[WPB-21366] Refactor brig-index cli to use brig.yaml

[fisx]

https://wearezeta.atlassian.net/browse/WPB-21366

brig-index needs to talk to postgres now. this PR adds --config-file brig.yaml to its cli.

this makes the already complicated cli even more complicated and also partially redundant. cleaning this up is left as FUTUREWORK.

This work is based on an earlier, failed attempt

Checklist

• Add a new entry in an appropriate subdirectory of changelog.d
• Read and follow the PR guidelines

[fisx]

manual testing

Apply this patch:

modified   integration/test/Test/Apps.hs
@@ -21,7 +21,9 @@ module Test.Apps where

 import API.Brig
 import qualified API.BrigInternal as BrigI
 import SetupHelpers
+import System.IO (appendFile)
 import Testlib.Prelude

 testCreateApp :: (HasCallStack) => App ()
@@ -53,6 +55,14 @@ testCreateApp = do
   bindResponse (getUser owner appIdObject) $ \resp -> do
     resp.status `shouldMatchInt` 200
     resp.json %. "type" `shouldMatch` "app"
+    ownerId <- owner %. "id" >>= asString
+    liftIO do
+      appendFile "/tmp/debug" ("OWNER=" <> ownerId <> "\n")
+      appendFile "/tmp/debug" ("TEAM=" <> tid <> "\n")
+      appendFile "/tmp/debug" ("APP=" <> appId <> "\n")

   -- Creator should have type "regular"
   bindResponse (getUser owner owner) $ \resp -> do

Then create an "old" app without the types field in UserDoc:

$ git co 3a9ffecd6a3ea7ba66c4967af7cfa6433dddeff1~1
$ rm -f /tmp/debug ; make ci-safe package=integration TEST_INCLUDE=testCreateApp

The "old" app is now generated, and /tmp/debug now contains team id, owner id, and app details. with that, we can first check if the app has been created properly, and then build the search query:

$ . /tmp/debug
$ make cr

make cr deletes ES indices, so we need to create them again from cassandra:

$ ./dist/brig-index reindex \
    --elasticsearch-server https://localhost:9200 \
    --elasticsearch-ca-cert ./libs/wire-subsystems/test/resources/elasticsearch-ca.pem \
    --elasticsearch-credentials ./libs/wire-subsystems/test/resources/elasticsearch-credentials.yaml
$ curl -X GET -H 'Z-Connection: conn' -H "Z-User: $OWNER" "http://127.0.0.1:8082/v15/teams/$TEAM/apps/$APP" | jq .
{
  "accent_id": 0,
  "assets": [],
  "category": "ai",
  "description": "some description of this app",
  "metadata": {},
  "name": "chappie",
  "picture": []
}

$ curl -v -X GET -H 'Z-Connection: conn' -H "Z-User: $OWNER" "http://127.0.0.1:8082/v15/users/example.com/$APP" | jq .type
"app"

$ curl -X GET -H 'Z-Connection: conn' -H "Z-User: $OWNER" "http://127.0.0.1:8082/v15/search/contacts?q=chappie&domain=example.com" | jq .documents
[
  {
    "accent_id": 0,
    "handle": null,
    "id": "660158b5-a535-4552-9eb6-bdfc262a5da1",
    "name": "chappie",
    "qualified_id": {
      "domain": "example.com",
      "id": "660158b5-a535-4552-9eb6-bdfc262a5da1"
    },
    "team": "894f3e29-5b1f-452d-bdbf-47fd5dfeb135"
  }
]

Note that there is no type field, that's expected because we're on an old commit. Now back to where ES indexing supports user type:

$ git co WPB-21366-refactor-brig-index-cli
$ make cr
$ curl -X GET -H 'Z-Connection: conn' -H "Z-User: $OWNER" "http://127.0.0.1:8082/v15/teams/$TEAM/apps/$APP" | jq .name
"chappie"
$ curl -X GET -H 'Z-Connection: conn' -H "Z-User: $OWNER" "http://127.0.0.1:8082/v15/search/contacts?q=chappie&domain=example.com" | jq .documents
[]

Damn, make cr destroyed the indices again!

$ make es-reindex
$ curl -X GET -H 'Z-Connection: conn' -H "Z-User: $OWNER" "http://127.0.0.1:8082/v15/search/contacts?q=chappie&domain=example.com" | jq .documents[0].type
"app"

[Copilot]

Pull request overview

This PR refactors the brig-index CLI tool to use the main brig.yaml configuration file instead of requiring redundant CLI parameters. This change was necessary because brig-index now needs to access PostgreSQL, which would have made the CLI even more bloated with additional parameters.

Changes:

• Added a --config-file flag to brig-index that reads brig.yaml for connection settings (Elasticsearch, Cassandra, PostgreSQL, Galley)
• Enabled the getUserType function to properly distinguish between regular users, bots, and app users by integrating AppStore access
• Updated all invocations of brig-index across tests, Makefile, integration scripts, and Helm charts to use the new config file approach

Reviewed changes

Copilot reviewed 14 out of 14 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
services/brig/index/src/Main.hs	Added config file parser and loading logic using --config-file flag
services/brig/src/Brig/Index/Eval.hs	Updated runCommand and runSem to accept Brig.Opts; added PostgreSQL pool initialization and AppStore interpreter
services/brig/src/Brig/Index/Options.hs	Added FUTUREWORK comment about further simplification opportunities
services/brig/test/integration/Index/Create.hs	Updated test calls to runCommand with brigOpts parameter
services/brig/test/integration/API/Search.hs	Updated test calls to runCommand with opts parameter
libs/wire-subsystems/src/Wire/IndexedUserStore/Bulk/ElasticSearch.hs	Uncommented and enabled getUserType implementation with AppStore access
integration/test/Test/Apps.hs	Enhanced tests to verify user types across team boundaries
integration/scripts/integration-dynamic-backends-brig-index.sh	Updated to use --config-file flag
charts/elasticsearch-index/templates/migrate-data.yaml	Added brig config/secrets volume mounts and --config-file argument
charts/elasticsearch-index/templates/create-index.yaml	Added brig config/secrets volume mounts and config file arguments
Makefile	Refactored es-reset to use --config-file and deduplicated db-migrate target
services/brig/brig.cabal	Added yaml dependency to brig-index executable
changelog.d/5-internal/brig-index-config-dedup	Added internal changelog entry
changelog.d/0-release-notes/WPB-21366-refactor-brig-index-cli	Added release notes about PostgreSQL requirement

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.