Merge pull request #1895 from gaoren002/fix/codex-spark-limitations

fix(openai): handle codex spark model limitations

backend/internal/service/openai_codex_transform.go

@@ -48,6 +48,8 @@ type codexTransformResult struct {
 const (
 	codexImageGenerationBridgeMarker = "<sub2api-codex-image-generation>"
 	codexImageGenerationBridgeText   = codexImageGenerationBridgeMarker + "\nWhen the user asks for raster image generation or editing, use the OpenAI Responses native `image_generation` tool attached to this request. The local Codex client may not expose an `image_gen` namespace, but that does not mean image generation is unavailable. Do not ask the user to switch to CLI fallback solely because `image_gen` is absent.\n</sub2api-codex-image-generation>"
+	codexSparkImageUnsupportedMarker = "<sub2api-codex-spark-image-unsupported>"
+	codexSparkImageUnsupportedText   = codexSparkImageUnsupportedMarker + "\nThe current model is gpt-5.3-codex-spark, which does not support image generation, image editing, image input, the `image_generation` tool, or Codex `image_gen`/`$imagegen` workflows. If the user asks for image generation or image editing, clearly explain this model limitation and ask them to switch to a non-Spark Codex model such as gpt-5.3-codex or gpt-5.4. Do not claim that the local environment merely lacks image_gen tooling, and do not suggest CLI fallback as the primary fix while the model remains Spark.\n</sub2api-codex-spark-image-unsupported>"
 )
 
 func applyCodexOAuthTransform(reqBody map[string]any, isCodexCLI bool, isCompact bool) codexTransformResult {
@@ -165,6 +167,9 @@ func applyCodexOAuthTransform(reqBody map[string]any, isCodexCLI bool, isCompact
 	if applyInstructions(reqBody, isCodexCLI) {
 		result.Modified = true
 	}
+	if isCodexSparkModel(normalizedModel) && applyCodexSparkImageUnsupportedInstructions(reqBody) {
+		result.Modified = true
+	}
 
 	// 续链场景保留 item_reference 与 id，避免 call_id 上下文丢失。
 	if input, ok := reqBody["input"].([]any); ok {
@@ -244,6 +249,10 @@ func normalizeCodexModel(model string) string {
 	return "gpt-5.4"
 }
 
+func isCodexSparkModel(model string) bool {
+	return normalizeCodexModel(model) == "gpt-5.3-codex-spark"
+}
+
 func hasOpenAIImageGenerationTool(reqBody map[string]any) bool {
 	rawTools, ok := reqBody["tools"]
 	if !ok || rawTools == nil {
@@ -265,6 +274,40 @@ func hasOpenAIImageGenerationTool(reqBody map[string]any) bool {
 	return false
 }
 
+func hasOpenAIInputImage(reqBody map[string]any) bool {
+	if reqBody == nil {
+		return false
+	}
+	return hasOpenAIInputImageValue(reqBody["input"]) || hasOpenAIInputImageValue(reqBody["messages"])
+}
+
+func hasOpenAIInputImageValue(value any) bool {
+	switch v := value.(type) {
+	case []any:
+		for _, item := range v {
+			if hasOpenAIInputImageValue(item) {
+				return true
+			}
+		}
+	case map[string]any:
+		if strings.TrimSpace(firstNonEmptyString(v["type"])) == "input_image" {
+			return true
+		}
+		if _, ok := v["image_url"]; ok {
+			return true
+		}
+		return hasOpenAIInputImageValue(v["content"])
+	}
+	return false
+}
+
+func validateCodexSparkInput(reqBody map[string]any, model string) error {
+	if !isCodexSparkModel(model) || !hasOpenAIInputImage(reqBody) {
+		return nil
+	}
+	return fmt.Errorf("model %q does not support image input", strings.TrimSpace(model))
+}
+
 func normalizeOpenAIResponsesImageGenerationTools(reqBody map[string]any) bool {
 	rawTools, ok := reqBody["tools"]
 	if !ok || rawTools == nil {
@@ -309,6 +352,9 @@ func ensureOpenAIResponsesImageGenerationTool(reqBody map[string]any) bool {
 	if len(reqBody) == 0 {
 		return false
 	}
+	if isCodexSparkModel(firstNonEmptyString(reqBody["model"])) {
+		return false
+	}
 
