23 KiB
| related-to | |
|---|---|
|
In Running Unit Tests for status-go we provide general notes on running unit tests in the status-go project. And then we have a similar note about functional tests in Running functional tests in status-go.
Also, to learn the history archive creation/upload and download/processing flows (recorded AI conversation with some edits), please check:
- archive creation/upload: When are magnetlink messages sent
- archive download/processing: status-go processing magnet links. In this document I am including a link to an excalidraw diagram that can be helpful - for convenience: https://link.excalidraw.com/readonly/vSon9uiUhYJWrwXiKAsi
To grasp the concept of topics and filters - check Filters, Topics, Channels, and Chat IDs in status-go and waku.
In this document, we focus on our Codex extension to status-go and here we focus on the related unit and integration tests.
Notice that this is still a proof-of-concept and may deserve a bit more exhaustive testing if it is decided to use it in the production.
The most important tests we added:
- CodexClient - the proxy to the code library
libcodex: - CodexIndexDownloader:
- CodexArchiveDownloader:
- More end-2-end developer tests in protocol/communities_messenger_token_permissions_test.go:
- Functional tests in tests-functional/tests/test_wakuext_community_archives.py
In the remaining part of the document we give some handy information on how we can run those tests.
Regenerating artifacts
There are two artifacts that need to be updated:
- the protobuf
- the mocks
For the first one - protobuf - you need two components:
protoc- the Protocol Buffer compiler itselfprotoc-gen-go- the Go plugin for protoc that generates.pb.gofiles
Installing protoc
I have followed the instructions from Protocol Buffer Compiler Installation.
The following bash script (Arch Linux) can come in handy:
#!/usr/bin/env bash
set -euo pipefail
echo "installing go..."
sudo pacman -S --noconfirm --needed go
echo "installing go protoc compiler"
PB_REL="https://github.com/protocolbuffers/protobuf/releases"
VERSION="32.1"
FILE="protoc-${VERSION}-linux-x86_64.zip"
# 1. create a temp dir
TMP_DIR="$(mktemp -d)"
# ensure cleanup on exit
trap 'rm -rf "$TMP_DIR"' EXIT
echo "Created temp dir: $TMP_DIR"
# 2. download file into temp dir
curl -L -o "$TMP_DIR/$FILE" "$PB_REL/download/v$VERSION/$FILE"
# 3. unzip into ~/.local/share/go
mkdir -p "$HOME/.local/share/go"
unzip -o "$TMP_DIR/$FILE" -d "$HOME/.local/share/go"
# 4. cleanup handled automatically by trap
echo "protoc $VERSION installed into $HOME/.local/share/go"
After that make sure that $HOME/.local/share/go/bin is in your path, and you should get:
protoc --version
libprotoc 32.1
Installing protoc-gen-go
The protoc-gen-go plugin is required to generate Go code from .proto files.
Install it with:
go install google.golang.org/protobuf/cmd/protoc-gen-go@v1.34.1
Make sure $(go env GOPATH)/bin is in your $PATH so protoc can find the plugin.
Verify the installation:
which protoc-gen-go
protoc-gen-go --version
# Should output: protoc-gen-go v1.34.1
Installing mockgen
In order to regenerate mocks you will need mockgen.
You can install it with:
go install go.uber.org/mock/mockgen
Also make sure you have
$(go env GOPATH)/binin your PATH. Otherwise make sure you have something likeexport PATH="$PATH:$(go env GOPATH)/bin"in your~/.bashrc(adjusted to your SHELL and OS version). This should be part of your standard GO installation.
If everything works well, you should see something like:
❯ which mockgen && mockgen -version
/home/<your-user-name>/go/bin/mockgen
v0.6.0
If everything seems to be under control, we can now proceed with actual generation.
The easiest way is to regenerate all in one go:
go generate ./...
If you just need to regenerate the mocks:
go generate ./protocol/communities
If you just need to regenerate the protobuf:
go generate ./protobuf
If you run
make, e.g.make statusgo-library, the correctgeneratecommands for the protobuf will be run for you. So in practice, you may not need to rungo generate ./protobufmanually yourself - but for reference, why not... let's break something ;).
