Merge branch 'Wei-Shaw:main' into main

backend/internal/service/antigravity_gateway_service.go

@@ -1309,7 +1309,7 @@ func (s *AntigravityGatewayService) Forward(ctx context.Context, c *gin.Context,
 		return nil, s.writeClaudeError(c, http.StatusForbidden, "permission_error", fmt.Sprintf("model %s not in whitelist", claudeReq.Model))
 	}
 	// 应用 thinking 模式自动后缀：如果 thinking 开启且目标是 claude-sonnet-4-5，自动改为 thinking 版本
-	thinkingEnabled := claudeReq.Thinking != nil && claudeReq.Thinking.Type == "enabled"
+	thinkingEnabled := claudeReq.Thinking != nil && (claudeReq.Thinking.Type == "enabled" || claudeReq.Thinking.Type == "adaptive")
 	mappedModel = applyThinkingModelSuffix(mappedModel, thinkingEnabled)
 
 	// 获取 access_token

backend/internal/service/antigravity_gateway_service_test.go

@@ -591,6 +591,75 @@ func TestHandleClaudeStreamingResponse_NormalComplete(t *testing.T) {
 	require.NotContains(t, body, "event: error")
 }
 
+// TestHandleGeminiStreamingResponse_ThoughtsTokenCount
+// 验证：Gemini 流式转发时 thoughtsTokenCount 被计入 OutputTokens
+func TestHandleGeminiStreamingResponse_ThoughtsTokenCount(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+	svc := newAntigravityTestService(&config.Config{
+		Gateway: config.GatewayConfig{MaxLineSize: defaultMaxLineSize},
+	})
+
+	rec := httptest.NewRecorder()
+	c, _ := gin.CreateTestContext(rec)
+	c.Request = httptest.NewRequest(http.MethodPost, "/", nil)
+
+	pr, pw := io.Pipe()
+	resp := &http.Response{StatusCode: http.StatusOK, Body: pr, Header: http.Header{}}
+
+	go func() {
+		defer func() { _ = pw.Close() }()
+		fmt.Fprintln(pw, `data: {"candidates":[{"content":{"parts":[{"text":"Hello"}]}}],"usageMetadata":{"promptTokenCount":100,"candidatesTokenCount":20,"thoughtsTokenCount":50}}`)
+		fmt.Fprintln(pw, "")
+		fmt.Fprintln(pw, `data: {"candidates":[{"content":{"parts":[{"text":" world"}]},"finishReason":"STOP"}],"usageMetadata":{"promptTokenCount":100,"candidatesTokenCount":30,"thoughtsTokenCount":80,"cachedContentTokenCount":10}}`)
+		fmt.Fprintln(pw, "")
+	}()
+
+	result, err := svc.handleGeminiStreamingResponse(c, resp, time.Now())
+	_ = pr.Close()
+
+	require.NoError(t, err)
+	require.NotNil(t, result)
+	require.NotNil(t, result.usage)
+	// promptTokenCount=100, cachedContentTokenCount=10 → InputTokens=90
+	require.Equal(t, 90, result.usage.InputTokens)
+	// candidatesTokenCount=30 + thoughtsTokenCount=80 → OutputTokens=110
+	require.Equal(t, 110, result.usage.OutputTokens)
+	require.Equal(t, 10, result.usage.CacheReadInputTokens)
+}
+
+// TestHandleClaudeStreamingResponse_ThoughtsTokenCount
+// 验证：Gemini→Claude 流式转换时 thoughtsTokenCount 被计入 OutputTokens
+func TestHandleClaudeStreamingResponse_ThoughtsTokenCount(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+	svc := newAntigravityTestService(&config.Config{
+		Gateway: config.GatewayConfig{MaxLineSize: defaultMaxLineSize},
+	})
+
+	rec := httptest.NewRecorder()
+	c, _ := gin.CreateTestContext(rec)
+	c.Request = httptest.NewRequest(http.MethodPost, "/", nil)
+
+	pr, pw := io.Pipe()
+	resp := &http.Response{StatusCode: http.StatusOK, Body: pr, Header: http.Header{}}
+
+	go func() {
+		defer func() { _ = pw.Close() }()
+		fmt.Fprintln(pw, `data: {"response":{"candidates":[{"content":{"parts":[{"text":"Hi"}]},"finishReason":"STOP"}],"usageMetadata":{"promptTokenCount":50,"candidatesTokenCount":10,"thoughtsTokenCount":25}}}`)
+		fmt.Fprintln(pw, "")
+	}()
+
+	result, err := svc.handleClaudeStreamingResponse(c, resp, time.Now(), "gemini-2.5-pro")
+	_ = pr.Close()
+
+	require.NoError(t, err)
+	require.NotNil(t, result)
+	require.NotNil(t, result.usage)
+	// promptTokenCount=50 → InputTokens=50
+	require.Equal(t, 50, result.usage.InputTokens)
+	// candidatesTokenCount=10 + thoughtsTokenCount=25 → OutputTokens=35
+	require.Equal(t, 35, result.usage.OutputTokens)
+}
+
 // --- 流式客户端断开检测测试 ---
 
