doc: improve markdown link checker

- Test link definitions too
- Report all broken links in a file, not just the first one
- Avoid use of workers (launching ~90 workers has substantial
  overhead, and in my 4-core machine it is slower than not using
  them at all) & other simplifications

Refs: #32359

PR-URL: #32586
Reviewed-By: Anna Henningsen <[email protected]>
Reviewed-By: Colin Ihrig <[email protected]>
Reviewed-By: James M Snell <[email protected]>

Author: mildsunrise

tools/doc/checkLinks.js

@@ -1,60 +1,43 @@
 'use strict';
 
 const fs = require('fs');
-const { Worker, isMainThread, workerData: path } = require('worker_threads');
+const { extname, join, resolve } = require('path');
+const unified = require('unified');
+const { pathToFileURL } = require('url');
+const DIR = resolve(process.argv[2]);
+
+console.log('Running Markdown link checker...');
+findMarkdownFilesRecursively(DIR);
 
 function* getLinksRecursively(node) {
-  if (
-    (node.type === 'link' && !node.url.startsWith('#')) ||
-    node.type === 'image'
-  ) {
+  if (node.url && !node.url.startsWith('#')) {
     yield node;
   }
   for (const child of node.children || []) {
     yield* getLinksRecursively(child);
   }
 }
 
-if (isMainThread) {
-  const { extname, join, resolve } = require('path');
-  const DIR = resolve(process.argv[2]);
-
-  console.log('Running Markdown link checker...');
-
-  async function* findMarkdownFilesRecursively(dirPath) {
-    const fileNames = await fs.promises.readdir(dirPath);
-
-    for (const fileName of fileNames) {
-      const path = join(dirPath, fileName);
-
-      const stats = await fs.promises.stat(path);
-      if (
-        stats.isDirectory() &&
-        fileName !== 'api' &&
-        fileName !== 'deps' &&
-        fileName !== 'node_modules'
-      ) {
-        yield* findMarkdownFilesRecursively(path);
-      } else if (extname(fileName) === '.md') {
-        yield path;
-      }
+function findMarkdownFilesRecursively(dirPath) {
+  const entries = fs.readdirSync(dirPath, { withFileTypes: true });
+
+  for (const entry of entries) {
+    const path = join(dirPath, entry.name);
+
+    if (
+      entry.isDirectory() &&
+      entry.name !== 'api' &&
+      entry.name !== 'deps' &&
+      entry.name !== 'node_modules'
+    ) {
+      findMarkdownFilesRecursively(path);
+    } else if (entry.isFile() && extname(entry.name) === '.md') {
+      checkFile(path);
     }
   }
+}
 
-  function errorHandler(error) {
-    console.error(error);
-    process.exitCode = 1;
-  }
-
-  setImmediate(async () => {
-    for await (const path of findMarkdownFilesRecursively(DIR)) {
-      new Worker(__filename, { workerData: path }).on('error', errorHandler);
-    }
-  });
-} else {
-  const unified = require('unified');
-  const { pathToFileURL } = require('url');
-
+function checkFile(path) {
   const tree = unified()
     .use(require('remark-parse'))
     .parse(fs.readFileSync(path));
@@ -63,12 +46,9 @@ if (isMainThread) {
   for (const node of getLinksRecursively(tree)) {
     const targetURL = new URL(node.url, base);
     if (targetURL.protocol === 'file:' && !fs.existsSync(targetURL)) {
-      const error = new Error('Broken link in a Markdown document.');
-      const { start } = node.position;
-      error.stack =
-        error.stack.substring(0, error.stack.indexOf('\n') + 5) +
-        `at ${node.type} (${path}:${start.line}:${start.column})`;
-      throw error;
+      const { line, column } = node.position.start;
+      console.error(`Broken link at ${path}:${line}:${column} (${node.url})`);
+      process.exitCode = 1;
     }
   }
 }