refactor: remove legacy compatibility configuration and UI components

internal/httpapi/openai/chat/chat_history_test.go

@@ -57,7 +57,7 @@ func blockChatHistoryDetailDir(t *testing.T, detailDir string) func() {
 func TestChatCompletionsNonStreamPersistsHistory(t *testing.T) {
 	historyStore := newTestChatHistoryStore(t)
 	h := &Handler{
-		Store:       mockOpenAIConfig{wideInput: true},
+		Store:       mockOpenAIConfig{},
 		Auth:        streamStatusAuthStub{},
 		DS:          streamStatusDSStub{resp: makeOpenAISSEHTTPResponse(`data: {"p":"response/content","v":"hello world"}`, `data: [DONE]`)},
 		ChatHistory: historyStore,
@@ -216,7 +216,7 @@ func TestHandleStreamContextCancelledMarksHistoryStopped(t *testing.T) {
 func TestChatCompletionsSkipsAdminWebUISource(t *testing.T) {
 	historyStore := newTestChatHistoryStore(t)
 	h := &Handler{
-		Store:       mockOpenAIConfig{wideInput: true},
+		Store:       mockOpenAIConfig{},
 		Auth:        streamStatusAuthStub{},
 		DS:          streamStatusDSStub{resp: makeOpenAISSEHTTPResponse(`data: {"p":"response/content","v":"hello world"}`, `data: [DONE]`)},
 		ChatHistory: historyStore,
@@ -248,7 +248,7 @@ func TestChatCompletionsSkipsHistoryWhenDisabled(t *testing.T) {
 		t.Fatalf("disable history store failed: %v", err)
 	}
 	h := &Handler{
-		Store:       mockOpenAIConfig{wideInput: true},
+		Store:       mockOpenAIConfig{},
 		Auth:        streamStatusAuthStub{},
 		DS:          streamStatusDSStub{resp: makeOpenAISSEHTTPResponse(`data: {"p":"response/content","v":"hello world"}`, `data: [DONE]`)},
 		ChatHistory: historyStore,
@@ -278,7 +278,6 @@ func TestChatCompletionsCurrentInputFilePersistsNeutralPrompt(t *testing.T) {
 	ds := &inlineUploadDSStub{}
 	h := &Handler{
 		Store: mockOpenAIConfig{
-			wideInput:           true,
 			currentInputEnabled: true,
 		},
 		Auth:        streamStatusAuthStub{},

internal/httpapi/openai/chat/chat_stream_runtime.go

@@ -230,7 +230,6 @@ func (s *chatStreamRuntime) finalize(finishReason string, deferEmptyOutput bool)
 	s.finalThinking = turn.Thinking
 	s.finalText = turn.Text
 	if len(turn.ToolCalls) > 0 && !s.toolCallsDoneEmitted {
-		finishReason = "tool_calls"
 		s.sendDelta(map[string]any{
 			"tool_calls": formatFinalStreamToolCallsWithStableIDs(turn.ToolCalls, s.streamToolCallIDs, s.toolsRaw),
 		})
@@ -241,7 +240,6 @@ func (s *chatStreamRuntime) finalize(finishReason string, deferEmptyOutput bool)
 		for _, evt := range toolstream.Flush(&s.toolSieve, s.toolNames) {
 			if len(evt.ToolCalls) > 0 {
 				batch.flush()
-				finishReason = "tool_calls"
 				s.toolCallsEmitted = true
 				s.toolCallsDoneEmitted = true
 				s.sendDelta(map[string]any{
@@ -261,14 +259,11 @@ func (s *chatStreamRuntime) finalize(finishReason string, deferEmptyOutput bool)
 		batch.flush()
 	}
 
