Repo sync

.env.example

@@ -20,4 +20,9 @@ BUILD_RECORDS_MAX_CONCURRENT=100
 BUILD_RECORDS_MIN_TIME=
 
 # Set to true to enable the /fastly-cache-test route for debugging Fastly headers
-ENABLE_FASTLY_TESTING=
+ENABLE_FASTLY_TESTING=
+
+# Needed to auth for AI search
+CSE_COPILOT_SECRET=
+CSE_COPILOT_ENDPOINT=https://cse-copilot-staging.service.iad.github.net
+

.github/workflows/codeowners-content-strategy.yml → .github/workflows/codeowners-content-systems.yml

@@ -11,14 +11,14 @@ on:
       - 'content/contributing/**.md'
 
 jobs:
-  codeowners-content-strategy:
+  codeowners-content-systems:
     if: ${{ github.repository == 'github/docs-internal' }}
     runs-on: ubuntu-latest
     steps:
       - name: Check out repo
         uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # v4.1.1
 
-      - name: Add Content Strategy as a reviewer
+      - name: Add Content Systems as a reviewer
         env:
           GH_TOKEN: ${{ secrets.DOCS_BOT_PAT_WRITEORG_PROJECT }}
           PR: ${{ github.event.pull_request.html_url }}
@@ -29,5 +29,5 @@ jobs:
           )
           if ! $has_reviewer
           then
-            gh pr edit $PR --add-reviewer github/docs-content-strategy
+            gh pr edit $PR --add-reviewer github/docs-content-systems
           fi

package.json

@@ -33,7 +33,7 @@
     "delete-orphan-translation-files": "tsx src/workflows/delete-orphan-translation-files.ts",
     "deleted-features-pr-comment": "tsx src/data-directory/scripts/deleted-features-pr-comment.ts",
     "dev": "cross-env npm start",
-    "find-orphaned-assets": "node src/assets/scripts/find-orphaned-assets.js",
+    "find-orphaned-assets": "tsx src/assets/scripts/find-orphaned-assets.ts",
     "find-orphaned-features": "tsx src/data-directory/scripts/find-orphaned-features/index.ts",
     "find-past-built-pr": "tsx src/workflows/find-past-built-pr.ts",
     "find-unused-variables": "tsx src/content-linter/scripts/find-unsed-variables.ts",
@@ -259,7 +259,7 @@
     "express": "4.21.1",
     "express-rate-limit": "7.4.0",
     "fastest-levenshtein": "1.0.16",
-    "file-type": "19.4.1",
+    "file-type": "19.6.0",
     "flat": "^6.0.1",
     "github-slugger": "^2.0.0",
     "glob": "11.0.0",

src/assets/lib/image-density.d.ts

@@ -0,0 +1 @@
+export const IMAGE_DENSITY: Record<string, string>

src/assets/scripts/deleted-assets-pr-comment-1.js → src/assets/scripts/deleted-assets-pr-comment-1.ts

@@ -18,14 +18,19 @@
 // [end-readme]
 
 import { program } from 'commander'
-import main from './deleted-assets-pr-comment.js'
+import main from './deleted-assets-pr-comment'
 
 program
   .description('If applicable, print a snippet of Markdown about deleted assets')
-  .arguments('owner repo base_sha head_sha', 'Simulate what the Actions workflow does')
+  .arguments('owner repo base_sha head_sha')
   .parse(process.argv)
 
-const opts = program.opts()
-const args = program.args
+type MainArgs = {
+  owner: string
+  repo: string
+  baseSHA: string
+  headSHA: string
+}
+const opts = program.opts() as MainArgs
 
-console.log(await main(...args, { ...opts }))
+console.log(await main(opts))

src/assets/scripts/deleted-assets-pr-comment.js → src/assets/scripts/deleted-assets-pr-comment.ts

@@ -13,16 +13,22 @@ if (!GITHUB_TOKEN) {
 // When this file is invoked directly from action as opposed to being imported
 if (import.meta.url.endsWith(process.argv[1])) {
   const owner = context.repo.owner
-  const repo = context.payload.repository.name
-  const baseSHA = context.payload.pull_request.base.sha
-  const headSHA = context.payload.pull_request.head.sha
+  const repo = context.payload.repository?.name || ''
+  const baseSHA = context.payload.pull_request?.base.sha
+  const headSHA = context.payload.pull_request?.head.sha
 
