },
        site,
    };
}

function scrapeActor(data) {
    const actor = {};

    if (data.male === 1) actor.gender = 'male';
    if (data.female === 1) actor.gender = 'female';
    if (data.shemale === 1 || data.trans === 1) actor.gender = 'transsexual';

    if (data.description) actor.description = data.description.trim();

    if (data.attributes.ethnicity) actor.ethnicity = data.attributes.ethnicity;
    if (data.attributes.eye_color) actor.eyes = data.attributes.eye_color;
    if (data.attributes.hair_color) actor.hair = data.attributes.hair_color;

    const avatarPath = Object.values(data.pictures).reverse()[0];
    actor.avatar = `https://images01-evilangel.gammacdn.com/actors${avatarPath}`;

    return actor;
}

async function fetchApiCredentials() {
    const res = await bhttp.get('https://evilangel.com/en/videos');
    const body = res.body.toString();

    const apiLine = body.split('\n').find(bodyLine => bodyLine.match('apiKey'));
    const apiSerial = apiLine.slice(apiLine.indexOf('{'), apiLine.indexOf('};') + 1);
    const apiData = JSON.parse(apiSerial);

    const { applicationID: appId, apiKey } = apiData.api.algolia;
    const userAgent = 'Algolia for vanilla JavaScript (lite) 3.27.0;instantsearch.js 2.7.4;JS Helper 2.26.0';

    const apiUrl = `https://${appId.toLowerCase()}-dsn.algolia.net/1/indexes/*/queries?x-algolia-agent=${userAgent}&x-algolia-application-id=${appId}&x-algolia-api-key=${apiKey}`;

    return {
        appId,
        apiKey,
        userAgent,
        apiUrl,
    };
}

async function fetchLatest(site, page = 1, upcoming = false) {
    const { apiUrl } = await fetchApiCredentials();

    const res = await bhttp.post(apiUrl, {
        requests: [
            {
                indexName: 'all_scenes',
                params: `query=&hitsPerPage=36&maxValuesPerFacet=100&page=${page - 1}&facetFilters=[["lesbian:"],["bisex:"],["shemale:"],["upcoming:${upcoming ? 1 : 0}"]]`,
            },
        ],
    }, {
        headers: {
            Referer: 'https://www.evilangel.com/en/videos',
        },
        encodeJSON: true,
    });

    return scrape(res.body.results[0].hits, site);
}

async function fetchUpcoming(site) {
    return fetchLatest(site, 1, true);
}

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

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

async function fetchProfile(actorName) {
    const { apiUrl } = await fetchApiCredentials();
    const actorSlug = encodeURI(actorName);

    const res = await bhttp.post(apiUrl, {
        requests: [
            {
                indexName: 'all_actors',
                params: `query=${actorSlug}`,
            },
        ],
    }, {
        headers: {
            Referer: `https://www.evilangel.com/en/search?query=${actorSlug}&tab=actors`,
        },
        encodeJSON: true,
    });

    if (res.statusCode === 200 && res.body.results[0].hits.length > 0) {
        const actorData = res.body.results[0].hits.find(actor => actor.name === actorName);

        if (actorData) {
            return scrapeActor(actorData);
        }
    }

    return null;
}

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