src/search refactor + new endpoint: AI Search Autocomplete (#52822)

.github/workflows/index-autocomplete-elasticsearch.yml → .github/workflows/index-autocomplete-search.yml

@@ -1,7 +1,7 @@
-name: Index autocomplete Elasticsearch
+name: Index autocomplete search in Elasticsearch
 
-# **What it does**: Indexes autocomplete data into Elasticsearch.
-# **Why we have it**: So we can power the API for autocomplete.
+# **What it does**: Indexes autocomplete data (general and AI search) into Elasticsearch.
+# **Why we have it**: So we can power the APIs for autocomplete.
 # **Who does it impact**: docs-engineering
 
 on:
@@ -10,7 +10,7 @@ on:
     - cron: '20 16 * * *' # Run every day at 16:20 UTC / 8:20 PST
   pull_request:
     paths:
-      - .github/workflows/index-autocomplete-elasticsearch.yml
+      - .github/workflows/index-autocomplete-search.yml
       - 'src/search/scripts/index/**'
       - 'package*.json'
 
@@ -40,10 +40,15 @@ jobs:
         if: ${{ github.event_name == 'pull_request' }}
         run: curl --fail --retry-connrefused --retry 5 -I http://localhost:9200
 
-      - name: Run indexing
+      - name: Run general auto-complete indexing
         env:
           ELASTICSEARCH_URL: ${{ github.event_name == 'pull_request' && 'http://localhost:9200' || secrets.ELASTICSEARCH_URL }}
-        run: npm run index -- autocomplete docs-internal-data
+        run: npm run index-general-autocomplete -- docs-internal-data
+
+      - name: Run AI search auto-complete indexing
+        env:
+          ELASTICSEARCH_URL: ${{ github.event_name == 'pull_request' && 'http://localhost:9200' || secrets.ELASTICSEARCH_URL }}
+        run: npm run index-ai-search-autocomplete -- docs-internal-data
 
       - uses: ./.github/actions/slack-alert
         if: ${{ failure() && github.event_name == 'schedule' }}

.github/workflows/sync-search-pr.yml → .github/workflows/index-general-search-pr.yml

@@ -1,6 +1,6 @@
-name: Sync search - PR
+name: Index general search in Elasticsearch on PR
 
-# **What it does**: This does what `sync-sarch-elasticsearch.yml` does but
+# **What it does**: This does what `index-general-search-elasticsearch.yml` does but
 #                   with a localhost Elasticsearch and only for English.
 # **Why we have it**: To test that the script works and the popular pages json is valid.
 # **Who does it impact**: Docs engineering
@@ -11,8 +11,8 @@ on:
     paths:
       - 'src/search/**'
       - 'package*.json'
-      # Ultimately, for debugging this workflow itself
-      - .github/workflows/sync-search-pr.yml
+      # For debugging this workflow
+      - .github/workflows/index-general-search-pr.yml
       # Make sure we run this if the composite action changes
       - .github/actions/setup-elasticsearch/action.yml
 
@@ -25,9 +25,6 @@ concurrency:
   cancel-in-progress: true
 
 env:
-  # Yes, it's hardcoded but it makes all the steps look exactly the same
-  # as they do in `sync-search-elasticsearch.yml` where it uses
-  # that `${{ env.ELASTICSEARCH_URL }}`
   ELASTICSEARCH_URL: http://localhost:9200
   # Since we'll run in NDOE_ENV=production, we need to be explicit that
   # we don't want Hydro configured.
@@ -63,7 +60,7 @@ jobs:
         env:
           ENABLE_DEV_LOGGING: false
         run: |
-          npm run sync-search-server > /tmp/stdout.log 2> /tmp/stderr.log &
+          npm run general-search-scrape-server > /tmp/stdout.log 2> /tmp/stderr.log &
 
           # first sleep to give it a chance to start
           sleep 6
@@ -88,15 +85,13 @@ jobs:
           # let's just accept an empty string instead.
           THROW_ON_EMPTY: false
 
-          # The sync-search-index recognizes this env var if you don't
-          # use the `--docs-internal-data <PATH>` option.
           DOCS_INTERNAL_DATA: docs-internal-data
 
         run: |
           mkdir /tmp/records
-          npm run sync-search-indices -- /tmp/records \
+          npm run general-search-scrape -- /tmp/records \
             --language en \
-            --version dotcom
+            --version fpt
 
           ls -lh /tmp/records
 
@@ -106,9 +101,9 @@ jobs:
 