-  const markdown = await main(owner, repo, baseSHA, headSHA)
+  const markdown = await main({ owner, repo, baseSHA, headSHA })
   core.setOutput('markdown', markdown)
 }
 
-async function main(owner, repo, baseSHA, headSHA) {
-  const octokit = github.getOctokit(GITHUB_TOKEN)
+type MainArgs = {
+  owner: string
+  repo: string
+  baseSHA: string
+  headSHA: string
+}
+async function main({ owner, repo, baseSHA, headSHA }: MainArgs) {
+  const octokit = github.getOctokit(GITHUB_TOKEN as string)
   // get the list of file changes from the PR
   const response = await octokit.rest.repos.compareCommitsWithBasehead({
     owner,
@@ -32,6 +38,10 @@ async function main(owner, repo, baseSHA, headSHA) {
 
   const { files } = response.data
 
+  if (!files) {
+    throw new Error('No files found in the PR')
+  }
+
   const oldFilenames = []
   for (const file of files) {
     const { filename, status } = file

src/assets/scripts/find-orphaned-assets.js → src/assets/scripts/find-orphaned-assets.ts

@@ -32,7 +32,7 @@ const EXCEPTIONS = new Set([
   'assets/images/site/apple-touch-icon-76x76.png',
 ])
 
-function isExceptionPath(imagePath) {
+function isExceptionPath(imagePath: string) {
   // We also check for .DS_Store because any macOS user that has opened
   // a folder with images will have this on disk. It won't get added
   // to git anyway thanks to our .DS_Store.
@@ -53,9 +53,15 @@ program
   .option('--exclude-translations', "Don't search in translations/")
   .parse(process.argv)
 
-main(program.opts(), program.args)
+type MainOptions = {
+  json: boolean
+  verbose: boolean
+  exit: boolean
+  excludeTranslations: boolean
+}
+main(program.opts())
 
-async function main(opts) {
+async function main(opts: MainOptions) {
   const { json, verbose, exit, excludeTranslations } = opts
 
   const walkOptions = {
@@ -164,7 +170,7 @@ async function main(opts) {
   }
 }
 
-function getTotalDiskSize(filePaths) {
+function getTotalDiskSize(filePaths: Set<string>) {
   let sum = 0
   for (const filePath of filePaths) {
     sum += fs.statSync(filePath).size

src/assets/scripts/list-image-sizes.js → src/assets/scripts/list-image-sizes.ts

@@ -10,32 +10,26 @@ import { fileURLToPath } from 'url'
 import path from 'path'
 import walk from 'walk-sync'
 import sharp from 'sharp'
-import { chain } from 'lodash-es'
 const __dirname = path.dirname(fileURLToPath(import.meta.url))
 
 const imagesPath = path.join(__dirname, '../assets/images')
 const imagesExtensions = ['.jpg', '.jpeg', '.png', '.gif']
 
-const files = chain(walk(imagesPath, { directories: false })).filter((relativePath) => {
+const files = walk(imagesPath, { directories: false }).filter((relativePath) => {
   return imagesExtensions.includes(path.extname(relativePath.toLowerCase()))
 })
-const infos = await Promise.all(
+const images = await Promise.all(
   files.map(async (relativePath) => {
     const fullPath = path.join(imagesPath, relativePath)
     const image = sharp(fullPath)
     const { width, height } = await image.metadata()
-    const size = width * height
+    const size = (width || 0) * (height || 0)
     return { relativePath, width, height, size }
   }),
 )
-const images = files
-  .map((relativePath, i) => {
-    return { relativePath, ...infos[i] }
+images
+  .sort((a, b) => b.size - a.size)
+  .forEach((image) => {
+    const { relativePath, width, height } = image
+    console.log(`${width} x ${height} - ${relativePath}`)
   })
-  .orderBy('size', 'desc')
-  .value()
-
-images.forEach((image) => {
-  const { relativePath, width, height } = image
-  console.log(`${width} x ${height} - ${relativePath}`)
-})

src/assets/scripts/validate-asset-images.js → src/assets/scripts/validate-asset-images.ts

@@ -19,6 +19,7 @@ import path from 'path'
 import { program } from 'commander'
 import chalk from 'chalk'
 import cheerio from 'cheerio'
+// @ts-ignore see https://github.com/sindresorhus/file-type/issues/652
 import { fileTypeFromFile } from 'file-type'
 import walk from 'walk-sync'
 import isSVG from 'is-svg'
@@ -43,7 +44,7 @@ const EXPECT = {
   '.ico': 'image/x-icon',
   '.pdf': 'application/pdf',
   '.webp': 'image/webp',
-}
+} as Record<string, string>
 
 const CRITICAL = 'critical'
 const WARNING = 'warning'
@@ -56,7 +57,7 @@ program
 
 main(program.opts())
 
