traxxx/src/scrapers/blowpass.js

'use strict';

/* eslint-disable newline-per-chained-call */
const bhttp = require('bhttp');
const cheerio = require('cheerio');
const moment = require('moment');

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

function scrape(html, site) {
    const $ = cheerio.load(html, { normalizeWhitespace: true });
    const sceneElements = $('.sceneList .scene').toArray();

    return sceneElements.map((element) => {
        const entryId = $(element).attr('data-itemid');

        const sceneLinkElement = $(element).find('.sceneTitle a');
        const title = sceneLinkElement.attr('title');
        const url = `${site.url}/en/scene/${sceneLinkElement.attr('href').split('/').slice(-2).join('/')}`;

        const date = moment.utc($(element).find('.sceneDate').text(), 'MM-DD-YYYY').toDate();
        const actors = $(element).find('.sceneActors a').map((actorIndex, actorElement) => $(actorElement).text()).toArray();

        const likes = Number($(element).find('.rating .state_1 .value').text());

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

async function scrapeScene(html, url, site) {
    const $ = cheerio.load(html, { normalizeWhitespace: true });
    const json = $('script[type="application/ld+json"]').html();
    const data = JSON.parse(json).slice(-1)[0];
    const sceneElement = $('#wrapper');

    const workName = data.isPartOf.name.split(' - ');
    const shootId = workName.length > 1 ? workName[0] : null;
    const entryId = url.split('/').slice(-1)[0];
    const title = data.isPartOf ? data.isPartOf.name : data.name;
    const { description } = data;
    const date = moment.utc(data.isPartOf.datePublished, 'YYYY-MM-DD').toDate();

    // const actors = sceneElement.find('.sceneActors a').map((actorIndex, actorElement) => $(actorElement).text().trim()).toArray();
    const actors = data.actor
        .sort(({ gender: genderA }, { gender: genderB }) => {
            if (genderA === 'female' && genderB === 'male') return -1;
            if (genderA === 'male' && genderB === 'female') return 1;

            return 0;
        })
        .map(actor => actor.name);

    const likes = Number(sceneElement.find('.rating .state_1 .value').text());
    const dislikes = Number(sceneElement.find('.rating .state_2 .value').text());

    const duration = moment.duration(data.duration.slice(2)).asSeconds();

    const rawTags = data.keywords.split(', ');
    const tags = await matchTags(rawTags);

    return {
        url,
        shootId,
        entryId,
        title,
        description,
        actors,
        date,
        duration,
        tags,
        rating: {
            likes,
            dislikes,
        },
        site,
    };
}

async function fetchLatest(site, page = 1) {
    const res = await bhttp.get(`https://www.blowpass.com/en/videos/${site.id}/latest/All-Categories/0/All-Pornstars/0/${page}`);

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

async function fetchUpcoming(site) {
    const res = await bhttp.get(`https://www.blowpass.com/en/videos/${site.id}/upcoming`);

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

async function fetchScene(url, site) {
    const res = await bhttp.get(`https://www.blowpass.com/en/video/${site.id}/${new URL(url).pathname.split('/').slice(-2).join('/')}`);

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

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