diff --git a/frontend/src/views/chat/components/AgentStreamDisplay.vue b/frontend/src/views/chat/components/AgentStreamDisplay.vue index 551b2ee9..39d04fcd 100644 --- a/frontend/src/views/chat/components/AgentStreamDisplay.vue +++ b/frontend/src/views/chat/components/AgentStreamDisplay.vue @@ -865,6 +865,12 @@ const isConversationDone = computed(() => { console.log('[Collapse] Found stop event, conversation done'); return true; } + + const completeEvent = stream.find((e: any) => e.type === 'agent_complete'); + if (completeEvent) { + console.log('[Collapse] Found complete event, conversation done'); + return true; + } // Check for answer event with done=true const answerEvents = stream.filter((e: any) => e.type === 'answer'); @@ -3435,4 +3441,4 @@ html[theme-mode="dark"] .tree-root-title img { filter: invert(1); opacity: 0.55; } - \ No newline at end of file + diff --git a/frontend/src/views/chat/index.vue b/frontend/src/views/chat/index.vue index 5b7f099a..642b55c9 100644 --- a/frontend/src/views/chat/index.vue +++ b/frontend/src/views/chat/index.vue @@ -1095,12 +1095,15 @@ const handleAgentChunk = (data) => { console.log('[Agent] Complete event received'); loading.value = false; isReplying.value = false; + message.is_completed = true; + fullContent.value = ''; + currentAssistantMessageId.value = ''; // 将 total_duration_ms 存入事件流供 AgentStreamDisplay 使用 - if (data.data?.total_duration_ms && message.agentEventStream) { + if (message.agentEventStream) { message.agentEventStream.push({ type: 'agent_complete', - total_duration_ms: data.data.total_duration_ms, - total_steps: data.data.total_steps, + total_duration_ms: data.data?.total_duration_ms || 0, + total_steps: data.data?.total_steps || 0, }); } break; @@ -1538,4 +1541,4 @@ onBeforeRouteUpdate((to, from, next) => { color: var(--td-success-color); } } - \ No newline at end of file + diff --git a/internal/agent/engine_test.go b/internal/agent/engine_test.go index cd4857b3..57f2ed40 100644 --- a/internal/agent/engine_test.go +++ b/internal/agent/engine_test.go @@ -219,3 +219,34 @@ func TestStreamThinkingToEventBus_PropagatesFinishReason(t *testing.T) { }) } } + +func TestStreamFinalAnswerToEventBus_EmitsDoneWhenProviderEndsWithEmptyChunk(t *testing.T) { + mock := &mockChat{ + responses: []mockResponse{ + {chunks: []types.StreamResponse{ + {ResponseType: types.ResponseTypeAnswer, Content: "final answer", Done: false}, + {ResponseType: types.ResponseTypeAnswer, Done: true, FinishReason: "stop"}, + }}, + }, + } + + engine := newTestEngine(t, mock) + var finalAnswerEvents []event.AgentFinalAnswerData + engine.eventBus.On(event.EventAgentFinalAnswer, func(_ context.Context, evt event.Event) error { + data, ok := evt.Data.(event.AgentFinalAnswerData) + require.True(t, ok) + finalAnswerEvents = append(finalAnswerEvents, data) + return nil + }) + + state := &types.AgentState{} + err := engine.streamFinalAnswerToEventBus(context.Background(), "test query", state, "sess-1") + + require.NoError(t, err) + require.Len(t, finalAnswerEvents, 2) + assert.Equal(t, "final answer", finalAnswerEvents[0].Content) + assert.False(t, finalAnswerEvents[0].Done) + assert.Empty(t, finalAnswerEvents[1].Content) + assert.True(t, finalAnswerEvents[1].Done) + assert.Equal(t, "final answer", state.FinalAnswer) +} diff --git a/internal/agent/finalize.go b/internal/agent/finalize.go index 28ef967b..329c171e 100644 --- a/internal/agent/finalize.go +++ b/internal/agent/finalize.go @@ -5,6 +5,7 @@ import ( "fmt" "time" + agenttools "github.com/Tencent/WeKnora/internal/agent/tools" "github.com/Tencent/WeKnora/internal/common" "github.com/Tencent/WeKnora/internal/event" "github.com/Tencent/WeKnora/internal/logger" @@ -86,12 +87,17 @@ Now generate the final answer:`, query) // Generate a single ID for this entire final answer stream answerID := generateEventID("answer") logger.Debugf(ctx, "[Agent][FinalAnswer] AnswerID: %s", answerID) + answerDoneEmitted := false llmResult, err := e.streamLLMToEventBus( ctx, messages, - &chat.ChatOptions{Temperature: e.config.Temperature, Thinking: e.config.Thinking}, + &chat.ChatOptions{Temperature: e.config.Temperature}, // Thinking disabled for final answer synthesis func(chunk *types.StreamResponse, fullContent string) { + // Defensive filter: only emit answer content, skip thinking chunks + if chunk.ResponseType == types.ResponseTypeThinking { + return + } if chunk.Content != "" { logger.Debugf(ctx, "[Agent][FinalAnswer] Emitting answer chunk: %d chars", len(chunk.Content)) e.eventBus.Emit(ctx, event.Event{ @@ -103,6 +109,9 @@ Now generate the final answer:`, query) Done: chunk.Done, }, }) + if chunk.Done { + answerDoneEmitted = true + } } }, ) @@ -115,7 +124,20 @@ Now generate the final answer:`, query) return err } - fullAnswer := llmResult.Content + if !answerDoneEmitted { + e.eventBus.Emit(ctx, event.Event{ + ID: answerID, + Type: event.EventAgentFinalAnswer, + SessionID: sessionID, + Data: event.AgentFinalAnswerData{ + Content: "", + Done: true, + }, + }) + } + + // Safety net: strip any residual blocks that may have leaked through + fullAnswer := agenttools.StripThinkBlocks(llmResult.Content) logger.Infof(ctx, "[Agent][FinalAnswer] Final answer generated: %d characters", len(fullAnswer)) common.PipelineInfo(ctx, "Agent", "final_answer_done", map[string]interface{}{ "session_id": sessionID, diff --git a/internal/agent/think.go b/internal/agent/think.go index 102adb2d..1b1f1a44 100644 --- a/internal/agent/think.go +++ b/internal/agent/think.go @@ -15,11 +15,12 @@ import ( // streamLLMResult holds accumulated output from a streaming LLM call. type streamLLMResult struct { - Content string - ToolCalls []types.LLMToolCall - Usage *types.TokenUsage - FinishReason string // actual finish_reason from LLM (captured from last stream chunk) - StreamError string // error message from stream (e.g., timeout), kept separate from Content + Content string + ThinkingContent string // accumulated thinking/reasoning content, kept separate from answer + ToolCalls []types.LLMToolCall + Usage *types.TokenUsage + FinishReason string // actual finish_reason from LLM (captured from last stream chunk) + StreamError string // error message from stream (e.g., timeout), kept separate from Content } // streamLLMToEventBus streams LLM response through EventBus (generic method) @@ -64,7 +65,11 @@ func (e *AgentEngine) streamLLMToEventBus( if chunk.Content != "" { isExtracted := chunk.Data != nil && chunk.Data["source"] != nil if !isExtracted { - result.Content += chunk.Content + if chunk.ResponseType == types.ResponseTypeThinking { + result.ThinkingContent += chunk.Content + } else { + result.Content += chunk.Content + } } }