 // TestStreamUpstreamResponse_ClientDisconnectDrainsUsage

backend/internal/service/antigravity_oauth_service.go

@@ -192,6 +192,43 @@ func (s *AntigravityOAuthService) RefreshToken(ctx context.Context, refreshToken
 	return nil, fmt.Errorf("token 刷新失败 (重试后): %w", lastErr)
 }
 
+// ValidateRefreshToken 用 refresh token 验证并获取完整的 token 信息（含 email 和 project_id）
+func (s *AntigravityOAuthService) ValidateRefreshToken(ctx context.Context, refreshToken string, proxyID *int64) (*AntigravityTokenInfo, error) {
+	var proxyURL string
+	if proxyID != nil {
+		proxy, err := s.proxyRepo.GetByID(ctx, *proxyID)
+		if err == nil && proxy != nil {
+			proxyURL = proxy.URL()
+		}
+	}
+
+	// 刷新 token
+	tokenInfo, err := s.RefreshToken(ctx, refreshToken, proxyURL)
+	if err != nil {
+		return nil, err
+	}
+
+	// 获取用户信息（email）
+	client := antigravity.NewClient(proxyURL)
+	userInfo, err := client.GetUserInfo(ctx, tokenInfo.AccessToken)
+	if err != nil {
+		fmt.Printf("[AntigravityOAuth] 警告: 获取用户信息失败: %v\n", err)
+	} else {
+		tokenInfo.Email = userInfo.Email
+	}
+
+	// 获取 project_id（容错，失败不阻塞）
+	projectID, loadErr := s.loadProjectIDWithRetry(ctx, tokenInfo.AccessToken, proxyURL, 3)
+	if loadErr != nil {
+		fmt.Printf("[AntigravityOAuth] 警告: 获取 project_id 失败（重试后）: %v\n", loadErr)
+		tokenInfo.ProjectIDMissing = true
+	} else {
+		tokenInfo.ProjectID = projectID
+	}
+
+	return tokenInfo, nil
+}
+
 func isNonRetryableAntigravityOAuthError(err error) bool {
 	msg := err.Error()
 	nonRetryable := []string{

backend/internal/service/gateway_request.go

@@ -101,9 +101,9 @@ func ParseGatewayRequest(body []byte, protocol string) (*ParsedRequest, error) {
 		}
 	}
 
-	// thinking: {type: "enabled"}
+	// thinking: {type: "enabled" | "adaptive"}
 	if rawThinking, ok := req["thinking"].(map[string]any); ok {
-		if t, ok := rawThinking["type"].(string); ok && t == "enabled" {
+		if t, ok := rawThinking["type"].(string); ok && (t == "enabled" || t == "adaptive") {
 			parsed.ThinkingEnabled = true
 		}
 	}
@@ -161,9 +161,9 @@ func parseIntegralNumber(raw any) (int, bool) {
 // Returns filtered body or original body if filtering fails (fail-safe)
 // This prevents 400 errors from invalid thinking block signatures
 //
-// Strategy:
-//   - When thinking.type != "enabled": Remove all thinking blocks
-//   - When thinking.type == "enabled": Only remove thinking blocks without valid signatures
+// 策略：
+//   - 当 thinking.type 不是 "enabled"/"adaptive"：移除所有 thinking 相关块
+//   - 当 thinking.type 是 "enabled"/"adaptive"：仅移除缺失/无效 signature 的 thinking 块（避免 400）
 //     (blocks with missing/empty/dummy signatures that would cause 400 errors)
 func FilterThinkingBlocks(body []byte) []byte {
 	return filterThinkingBlocksInternal(body, false)
@@ -489,9 +489,9 @@ func FilterSignatureSensitiveBlocksForRetry(body []byte) []byte {
 }
 
