X-Git-Url: https://git.giacomolli.uk/onechat/blobdiff_plain/b2873a2a0bf8b0ba648aebac51759e5fed89a2e2..9bd91a76e6277f7716e206ddb665dc5fe6f5b583:/turn_test.go?ds=sidebyside diff --git a/turn_test.go b/turn_test.go index 0de7dd7..b89a0ec 100644 --- a/turn_test.go +++ b/turn_test.go @@ -15,21 +15,21 @@ import ( ) func TestTurnRunsToolsUntilTheFinalAnswer(t *testing.T) { - server := newFakeResponses(t) - server.reply = func(w http.ResponseWriter, call fakeCall) { + server := newModelFake(t) + server.reply = func(w http.ResponseWriter, call ModelFakeCall) { if call.number == 0 { - writeEvents(w, completedEvent([]any{ - reasoningItem("rs_1", "opaque thought"), - functionCallItem("fc_1", "call_1", "shell", `{"command":"echo hi"}`), + modelWriteEvents(w, modelCompletedEvent([]any{ + modelReasoningItem("rs_1", "opaque thought"), + modelFunctionCallItem("fc_1", "call_1", "shell", `{"command":"echo hi"}`), }, 100, 0, 5)) return } - writeEvents(w, completedEvent([]any{messageItem("It said hi.")}, 120, 100, 5)) + modelWriteEvents(w, modelCompletedEvent([]any{modelMessageItem("It said hi.")}, 120, 100, 5)) } - test := newTestLoop(t, server) + test := newTestTurnLoop(t, server) _, err := test.loop.Submit(terminalSource, "Run echo hi.") require.NoError(t, err) test.events.WaitFor(t, "idle", 1) @@ -61,7 +61,7 @@ func TestTurnRunsToolsUntilTheFinalAnswer(t *testing.T) { require.Equal(t, []string{"user", "busy", "tool", "echo", "reply", "idle"}, test.events.Kinds()) var turns []string - for _, record := range readUsage(t, test.usagePath) { + for _, record := range modelReadUsage(t, test.usagePath) { turns = append(turns, record.Turn) } @@ -69,8 +69,8 @@ func TestTurnRunsToolsUntilTheFinalAnswer(t *testing.T) { } func TestFirstRequestHoldsTheViewTheStateAndTheMessage(t *testing.T) { - server := newFakeResponses(t) - test := newTestLoop(t, server) + server := newModelFake(t) + test := newTestTurnLoop(t, server) _, err := test.memory.Append("user", "an earlier message") require.NoError(t, err) @@ -82,26 +82,26 @@ func TestFirstRequestHoldsTheViewTheStateAndTheMessage(t *testing.T) { "\n0+1|user: an earlier message\n", "\n\nNow: Thursday 2026-01-01 12:00 +00:00", "The new message.", - }, inputTexts(t, server.Call(0))) + }, turnInputTexts(t, server.Call(0))) } func TestMessageDuringATurnJoinsTheNextRequest(t *testing.T) { - server := newFakeResponses(t) + server := newModelFake(t) arrived := make(chan struct{}) release := make(chan struct{}) - server.reply = func(w http.ResponseWriter, call fakeCall) { + server.reply = func(w http.ResponseWriter, call ModelFakeCall) { if call.number == 0 { close(arrived) <-release - writeEvents(w, completedEvent([]any{functionCallItem("fc_1", "call_1", "date", `{"id":0}`)}, 100, 0, 5)) + modelWriteEvents(w, modelCompletedEvent([]any{modelFunctionCallItem("fc_1", "call_1", "date", `{"id":0}`)}, 100, 0, 5)) return } - writeEvents(w, completedEvent([]any{messageItem("done")}, 100, 0, 5)) + modelWriteEvents(w, modelCompletedEvent([]any{modelMessageItem("done")}, 100, 0, 5)) } - test := newTestLoop(t, server) + test := newTestTurnLoop(t, server) _, err := test.loop.Submit(terminalSource, "first") require.NoError(t, err) @@ -123,19 +123,19 @@ func TestMessageDuringATurnJoinsTheNextRequest(t *testing.T) { } func TestMessageDuringTheFinalRequestStartsANewTurn(t *testing.T) { - server := newFakeResponses(t) + server := newModelFake(t) arrived := make(chan struct{}) release := make(chan struct{}) - server.reply = func(w http.ResponseWriter, call fakeCall) { + server.reply = func(w http.ResponseWriter, call ModelFakeCall) { if call.number == 0 { close(arrived) <-release } - writeEvents(w, completedEvent([]any{messageItem("answer")}, 100, 0, 5)) + modelWriteEvents(w, modelCompletedEvent([]any{modelMessageItem("answer")}, 100, 0, 5)) } - test := newTestLoop(t, server) + test := newTestTurnLoop(t, server) _, err := test.loop.Submit(terminalSource, "first") require.NoError(t, err) @@ -147,16 +147,16 @@ func TestMessageDuringTheFinalRequestStartsANewTurn(t *testing.T) { test.events.WaitFor(t, "idle", 2) require.Equal(t, 2, server.Calls()) - texts := inputTexts(t, server.Call(1)) + texts := turnInputTexts(t, server.Call(1)) require.Equal(t, "second", texts[len(texts)-1]) require.Contains(t, strings.Join(texts, ""), "|reply: answer\n") } func TestTurnWaitsForTheSummaries(t *testing.T) { - server := newFakeResponses(t) + server := newModelFake(t) summarizer := newFakeSummarizer() summarizer.gate = make(chan struct{}) - test := newTestLoopWith(t, server, summarizer) + test := newTestTurnLoopWith(t, server, summarizer) _, err := test.memory.Append("echo", strings.Repeat("long output ", 100)) require.NoError(t, err) @@ -170,23 +170,23 @@ func TestTurnWaitsForTheSummaries(t *testing.T) { close(summarizer.gate) test.events.WaitFor(t, "idle", 1) require.Equal(t, 1, server.Calls()) - require.NotContains(t, inputTexts(t, server.Call(0))[0], pending) + require.NotContains(t, turnInputTexts(t, server.Call(0))[0], viewPending) } func TestInterruptEndsTheTurn(t *testing.T) { - server := newFakeResponses(t) + server := newModelFake(t) arrived := make(chan struct{}) release := make(chan struct{}) - server.reply = func(w http.ResponseWriter, call fakeCall) { + server.reply = func(w http.ResponseWriter, call ModelFakeCall) { if call.number == 0 { close(arrived) <-release } - writeEvents(w, completedEvent([]any{messageItem("answer")}, 100, 0, 5)) + modelWriteEvents(w, modelCompletedEvent([]any{modelMessageItem("answer")}, 100, 0, 5)) } - test := newTestLoop(t, server) + test := newTestTurnLoop(t, server) _, err := test.loop.Submit(terminalSource, "a long task") require.NoError(t, err) @@ -206,12 +206,12 @@ func TestInterruptEndsTheTurn(t *testing.T) { } func TestFailedCallEndsTheTurnWithoutARetry(t *testing.T) { - server := newFakeResponses(t) - server.reply = func(w http.ResponseWriter, call fakeCall) { + server := newModelFake(t) + server.reply = func(w http.ResponseWriter, call ModelFakeCall) { http.Error(w, "busy", http.StatusServiceUnavailable) } - test := newTestLoop(t, server) + test := newTestTurnLoop(t, server) _, err := test.loop.Submit(terminalSource, "hello") require.NoError(t, err) test.events.WaitFor(t, "idle", 1) @@ -222,8 +222,8 @@ func TestFailedCallEndsTheTurnWithoutARetry(t *testing.T) { } func TestViewMessagesOnlyGrowBetweenTurns(t *testing.T) { - server := newFakeResponses(t) - test := newTestLoop(t, server) + server := newModelFake(t) + test := newTestTurnLoop(t, server) for n, text := range []string{"one", "two", "three"} { _, err := test.loop.Submit(terminalSource, text) require.NoError(t, err) @@ -241,20 +241,20 @@ func TestViewMessagesOnlyGrowBetweenTurns(t *testing.T) { "2+1|user: two\n3+1|reply: a line\n", "\n\nNow: Thursday 2026-01-01 12:00 +00:00", "three", - }, inputTexts(t, server.Call(2))) + }, turnInputTexts(t, server.Call(2))) } -// testLoop is a loop that runs in the background of a test, with its parts. -type testLoop struct { +// TurnTestLoop is a loop that runs in the background of a test, with its parts. +type TurnTestLoop struct { agent *Agent memory *Memory - loop *loop - events *eventRecorder + loop *TurnLoop + events *TurnEventRecorder usagePath string } // Log returns the logged messages as "kind: text". -func (l *testLoop) Log(t *testing.T) []string { +func (l *TurnTestLoop) Log(t *testing.T) []string { t.Helper() var lines []string @@ -268,20 +268,22 @@ func (l *testLoop) Log(t *testing.T) []string { return lines } -// eventRecorder keeps the events of a loop. -type eventRecorder struct { +// TurnEventRecorder keeps the events of a loop. +type TurnEventRecorder struct { mutex sync.Mutex events []Event } -func (r *eventRecorder) Add(event Event) { - r.mutex.Lock() - defer r.mutex.Unlock() +// WaitFor waits until count events of kind came. +func (r *TurnEventRecorder) WaitFor(t *testing.T, kind string, count int) { + t.Helper() - r.events = append(r.events, event) + require.Eventually(t, func() bool { + return r.Count(kind) >= count + }, 10*time.Second, time.Millisecond) } -func (r *eventRecorder) Kinds() []string { +func (r *TurnEventRecorder) Kinds() []string { r.mutex.Lock() defer r.mutex.Unlock() @@ -293,7 +295,7 @@ func (r *eventRecorder) Kinds() []string { return kinds } -func (r *eventRecorder) Count(kind