 	tool := map[string]any{
 		"type":          "image_generation",
@@ -344,6 +390,9 @@ func applyCodexImageGenerationBridgeInstructions(reqBody map[string]any) bool {
 	if len(reqBody) == 0 || !hasOpenAIImageGenerationTool(reqBody) {
 		return false
 	}
+	if isCodexSparkModel(firstNonEmptyString(reqBody["model"])) {
+		return false
+	}
 
 	existing, _ := reqBody["instructions"].(string)
 	if strings.Contains(existing, codexImageGenerationBridgeMarker) {
@@ -360,6 +409,23 @@ func applyCodexImageGenerationBridgeInstructions(reqBody map[string]any) bool {
 	return true
 }
 
+func applyCodexSparkImageUnsupportedInstructions(reqBody map[string]any) bool {
+	if len(reqBody) == 0 {
+		return false
+	}
+	existing, _ := reqBody["instructions"].(string)
+	if strings.Contains(existing, codexSparkImageUnsupportedMarker) {
+		return false
+	}
+	existing = strings.TrimRight(existing, " \t\r\n")
+	if strings.TrimSpace(existing) == "" {
+		reqBody["instructions"] = codexSparkImageUnsupportedText
+		return true
+	}
+	reqBody["instructions"] = existing + "\n\n" + codexSparkImageUnsupportedText
+	return true
+}
+
 func validateOpenAIResponsesImageModel(reqBody map[string]any, model string) error {
 	if !hasOpenAIImageGenerationTool(reqBody) {
 		return nil

backend/internal/service/openai_codex_transform_test.go

@@ -333,6 +333,17 @@ func TestEnsureOpenAIResponsesImageGenerationTool_NoTools(t *testing.T) {
 	require.Equal(t, "png", tool["output_format"])
 }
 
+func TestEnsureOpenAIResponsesImageGenerationTool_SkipsSpark(t *testing.T) {
+	reqBody := map[string]any{
+		"model": "gpt-5.3-codex-spark",
+		"input": "draw a cat",
+	}
+
+	modified := ensureOpenAIResponsesImageGenerationTool(reqBody)
+	require.False(t, modified)
+	require.NotContains(t, reqBody, "tools")
+}
+
 func TestEnsureOpenAIResponsesImageGenerationTool_AppendsToExistingTools(t *testing.T) {
 	reqBody := map[string]any{
 		"model": "gpt-5.4",
@@ -378,6 +389,7 @@ func TestEnsureOpenAIResponsesImageGenerationTool_PreservesExistingImageTool(t *
 
 func TestApplyCodexImageGenerationBridgeInstructions_AppendsBridgeOnce(t *testing.T) {
 	reqBody := map[string]any{
+		"model":        "gpt-5.4",
 		"instructions": "existing instructions",
 		"tools": []any{
 			map[string]any{"type": "image_generation", "output_format": "png"},
@@ -397,6 +409,20 @@ func TestApplyCodexImageGenerationBridgeInstructions_AppendsBridgeOnce(t *testin
 	require.False(t, modified)
 }
 
+func TestApplyCodexImageGenerationBridgeInstructions_SkipsSpark(t *testing.T) {
+	reqBody := map[string]any{
+		"model":        "gpt-5.3-codex-spark",
+		"instructions": "existing instructions",
+		"tools": []any{
+			map[string]any{"type": "image_generation", "output_format": "png"},
+		},
+	}
+
+	modified := applyCodexImageGenerationBridgeInstructions(reqBody)
+	require.False(t, modified)
+	require.Equal(t, "existing instructions", reqBody["instructions"])
+}
+
 func TestApplyCodexImageGenerationBridgeInstructions_SkipsWithoutImageTool(t *testing.T) {
 	reqBody := map[string]any{
 		"instructions": "existing instructions",
@@ -410,6 +436,91 @@ func TestApplyCodexImageGenerationBridgeInstructions_SkipsWithoutImageTool(t *te
 	require.Equal(t, "existing instructions", reqBody["instructions"])
 }
 