 // filterThinkingBlocksInternal removes invalid thinking blocks from request
-// Strategy:
-//   - When thinking.type != "enabled": Remove all thinking blocks
-//   - When thinking.type == "enabled": Only remove thinking blocks without valid signatures
+// 策略：
+//   - 当 thinking.type 不是 "enabled"/"adaptive"：移除所有 thinking 相关块
+//   - 当 thinking.type 是 "enabled"/"adaptive"：仅移除缺失/无效 signature 的 thinking 块
 func filterThinkingBlocksInternal(body []byte, _ bool) []byte {
 	// Fast path: if body doesn't contain "thinking", skip parsing
 	if !bytes.Contains(body, []byte(`"type":"thinking"`)) &&
@@ -511,7 +511,7 @@ func filterThinkingBlocksInternal(body []byte, _ bool) []byte {
 	// Check if thinking is enabled
 	thinkingEnabled := false
 	if thinking, ok := req["thinking"].(map[string]any); ok {
-		if thinkType, ok := thinking["type"].(string); ok && thinkType == "enabled" {
+		if thinkType, ok := thinking["type"].(string); ok && (thinkType == "enabled" || thinkType == "adaptive") {
 			thinkingEnabled = true
 		}
 	}

backend/internal/service/gateway_request_test.go

@@ -29,6 +29,14 @@ func TestParseGatewayRequest_ThinkingEnabled(t *testing.T) {
 	require.True(t, parsed.ThinkingEnabled)
 }
 
+func TestParseGatewayRequest_ThinkingAdaptiveEnabled(t *testing.T) {
+	body := []byte(`{"model":"claude-sonnet-4-5","thinking":{"type":"adaptive"},"messages":[{"content":"hi"}]}`)
+	parsed, err := ParseGatewayRequest(body, "")
+	require.NoError(t, err)
+	require.Equal(t, "claude-sonnet-4-5", parsed.Model)
+	require.True(t, parsed.ThinkingEnabled)
+}
+
 func TestParseGatewayRequest_MaxTokens(t *testing.T) {
 	body := []byte(`{"model":"claude-haiku-4-5","max_tokens":1}`)
 	parsed, err := ParseGatewayRequest(body, "")
@@ -209,6 +217,16 @@ func TestFilterThinkingBlocks(t *testing.T) {
 			input:        `{"model":"claude-3-5-sonnet-20241022","messages":[{"role":"user","content":[{"type":"text","text":"Hello"},{"type":"thinking","thinking":"internal","signature":"invalid"},{"type":"text","text":"World"}]}]}`,
 			shouldFilter: true,
 		},
+		{
+			name:         "does not filter signed thinking blocks when thinking adaptive",
+			input:        `{"thinking":{"type":"adaptive"},"messages":[{"role":"assistant","content":[{"type":"thinking","thinking":"ok","signature":"sig_real_123"},{"type":"text","text":"B"}]}]}`,
+			shouldFilter: false,
+		},
+		{
+			name:         "filters unsigned thinking blocks when thinking adaptive",
+			input:        `{"thinking":{"type":"adaptive"},"messages":[{"role":"assistant","content":[{"type":"thinking","thinking":"internal","signature":""},{"type":"text","text":"B"}]}]}`,
+			shouldFilter: true,
+		},
 		{
 			name:         "handles no thinking blocks",
 			input:        `{"model":"claude-3-5-sonnet-20241022","messages":[{"role":"user","content":[{"type":"text","text":"Hello"}]}]}`,

