pyseoa-ts/dist/crawler/extractLinks.js

SEO analysis tools for the web, ported from pyseoa (Python) to TypeScript

import * as cheerio from 'cheerio';
export function extractLinks(html, baseUrl) {
    const $ = cheerio.load(html);
    const base = new URL(baseUrl);
    const links = [];
    $("a[href]").each((_, el) => {
        let href = $(el).attr("href");
        if (!href)
            return;
        try {
            const url = new URL(href, base);
            if (url.hostname === base.hostname) {
                links.push(url.toString());
            }
        }
        catch {
            // ignore malformed URLs
        }
    });
    return [...new Set(links)];
}