custom Fastly surrogate key for api search (#32536)

.github/workflows/sync-search-elasticsearch.yml

@@ -126,6 +126,13 @@ jobs:
           curl --retry-connrefused --retry 5 ${{ env.ELASTICSEARCH_URL }}/_cat/indices?v
           curl --retry-connrefused --retry 5 ${{ env.ELASTICSEARCH_URL }}/_cat/indices?v
 
+      - name: Purge Fastly edge cache
+        env:
+          FASTLY_TOKEN: ${{ secrets.FASTLY_TOKEN }}
+          FASTLY_SERVICE_ID: ${{ secrets.FASTLY_SERVICE_ID }}
+          FASTLY_SURROGATE_KEY: api-search:${{ matrix-language }}
+        run: .github/actions-scripts/purge-fastly-edge-cache.js
+
       - name: Send Slack notification if workflow fails
         uses: someimportantcompany/github-actions-slack-message@f8d28715e7b8a4717047d23f48c39827cacad340
         if: failure()

middleware/api/search.js

@@ -5,14 +5,22 @@ import FailBot from '../../lib/failbot.js'
 import languages from '../../lib/languages.js'
 import { allVersions } from '../../lib/all-versions.js'
 import statsd from '../../lib/statsd.js'
-import { defaultCacheControl } from '../cache-control.js'
+import { cacheControlFactory } from '../cache-control.js'
 import catchMiddlewareError from '../catch-middleware-error.js'
+import { setFastlySurrogateKey } from '../set-fastly-surrogate-key.js'
 import {
   getSearchResults,
   POSSIBLE_HIGHLIGHT_FIELDS,
   DEFAULT_HIGHLIGHT_FIELDS,
 } from './es-search.js'
 
+// This means we tell the browser to cache the XHR request for 1h
+const browserCacheControl = cacheControlFactory(60 * 60)
+// This tells the CDN to cache the response for 4 hours
+const cdnCacheControl = cacheControlFactory(60 * 60 * 4, {
+  key: 'surrogate-control',
+})
+
 // Used by the legacy search
 const versions = new Set(Object.values(searchVersions))
 const languagesSet = new Set(Object.keys(languages))
@@ -149,7 +157,9 @@ router.get(
       }
     })
     if (process.env.NODE_ENV !== 'development') {
-      defaultCacheControl(res)
+      browserCacheControl(res)
+      cdnCacheControl(res)
+      setFastlySurrogateKey(res, `api-search:${language}`, true)
     }
 
     res.setHeader('x-search-legacy', 'yes')
@@ -248,7 +258,8 @@ router.get(
   '/v1',
   validationMiddleware,
   catchMiddlewareError(async function search(req, res) {
-    const { indexName, query, autocomplete, page, size, debug, sort, highlights } = req.search
+    const { indexName, language, query, autocomplete, page, size, debug, sort, highlights } =
+      req.search
 
     // The getSearchResults() function is a mix of preparing the search,
     // sending & receiving it, and post-processing the response from the
@@ -276,12 +287,9 @@ router.get(
       statsd.timing('api.search.query', meta.took.query_msec, tags)
 
       if (process.env.NODE_ENV !== 'development') {
-        // The assumption, at the moment is that searches are never distinguished
-        // differently depending on a cookie or a request header.
-        // So the only distinguishing key is the request URL.
-        // Because of that, it's safe to allow the reverse proxy (a.k.a the CDN)
-        // cache and hold on to this.
-        defaultCacheControl(res)
+        browserCacheControl(res)
+        cdnCacheControl(res)
+        setFastlySurrogateKey(res, `api-search:${language}`, true)
       }
 
       // The v1 version of the output matches perfectly what comes out

middleware/set-fastly-surrogate-key.js

@@ -14,11 +14,13 @@ export const SURROGATE_ENUMS = {
   MANUAL: 'manual-purge',
 }
 
-export function setFastlySurrogateKey(res, enumKey) {
-  if (!Object.values(SURROGATE_ENUMS).includes(enumKey)) {
-    throw new Error(
-      `Unrecognizes surrogate enumKey. ${enumKey} is not one of ${Object.values(SURROGATE_ENUMS)}`
-    )
+export function setFastlySurrogateKey(res, enumKey, isCustomKey = false) {
+  if (process.env.NODE_ENV !== 'production') {
+    if (!isCustomKey && !Object.values(SURROGATE_ENUMS).includes(enumKey)) {
+      throw new Error(
+        `Unrecognizes surrogate enumKey. ${enumKey} is not one of ${Object.values(SURROGATE_ENUMS)}`
+      )
+    }
   }
   res.set(KEY, enumKey)
 }