backend/internal/service/gateway_service.go

@@ -3631,7 +3631,8 @@ func requestNeedsBetaFeatures(body []byte) bool {
 	if tools.Exists() && tools.IsArray() && len(tools.Array()) > 0 {
 		return true
 	}
-	if strings.EqualFold(gjson.GetBytes(body, "thinking.type").String(), "enabled") {
+	thinkingType := gjson.GetBytes(body, "thinking.type").String()
+	if strings.EqualFold(thinkingType, "enabled") || strings.EqualFold(thinkingType, "adaptive") {
 		return true
 	}
 	return false

backend/internal/service/gemini_messages_compat_service.go

@@ -2663,11 +2663,12 @@ func extractGeminiUsage(geminiResp map[string]any) *ClaudeUsage {
 	prompt, _ := asInt(usageMeta["promptTokenCount"])
 	cand, _ := asInt(usageMeta["candidatesTokenCount"])
 	cached, _ := asInt(usageMeta["cachedContentTokenCount"])
+	thoughts, _ := asInt(usageMeta["thoughtsTokenCount"])
 	// 注意：Gemini 的 promptTokenCount 包含 cachedContentTokenCount，
 	// 但 Claude 的 input_tokens 不包含 cache_read_input_tokens，需要减去
 	return &ClaudeUsage{
 		InputTokens:          prompt - cached,
-		OutputTokens:         cand,
+		OutputTokens:         cand + thoughts,
 		CacheReadInputTokens: cached,
 	}
 }

backend/internal/service/gemini_messages_compat_service_test.go

@@ -4,6 +4,8 @@ import (
 	"encoding/json"
 	"strings"
 	"testing"
+
+	"github.com/stretchr/testify/require"
 )
 
 // TestConvertClaudeToolsToGeminiTools_CustomType 测试custom类型工具转换
@@ -203,3 +205,70 @@ func TestEnsureGeminiFunctionCallThoughtSignatures_InsertsWhenMissing(t *testing
 		t.Fatalf("expected injected thoughtSignature %q, got: %s", geminiDummyThoughtSignature, s)
 	}
 }
+
+func TestExtractGeminiUsage_ThoughtsTokenCount(t *testing.T) {
+	tests := []struct {
+		name          string
+		resp          map[string]any
+		wantInput     int
+		wantOutput    int
+		wantCacheRead int
+		wantNil       bool
+	}{
+		{
+			name: "with thoughtsTokenCount",
+			resp: map[string]any{
+				"usageMetadata": map[string]any{
+					"promptTokenCount":     float64(100),
+					"candidatesTokenCount": float64(20),
+					"thoughtsTokenCount":   float64(50),
+				},
+			},
+			wantInput:  100,
+			wantOutput: 70,
+		},
+		{
+			name: "with thoughtsTokenCount and cache",
+			resp: map[string]any{
+				"usageMetadata": map[string]any{
+					"promptTokenCount":        float64(100),
+					"candidatesTokenCount":    float64(20),
+					"cachedContentTokenCount": float64(30),
+					"thoughtsTokenCount":      float64(50),
+				},
+			},
+			wantInput:     70,
+			wantOutput:    70,
+			wantCacheRead: 30,
+		},
+		{
+			name: "without thoughtsTokenCount (old model)",
+			resp: map[string]any{
+				"usageMetadata": map[string]any{
+					"promptTokenCount":     float64(100),
+					"candidatesTokenCount": float64(20),
+				},
+			},
+			wantInput:  100,
+			wantOutput: 20,
+		},
+		{
+			name:    "no usageMetadata",
+			resp:    map[string]any{},
+			wantNil: true,
+		},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			usage := extractGeminiUsage(tt.resp)
+			if tt.wantNil {
+				require.Nil(t, usage)
+				return
+			}
+			require.NotNil(t, usage)
+			require.Equal(t, tt.wantInput, usage.InputTokens)
+			require.Equal(t, tt.wantOutput, usage.OutputTokens)
+			require.Equal(t, tt.wantCacheRead, usage.CacheReadInputTokens)
+		})
+	}
+}