diff --git a/controller/midjourney.go b/controller/midjourney.go
index bf52314a7581..dbc7fb9d7360 100644
--- a/controller/midjourney.go
+++ b/controller/midjourney.go
@@ -14,6 +14,7 @@ import (
"github.com/QuantumNous/new-api/model"
"github.com/QuantumNous/new-api/service"
"github.com/QuantumNous/new-api/setting"
+ "github.com/QuantumNous/new-api/setting/operation_setting"
"github.com/QuantumNous/new-api/setting/system_setting"
"github.com/gin-gonic/gin"
@@ -332,11 +333,13 @@ func GetUserMidjourney(c *gin.Context) {
items := model.GetAllUserTask(userId, pageInfo.GetStartIdx(), pageInfo.GetPageSize(), queryParams)
total := model.CountAllUserTask(userId, queryParams)
- if setting.MjForwardUrlEnabled {
- for i, midjourney := range items {
+ for _, midjourney := range items {
+ if setting.MjForwardUrlEnabled {
midjourney.ImageUrl = system_setting.ServerAddress + "/mj/image/" + midjourney.MjId
- items[i] = midjourney
}
+ // 面向用户的 MJ 任务列表同样是上游失败原因的出口，与 /mj/task 查询
+ // （coverMidjourneyTaskDto）保持一致；管理员列表 GetAllMidjourney 保留原文。
+ midjourney.FailReason = operation_setting.OverrideUpstreamMessage(midjourney.FailReason)
}
pageInfo.SetTotal(int(total))
pageInfo.SetItems(items)
diff --git a/controller/midjourney_fail_reason_override_test.go b/controller/midjourney_fail_reason_override_test.go
new file mode 100644
index 000000000000..09297cd33e1b
--- /dev/null
+++ b/controller/midjourney_fail_reason_override_test.go
@@ -0,0 +1,78 @@
+package controller
+
+import (
+ "net/http"
+ "net/http/httptest"
+ "testing"
+
+ "github.com/QuantumNous/new-api/common"
+ "github.com/QuantumNous/new-api/model"
+ "github.com/QuantumNous/new-api/setting/operation_setting"
+
+ "github.com/gin-gonic/gin"
+ "github.com/glebarez/sqlite"
+ "github.com/stretchr/testify/assert"
+ "github.com/stretchr/testify/require"
+ "gorm.io/gorm"
+)
+
+// The Midjourney task list is a user-facing outlet for the same upstream failure text
+// the /mj/task fetch path already masks (relay.coverMidjourneyTaskDto). The user list
+// must mask it while the admin list keeps the original for diagnosis.
+func TestMidjourneyListFailReasonOverride(t *testing.T) {
+ enableTaskErrorOverride(t)
+ gin.SetMode(gin.TestMode)
+
+ previousDB := model.DB
+ db, err := gorm.Open(sqlite.Open(":memory:"), &gorm.Config{})
+ require.NoError(t, err)
+ require.NoError(t, db.AutoMigrate(&model.Midjourney{}))
+ model.DB = db
+ t.Cleanup(func() { model.DB = previousDB })
+
+ const userID = 7
+ const upstreamReason = "insufficient credits, please top-up your account"
+ const localReason = "获取渠道信息失败，请联系管理员，渠道ID：3"
+
+ require.NoError(t, db.Create(&model.Midjourney{
+ UserId: userID, MjId: "mj-upstream", Status: "FAILURE", FailReason: upstreamReason,
+ }).Error)
+ require.NoError(t, db.Create(&model.Midjourney{
+ UserId: userID, MjId: "mj-local", Status: "FAILURE", FailReason: localReason,
+ }).Error)
+
+ failReasons := func(handler gin.HandlerFunc) map[string]string {
+ t.Helper()
+ recorder := httptest.NewRecorder()
+ c, _ := gin.CreateTestContext(recorder)
+ c.Request = httptest.NewRequest(http.MethodGet, "/api/mj/self", nil)
+ c.Set("id", userID)
+
+ handler(c)
+ require.Equal(t, http.StatusOK, recorder.Code)
+
+ var response struct {
+ Data struct {
+ Items []struct {
+ MjId string `json:"mj_id"`
+ FailReason string `json:"fail_reason"`
+ } `json:"items"`
+ } `json:"data"`
+ }
+ require.NoError(t, common.Unmarshal(recorder.Body.Bytes(), &response))
+ reasons := make(map[string]string, len(response.Data.Items))
+ for _, item := range response.Data.Items {
+ reasons[item.MjId] = item.FailReason
+ }
+ return reasons
+ }
+
+ userReasons := failReasons(GetUserMidjourney)
+ assert.Equal(t, operation_setting.ErrorOverrideMessage, userReasons["mj-upstream"])
+ // This site's own failure text must survive the user-facing path.
