extract-hrefs/dist/index.mjs

This library is used to extract hrefs from a given HTML string.

// src/index.ts
import * as cheerio from "cheerio";
import normalizeUrl from "normalize-url";
var normalizeUrls = (links) => {
  return links.map((link) => {
    try {
      return normalizeUrl(link, { stripWWW: false });
    } catch (error) {
      console.error(`Error normalizing URL: ${link}`, error);
      return link;
    }
  });
};
var extractHrefs = (htmlString) => {
  if (!htmlString || typeof htmlString !== "string") {
    throw new Error("HTML string is required");
  }
  const $ = cheerio.load(htmlString);
  const urls = [];
  $("a").each((index, element) => {
    const href = $(element).attr("href");
    if (href) {
      urls.push(href);
    }
  });
  return normalizeUrls(urls);
};
var index_default = extractHrefs;
export {
  index_default as default
};
//# sourceMappingURL=index.mjs.map