traxxx/src/scrapers/julesjordan.js

'use strict';

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

const tagMap = {
    Anal: 'anal',
    Asian: 'asian',
    'Ass To Mouth': 'ATM',
    'Big Cocks': 'big cock',
    Black: 'BBC',
    Blondes: 'blonde',
    Brunettes: 'brunette',
    Blowjobs: 'blowjob',
    Creampie: 'creampie',
    'Deep Throat': 'deepthroat',
    Facial: 'facial',
    Interracial: 'interracial',
    Lingerie: 'lingerie',
    Natural: 'natural',
    'Red Head': 'readhead',
    'School Girl': 'schoolgirl',
    Tattoo: 'tattoo',
    Teen: 'teen',
};

function scrapeLatest(html, site) {
    const $ = cheerio.load(html, { normalizeWhitespace: true });
    const scenesElements = $('.update_details').toArray();

    return scenesElements.map((element) => {
        const sceneLinkElement = $(element).children('a').eq(1);
        const url = sceneLinkElement.attr('href');
        const title = sceneLinkElement.text();

        const date = moment
            .utc($(element).find('.update_date').text(), 'MM/DD/YYYY')
            .toDate();

        const actors = $(element).find('.update_models a')
            .map((actorIndex, actorElement) => $(actorElement).text())
            .toArray();

        return {
            url,
            title,
            actors,
            date,
            site,
        };
    });
}

function scrapeUpcoming(html, site) {
    const $ = cheerio.load(html, { normalizeWhitespace: true });
    const scenesElements = $('#coming_soon_carousel').find('.table').toArray();

    return scenesElements.map((element) => {
        const details = $(element).find('.update_details_comingsoon')
            .eq(1)
            .children()
            .remove();

        const title = details
            .end()
            .text()
            .trim();

        const actors = details
            .text()
            .trim()
            .split(', ');

        const date = moment
            .utc($(element).find('.update_date_comingsoon').text().slice(7), 'MM/DD/YYYY')
            .toDate();

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

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

    const title = $('.title_bar_hilite').text();
    const date = moment
        .utc($('.update_date').text(), 'MM/DD/YYYY')
        .toDate();

    const actors = $('.update_description + .update_models a')
        .map((_actorIndex, actorElement) => $(actorElement).text())
        .toArray();

    const description = $('.update_description').text().trim();

    const stars = Number($('.avg_rating').text().trim().replace(/[\s|Avg Rating:]/g, ''));

    const rawTags = $('.update_tags a').map((tagIndex, tagElement) => $(tagElement).text()).toArray();
    const tags = rawTags.reduce((accTags, tag) => (tagMap[tag] ? [...accTags, tagMap[tag]] : accTags), []);

    return {
        url,
        title,
        date,
        actors,
        description,
        rating: {
            stars,
        },
        tags,
        site,
    };
}

async function fetchLatest(site) {
    const res = await bhttp.get(`${site.url}/trial/categories/movies_1_d.html`);

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

async function fetchUpcoming(site) {
    const res = await bhttp.get(`${site.url}/trial/index.php`);

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

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

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

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