site-analyzer/index.js

import { Parser } from "htmlparser2";
import { DomHandler } from "domhandler";
import { readdir, readFile } from "fs/promises";
import { join } from "path";

const { argv } = process;
const dirPath = argv[2] || ".";

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

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

function checkUrl(report, type, el, url) {
  if (isHttp(url)) {
    report.warnings.push(`HTTP/S ${type}: ${getHtml(el)}`);
  } else if (isAbsolute(url)) {
    report.warnings.push(`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;
}

function processFile(content) {
  return new Promise((resolve, reject) => {
    const handler = new DomHandler(
      (error, dom) => {
        if (error) {
          reject(error);
        } else {
          let report = {
            warnings: [],
          };
          for (const el of dom) {
            if (el.type === "tag") {
              recurseElement(report, el);
            }
          }
          resolve(report);
        }
      },
      { withEndIndices: true, withStartIndices: true }
    );
    const parser = new Parser(handler);
    parser.parseComplete(content);
  });
}

async function recurseDirectory(path) {
  const dir = await readdir(path, { withFileTypes: true });
  for (const file of dir) {
    const filePath = join(path, file.name);
    if (file.isDirectory()) recurseDirectory(filePath);
    else {
      if (!file.name.endsWith(".html")) continue;
      const content = await readFile(filePath, "utf-8");
      console.time(filePath);
      await processFile(content);
      console.timeEnd(filePath);
    }
  }
}
await recurseDirectory(dirPath);