add search_client to search events when the host is not docs.github.com (#56458)

Co-authored-by: Copilot <[email protected]>

Author: Ebonsignori

src/events/lib/schema.ts

@@ -371,6 +371,10 @@ const search = {
       type: 'string',
       description: 'Any additional search context, such as component searched.',
     },
+    search_client: {
+      type: 'string',
+      description: 'The client name identifier when the request is not from docs.github.com.',
+    },
   },
 }
 

src/events/types.ts

@@ -104,6 +104,7 @@ export type EventPropsByType = {
   [EventType.search]: {
     search_query: string
     search_context?: string
+    search_client?: string
   }
   [EventType.searchResult]: {
     search_result_query: string

src/search/lib/ai-search-proxy.ts

@@ -4,6 +4,7 @@ import got from 'got'
 import { getHmacWithEpoch } from '@/search/lib/helpers/get-cse-copilot-auth'
 import { getCSECopilotSource } from '@/search/lib/helpers/cse-copilot-docs-versions'
 import type { ExtendedRequest } from '@/types'
+import { handleExternalSearchAnalytics } from '@/search/lib/helpers/external-search-analytics'
 
 export const aiSearchProxy = async (req: ExtendedRequest, res: Response) => {
   const { query, version } = req.body
@@ -29,6 +30,15 @@ export const aiSearchProxy = async (req: ExtendedRequest, res: Response) => {
     return
   }
 
+  // Handle search analytics and client_name validation
+  const analyticsError = await handleExternalSearchAnalytics(req, 'ai-search')
+  if (analyticsError) {
+    res.status(analyticsError.status).json({
+      errors: [{ message: analyticsError.error }],
+    })
+    return
+  }
+
   const diagnosticTags = [
     `version:${version}`.slice(0, 200),
     `language:${req.language}`.slice(0, 200),

src/search/lib/helpers/external-search-analytics.ts

@@ -0,0 +1,95 @@
+import { publish } from '@/events/lib/hydro'
+import { hydroNames } from '@/events/lib/schema'
+
+/**
+ * Handles search analytics and client_name validation for external requests
+ * Returns null if the request should continue, or an error response object if validation failed
+ */
+export async function handleExternalSearchAnalytics(
+  req: any,
+  searchContext: string,
+): Promise<{ error: string; status: number } | null> {
+  const host = req.headers['x-host'] || req.headers.host
+  const normalizedHost = stripPort(host as string)
+
+  // Skip analytics entirely for production and internal staging environments
+  if (
+    normalizedHost === 'docs.github.com' ||
+    normalizedHost.endsWith('.github.net') ||
+    normalizedHost.endsWith('.githubapp.com')
+  ) {
+    return null
+  }
+
+  // For localhost, send analytics but auto-set client_name if not provided
+  let client_name = req.query.client_name || req.body?.client_name
+  if (normalizedHost === 'localhost' && !client_name) {
+    client_name = 'localhost'
+  }
+
+  // For all other external requests, require explicit client_name
+  if (!client_name) {
+    return {
+      status: 400,
+      error: "Missing required parameter 'client_name' for external requests",
+    }
+  }
+
+  // Send search event with client identifier
+  try {
+    await publish({
+      schema: hydroNames.search,
+      value: {
+        type: 'search',
+        version: '1.0.0',
+        context: {
+          event_id: crypto.randomUUID(),
+          user: 'server-side',
+          version: '1.0.0',
+          created: new Date().toISOString(),
+          hostname: normalizedHost,
+          path: '',
+          search: '',
+          hash: '',
+          path_language: 'en',
+          path_version: '',
+          path_product: '',
+          path_article: '',
+        },
+        search_query: 'REDACTED',
+        search_context: searchContext,
+        search_client: client_name as string,
+      },
+    })
+  } catch (error) {
+    // Don't fail the request if analytics fails
+    console.error('Failed to send search analytics:', error)
+  }
+
+  return null
+}
+
+/**
+ * Determines if a host should bypass client_name requirement for analytics
+ * Returns true if the host is docs.github.com or ends with github.net or githubapp.com
+ * (for production and internal staging environments)
+ * Note: localhost is NOT included here as it should send analytics with auto-set client_name
+ */
+export function shouldBypassClientNameRequirement(host: string | undefined): boolean {
+  if (!host) return false
+
+  const normalizedHost = stripPort(host)
+  return (
+    normalizedHost === 'docs.github.com' ||
+    normalizedHost.endsWith('.github.net') ||
+    normalizedHost.endsWith('.githubapp.com')
+  )
+}
+
+/**
+ * Strips port number from host string
+ */
+function stripPort(host: string): string {
+  const [hostname] = host.split(':')
+  return hostname
+}

src/search/lib/routes/combined-search-route.ts

@@ -3,6 +3,7 @@ import { getAISearchAutocompleteResults } from '@/search/lib/get-elasticsearch-r
 import { searchCacheControl } from '@/frame/middleware/cache-control'
 import { SURROGATE_ENUMS, setFastlySurrogateKey } from '@/frame/middleware/set-fastly-surrogate-key'
 import { handleGetSearchResultsError } from '@/search/middleware/search-routes'
+import { handleExternalSearchAnalytics } from '@/search/lib/helpers/external-search-analytics'
 
 import type { Request, Response } from 'express'
 import type { CombinedSearchResponse, GeneralSearchResponse } from '@/search/types'
@@ -35,6 +36,14 @@ export async function combinedSearchRoute(req: Request, res: Response) {
     return res.status(400).json(combinedValidationErrors[0])
   }
 
+  // Handle search analytics and client_name validation
+  const analyticsError = await handleExternalSearchAnalytics(req, 'combined-search')
+  if (analyticsError) {
+    return res.status(analyticsError.status).json({
+      error: analyticsError.error,
+    })
+  }
+
   try {
     const autocompletePromise = getAISearchAutocompleteResults({
       indexName: aiIndexName,

src/search/middleware/search-routes.ts

@@ -17,6 +17,7 @@ import { getAISearchAutocompleteResults } from '@/search/lib/get-elasticsearch-r
 import { getSearchFromRequestParams } from '@/search/lib/search-request-params/get-search-from-request-params'
 import { getGeneralSearchResults } from '@/search/lib/get-elasticsearch-results/general-search'
 import { combinedSearchRoute } from '@/search/lib/routes/combined-search-route'
+import { handleExternalSearchAnalytics } from '@/search/lib/helpers/external-search-analytics'
 
 const router = express.Router()
 
@@ -36,6 +37,14 @@ router.get(
       return res.status(400).json(validationErrors[0])
     }
 
+    // Handle search analytics and client_name validation
+    const analyticsError = await handleExternalSearchAnalytics(req, 'general-search')
+    if (analyticsError) {
+      return res.status(analyticsError.status).json({
+        error: analyticsError.error,
+      })
+    }
+
     const getResultOptions = {
       indexName,
       searchParams,