-	if len(turn.ToolCalls) > 0 || s.toolCallsEmitted {
-		finishReason = "tool_calls"
-	}
-	if len(turn.ToolCalls) == 0 && !s.toolCallsEmitted && strings.TrimSpace(turn.Text) == "" {
-		status, message, code := upstreamEmptyOutputDetail(finishReason == "content_filter", turn.Text, turn.Thinking)
-		if turn.Error != nil {
-			status, message, code = turn.Error.Status, turn.Error.Message, turn.Error.Code
-		}
+	outcome := assistantturn.FinalizeTurn(turn, assistantturn.FinalizeOptions{
+		AlreadyEmittedToolCalls: s.toolCallsEmitted || s.toolCallsDoneEmitted,
+	})
+	if outcome.ShouldFail {
+		status, message, code := outcome.Error.Status, outcome.Error.Message, outcome.Error.Code
 		if deferEmptyOutput {
 			s.finalErrorStatus = status
 			s.finalErrorMessage = message
@@ -278,31 +273,20 @@ func (s *chatStreamRuntime) finalize(finishReason string, deferEmptyOutput bool)
 		s.sendFailedChunk(status, message, code)
 		return true
 	}
-	usage := chatUsageFromTurn(turn)
-	s.finalFinishReason = finishReason
+	usage := assistantturn.OpenAIChatUsage(turn)
+	s.finalFinishReason = outcome.FinishReason
 	s.finalUsage = usage
 	s.sendChunk(openaifmt.BuildChatStreamChunk(
 		s.completionID,
 		s.created,
 		s.model,
-		[]map[string]any{openaifmt.BuildChatStreamFinishChoice(0, finishReason)},
+		[]map[string]any{openaifmt.BuildChatStreamFinishChoice(0, outcome.FinishReason)},
 		usage,
 	))
 	s.sendDone()
 	return true
 }
 
-func chatUsageFromTurn(turn assistantturn.Turn) map[string]any {
-	return map[string]any{
-		"prompt_tokens":     turn.Usage.InputTokens,
-		"completion_tokens": turn.Usage.OutputTokens,
-		"total_tokens":      turn.Usage.TotalTokens,
-		"completion_tokens_details": map[string]any{
-			"reasoning_tokens": turn.Usage.ReasoningTokens,
-		},
-	}
-}
-
 func (s *chatStreamRuntime) onParsed(parsed sse.LineResult) streamengine.ParsedDecision {
 	if !parsed.Parsed {
 		return streamengine.ParsedDecision{}

internal/httpapi/openai/chat/empty_retry_runtime.go

@@ -93,7 +93,7 @@ func (h *Handler) collectChatNonStreamAttempt(w http.ResponseWriter, resp *http.
 		Model:                 model,
 		Prompt:                usagePrompt,
 		SearchEnabled:         searchEnabled,
-		StripReferenceMarkers: h.compatStripReferenceMarkers(),
+		StripReferenceMarkers: stripReferenceMarkersEnabled(),
 		ToolNames:             toolNames,
 		ToolsRaw:              toolsRaw,
 	})
@@ -223,7 +223,7 @@ func (h *Handler) prepareChatStreamRuntime(w http.ResponseWriter, resp *http.Res
 	}
 	streamRuntime := newChatStreamRuntime(
 		w, rc, canFlush, completionID, time.Now().Unix(), model, finalPrompt,
-		thinkingEnabled, searchEnabled, h.compatStripReferenceMarkers(), toolNames, toolsRaw,
+		thinkingEnabled, searchEnabled, stripReferenceMarkersEnabled(), toolNames, toolsRaw,
 		toolChoice,
 		len(toolNames) > 0, h.toolcallFeatureMatchEnabled() && h.toolcallEarlyEmitHighConfidence(),
 	)

internal/httpapi/openai/chat/handler.go

@@ -35,11 +35,8 @@ type streamLease struct {
 	ExpiresAt time.Time
 }
 
-func (h *Handler) compatStripReferenceMarkers() bool {
-	if h == nil {
-		return true
-	}
-	return shared.CompatStripReferenceMarkers(h.Store)
+func stripReferenceMarkersEnabled() bool {
+	return true
 }
 
 func (h *Handler) applyCurrentInputFile(ctx context.Context, a *auth.RequestAuth, stdReq promptcompat.StandardRequest) (promptcompat.StandardRequest, error) {
@@ -108,22 +105,10 @@ func cleanVisibleOutput(text string, stripReferenceMarkers bool) string {
 	return shared.CleanVisibleOutput(text, stripReferenceMarkers)
 }
 