Environment variables to run unit tests
After Codex library (libcodex) has been added to the project the build system depends on it. Thus, to run the tests directly using go test or gotestsum command you need to make sure that the following environment variables are set:
export LIBS_DIR="$(realpath ./libs)"
export CGO_CFLAGS=-I$LIBS_DIR
export CGO_LDFLAGS="-L$LIBS_DIR -lcodex -Wl,-rpath,$LIBS_DIR"
Additionally, since status-go added the SDS library, there are even more environment variables that need to be setup. We have a handy script to do just that:
#!/usr/bin/env bash
# Source this file to set up environment variables for running tests with gotestsum
# Usage: source ./set-test-env.sh
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
export LIBS_DIR="$(realpath "$SCRIPT_DIR/libs")"
export NIM_SDS_LIB_DIR="$(realpath "$SCRIPT_DIR/../nim-sds/build")"
export NIM_SDS_INC_DIR="$(realpath "$SCRIPT_DIR/../nim-sds/library")"
export CGO_CFLAGS="-I$LIBS_DIR -I$NIM_SDS_INC_DIR"
export CGO_LDFLAGS="-L$LIBS_DIR -lcodex -Wl,-rpath,$LIBS_DIR -L$NIM_SDS_LIB_DIR -lsds"
# Detect OS and set library path accordingly
if [[ "$OSTYPE" == "darwin"* ]]; then
export DYLD_LIBRARY_PATH="$LIBS_DIR:$NIM_SDS_LIB_DIR:$DYLD_LIBRARY_PATH"
echo "Environment configured for macOS"
else
export LD_LIBRARY_PATH="$LIBS_DIR:$NIM_SDS_LIB_DIR:$LD_LIBRARY_PATH"
echo "Environment configured for Linux"
fi
echo "Test environment variables set:"
echo " LIBS_DIR=$LIBS_DIR"
echo " NIM_SDS_LIB_DIR=$NIM_SDS_LIB_DIR"
echo " NIM_SDS_INC_DIR=$NIM_SDS_INC_DIR"
echo ""
echo "You can now run tests with gotestsum, for example:"
echo ' gotestsum --packages="./protocol/communities" -f testname -- -count 1 -tags "gowaku_no_rln gowaku_skip_migrations" -run CodexArchiveManagerSuite'
Thus, just run:
source ./set-test-env.sh
in the top level directory and you should be good to go.
Running developer-facing tests for Codex abstractions
TL;DR
Install gotestsum: go install gotest.tools/gotestsum@v1.13.0
Then to run ./protocol/community tests:
gotestsum --packages="./protocol/communities" -f testname --rerun-fails -- -count 1 -timeout "5m" -tags "gowaku_no_rln gowaku_skip_migrations"
To run all ./protocol tests:
gotestsum --packages="./protocol" -f testname --rerun-fails -- -count 1 -timeout "45m" -tags "gowaku_no_rln gowaku_skip_migrations"
Of course, you can also just run:
make test
I found running tests using gotestsum more reliable and more appropriate for development.
To be more selective, e.g. in order to run all the tests from
CodexArchiveDownloaderSuite, run:
gotestsum --packages="./protocol/communities" -f testname --rerun-fails -- -run CodexArchiveDownloader -count 1
or for an individual test from that suite:
gotestsum --packages="./protocol/communities" -f testname --rerun-fails -- -run CodexArchiveDownloaderSuite/TestCancellationDuringPolling -count 1
Notice, that the -run flag accepts a regular expression that matches against the full test path, so you can be more concise in naming if necessary, e.g.:
gotestsum --packages="./protocol/communities" -f testname --rerun-fails -- -run CodexArchiveDownloader/Cancellation -count 1
This also applies to native go test command, e.g.:
go test -v ./protocol/communities -count 1 -run CodexArchiveDownloaderSuite/TestCancellationDuringPolling
For a more verbose output including logs use -f standard-verbose, e.g.:
gotestsum --packages="./protocol/communities" -f standard-verbose --rerun-fails -- -v -count 1 -run CodexArchiveDownloaderSuite/TestCancellationDuringPolling
Finally, the relevant Codex integration tests from the ./protocol module run:
gotestsum --packages="./protocol" -f testname -- -count 1 -tags "gowaku_no_rln gowaku_skip_migrations" -run TestMessengerCommunitiesTokenPermissionsSuite/Codex
PASS protocol.TestMessengerCommunitiesTokenPermissionsSuite/TestUploadDownloadCodexHistoryArchives (5.44s)
PASS protocol.TestMessengerCommunitiesTokenPermissionsSuite/TestUploadDownloadCodexHistoryArchives_withSharedCodexClient (5.50s)
PASS protocol.TestMessengerCommunitiesTokenPermissionsSuite (10.94s)
PASS protocol
DONE 3 tests in 10.968s
Running functional tests
Here we use our convenience scripts. First we build the docker image running:
_assets/scripts/build_status_go_docker.sh
We run all the commands from the top-level project directory
And then:
./_assets/scripts/run_functional_tests_dev.sh TestCommunityArchives
Using existing virtual environment
Installing dependencies
Discovering tests to be run...
