traxxx/src/scrapers/template.js

'use strict';

/* eslint-disable newline-per-chained-call */
const bhttp = require('bhttp');
const { JSDOM } = require('jsdom');
const moment = require('moment');

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

function scrapeLatest(html, site) {
    const { document } = new JSDOM(html).window;
    const sceneElements = $('.scenes-latest').toArray();

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

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

function scrapeUpcoming(html, site) {
    const { document } = new JSDOM(html).window;
    const sceneElements = $('.scenes-upcoming').toArray();

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

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

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

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

    const rawTags = $('.tags a').map((tagIndex, tagElement) => $(tagElement).text()).toArray();
    const tags = await matchTags(rawTags);

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

async function fetchLatest(site, page = 1) {
    const res = await bhttp.get(`${site.url}/url`);

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

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

    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,
};