-func replaceCitationMarkersWithLinks(text string, links map[int]string) string {
-	return shared.ReplaceCitationMarkersWithLinks(text, links)
-}
-
-func shouldWriteUpstreamEmptyOutputError(text, thinking string) bool {
-	return shared.ShouldWriteUpstreamEmptyOutputError(text, thinking)
-}
-
 func upstreamEmptyOutputDetail(contentFilter bool, text, thinking string) (int, string, string) {
 	return shared.UpstreamEmptyOutputDetail(contentFilter, text, thinking)
 }
 
-func writeUpstreamEmptyOutputError(w http.ResponseWriter, text, thinking string, contentFilter bool) bool {
-	return shared.WriteUpstreamEmptyOutputError(w, text, thinking, contentFilter)
-}
-
 func emptyOutputRetryEnabled() bool {
 	return shared.EmptyOutputRetryEnabled()
 }

internal/httpapi/openai/chat/handler_chat.go

@@ -8,6 +8,7 @@ import (
 	"strings"
 	"time"
 
+	"ds2api/internal/assistantturn"
 	"ds2api/internal/auth"
 	"ds2api/internal/completionruntime"
 	"ds2api/internal/config"
@@ -79,7 +80,7 @@ func (h *Handler) ChatCompletions(w http.ResponseWriter, r *http.Request) {
 
 	if !stdReq.Stream {
 		result, outErr := completionruntime.ExecuteNonStreamWithRetry(r.Context(), h.DS, a, stdReq, completionruntime.Options{
-			StripReferenceMarkers: h.compatStripReferenceMarkers(),
+			StripReferenceMarkers: stripReferenceMarkersEnabled(),
 			RetryEnabled:          true,
 			CurrentInputFile:      h.Store,
 		})
@@ -92,10 +93,10 @@ func (h *Handler) ChatCompletions(w http.ResponseWriter, r *http.Request) {
 			return
 		}
 		respBody := openaifmt.BuildChatCompletionWithToolCalls(result.SessionID, stdReq.ResponseModel, result.Turn.Prompt, result.Turn.Thinking, result.Turn.Text, result.Turn.ToolCalls, stdReq.ToolsRaw)
-		respBody["usage"] = chatUsageFromTurn(result.Turn)
-		finishReason := chatFinishReason(respBody)
+		respBody["usage"] = assistantturn.OpenAIChatUsage(result.Turn)
+		finishReason := assistantturn.FinalizeTurn(result.Turn, assistantturn.FinalizeOptions{}).FinishReason
 		if historySession != nil {
-			historySession.success(http.StatusOK, result.Turn.Thinking, result.Turn.Text, finishReason, chatUsageFromTurn(result.Turn))
+			historySession.success(http.StatusOK, result.Turn.Thinking, result.Turn.Text, finishReason, assistantturn.OpenAIChatUsage(result.Turn))
 		}
 		writeJSON(w, http.StatusOK, respBody)
 		return
@@ -162,33 +163,29 @@ func (h *Handler) handleNonStream(w http.ResponseWriter, resp *http.Response, co
 	}
 	result := sse.CollectStream(resp, thinkingEnabled, true)
 
