traxxx/src/fetch-releases.js

'use strict';

const config = require('config');
const moment = require('moment');

const knex = require('./knex');
const scrapers = require('./scrapers');

function destructConfigNetworks(networks) {
    return networks.reduce((acc, network) => {
        if (Array.isArray(network)) {
            // network specifies sites
            return {
                ...acc,
                sites: [...acc.sites, ...network[1]],
            };
        }

        return {
            ...acc,
            networks: [...acc.networks, network],
        };
    }, {
        networks: [],
        sites: [],
    });
}

function curateSites(sites) {
    return sites.map(site => ({
        id: site.id,
        name: site.name,
        description: site.description,
        url: site.url,
        networkId: site.network_id,
        parameters: JSON.parse(site.parameters),
    }));
}

async function accumulateIncludedSites() {
    const included = destructConfigNetworks(config.include);

    const rawSites = await knex('sites')
        .whereIn('id', included.sites)
        .orWhereIn('network_id', included.networks);

    return curateSites(rawSites);
}

async function getExistingReleases() {
    return knex('releases');
    // .where('date', '>', new Date(2019, 2, 26));
}

async function storeReleases(releases) {
    const curatedReleases = releases.map(release => ({
        site_id: release.site.id,
        shoot_id: release.shootId || null,
        url: release.url,
        title: release.title,
        date: release.date,
        description: release.description,
        duration: release.duration,
        likes: release.rating && release.rating.likes,
        dislikes: release.rating && release.rating.dislikes,
        rating: release.rating && release.rating.stars,
    }));

    if (curatedReleases.length) {
        console.log(`Adding ${curatedReleases.length} releases to database (if unique)`);

        const insertQuery = knex('releases').insert(curatedReleases).toString();
        await knex.raw(insertQuery.replace('insert', 'INSERT OR IGNORE'));

        return curatedReleases;
    }

    return [];
}

async function fetchReleases() {
    const sites = await accumulateIncludedSites();
    const releases = await getExistingReleases();

    const scenesPerSite = await Promise.all(sites.map(async (site) => {
        const scraper = scrapers[site.id] || scrapers[site.networkId];

        if (scraper) {
            const [latest, upcoming] = await Promise.all([
                scraper.fetchLatest(site),
                scraper.fetchUpcoming ? scraper.fetchUpcoming(site) : [],
            ]);

            console.log(`${latest.length} published releases and ${upcoming.length} upcoming releases found`);

            await storeReleases(latest, releases);

            return [...latest, ...upcoming];
        }

        return [];
    }));

    const accumulatedScenes = scenesPerSite.reduce((acc, siteScenes) => ([...acc, ...siteScenes]), []);
    const sortedScenes = accumulatedScenes.sort(({ date: dateA }, { date: dateB }) => moment(dateB).diff(dateA));

    return sortedScenes;
}

module.exports = fetchReleases;