       - name: Index into Elasticsearch
         run: |
-          npm run index-elasticsearch -- /tmp/records \
+          npm run index-general-search -- /tmp/records \
             --language en \
-            --version dotcom
+            --version fpt
 
       - name: Check created indexes and aliases
         run: |

.github/workflows/sync-search-elasticsearch.yml → .github/workflows/index-general-search.yml

@@ -1,4 +1,4 @@
-name: Sync search Elasticsearch
+name: Index general search in Elasticsearch
 
 # **What it does**: It scrapes the whole site and dumps the records in a
 #                   temp directory. Then it indexes that into Elasticsearch.
@@ -140,7 +140,7 @@ jobs:
         env:
           ENABLE_DEV_LOGGING: false
         run: |
-          npm run sync-search-server > /tmp/stdout.log 2> /tmp/stderr.log &
+          npm run general-search-scrape-server > /tmp/stdout.log 2> /tmp/stderr.log &
 
           # first sleep to give it a chance to start
           sleep 6
@@ -169,13 +169,11 @@ jobs:
           # the same as not set within the script.
           VERSION: ${{ inputs.version }}
 
-          # The sync-search-index recognizes this env var if you don't
-          # use the `--docs-internal-data <PATH>` option.
           DOCS_INTERNAL_DATA: docs-internal-data
 
         run: |
           mkdir /tmp/records
-          npm run sync-search-indices -- /tmp/records \
+          npm run general-search-scrape -- /tmp/records \
             --language ${{ matrix.language }}
 
           ls -lh /tmp/records
@@ -186,12 +184,12 @@ jobs:
 
       - name: Index into Elasticsearch
         env:
-          # Must match what we used when scraping (npm run sync-search-indices)
+          # Must match what we used when scraping (npm run general-search-scrape)
           # otherwise the script will seek other versions from disk that might
           # not exist.
           VERSION: ${{ inputs.version }}
         run: |
-          npm run index-elasticsearch -- /tmp/records \
+          npm run index-general-search -- /tmp/records \
             --language ${{ matrix.language }} \
             --stagger-seconds 5 \
             --retries 5

.gitignore

@@ -51,3 +51,9 @@ assets/images/help/writing/unordered-list-rendered (1).png
 
 # Used by precompute-pageinfo
 .pageinfo-cache.json.br
+
+# Cloned and used for indexing Elasticsearch data
+docs-internal-data/
+
+# For intermediate data (like scraping for Elasticsearch indexing)
+tmp/

