traxxx/src/scrapers/bang.js

'use strict';

const bhttp = require('bhttp');

const slugify = require('../utils/slugify');

const clusterId = '617fb597b659459bafe6472470d9073a';
const authKey = 'YmFuZy1yZWFkOktqVDN0RzJacmQ1TFNRazI=';

const genderMap = {
    M: 'male',
    F: 'female',
};

function getScreenUrl(item, scene) {
    return `https://i.bang.com/screenshots/${scene.dvd.id}/movie/${scene.order}/${item.screenId}.jpg`;
}

function encodeId(id) {
    return Buffer
        .from(id, 'hex')
        .toString('base64')
        .replace(/\+/g, '-')
        .replace(/\//g, '_')
        .replace(/=/g, ',');
}

function decodeId(id) {
    const restoredId = id
        .replace(/-/g, '+')
        .replace(/_/g, '/')
        .replace(/,/g, '=');

    return Buffer
        .from(restoredId, 'base64')
        .toString('hex');
}

function scrapeScene(scene, site) {
    const release = {
        site,
        entryId: scene.id,
        title: scene.name,
        description: scene.description,
        tags: scene.genres.concat(scene.actions).map(genre => genre.name),
        duration: scene.duration,
    };

    const slug = slugify(release.title);
    release.url = `https://www.bang.com/video/${encodeId(release.entryId)}/${slug}`;

    const date = new Date(scene.releaseDate);
    release.date = new Date(Date.UTC(date.getUTCFullYear(), date.getUTCMonth(), date.getUTCDate()));

    release.actors = scene.actors.map(actor => ({ name: actor.name, gender: genderMap[actor.gender] }));

    if (scene.is4k) release.tags.push('4k');
    if (scene.gay) release.tags.push('gay');

    const defaultPoster = scene.screenshots.find(photo => photo.default === true);
    const photoset = scene.screenshots.filter(photo => photo.default === false);

    const photos = defaultPoster ? photoset : photoset.slice(1);
    const poster = defaultPoster || photoset[0];

    release.poster = getScreenUrl(poster, scene);
    release.photos = photos.map(photo => getScreenUrl(photo, scene));

    release.trailer = {
        src: `https://i.bang.com/v/${scene.dvd.id}/${scene.identifier}/preview.mp4`,
    };

    release.channel = scene.series.name
        .replace(/[! .]/g, '')
        .replace('&', 'and');

    return release;
}

function scrapeLatest(scenes, site) {
    return scenes.map(({ _source: scene }) => scrapeScene(scene, site));
}

async function fetchLatest(site, page = 1) {
    const res = await bhttp.post(`https://${clusterId}.us-east-1.aws.found.io/videos/video/_search`, {
        size: 50,
        from: (page - 1) * 50,
        query: {
            bool: {
                must: [
                    {
                        match: {
                            status: 'ok',
                        },
                    },
                    {
                        range: {
                            releaseDate: {
                                lte: 'now',
                            },
                        },
                    },
                    /*
                     * global fetch
                    {
                        nested: {
                            path: 'studio',
                            query: {
                                bool: {
                                    must: [
                                        {
                                            match: {
                                                'studio.name': {
                                                    operator: 'AND',
                                                    query: 'bang! originals',
                                                },
                                            },
                                        },
                                    ],
                                },
                            },
                        },
                    },
                    */
                    {
                        nested: {
                            path: 'series',
                            query: {
                                bool: {
                                    must: [
                                        {
                                            match: {
                                                'series.id': {
                                                    operator: 'AND',
                                                    query: site.parameters.siteId,
                                                },
                                            },
                                        },
                                    ],
                                },
                            },
                        },
                    },
                ],
                must_not: [
                    {
                        match: {
                            type: 'trailer',
                        },
                    },
                ],
            },
        },
        sort: [
            {
                releaseDate: {
                    order: 'desc',
                },
            },
        ],
    }, {
        encodeJSON: true,
        headers: {
            Authorization: `Basic ${authKey}`,
        },
    });

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

async function fetchScene(url, site) {
    const encodedId = new URL(url).pathname.split('/')[2];
    const entryId = decodeId(encodedId);

    const res = await bhttp.get(`https://${clusterId}.us-east-1.aws.found.io/videos/video/${entryId}`, {
        headers: {
            Authorization: `Basic ${authKey}`,
        },
    });

    return scrapeScene(res.body._source, site); // eslint-disable-line no-underscore-dangle
}

module.exports = {
    fetchLatest,
    fetchScene,
};