workflow_dispatch version input for sync search (#32207)

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

@@ -7,6 +7,11 @@ name: Sync search Elasticsearch
 
 on:
   workflow_dispatch:
+    inputs:
+      version:
+        description: "Version to exclusively generate the search index for. E.g. 'dotcom', 'ghes-3.7', 'ghae'"
+        required: false
+        default: ''
   schedule:
     - cron: '23 */4 * * *' # Run every 4 hours at 23 minutes past the hour
 
@@ -88,6 +93,10 @@ jobs:
           # let's just accept an empty string instead.
           THROW_ON_EMPTY: false
 
+          # Note that by default, this is '' (empty string) and that means
+          # the same as not set within the script.
+          VERSION: ${{ github.event.inputs.version }}
+
         run: |
           mkdir /tmp/records
           npm run sync-search-indices -- \
@@ -102,6 +111,11 @@ jobs:
           curl --fail --retry-connrefused --retry 5 -I ${{ env.ELASTICSEARCH_URL }}
 
       - name: Index into Elasticsearch
+        env:
+          # Must match what we used when scraping (npm run sync-search-indices)
+          # otherwise the script will seek other versions from disk that might
+          # not exist.
+          VERSION: ${{ github.event.inputs.version }}
         run: |
           ./script/search/index-elasticsearch.js \
             --language ${{ matrix.language }} -- /tmp/records

script/search/index-elasticsearch.js

@@ -137,12 +137,40 @@ async function main(opts, args) {
 async function indexAll(node, sourceDirectory, opts) {
   const client = new Client({ node })
 
-  const { version, language, verbose, notLanguage, indexPrefix } = opts
+  const { language, verbose, notLanguage, indexPrefix } = opts
+
+  let version
+  if ('version' in opts) {
+    version = opts.version
+    if (process.env.VERSION) {
+      console.warn(
+        `'version' specified as argument ('${version}') AND environment variable ('${process.env.VERSION}')`
+      )
+    }
+  } else {
+    if (process.env.VERSION && process.env.VERSION !== 'all') {
+      version = process.env.VERSION
+      if (!allVersionKeys.includes(version)) {
+        throw new Error(
+          `Environment variable 'VERSION' (${version}) is not recognized. Must be one of ${allVersionKeys}`
+        )
+      }
+    }
+  }
+  let versionKeys = allVersionKeys
+  // If it came from the `--version` argument parsing, it might be a string
+  // or an array of strings because it uses `--version [VERSION...]`.
+  if (version) {
+    if (Array.isArray(version)) {
+      versionKeys = version
+    } else {
+      versionKeys = [version]
+    }
+  }
 
   // This will throw if it can't ping
   await client.ping()
 
-  const versionKeys = version || allVersionKeys
   const languages =
     language || languageKeys.filter((lang) => !notLanguage || !notLanguage.includes(lang))
   if (verbose) {