+func TestValidateCodexSparkInputRejectsInputImage(t *testing.T) {
+	reqBody := map[string]any{
+		"model": "gpt-5.3-codex-spark",
+		"input": []any{
+			map[string]any{
+				"role": "user",
+				"content": []any{
+					map[string]any{"type": "input_text", "text": "describe"},
+					map[string]any{"type": "input_image", "image_url": "data:image/png;base64,aGVsbG8="},
+				},
+			},
+		},
+	}
+
+	err := validateCodexSparkInput(reqBody, "gpt-5.3-codex-spark")
+	require.Error(t, err)
+	require.Contains(t, err.Error(), "does not support image input")
+}
+
+func TestValidateCodexSparkInputRejectsChatImageURL(t *testing.T) {
+	reqBody := map[string]any{
+		"model": "gpt-5.3-codex-spark",
+		"messages": []any{
+			map[string]any{
+				"role": "user",
+				"content": []any{
+					map[string]any{"type": "text", "text": "describe"},
+					map[string]any{"type": "image_url", "image_url": map[string]any{"url": "data:image/png;base64,aGVsbG8="}},
+				},
+			},
+		},
+	}
+
+	err := validateCodexSparkInput(reqBody, "gpt-5.3-codex-spark")
+	require.Error(t, err)
+}
+
+func TestValidateCodexSparkInputAllowsTextOnly(t *testing.T) {
+	reqBody := map[string]any{
+		"model": "gpt-5.3-codex-spark",
+		"input": []any{
+			map[string]any{
+				"role": "user",
+				"content": []any{
+					map[string]any{"type": "input_text", "text": "hello"},
+				},
+			},
+		},
+	}
+
+	require.NoError(t, validateCodexSparkInput(reqBody, "gpt-5.3-codex-spark"))
+}
+
+func TestApplyCodexOAuthTransform_AddsSparkImageUnsupportedInstructions(t *testing.T) {
+	reqBody := map[string]any{
+		"model":        "gpt-5.3-codex-spark",
+		"instructions": "existing instructions",
+		"input":        "hello",
+	}
+
+	result := applyCodexOAuthTransform(reqBody, true, false)
+	require.True(t, result.Modified)
+
+	instructions, ok := reqBody["instructions"].(string)
+	require.True(t, ok)
+	require.Contains(t, instructions, "existing instructions")
+	require.Contains(t, instructions, codexSparkImageUnsupportedMarker)
+	require.Contains(t, instructions, "does not support image generation")
+	require.Contains(t, instructions, "switch to a non-Spark Codex model")
+	require.NotContains(t, instructions, codexImageGenerationBridgeMarker)
+}
+
+func TestApplyCodexOAuthTransform_DoesNotAddSparkImageUnsupportedForNonSpark(t *testing.T) {
+	reqBody := map[string]any{
+		"model":        "gpt-5.4",
+		"instructions": "existing instructions",
+		"input":        "hello",
+	}
+
+	applyCodexOAuthTransform(reqBody, true, false)
+	instructions, ok := reqBody["instructions"].(string)
+	require.True(t, ok)
+	require.NotContains(t, instructions, codexSparkImageUnsupportedMarker)
+}
+
 func TestNormalizeOpenAIResponsesImageOnlyModel_BuildsImageToolRequest(t *testing.T) {
 	reqBody := map[string]any{
 		"model":         "gpt-image-2",

backend/internal/service/openai_gateway_service.go

@@ -1995,6 +1995,17 @@ func (s *OpenAIGatewayService) Forward(ctx context.Context, c *gin.Context, acco
 			account.Type,
 		)
 	}
+	if err := validateCodexSparkInput(reqBody, upstreamModel); err != nil {
+		setOpsUpstreamError(c, http.StatusBadRequest, err.Error(), "")
+		c.JSON(http.StatusBadRequest, gin.H{
+			"error": gin.H{
+				"type":    "invalid_request_error",
+				"message": err.Error(),
+				"param":   "input",
+			},
+		})
+		return nil, err
+	}
 
