sisa-crawl/coursesEnrolled.js

108 lines
4.2 KiB
JavaScript

const puppeteer = require('puppeteer');
const IFRAME_SELECTOR = 'iframe[title="Hoofdinhoud"]';
const wait = require("./wait");
const config = require("./config.js");
//Er zijn een aantal helperfuncties die nog verwijderd moeten worden
//zo is de functie van de contentFrame te verwijderen want staat dubbel in evaluatestartpakket.js
//de iframeselector is eveneens van belang om te verwijderen want staat ook dubbel in evaluatestartpakket.js
async function coursesEnrolled(page) {
/*connection to local host
const browser = await puppeteer.connect({
browserURL: "http://localhost:9222", // Connect to the browser's debugging address
});
// Get all open pages (tabs)
const pages = await browser.pages();
console.log(`Found ${pages.length} open pages on the browser.`);
// Interact with the last tab (or any tab of choice)
let page = pages[pages.length - 1]; // Choose the last opened tab
console.log("Current URL:", page.url()); */
let contentFrame= await getContentFrame(page);
const uitvouwenButton = await contentFrame.$(config.uitvouwenSelector);
if (uitvouwenButton) {
await uitvouwenButton.click();
console.log("uitvouwen button clicked");
await wait(2000);
contentFrame= await getContentFrame(page)
}
//logic to collapse all sections except the one with the excluded title
const collapseArrows = await contentFrame.$$(config.collapsArrowSelector);
const excludedArrow = await contentFrame.$(config.excludeFromCollapsingSelector);
const excludedTbody = await excludedArrow.evaluate(el => el.closest('tbody'));
for (const arrow of collapseArrows) {
const isExcluded = await arrow.evaluate((el, excludedTbody) => {
const title = el.getAttribute('title');
const tbody = el.closest('tbody');
return title === 'Sectie samenvouwen AFSTUDEERVEREISTEN BACHELOR GESCHIEDENIS' || tbody === excludedTbody;
}, excludedTbody);
if (!isExcluded) {
const isStillInDOM = await arrow.evaluate(el => document.body.contains(el));
if (isStillInDOM) {
await arrow.click();
await wait(500); // Adjust the wait time as needed
}
}
}
contentFrame= await getContentFrame(page)
const rowData = await contentFrame.$$eval(
'span',
(spans) => {
const uniqueEntries = new Set();
return spans
.filter((span) => span.innerText.trim().includes('Academiejaar'))
.map((span) => {
const tr = span.closest('tr');
if (!tr) return null;
const tds = tr.querySelectorAll('td');
const getInnerHtml = (element) => {
const aElement = element.querySelector('a');
return aElement ? aElement.innerHTML.trim() : element.innerHTML.trim();
};
const studiegidsNummer = tds[0] ? getInnerHtml(tds[0].querySelector('span')) : '';
const opleidingsOnderdeel = tds[1] ? getInnerHtml(tds[1].querySelector('span')) : '';
const studiepunten = tds[2] ? getInnerHtml(tds[2].querySelector('span')) : '';
if (uniqueEntries.has(studiegidsNummer)) return null;
uniqueEntries.add(studiegidsNummer);
return {
studiegidsNummer,
opleidingsOnderdeel,
studiepunten,
};
})
.filter((entry) => entry);
}
);
const totalStudiepunten = sumStudiepunten(rowData);
return {rowData, totalStudiepunten};
}
function sumStudiepunten(rowData) {
return rowData.reduce((total, row) => {
const studiepunten = parseFloat(row.studiepunten) || 0;
return total + studiepunten;
}, 0);
}
async function getContentFrame(page) {
await page.waitForSelector(IFRAME_SELECTOR);
const iframe = await page.$(IFRAME_SELECTOR);
if (!iframe) {
throw new Error("Could not find iframe on the AA-page");
}
console.log("iframe found on AA-page");
return await iframe.contentFrame();
}
module.exports = coursesEnrolled;