Found 5 tests matching: TestCommunityArchives
Tests to execute:
1) test_community_archive_index_exists
2) test_community_archive_exists_for_default_chat
3) test_archive_is_not_created_without_messages
4) test_different_archives_are_created_with_multiple_messages
5) test_archive_is_downloaded_after_logout_login
Continue with execution? (y/n): y
...
tests-functional/tests/test_wakuext_community_archives.py::TestCommunityArchives::test_community_archive_exists_for_default_chat
tests-functional/tests/test_wakuext_community_archives.py::TestCommunityArchives::test_community_archive_index_exists
tests-functional/tests/test_wakuext_community_archives.py::TestCommunityArchives::test_archive_is_downloaded_after_logout_login
tests-functional/tests/test_wakuext_community_archives.py::TestCommunityArchives::test_different_archives_are_created_with_multiple_messages
tests-functional/tests/test_wakuext_community_archives.py::TestCommunityArchives::test_archive_is_not_created_without_messages
[gw1] [ 20%] PASSED tests-functional/tests/test_wakuext_community_archives.py::TestCommunityArchives::test_community_archive_exists_for_default_chat
[gw2] [ 40%] PASSED tests-functional/tests/test_wakuext_community_archives.py::TestCommunityArchives::test_archive_is_not_created_without_messages
[gw3] [ 60%] PASSED tests-functional/tests/test_wakuext_community_archives.py::TestCommunityArchives::test_different_archives_are_created_with_multiple_messages
[gw4] [ 80%] PASSED tests-functional/tests/test_wakuext_community_archives.py::TestCommunityArchives::test_archive_is_downloaded_after_logout_login
[gw0] [100%] PASSED tests-functional/tests/test_wakuext_community_archives.py::TestCommunityArchives::test_community_archive_index_exists
======================================================================== 5 passed in 141.69s (0:02:21) =========================================================================
Testing finished
Running cleanup...
You can also run a single test as follow:
./_assets/scripts/run_functional_tests_dev.sh test_community_archive_index_exists
The logs are in tests-functional/logs.
More notes
For the Codex-related tests in protocol/communities_messenger_token_permissions_test.go, in we add some more context below.
Our two "Codex"-related tests are based on TestImportDecryptedArchiveMessages test.
This test produces lots of output - with lot's warnings and errors - so looking at the log to judge the success would be a challenge. Yet, the test passes:
❯ gotestsum --packages="./protocol" -f testname -- -run "TestMessengerCommunitiesTokenPermissionsSuite/TestImportDecryptedArchiveMessages" -count 1 -tags "gowaku_no_rln gowaku_skip_migrations"
PASS protocol.TestMessengerCommunitiesTokenPermissionsSuite/TestImportDecryptedArchiveMessages (1.88s)
PASS protocol.TestMessengerCommunitiesTokenPermissionsSuite (1.88s)
PASS protocol
DONE 2 tests in 1.900s
If you want to take a look at the logs you can use the more verbose version of the above command:
❯ gotestsum --packages="./protocol" -f standard-verbose -- -run "TestMessengerCommunitiesTokenPermissionsSuite/TestImportDecryptedArchiveMessages" -v -count 1 -tags "gowaku_no_rln gowaku_skip_migrations"
and you can use tee to redirect all the output to a file:
❯ gotestsum --packages="./protocol" -f standard-verbose -- -run "TestMessengerCommunitiesTokenPermissionsSuite/TestImportDecryptedArchiveMessages" -v -count 1 -tags "gowaku_no_rln gowaku_skip_migrations" | tee "test_1.log"
The test first creates a community and sets up the corresponding permissions. Then the community owner sends a message to the community and then immediately retrieves it so that it is now recorded in the DB.
After that it prepares archive parameters: startDate, endDate, partition, and community topics. All those will be passed to CreateHistoryArchiveTorrentFromDB - our entry point to creating history archive torrent.
// 1.1. Create community
community, chat := s.createCommunity()
// ...
