hln/hln.js

const got = require('got');
const xmlParser = require('fast-xml-parser');
const htmlparser = require('htmlparser2');
const pRetry = require('p-retry');
const categories = [
    'nieuws/binnenland',
    'nieuws/buitenland',
    'bizar',
    'geld',
];

const hln = {};

hln.findRecentArticles = async (category) => {
        const response = await got(`https://www.hln.be/${category}/rss.xml`);
        return xmlParser.parse(response.body).rss.channel.item
            .map((rssItem) => {
                const { title, description, pubDate, link } = rssItem;
                const id = link.split('~a')[1].split('/')[0];
                return {
                    id,
                    title,
                    description,
                    link,
                    pubDate,
                };
            });
    };

hln.findRecentCommentsForArticle = async (articleId) => {
    const time = new Date().getTime();
    const response = await got(`https://www.hln.be/ajax/comments/${articleId}/start/ts_${time}`, { headers: {
            cookie: 'pwv=1; pws=functional|analytics|content_recommendation|targeted_advertising|social_media; ' +
                '_wingify_pc_uuid=333923217b0a447b9b96dbab5bc683dc; gr_reco=167318f92c4-a99ca317c85daf53; ' +
                'gr_rt=167318f92c4-a99ca317c85daf53; ' +
                'gig_hasGmid=accounts; ' +
                'paywall_tracking_id="mVyiKnYBTC-PM_jlbd_fgg=="'
        }});

    let parsing;
    let comment = {};
    const comments = [];
    var parser = new htmlparser.Parser({
        onopentag(name){
            if(name === 'h2') {
                parsing = 'author';
            } else if (name === 'p') {
                parsing = 'comment';
            } else if (name === 'time') {
                parsing = 'time';
            }
        },
        ontext(text){
            if (parsing) {
                comment[parsing] = text;
                if (parsing === 'time') {
                    comments.push(comment);
                    comment = {};
                }
            }
        },
        onclosetag(){
            parsing = undefined;
        }
    }, {decodeEntities: true});

    parser.write(response.body);
    parser.end();
    return comments;
};

async function randomAttempt() {
    const randomCategory = random(categories);

    const recentArticles = await hln.findRecentArticles(randomCategory);

    const randomArticle = random(recentArticles);

    const comments = await hln.findRecentCommentsForArticle(randomArticle.id);
    const randomComment = random(comments);
    if (randomComment) {
        randomComment.article = randomArticle;
        return randomComment;
    }
    throw new Error('Nope');
}


function random(items) {
    return items[Math.floor(Math.random()*items.length)];
}

hln.random = function () {
    const run = () => randomAttempt();
    pRetry(run, { retries: 10}).then((result) => {
        if (result) {
            console.log(`"${result.comment}"`);
            console.log(`--${result.author.toUpperCase()}, reactie op '${result.article.title}'`);
        }
    });
};

module.exports = hln;