package.json

@@ -17,7 +17,7 @@
   "exports": "./src/frame/server.ts",
   "scripts": {
     "all-documents": "tsx src/content-render/scripts/all-documents/cli.ts",
-    "analyze-text": "node src/search/scripts/analyze-text.js",
+    "analyze-text": "tsx src/search/scripts/analyze-text.ts",
     "analyze-comment": "tsx src/events/scripts/analyze-comment-cli.ts",
     "archive-version": "tsx --max-old-space-size=16384 src/ghes-releases/scripts/archive-version.ts",
     "audit-log-sync": "tsx src/audit-logs/scripts/sync.ts",
@@ -39,8 +39,14 @@
     "find-unused-variables": "tsx src/content-linter/scripts/find-unsed-variables.ts",
     "fixture-dev": "cross-env ROOT=src/fixtures/fixtures npm start",
     "fixture-test": "cross-env ROOT=src/fixtures/fixtures npm test -- src/fixtures/tests",
-    "index": "tsx src/search/scripts/index/index.ts",
-    "index-elasticsearch": "node src/search/scripts/index-elasticsearch.js",
+    "general-search-scrape": "tsx src/search/scripts/scrape/scrape-cli.ts",
+    "general-search-scrape-server": "cross-env NODE_ENV=production PORT=4002 MINIMAL_RENDER=true CHANGELOG_DISABLED=true tsx src/frame/server.ts",
+    "ghes-release-scrape-with-server": "cross-env GHES_RELEASE=1 start-server-and-test general-search-scrape-server 4002 general-search-scrape",
+    "general-search-scrape-with-server": "cross-env NODE_OPTIONS='--max_old_space_size=8192' start-server-and-test general-search-scrape-server 4002 general-search-scrape",
+    "index": "tsx src/search/scripts/index/index-cli autocomplete docs-internal-data",
+    "index-ai-search-autocomplete": "tsx src/search/scripts/index/index-cli ai-search-autocomplete",
+    "index-general-autocomplete": "tsx src/search/scripts/index/index-cli general-autocomplete",
+    "index-general-search": "tsx src/search/scripts/index/index-cli general-search",
     "index-test-fixtures": "./src/search/scripts/index-test-fixtures.sh",
     "lint": "eslint '**/*.{js,mjs,ts,tsx}'",
     "lint-content": "node src/content-linter/scripts/lint-content.js",
@@ -70,10 +76,6 @@
     "start-for-playwright": "cross-env ROOT=src/fixtures/fixtures TRANSLATIONS_FIXTURE_ROOT=src/fixtures/fixtures/translations ENABLED_LANGUAGES=en,ja NODE_ENV=test tsx src/frame/server.ts",
     "symlink-from-local-repo": "node src/early-access/scripts/symlink-from-local-repo.js",
     "sync-rest": "tsx src/rest/scripts/update-files.ts",
-    "sync-search": "cross-env NODE_OPTIONS='--max_old_space_size=8192' start-server-and-test sync-search-server 4002 sync-search-indices",
-    "sync-search-ghes-release": "cross-env GHES_RELEASE=1 start-server-and-test sync-search-server 4002 sync-search-indices",
-    "sync-search-indices": "node src/search/scripts/sync-search-indices.js",
-    "sync-search-server": "cross-env NODE_ENV=production PORT=4002 MINIMAL_RENDER=true CHANGELOG_DISABLED=true tsx src/frame/server.ts",
     "sync-secret-scanning": "tsx src/secret-scanning/scripts/sync.ts",
     "sync-webhooks": "npx tsx src/rest/scripts/update-files.ts -o webhooks",
     "test": "vitest",
@@ -222,6 +224,7 @@
       "src/open-source/scripts/add-pr-links.js",
       "src/open-source/scripts/pr-link-source.js",
       "rest-api-description/",
+      "docs-internal-data/",
       "src/code-scanning/scripts/generate-code-scanning-query-list.ts"
     ]
   },
@@ -327,10 +330,13 @@
     "@octokit/rest": "21.0.2",
     "@playwright/test": "^1.48.1",
     "@types/accept-language-parser": "1.5.6",
+    "@types/cheerio": "^0.22.35",
     "@types/connect-datadog": "0.0.10",
     "@types/connect-timeout": "0.0.39",
     "@types/cookie": "0.6.0",
     "@types/cookie-parser": "1.4.7",
+    "@types/elasticsearch": "^5.0.43",
+    "@types/event-to-promise": "^0.7.5",
     "@types/express": "4.17.21",
     "@types/imurmurhash": "^0.1.4",
     "@types/js-cookie": "^3.0.6",

src/fixtures/tests/breadcrumbs.ts

@@ -68,7 +68,7 @@ describe('breadcrumbs', () => {
 
     expect($breadcrumbTitles.length).toBe(0)
     expect($breadcrumbLinks.length).toBe(2)
-    expect($breadcrumbLinks[0].attribs.title).toBe('Deeper secrets')
-    expect($breadcrumbLinks[1].attribs.title).toBe('Mariana Trench')
+    expect(($breadcrumbLinks[0] as cheerio.TagElement).attribs.title).toBe('Deeper secrets')
+    expect(($breadcrumbLinks[1] as cheerio.TagElement).attribs.title).toBe('Mariana Trench')
   })
 })

src/frame/middleware/api.ts

@@ -3,7 +3,7 @@ import { createProxyMiddleware } from 'http-proxy-middleware'
 
 import events from '@/events/middleware.js'
 import anchorRedirect from '@/rest/api/anchor-redirect.js'
-import search from '@/search/middleware/search.js'
+import search from '@/search/middleware/search-routes.js'
 import pageInfo from '@/pageinfo/middleware'
 import pageList from '@/pagelist/middleware'
 import webhooks from '@/webhooks/middleware/webhooks.js'

src/frame/middleware/index.ts

@@ -61,7 +61,7 @@ import fastlyCacheTest from './fastly-cache-test'
 import trailingSlashes from './trailing-slashes'
 import mockVaPortal from './mock-va-portal'
 import dynamicAssets from '@/assets/middleware/dynamic-assets'
-import contextualizeSearch from '@/search/middleware/contextualize.js'
+import generalSearchMiddleware from '@/search/middleware/general-search-middleware'
 import shielding from '@/shielding/middleware'
 import tracking from '@/tracking/middleware'
 import { MAX_REQUEST_TIMEOUT } from '@/frame/lib/constants.js'
@@ -275,7 +275,7 @@ export default function (app: Express) {
   app.use(asyncMiddleware(productExamples))
   app.use(asyncMiddleware(productGroups))
   app.use(asyncMiddleware(glossaries))
-  app.use(asyncMiddleware(contextualizeSearch))
+  app.use(asyncMiddleware(generalSearchMiddleware))
   app.use(asyncMiddleware(featuredLinks))
   app.use(asyncMiddleware(learningTrack))
 

