speed up repository reference checks (#23178)

* speed up repository reference checks

Part of #1246

* feedbacked

* microoptimization

Author: Peter Bengtsson

tests/meta/repository-references.js

@@ -1,5 +1,6 @@
+import fs from 'fs'
+
 import walkSync from 'walk-sync'
-import readFileAsync from '../../lib/readfile-async.js'
 import minimatch from 'minimatch'
 
 /*
@@ -75,6 +76,7 @@ const REPO_REGEXP = /\/\/github\.com\/github\/(?!docs[/'"\n])([\w-.]+)/gi
 const IGNORE_PATHS = [
   '.git',
   '.next',
+  '.vscode', // Not part of the repo but could be for a developer locally
   'node_modules',
   'translations',
   '**/*.png', // Do not check images or font files.
@@ -83,25 +85,46 @@ const IGNORE_PATHS = [
   '**/*.pdf',
   '**/*.ico',
   '**/*.woff',
+  '**/*.csv',
+  '**/*.br', // E.g. the search index .json.br files
+  '**/*.graphql', // E.g. data/graphql/ghec/schema.docs.graphql
+  'package-lock.json', // At the time of writing it's 1.5MB!
+  '.linkinator/full.log', // Only present if you've run linkinator
+  'lib/search/popular-pages.json', // used to build search indexes
+  'tests/**/*.json',
 
   'content/early-access', // Not committed to public repository.
   'data/early-access', // Not committed to public repository.
   'data/release-notes', // These include links to many internal issues in Liquid comments.
+  'lib/redirects/.redirects-cache*',
 ]
 
 describe('check if a GitHub-owned private repository is referenced', () => {
   const filenames = walkSync(process.cwd(), {
     directories: false,
     ignore: IGNORE_PATHS,
-  })
+  }).filter(
+    (filename) =>
+      // Skip the large static json files because they're not code.
+      !(
+        filename.includes('static') &&
+        (filename.endsWith('.json') || filename.endsWith('.json.br'))
+      )
+  )
 
-  test.each(filenames)('in file %s', async (filename) => {
-    const file = await readFileAsync(filename, 'utf8')
-    const allowDocs = ALLOW_DOCS_PATHS.some((path) => minimatch(filename, path))
+  test.each(filenames)('in file %s', (filename) => {
+    // When you're reading many small files, it's faster to do it
+    // *synchronously* because the event-loop overhead is less since
+    // the disk I/O is sufficiently small.
+    const file = fs.readFileSync(filename, 'utf8')
     const matches = Array.from(file.matchAll(REPO_REGEXP))
       .map(([, repoName]) => repoName)
       .filter((repoName) => !PUBLIC_REPOS.has(repoName))
-      .filter((repoName) => !(allowDocs && repoName.startsWith('docs')))
+      .filter((repoName) => {
+        return !(
+          repoName.startsWith('docs') && ALLOW_DOCS_PATHS.some((path) => minimatch(filename, path))
+        )
+      })
     expect(
       matches,
       `Please edit ${filename} to remove references to ${matches.join(', ')}`