-async function main(opts) {
+async function main(opts: { dryRun: boolean; verbose: boolean }) {
   let errors = 0
 
   const files = walk(ASSETS_ROOT, { includeBasePath: true, directories: false }).filter(
@@ -71,7 +72,11 @@ async function main(opts) {
       )
     },
   )
-  const results = (await Promise.all(files.map(checkFile))).filter(Boolean)
+  const results = (await Promise.all(files.map(checkFile))).filter(Boolean) as [
+    level: string,
+    filePath: string,
+    error: string,
+  ][]
   for (const [level, filePath, error] of results) {
     console.log(
       level === CRITICAL ? chalk.red(level) : chalk.yellow(level),
@@ -94,7 +99,7 @@ async function main(opts) {
   process.exitCode = errors
 }
 
-async function checkFile(filePath) {
+async function checkFile(filePath: string) {
   const ext = path.extname(filePath)
 
   const { size } = await fs.stat(filePath)
@@ -113,7 +118,7 @@ async function checkFile(filePath) {
     }
     try {
       checkSVGContent(content)
-    } catch (error) {
+    } catch (error: any) {
       return [CRITICAL, filePath, error.message]
     }
   } else if (EXPECT[ext]) {
@@ -135,15 +140,15 @@ async function checkFile(filePath) {
   // All is well. Nothing to complain about.
 }
 
-function checkSVGContent(content) {
+function checkSVGContent(content: string) {
   const $ = cheerio.load(content)
   const disallowedTagNames = new Set(['script', 'object', 'iframe', 'embed'])
   $('*').each((i, element) => {
-    const { tagName } = element
+    const { tagName } = $(element).get(0)
     if (disallowedTagNames.has(tagName)) {
       throw new Error(`contains a <${tagName}> tag`)
     }
-    for (const key in element.attribs) {
+    for (const key in $(element).get(0).attribs) {
       // Looks for suspicious event handlers on tags.
       // For example `<path oNload="alert(1)"" d="M28 0l4.59 4.59-9.76`
       // We don't need to do a case-sensitive regex here because cheerio

src/assets/tests/static-assets.js → src/assets/tests/static-assets.ts

@@ -6,7 +6,7 @@ import { describe, expect, test, vi } from 'vitest'
 import { get } from '#src/tests/helpers/e2etest.js'
 import { checkCachingHeaders } from '#src/tests/helpers/caching-headers.js'
 
-function getNextStaticAsset(directory) {
+function getNextStaticAsset(directory: string) {
   const root = path.join('.next', 'static', directory)
   const files = fs.readdirSync(root)
   if (!files.length) throw new Error(`Can't find any files in ${root}`)

src/frame/middleware/api.ts

@@ -3,6 +3,7 @@ import { createProxyMiddleware } from 'http-proxy-middleware'
 
 import events from '@/events/middleware.js'
 import anchorRedirect from '@/rest/api/anchor-redirect.js'
+import aiSearch from '@/search/middleware/ai-search'
 import search from '@/search/middleware/search-routes.js'
 import pageInfo from '@/pageinfo/middleware'
 import pageList from '@/pagelist/middleware'
@@ -23,6 +24,23 @@ router.use('/pagelist', pageList)
 // local laptop, they don't have an Elasticsearch. Neither a running local
 // server or the known credentials to a remote Elasticsearch. Whenever
 // that's the case, they can just HTTP proxy to the production server.
+if (process.env.CSE_COPILOT_ENDPOINT || process.env.NODE_ENV === 'test') {
+  router.use('/ai-search', aiSearch)
+} else {
+  console.log(
+    'Proxying AI Search requests to docs.github.com. To use the cse-copilot endpoint, set the CSE_COPILOT_ENDPOINT environment variable.',
+  )
+  router.use(
+    '/ai-search',
+    createProxyMiddleware({
+      target: 'https://docs.github.com',
+      changeOrigin: true,
+      pathRewrite: function (path, req: ExtendedRequest) {
+        return req.originalUrl
+      },
+    }),
+  )
+}
 if (process.env.ELASTICSEARCH_URL) {
   router.use('/search', search)
 } else {