MLZ Telegram Deno Bot

script.js

// Lazier Debugging
const log = console.log;

// Constants
const url = 'https://www.muensterlandzeitung.de/';
const botUrl = 'https://api.telegram.org/<bot-token>/sendMessage?chat_id=@muensterlandzeitung&text=';

const historyFile = './history.json';
let history = [];

try { history = JSON.parse(readHistory()); } catch(e) {}

// RegEx's are defined here
const regExArticleHeadersFromHomePage = /<h2 class="article-heading.*?">(?<content>.*?)<\/h2>/gms;
const regExAnchors = /<a title="(?<title>.*?)" href="(?<href>.*?)">/gms;

// Think of this like a main() function
let homePage = fetch('https://www.muensterlandzeitung.de/')
    .then(res => res.text())
    .then(data => {
        let articles = getGroupsFromRegex(data, regExArticleHeadersFromHomePage);

        let links = articles.map(article => {
            return getGroupsFromRegex(article.content, regExAnchors)[0]; // As getGroupsFromRegex returns an array, we only pass the first element from it
        });

        // We need the clean link array
        links = removeDuplicateLinks(links);

        // Loop over links and check against the history
        links.map(link => {
            let boolFoundInHistory = false;

            // Now lets loop over the history and see if any links are not yet in there
            history.map(el => {
                if(link.title === el.title) boolFoundInHistory = true;
            });

            if(!boolFoundInHistory) {
                sendTelegramRequest(link);
                history.push(link); // Put into history so it does not get send again 
            }
        });

        // Now after all that, write the history again
        writeHistory(history);
    });

// Helper functions down here
function getGroupsFromRegex(data, regex) {
    let result = '';
    let resultArray = [];

    while((result = regex.exec(data)) != null) {
        resultArray.push(result.groups); 
    }

    return resultArray;
}

function removeDuplicateLinks(links) {
    let returnArray = [];

    // The easy way out: just throw every second link away
    for(let i = 0; i < links.length; i += 2) {
        returnArray.push(links[i]);
    }

    // Another thing that happens are empty array elements. Remove them here
    return returnArray.filter(el => {
        if(el) return true;
    });
}

// Reads History from JSON
function readHistory() {
    return Deno.readTextFileSync(historyFile);
}

// Writes history into a json
async function writeHistory(data) {
    await Deno.writeTextFileSync(historyFile, JSON.stringify(data));
}

function sendTelegramRequest(link) {
    fetch(botUrl + encodeURI(link.title) + "%20" + encodeURI(url + link.href));
}