-	stripReferenceMarkers := h.compatStripReferenceMarkers()
-	finalThinking := cleanVisibleOutput(result.Thinking, stripReferenceMarkers)
-	finalText := cleanVisibleOutput(result.Text, stripReferenceMarkers)
-	if searchEnabled {
-		finalText = replaceCitationMarkersWithLinks(finalText, result.CitationLinks)
-	}
-	detected := detectAssistantToolCalls(result.Text, finalText, result.Thinking, result.ToolDetectionThinking, toolNames)
-	if shouldWriteUpstreamEmptyOutputError(finalText, finalThinking) && len(detected.Calls) == 0 {
-		status, message, code := upstreamEmptyOutputDetail(result.ContentFilter, finalText, finalThinking)
+	turn := assistantturn.BuildTurnFromCollected(result, assistantturn.BuildOptions{
+		Model:                 model,
+		Prompt:                finalPrompt,
+		RefFileTokens:         refFileTokens,
+		SearchEnabled:         searchEnabled,
+		StripReferenceMarkers: stripReferenceMarkersEnabled(),
+		ToolNames:             toolNames,
+		ToolsRaw:              toolsRaw,
+		ToolChoice:            promptcompat.DefaultToolChoicePolicy(),
+	})
+	outcome := assistantturn.FinalizeTurn(turn, assistantturn.FinalizeOptions{})
+	if outcome.ShouldFail {
+		status, message, code := outcome.Error.Status, outcome.Error.Message, outcome.Error.Code
 		if historySession != nil {
-			historySession.error(status, message, code, finalThinking, finalText)
+			historySession.error(status, message, code, turn.Thinking, turn.Text)
 		}
-		writeUpstreamEmptyOutputError(w, finalText, finalThinking, result.ContentFilter)
+		writeOpenAIErrorWithCode(w, status, message, code)
 		return
 	}
-	respBody := openaifmt.BuildChatCompletionWithToolCalls(completionID, model, finalPrompt, finalThinking, finalText, detected.Calls, toolsRaw)
-	if refFileTokens > 0 {
-		addRefFileTokensToUsage(respBody, refFileTokens)
-	}
-	finishReason := "stop"
-	if choices, ok := respBody["choices"].([]map[string]any); ok && len(choices) > 0 {
-		if fr, _ := choices[0]["finish_reason"].(string); strings.TrimSpace(fr) != "" {
-			finishReason = fr
-		}
-	}
+	respBody := openaifmt.BuildChatCompletionWithToolCalls(completionID, model, finalPrompt, turn.Thinking, turn.Text, turn.ToolCalls, toolsRaw)
+	respBody["usage"] = assistantturn.OpenAIChatUsage(turn)
 	if historySession != nil {
-		historySession.success(http.StatusOK, finalThinking, finalText, finishReason, openaifmt.BuildChatUsageForModel(model, finalPrompt, finalThinking, finalText, refFileTokens))
+		historySession.success(http.StatusOK, turn.Thinking, turn.Text, outcome.FinishReason, assistantturn.OpenAIChatUsage(turn))
 	}
 	writeJSON(w, http.StatusOK, respBody)
 }
@@ -216,7 +213,7 @@ func (h *Handler) handleStream(w http.ResponseWriter, r *http.Request, resp *htt
 	created := time.Now().Unix()
 	bufferToolContent := len(toolNames) > 0
 	emitEarlyToolDeltas := h.toolcallFeatureMatchEnabled() && h.toolcallEarlyEmitHighConfidence()
-	stripReferenceMarkers := h.compatStripReferenceMarkers()
+	stripReferenceMarkers := stripReferenceMarkersEnabled()
 	initialType := "text"
 	if thinkingEnabled {
 		initialType = "thinking"

internal/httpapi/openai/chat/handler_chat_auto_delete_test.go

@@ -75,7 +75,6 @@ func TestChatCompletionsAutoDeleteModes(t *testing.T) {
 			}
 			h := &Handler{
 				Store: mockOpenAIConfig{
-					wideInput:      true,
 					autoDeleteMode: tc.mode,
 				},
 				Auth: streamStatusAuthStub{},
@@ -123,7 +122,6 @@ func TestAutoDeleteRemoteSessionIgnoresCanceledParentContext(t *testing.T) {
 	ds := &autoDeleteCtxDSStub{}
 	h := &Handler{
 		Store: mockOpenAIConfig{
-			wideInput:      true,
 			autoDeleteMode: "single",
 		},
 		DS: ds,

internal/httpapi/openai/chat/handler_toolcall_test.go

@@ -133,18 +133,13 @@ func TestHandleNonStreamReturns429WhenUpstreamHasOnlyThinking(t *testing.T) {
 	rec := httptest.NewRecorder()
 