package.json

@@ -182,7 +182,7 @@
     "build": "next build",
     "debug": "cross-env NODE_ENV=development ENABLED_LANGUAGES='en,ja,ru' nodemon --inspect server.js",
     "dev": "cross-env npm start",
-    "index-test-fixtures": "node script/search/index-elasticsearch.js -l en -V ghae -V dotcom --index-prefix tests -- tests/content/fixtures/search-indexes",
+    "index-test-fixtures": "node script/search/index-elasticsearch.js -l en -l ja -V ghae -V dotcom --index-prefix tests -- tests/content/fixtures/search-indexes",
     "lint": "eslint '**/*.{js,mjs,ts,tsx}'",
     "lint-translation": "cross-env NODE_OPTIONS=--experimental-vm-modules TEST_TRANSLATION=true jest tests/linting/lint-files.js",
     "prepare": "husky install",

tests/content/api-search.js

@@ -15,7 +15,6 @@ import { jest, test, expect } from '@jest/globals'
 
 import { describeIfElasticsearchURL } from '../helpers/conditional-runs.js'
 import { get } from '../helpers/e2etest.js'
-import { SURROGATE_ENUMS } from '../../middleware/set-fastly-surrogate-key.js'
 
 if (!process.env.ELASTICSEARCH_URL) {
   console.warn(
@@ -69,7 +68,47 @@ describeIfElasticsearchURL('search v1 middleware', () => {
     expect(res.headers['cache-control']).toMatch(/max-age=[1-9]/)
     expect(res.headers['surrogate-control']).toContain('public')
     expect(res.headers['surrogate-control']).toMatch(/max-age=[1-9]/)
-    expect(res.headers['surrogate-key']).toBe(SURROGATE_ENUMS.DEFAULT)
+    expect(res.headers['surrogate-key']).toBe('api-search:en')
+  })
+
+  test('basic search in Japanese', async () => {
+    const sp = new URLSearchParams()
+    // To see why this will work,
+    // see tests/content/fixtures/search-indexes/github-docs-dotcom-en-records.json
+    // which clearly has a record with the title "Foo"
+    sp.set('query', 'foo')
+    sp.set('language', 'ja')
+    const res = await get('/api/search/v1?' + sp)
+    expect(res.statusCode).toBe(200)
+    const results = JSON.parse(res.text)
+
+    expect(results.meta).toBeTruthy()
+    expect(results.meta.found.value).toBeGreaterThanOrEqual(1)
+    expect(results.meta.found.relation).toBeTruthy()
+    expect(results.meta.page).toBe(1)
+    expect(results.meta.size).toBeGreaterThanOrEqual(1)
+    expect(results.meta.took.query_msec).toBeGreaterThanOrEqual(0)
+    expect(results.meta.took.total_msec).toBeGreaterThanOrEqual(0)
+
+    // Might be empty but at least an array
+    expect(results.hits).toBeTruthy()
+    // The word 'foo' appears in more than 1 document in the fixtures.
+    expect(results.hits.length).toBeGreaterThanOrEqual(1)
+    // ...but only one has the word "foo" in its title so we can
+    // be certain it comes first.
+    const hit = results.hits[0]
+    // This specifically checks what we expect of version v1
+    expect(hit.url).toBe('/ja/foo')
+    expect(hit.title).toBe('フー')
+    expect(hit.breadcrumbs).toBe('fooing')
+
+    // Check that it can be cached at the CDN
+    expect(res.headers['set-cookie']).toBeUndefined()
+    expect(res.headers['cache-control']).toContain('public')
+    expect(res.headers['cache-control']).toMatch(/max-age=[1-9]/)
+    expect(res.headers['surrogate-control']).toContain('public')
+    expect(res.headers['surrogate-control']).toMatch(/max-age=[1-9]/)
+    expect(res.headers['surrogate-key']).toBe('api-search:ja')
   })
 
   test('debug search', async () => {

tests/content/fixtures/search-indexes/github-docs-dotcom-ja-records.json

@@ -0,0 +1,29 @@
+{
+    "/ja/foo": {
+        "objectID": "/ja/foo",
+        "breadcrumbs": "fooing",
+        "title": "フー",
+        "headings": "",
+        "content": "これは、ばかげた foo ワードのフィクスチャです。",
+        "topics": ["Test", "Fixture"],
+        "popularity": 0.5
+    },
+    "/ja/bar": {
+        "objectID": "/ja/bar",
+        "breadcrumbs": "baring",
+        "title": "バー",
+        "headings": "",
+        "content": "bar も持っていない場合、foo を持つことはできません。",
+        "topics": ["Test", "Fixture", "Get started"],
+        "popularity": 0.6
+    },
+    "/ja/breadcrumbless": {
+        "objectID": "/ja/bar",
+        "breadcrumbs": "",
+        "title": "パンくずなし",
+        "headings": "",
+        "content": "すべてのページにブレッドクラムがあるわけではありません。人生の事実。",
+        "topics": [],
+        "popularity": 0.1
+    }
+}

tests/content/fixtures/search-indexes/github-docs-ghae-ja-records.json

@@ -0,0 +1,10 @@
+{
+    "/ja/github-ae@latest/foo": {
+        "objectID": "/ja/github-ae@latest/foo",
+        "breadcrumbs": "foo",
+        "title": "フー",
+        "headings": "",
+        "content": "これは、ばかげた foo ワードのフィクスチャです。 GHAE専用。",
+        "topics": ["Test", "Fixture"]
+    }
+}