src/frame/tests/favicons.ts

@@ -15,7 +15,10 @@ describe('favicon assets', () => {
     expect(res.headers['cache-control']).toContain('public')
     expect(res.headers['cache-control']).toContain('immutable')
     expect(res.headers['cache-control']).toMatch(/max-age=\d+/)
-    const maxAgeSeconds = parseInt(res.headers['cache-control'].match(/max-age=(\d+)/)[1], 10)
+    const maxAgeSeconds = parseInt(
+      (res.headers['cache-control'] || '').match(/max-age=(\d+)/)?.[1] || '',
+      10,
+    )
     // Let's not be too specific in the tests, just as long as it's testing
     // that it's a reasonably large number of seconds.
     expect(maxAgeSeconds).toBeGreaterThanOrEqual(60 * 60)
@@ -25,13 +28,16 @@ describe('favicon assets', () => {
   test('should serve a valid and aggressively caching /apple-touch-icon.png', async () => {
     const res = await get('/apple-touch-icon.png')
     expect(res.statusCode).toBe(200)
-    expect(parseInt(res.headers['content-length'], 10)).toBeGreaterThan(0)
+    expect(parseInt(res.headers['content-length'] || '', 10)).toBeGreaterThan(0)
     expect(res.headers['content-type']).toBe('image/png')
     expect(res.headers['set-cookie']).toBeUndefined()
     expect(res.headers['cache-control']).toContain('public')
     expect(res.headers['cache-control']).toContain('immutable')
     expect(res.headers['cache-control']).toMatch(/max-age=\d+/)
-    const maxAgeSeconds = parseInt(res.headers['cache-control'].match(/max-age=(\d+)/)[1], 10)
+    const maxAgeSeconds = parseInt(
+      (res.headers['cache-control'] || '').match(/max-age=(\d+)/)?.[1] || '',
+      10,
+    )
     // Let's not be too specific in the tests, just as long as it's testing
     // that it's a reasonably large number of seconds.
     expect(maxAgeSeconds).toBeGreaterThanOrEqual(60 * 60)

src/frame/tests/manifest.ts

@@ -20,6 +20,9 @@ describe('manifest', () => {
   test('download manifest from HTML and check content', async () => {
     const $ = await getDOM('/')
     const url = $('link[rel="manifest"]').attr('href')
+    if (!url) {
+      throw new Error('No manifest URL found')
+    }
     const res = await get(url)
     expect(res.statusCode).toBe(200)
 

src/ghes-releases/lib/release-templates/release-steps-1.md

@@ -17,7 +17,7 @@ labels:
 - [Prerequisites](#prerequisites)
 - [Create publication branch for a new version of GHES](#creation)
 - [Resolve check failures](#check-failures)
-- [Sync the search indices](#sync-search-indices)
+- [Scrape the search indices](#scrape-search-indices)
 - [Maintain the publication branch](#maintenance)
 - [Complete preparation for the RC and publish the docset](#publication)
 
@@ -110,11 +110,11 @@ For content from the OpenAPI schema, note the affected content with broken links
 
 <br/>
 
-<a name="sync-search-indices">
+<a name="scrape-search-indices">
 
-### [🔎](#sync-search-indices) Sync the search indices
+### [🔎](#scrape-search-indices) Scrape the search indices
 
-1. Go to the [`sync-search-elasticsearch` workflow](https://github.com/github/docs-internal/actions/workflows/sync-search-elasticsearch.yml) ([permalink](https://github.com/github/docs-internal/blob/f8ca45703c48c7d1976a278337bc3391fb14fe9e/.github/workflows/sync-search-elasticsearch.yml) in case it moves)
+1. Go to the [`index-general-search.yml` workflow](https://github.com/github/docs-internal/actions/workflows/index-general-search.yml)
 1. Click on the **Run workflow** drop down and set the following parameters:
     - `Branch:` set to the name of the publication branch
     - `Version` set to the version you're publishing (e.g., `ghes-3.12` if you're publishing GHES 3.12)