 	h.handleNonStream(rec, resp, "cid-thinking-only", "deepseek-v4-pro", "prompt", 0, true, false, nil, nil, nil)
-	if rec.Code != http.StatusOK {
-		t.Fatalf("expected status 200 for thinking-only upstream output, got %d body=%s", rec.Code, rec.Body.String())
+	if rec.Code != http.StatusTooManyRequests {
+		t.Fatalf("expected status 429 for thinking-only upstream output, got %d body=%s", rec.Code, rec.Body.String())
 	}
 	out := decodeJSONBody(t, rec.Body.String())
-	choices, _ := out["choices"].([]any)
-	if len(choices) == 0 {
-		t.Fatal("expected at least one choice")
-	}
-	first, _ := choices[0].(map[string]any)
-	msg, _ := first["message"].(map[string]any)
-	if asString(msg["reasoning_content"]) != "Only thinking" {
-		t.Fatalf("expected reasoning_content='Only thinking', got %#v", msg)
+	errObj, _ := out["error"].(map[string]any)
+	if asString(errObj["code"]) != "upstream_empty_output" {
+		t.Fatalf("expected code=upstream_empty_output, got %#v", out)
 	}
 }
 

internal/httpapi/openai/chat/test_helpers_test.go

@@ -12,7 +12,6 @@ import (
 
 type mockOpenAIConfig struct {
 	aliases             map[string]string
-	wideInput           bool
 	autoDeleteMode      string
 	toolMode            string
 	earlyEmit           string
@@ -24,11 +23,7 @@ type mockOpenAIConfig struct {
 	thinkingPrompt      string
 }
 
-func (m mockOpenAIConfig) ModelAliases() map[string]string { return m.aliases }
-func (m mockOpenAIConfig) CompatWideInputStrictOutput() bool {
-	return m.wideInput
-}
-func (m mockOpenAIConfig) CompatStripReferenceMarkers() bool   { return true }
+func (m mockOpenAIConfig) ModelAliases() map[string]string     { return m.aliases }
 func (m mockOpenAIConfig) ToolcallMode() string                { return m.toolMode }
 func (m mockOpenAIConfig) ToolcallEarlyEmitConfidence() string { return m.earlyEmit }
 func (m mockOpenAIConfig) ResponsesStoreTTLSeconds() int       { return m.responsesTTL }

internal/httpapi/openai/chat/vercel_prepare_test.go

@@ -94,7 +94,6 @@ func TestHandleVercelStreamPrepareAppliesCurrentInputFile(t *testing.T) {
 	ds := &inlineUploadDSStub{}
 	h := &Handler{
 		Store: mockOpenAIConfig{
-			wideInput:           true,
 			currentInputEnabled: true,
 		},
 		Auth: streamStatusAuthStub{},
@@ -151,7 +150,6 @@ func TestHandleVercelStreamPrepareMapsCurrentInputFileManagedAuthFailureTo401(t
 	}
 	h := &Handler{
 		Store: mockOpenAIConfig{
-			wideInput:           true,
 			currentInputEnabled: true,
 		},
 		Auth: streamStatusManagedAuthStub{},

internal/httpapi/openai/chat/vercel_stream.go

@@ -109,13 +109,10 @@ func (h *Handler) handleVercelStreamPrepare(w http.ResponseWriter, r *http.Reque
 		"final_prompt":     stdReq.FinalPrompt,
 		"thinking_enabled": stdReq.Thinking,
 		"search_enabled":   stdReq.Search,
-		"compat": map[string]any{
-			"strip_reference_markers": h.compatStripReferenceMarkers(),
-		},
-		"tool_names":     stdReq.ToolNames,
-		"deepseek_token": a.DeepSeekToken,
-		"pow_header":     powHeader,
-		"payload":        payload,
+		"tool_names":       stdReq.ToolNames,
+		"deepseek_token":   a.DeepSeekToken,
+		"pow_header":       powHeader,
+		"payload":          payload,
 	})
 }