// 1.2. Setup permissions
// ...
// 2. Owner: Send a message A
messageText1 := RandomLettersString(10)
message1 := s.sendChatMessage(s.owner, chat.ID, messageText1)
// 2.2. Retrieve own message (to make it stored in the archive later)
_, err = s.owner.RetrieveAll()
s.Require().NoError(err)
// 3. Owner: Create community archive
const partition = 2 * time.Minute
messageDate := time.UnixMilli(int64(message1.Timestamp))
startDate := messageDate.Add(-time.Minute)
endDate := messageDate.Add(time.Minute)
topic := messagingtypes.BytesToContentTopic(messaging.ToContentTopic(chat.ID))
communityCommonTopic := messagingtypes.BytesToContentTopic(messaging.ToContentTopic(community.UniversalChatID()))
topics := []messagingtypes.ContentTopic{topic, communityCommonTopic}
torrentConfig := params.TorrentConfig{
Enabled: true,
DataDir: os.TempDir() + "/archivedata",
TorrentDir: os.TempDir() + "/torrents",
Port: 0,
}
// Share archive directory between all users
s.owner.archiveManager.SetTorrentConfig(&torrentConfig)
s.bob.archiveManager.SetTorrentConfig(&torrentConfig)
s.owner.config.messengerSignalsHandler = &MessengerSignalsHandlerMock{}
s.bob.config.messengerSignalsHandler = &MessengerSignalsHandlerMock{}
Finally we call the CreateHistoryArchiveTorrentFromDB:
archiveIDs, err := s.owner.archiveManager.CreateHistoryArchiveTorrentFromDB(community.ID(), topics, startDate, endDate, partition, community.Encrypted())
s.Require().NoError(err)
s.Require().Len(archiveIDs, 1)
Notice, there is one archive expected.
The CreateHistoryArchiveTorrentFromDB is called directly here, in a way bypassing the torrent seeding: in normal flow CreateHistoryArchiveTorrentFromDB is called in CreateAndSeedHistoryArchive which immediately after creating the archive, calls SeedHistoryArchiveTorrent. CreateHistoryArchiveTorrentFromDB calls createHistoryArchiveTorrent - which is central to the archive creating.
The "Codex" version of the CreateHistoryArchiveTorrentFromDB is CreateHistoryArchiveCodexFromDB which will call createHistoryArchiveCodex - this is where archives are created and uploaded to Codex.
Another function that this test "touches" is LoadHistoryArchiveIndexFromFile, for which the "Codex" version is CodexLoadHistoryArchiveIndex. Notice, we do not need to load it from a file anymore - Codex takes care.
Thus, this test focuses on an important cut of the whole end-to-end flow and in our case, where we use Codex, it also indirectly tests seeding and retrieving the archives from the network.
Other places we can consider to test
The integration test described above does not cover actual publishing of the generated archives over waku channel. This normally happens in CreateAndSeedHistoryArchive:
CreateAndSeedHistoryArchive
|- if distributionPreference == Torrent
|- CreateHistoryArchiveTorrentFromDB
|- SeedHistoryArchiveTorrent
|- if distributionPreference == Codex
|- CreateHistoryArchiveCodexFromDB
|- SeedHistoryArchiveIndexCid (only in some cases: Codex takes care for seeding)
|- publish: HistoryArchiveSeedingSignal
Depending on the distribution preference (Codex or Torrent) we either seed the index file and the archives for torrent or for Codex and when at least one publishing method succeeds, we do:
m.publisher.publish(&Subscription{
HistoryArchivesSeedingSignal: &signal.HistoryArchivesSeedingSignal{
CommunityID: communityID.String(),
MagnetLink: archiveTorrentCreatedSuccessfully, // true if torrent created successfully
IndexCid: archiveCodexCreatedSuccessfully, // true if codex created successfully
},
})
This signal is subsequently received in handleCommunitiesHistoryArchivesSubscription, where we find:
if c.IsControlNode() {
if sub.HistoryArchivesSeedingSignal.MagnetLink {
err := m.dispatchMagnetlinkMessage(sub.HistoryArchivesSeedingSignal.CommunityID)
if err != nil {
m.logger.Debug("failed to dispatch magnetlink message", zap.Error(err))
}
}
if sub.HistoryArchivesSeedingSignal.IndexCid {
err := m.dispatchIndexCidMessage(sub.HistoryArchivesSeedingSignal.CommunityID)
if err != nil {
m.logger.Debug("failed to dispatch index cid message", zap.Error(err))
}
}
}
The dispatchMagnetlinkMessage and dispatchIndexCidMessage is where dispatching is happening. For Codex the MessageType used is protobuf.ApplicationMetadataMessage_COMMUNITY_MESSAGE_ARCHIVE_INDEX_CID.