 	// OpenAI OAuth 账号走 ChatGPT internal Codex endpoint，需要将模型名规范化为
 	// 上游可识别的 Codex/GPT 系列。API Key 账号则应保留原始/映射后的模型名，

backend/internal/service/openai_model_mapping.go

@@ -1,5 +1,7 @@
 package service
 
+import "strings"
+
 // resolveOpenAIForwardModel determines the upstream model for OpenAI-compatible
 // forwarding. Group-level default mapping only applies when the account itself
 // did not match any explicit model_mapping rule.
@@ -12,8 +14,28 @@ func resolveOpenAIForwardModel(account *Account, requestedModel, defaultMappedMo
 	}
 
 	mappedModel, matched := account.ResolveMappedModel(requestedModel)
-	if !matched && defaultMappedModel != "" {
+	if !matched && defaultMappedModel != "" && !isExplicitCodexModel(requestedModel) {
 		return defaultMappedModel
 	}
 	return mappedModel
 }
+
+func isExplicitCodexModel(model string) bool {
+	model = strings.TrimSpace(model)
+	if model == "" {
+		return false
+	}
+	if strings.Contains(model, "/") {
+		parts := strings.Split(model, "/")
+		model = parts[len(parts)-1]
+	}
+	model = strings.ToLower(strings.TrimSpace(model))
+	if getNormalizedCodexModel(model) != "" {
+		return true
+	}
+	if strings.HasSuffix(model, "-openai-compact") {
+		base := strings.TrimSuffix(model, "-openai-compact")
+		return getNormalizedCodexModel(base) != ""
+	}
+	return false
+}

backend/internal/service/openai_model_mapping_test.go

@@ -15,10 +15,19 @@ func TestResolveOpenAIForwardModel(t *testing.T) {
 			account: &Account{
 				Credentials: map[string]any{},
 			},
-			requestedModel:     "gpt-5.4",
+			requestedModel:     "claude-opus-4-6",
 			defaultMappedModel: "gpt-4o-mini",
 			expectedModel:      "gpt-4o-mini",
 		},
+		{
+			name: "preserves explicit gpt-5.4 instead of group default",
+			account: &Account{
+				Credentials: map[string]any{},
+			},
+			requestedModel:     "gpt-5.4",
+			defaultMappedModel: "gpt-4o-mini",
+			expectedModel:      "gpt-5.4",
+		},
 		{
 			name: "preserves exact passthrough mapping instead of group default",
 			account: &Account{
@@ -58,6 +67,42 @@ func TestResolveOpenAIForwardModel(t *testing.T) {
 			defaultMappedModel: "gpt-4o-mini",
 			expectedModel:      "gpt-5.4",
 		},
+		{
+			name: "preserves codex spark instead of group default",
+			account: &Account{
+				Credentials: map[string]any{},
+			},
+			requestedModel:     "gpt-5.3-codex-spark",
+			defaultMappedModel: "gpt-5.4",
+			expectedModel:      "gpt-5.3-codex-spark",
+		},
+		{
+			name: "preserves gpt-5.5 instead of group default",
+			account: &Account{
+				Credentials: map[string]any{},
+			},
+			requestedModel:     "gpt-5.5",
+			defaultMappedModel: "gpt-5.4",
+			expectedModel:      "gpt-5.5",
+		},
+		{
+			name: "preserves openai namespaced gpt-5.5 instead of group default",
+			account: &Account{
+				Credentials: map[string]any{},
+			},
+			requestedModel:     "openai/gpt-5.5",
+			defaultMappedModel: "gpt-5.4",
+			expectedModel:      "openai/gpt-5.5",
+		},
+		{
+			name: "preserves compact gpt-5.5 instead of group default",
+			account: &Account{
+				Credentials: map[string]any{},
+			},
+			requestedModel:     "gpt-5.5-openai-compact",
+			defaultMappedModel: "gpt-5.4",
+			expectedModel:      "gpt-5.5-openai-compact",
+		},
 	}
 
 	for _, tt := range tests {