string) int { +func (r *TurnEventRecorder) Count(kind string) int { count := 0 for _, each := range r.Kinds() { if each == kind { @@ -305,7 +307,7 @@ func (r *eventRecorder) Count(kind string) int { } // Last returns the text of the last event of kind. -func (r *eventRecorder) Last(kind string) string { +func (r *TurnEventRecorder) Last(kind string) string { r.mutex.Lock() defer r.mutex.Unlock() @@ -318,61 +320,21 @@ func (r *eventRecorder) Last(kind string) string { return "" } -// WaitFor waits until count events of kind came. -func (r *eventRecorder) WaitFor(t *testing.T, kind string, count int) { - t.Helper() +func (r *TurnEventRecorder) Add(event Event) { + r.mutex.Lock() + defer r.mutex.Unlock() - require.Eventually(t, func() bool { - return r.Count(kind) >= count - }, 10*time.Second, time.Millisecond) + r.events = append(r.events, event) } -func newTestLoop(t *testing.T, server *fakeResponses) *testLoop { +func newTestTurnLoop(t *testing.T, server *ModelFake) *TurnTestLoop { t.Helper() - return newTestLoopWith(t, server, newFakeSummarizer()) -} - -// newTestLoopWith returns a running loop on a new agent whose builds use -// summarizer and whose commands run in a temporary directory. -func newTestLoopWith(t *testing.T, server *fakeResponses, summarizer Summarizer) *testLoop { - t.Helper() - - agent, err := openAgent(t.TempDir()) - require.NoError(t, err) - - agent.config.Workdir = t.TempDir() - clock := newTestClock() - memory, err := openMemory(t.Context(), t.TempDir(), clock.Now, summarizer) - require.NoError(t, err) - - test := &testLoop{ - agent: agent, - memory: memory, - events: &eventRecorder{}, - usagePath: filepath.Join(t.TempDir(), "usage.jsonl"), - } - - client := newClient(server.URL, apiKey(testKey), agent.CacheKey(), test.usagePath) - test.loop, err = newLoop(agent, memory, client, clock.Now, test.events.Add) - require.NoError(t, err) - ctx, cancel := context.WithCancel(t.Context()) - done := make(chan error) - go func() { - done <- test.loop.Run(ctx) - }() - - t.Cleanup(func() { - cancel() - require.True(t, errors.Is(<-done, context.Canceled)) - require.NoError(t, memory.Close()) - }) - - return test + return newTestTurnLoopWith(t, server, newFakeSummarizer()) } -// inputTexts returns the text of each user message of the input of call. -func inputTexts(t *testing.T, call fakeCall) []string { +// turnInputTexts returns the text of each user message of the input of call. +func turnInputTexts(t *testing.T, call ModelFakeCall) []string { t.Helper() var body struct { @@ -403,21 +365,40 @@ func inputTexts(t *testing.T, call fakeCall) []string { return texts } -func reasoningItem(id, encrypted string) map[string]any { - return map[string]any{ - "type": "reasoning", - "id": id, - "encrypted_content": encrypted, - "summary": []any{}, - } -} +// newTestTurnLoopWith returns a running loop on a new agent whose builds use +// summarizer and whose commands run in a temporary directory. +func newTestTurnLoopWith(t *testing.T, server *ModelFake, summarizer Summarizer) *TurnTestLoop { + t.Helper() + + agent, err := openAgent(t.TempDir()) + require.NoError(t, err) -func functionCallItem(id, callID, name, arguments string) map[string]any { - return map[string]any{ - "type": "function_call", - "id": id, - "call_id": callID, - "name": name, - "arguments": arguments, + agent.config.Workdir = t.TempDir() + clock := newTestClock() + memory, err := openMemory(t.Context(), t.TempDir(), clock.Now, summarizer) + require.NoError(t, err) + + test := &TurnTestLoop{ + agent: agent, + memory: memory, + events: &TurnEventRecorder{}, + usagePath: filepath.Join(t.TempDir(), "usage.jsonl"), } + + client := newModelClient(server.URL, ModelAPIKey(modelTestKey), agent.CacheKey(), test.usagePath) + test.loop, err = newTurnLoop(agent, memory, client, clock.Now, test.events.Add) + require.NoError(t, err) + ctx, cancel := context.WithCancel(t.Context()) + done := make(chan error) + go func() { + done <- test.loop.Run(ctx) + }() + + t.Cleanup(func() { + cancel() + require.True(t, errors.Is(<-done, context.Canceled)) + require.NoError(t, memory.Close()) + }) + + return test }