feat(ops): propagate endpoint/request-type context in handlers; add UpstreamURL to upstream error events

backend/internal/handler/gateway_handler.go

@@ -178,6 +178,7 @@ func (h *GatewayHandler) Messages(c *gin.Context) {
 	c.Request = c.Request.WithContext(service.WithThinkingEnabled(c.Request.Context(), parsedReq.ThinkingEnabled, h.metadataBridgeEnabled()))
 
 	setOpsRequestContext(c, reqModel, reqStream, body)
+	setOpsEndpointContext(c, "", int16(service.RequestTypeFromLegacy(reqStream, false)))
 
 	// 验证 model 必填
 	if reqModel == "" {
@@ -1396,6 +1397,7 @@ func (h *GatewayHandler) CountTokens(c *gin.Context) {
 	}
 
 	setOpsRequestContext(c, parsedReq.Model, parsedReq.Stream, body)
+	setOpsEndpointContext(c, "", int16(service.RequestTypeFromLegacy(parsedReq.Stream, false)))
 
 	// 获取订阅信息（可能为nil）
 	subscription, _ := middleware2.GetSubscriptionFromContext(c)

backend/internal/handler/gemini_v1beta_handler.go

@@ -182,6 +182,7 @@ func (h *GatewayHandler) GeminiV1BetaModels(c *gin.Context) {
 	}
 
 	setOpsRequestContext(c, modelName, stream, body)
+	setOpsEndpointContext(c, "", int16(service.RequestTypeFromLegacy(stream, false)))
 
 	// Get subscription (may be nil)
 	subscription, _ := middleware.GetSubscriptionFromContext(c)

backend/internal/handler/openai_chat_completions.go

@@ -77,6 +77,7 @@ func (h *OpenAIGatewayHandler) ChatCompletions(c *gin.Context) {
 	reqLog = reqLog.With(zap.String("model", reqModel), zap.Bool("stream", reqStream))
 
 	setOpsRequestContext(c, reqModel, reqStream, body)
+	setOpsEndpointContext(c, "", int16(service.RequestTypeFromLegacy(reqStream, false)))
 
 	if h.errorPassthroughService != nil {
 		service.BindErrorPassthroughService(c, h.errorPassthroughService)

backend/internal/handler/openai_gateway_handler.go

@@ -183,6 +183,7 @@ func (h *OpenAIGatewayHandler) Responses(c *gin.Context) {
 	}
 
 	setOpsRequestContext(c, reqModel, reqStream, body)
+	setOpsEndpointContext(c, "", int16(service.RequestTypeFromLegacy(reqStream, false)))
 
 	// 提前校验 function_call_output 是否具备可关联上下文，避免上游 400。
 	if !h.validateFunctionCallOutputRequest(c, body, reqLog) {
@@ -545,6 +546,7 @@ func (h *OpenAIGatewayHandler) Messages(c *gin.Context) {
 	reqLog = reqLog.With(zap.String("model", reqModel), zap.Bool("stream", reqStream))
 
 	setOpsRequestContext(c, reqModel, reqStream, body)
+	setOpsEndpointContext(c, "", int16(service.RequestTypeFromLegacy(reqStream, false)))
 
 	// 绑定错误透传服务，允许 service 层在非 failover 错误场景复用规则。
 	if h.errorPassthroughService != nil {
@@ -1096,6 +1098,7 @@ func (h *OpenAIGatewayHandler) ResponsesWebSocket(c *gin.Context) {
 		zap.String("previous_response_id_kind", previousResponseIDKind),
 	)
 	setOpsRequestContext(c, reqModel, true, firstMessage)
+	setOpsEndpointContext(c, "", int16(service.RequestTypeWSV2))
 
 	var currentUserRelease func()
 	var currentAccountRelease func()

backend/internal/handler/sora_gateway_handler.go

@@ -159,6 +159,7 @@ func (h *SoraGatewayHandler) ChatCompletions(c *gin.Context) {
 	}
 
 	setOpsRequestContext(c, reqModel, clientStream, body)
+	setOpsEndpointContext(c, "", int16(service.RequestTypeFromLegacy(clientStream, false)))
 
 	platform := ""
 	if forced, ok := middleware2.GetForcePlatformFromContext(c); ok {