site-analyzer/index.js

import { Parser } from "htmlparser2";
import { DomHandler } from "domhandler";

const noop = () => {};
// const log = noop;
const { log } = console;

const stdin = await readStdin();
const rawHtml = stdin;

function recursive(el) {
  if (el.name === "a") {
    if (el.attribs.href) {
      checkUrl("link", el, el.attribs.href);
    } else {
      log("Link with no href:", getHtml(el));
    }
  }
  if (["audio", "video", "img"].includes(el.name)) {
    if (el.attribs.src) {
      checkUrl(el.name, el, el.attribs.src);
    } else {
      log(`${el.name} with no src:`, getHtml(el));
    }
  }

  for (const child of el.children) {
    if (child.type === "tag") {
      recursive(child);
    }
  }
}

function checkUrl(type, el, url) {
  if (isHttp(url)) {
    log(`HTTP/S ${type}:`, getHtml(el));
  } else if (isAbsolute(url)) {
    log(`Absolute ${type}:`, getHtml(el));
  }
}

function isHttp(url) {
  const r = /^(https?:\/\/|\/\/)/;
  return r.test(url);
}
function isAbsolute(url) {
  return url.startsWith("/");
}

function getHtml(el) {
  // return rawHtml.slice(el.startIndex, el.endIndex);
  let text = "";
  for (const child of el.children) {
    if (child.type === "text") text += child.data.trim();
    else if (child.type === "tag") text += getHtml(child);
  }
  return text;
}

const handler = new DomHandler(
  (error, dom) => {
    if (error) {
      // TODO: Handle error
    } else {
      console.time();
      for (const el of dom) {
        if (el.type === "tag") {
          recursive(el);
        }
      }
      console.timeEnd();
    }
  },
  { withEndIndices: true, withStartIndices: true }
);
const parser = new Parser(handler);
parser.parseComplete(rawHtml);

function readStdin() {
  return new Promise((resolve, reject) => {
    let buffer = "";
    process.stdin.resume();
    process.stdin.on("data", (d) => (buffer = buffer.concat(d.toString())));
    process.stdin.on("close", () => resolve(buffer));
  });
}