The message is sent as follows:
chatID := community.UniversalChatID()
rawMessage := common.RawMessage{
LocalChatID: chatID,
Sender: community.PrivateKey(),
Payload: encodedMessage,
MessageType: protobuf.ApplicationMetadataMessage_COMMUNITY_MESSAGE_ARCHIVE_INDEX_CID,
SkipGroupMessageWrap: true,
PubsubTopic: community.PubsubTopic(),
Priority: &messagingtypes.LowPriority,
}
_, err = m.messaging.SendPublic(context.Background(), chatID, rawMessage)
if err != nil {
return err
}
err = m.communitiesManager.UpdateCommunityDescriptionIndexCidMessageClock(community.ID(), indexCidMessage.Clock)
if err != nil {
return err
}
return m.communitiesManager.UpdateIndexCidMessageClock(community.ID(), indexCidMessage.Clock)
Here notice the call to UpdateCommunityDescriptionIndexCidMessageClock. As you can see, the clocks are also recorded in the community description. Community description is sent over waku to other community members periodically (every 5mins) or on some events. This how it is sent in publishOrg in protocol/messenger_communities.go:
rawMessage := common.RawMessage{
Payload: payload,
Sender: org.PrivateKey(),
// we don't want to wrap in an encryption layer message
SkipEncryptionLayer: true,
CommunityID: org.ID(),
MessageType: protobuf.ApplicationMetadataMessage_COMMUNITY_DESCRIPTION,
PubsubTopic: org.PubsubTopic(), // TODO: confirm if it should be sent in community pubsub topic
Priority: &messagingtypes.HighPriority,
}
if org.Encrypted() {
members := org.GetMemberPubkeys()
if err != nil {
return err
}
rawMessage.CommunityKeyExMsgType = messagingtypes.KeyExMsgRekey
// This should be the one that it was used to encrypt this community
rawMessage.HashRatchetGroupID = org.ID()
rawMessage.Recipients = members
}
messageID, err := m.messaging.SendPublic(context.Background(), org.IDString(), rawMessage)
When community members receive the community description, they compare the clocks in the community description with the most recent clocks of the most recent magnet link or index Cid. IF the clocks in the community description are newer (more recent), they update the local copies, so that when a new magnet link or index Cid message arrives, they know to ignore outdated messages.
To read more, check: History Archives and Community Description (AI session recording).
This naturally brought us to the reception of the magnet links/index Cid (see also status-go processing magnet links and status-go publishing magnet links).
Going from Waku and layers of status-go, the accumulated messages are periodically collected using RetrieveAll functions, which passes the message to handleRetrievedMessages for processing.
For each message, dispatchToHandler is called (in it in generated file message_handler.go generated with go generate ./cmd/generae_handlers/) where are magnet links/index Cid message are forwarded to their respective handlers: handleCommunityMessageArchiveMagnetlinkProtobuf and handleCommunityMessageArchiveIndexCidProtobuf. In the end they end up in HandleHistoryArchiveMagnetlinkMessage and HandleHistoryArchiveIndexCidMessage respectively. At the end, they trigger async processing in go routines calling downloadAndImportHistoryArchives and downloadAndImportCodexHistoryArchives. Just after the respective go routines are created, the clock are updated with m.communitiesManager.UpdateMagnetlinkMessageClock(id, clock) and m.communitiesManager.UpdateIndexCidMessageClock(id, clock).
But before even starting go routine, we check the distribution preference with GetArchiveDistributionPreference and the last seen index Cid m.communitiesManager.GetLastSeenIndexCid(id) (it will be updated after downloading the archives but before processing them in downloadAndImportCodexHistoryArchives) and the corresponding clock GetIndexCidMessageClock, to make sure we are not processing some outdated archives.
We can see that the above "integration test", in case of Codex, by calling RetrieveAll it effectively triggers archive download from Codex, yet it does not cover the dissemination of the indexCid - the test basically assumes that the index file has been stored, thus, if we want to cover more, we need to go beyond that test.
Perhaps, to cover the whole flow, it is best to build status-desktop with our status-go library and test it from there.