traxxx/src/scrapers/ddfnetwork.js

'use strict';

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

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

/* eslint-disable newline-per-chained-call */
function scrapeLatest(html, site) {
    const $ = cheerio.load(html, { normalizeWhitespace: true });
    const sceneElements = $('.card.m-1').toArray();

    return sceneElements.map((element) => {
        const sceneLinkElement = $(element).find('a').first();
        const title = sceneLinkElement.attr('title');
        const url = `${site.url}${sceneLinkElement.attr('href')}`;
        const entryId = url.split('/').slice(-1)[0];

        const date = moment.utc($(element).find('.card-footer .text-muted').text(), 'MMMM DD, YYYY').toDate();
        const actors = $(element).find('.card-subtitle a').map((actorIndex, actorElement) => $(actorElement).text().trim()).toArray().filter(actor => actor);

        const duration = Number($(element).find('.card-info div:nth-child(2) .card-text').text().slice(0, -4)) * 60;

        return {
            url,
            entryId,
            title,
            actors,
            date,
            duration,
            rating: null,
            site,
        };
    });
}

async function scrapeScene(html, url, site) {
    const $ = cheerio.load(html, { normalizeWhitespace: true });

    const entryId = url.split('/').slice(-1)[0];
    const title = $('.video-title h1').text();
    const description = $('.description-box .box-container').text();

    const date = moment.utc($('.video-title .remain time').text(), 'MMMM DD, YYYY').toDate();
    const actors = $('.pornstars-box .pornstar-card .card-title a').map((actorIndex, actorElement) => $(actorElement).text()).toArray();

    const likes = Number($('.info-panel.likes .likes').text());
    const duration = Number($('.info-panel.duration .duration').text().slice(0, -4)) * 60;

    const { origin } = new URL($('.pornstar-card meta[itemprop="url"]').first().attr('content'));
    const rawTags = $('#tagsBox .tags a').map((tagIndex, tagElement) => $(tagElement).text()).toArray();

    const [channelSite, tags] = await Promise.all([
        // don't find site if original is already specific
        site.isFallback ? knex('sites').where({ url: origin }).first() : site,
        matchTags(rawTags),
    ]);

    return {
        url: channelSite ? `${channelSite.url}${new URL(url).pathname}` : url,
        entryId,
        title,
        description,
        actors,
        date,
        duration,
        tags,
        rating: {
            likes,
        },
        site: channelSite || site,
    };
}

async function fetchLatest(site, page = 1) {
    const res = await bhttp.get(`https://ddfnetwork.com/videos/search/latest/ever/${new URL(site.url).hostname}/-/${page}`);

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

async function fetchScene(url, site) {
    const res = await bhttp.get(`https://ddfnetwork.com${new URL(url).pathname}`);

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

module.exports = {
    fetchLatest,
    fetchScene,
};