+ assert.Equal(t, localReason, userReasons["mj-local"])
+
+ adminReasons := failReasons(GetAllMidjourney)
+ assert.Equal(t, upstreamReason, adminReasons["mj-upstream"])
+ assert.Equal(t, localReason, adminReasons["mj-local"])
+}
diff --git a/controller/playground.go b/controller/playground.go
index 1c9c8d3b77d9..e434c892aade 100644
--- a/controller/playground.go
+++ b/controller/playground.go
@@ -8,6 +8,7 @@ import (
"github.com/QuantumNous/new-api/model"
relaycommon "github.com/QuantumNous/new-api/relay/common"
"github.com/QuantumNous/new-api/relaykit/types"
+ "github.com/QuantumNous/new-api/setting/operation_setting"
"github.com/gin-gonic/gin"
)
@@ -17,6 +18,7 @@ func Playground(c *gin.Context) {
defer func() {
if newAPIError != nil {
+ operation_setting.OverrideUpstreamError(newAPIError)
c.JSON(newAPIError.StatusCode, gin.H{
"error": newAPIError.ToOpenAIError(),
})
diff --git a/controller/relay.go b/controller/relay.go
index e48f60135e55..baa4e0f2ee11 100644
--- a/controller/relay.go
+++ b/controller/relay.go
@@ -92,7 +92,14 @@ func Relay(c *gin.Context, relayFormat types.RelayFormat) {
defer func() {
if newAPIError != nil {
logger.LogError(c, fmt.Sprintf("relay error: %s", common.LocalLogPreview(newAPIError.Error())))
- newAPIError.SetMessage(common.MessageWithRequestId(newAPIError.Error(), requestId))
+ // 必须在错误日志之后执行：日志与渠道禁用判定始终使用原始上游文案。
+ // 覆写文案与 request id 一次写入：上游错误的 ToOpenAIError() 直接返回 RelayError，
+ // 不读 Err，SetMessage 改不到响应体，必须走 ReplaceMessage
+ if operation_setting.ShouldOverrideUpstreamError(newAPIError) {
+ newAPIError.ReplaceMessage(common.MessageWithRequestId(operation_setting.ErrorOverrideMessage, requestId))
+ } else {
+ newAPIError.SetMessage(common.MessageWithRequestId(newAPIError.Error(), requestId))
+ }
switch relayFormat {
case types.RelayFormatOpenAIRealtime:
helper.WssError(c, ws, newAPIError.ToOpenAIError())
@@ -431,13 +438,26 @@ func processChannelError(c *gin.Context, channelError types.ChannelError, err *t
adminInfo["multi_key_index"] = common.GetContextKeyInt(c, constant.ContextKeyChannelMultiKeyIndex)
}
service.AppendChannelAffinityAdminInfo(c, adminInfo)
+ // 错误日志的 Content 会通过 /api/log/self 回显给发起请求的用户，因此它和 HTTP 响应
+ // 一样是对外出口。覆写生效时这里必须同步覆写，否则客户端在响应里看到
+ // Service Unavailable，转头在日志页仍能读到上游账务原文。原文改记到
+ // admin_info，model.formatUserLogs 会为普通用户剥离整个 admin_info。
+ logContent := err.MaskSensitiveErrorWithStatusCode()
+ if operation_setting.ShouldOverrideUpstreamError(err) {
+ adminInfo["original_error"] = logContent
+ logContent = operation_setting.ErrorOverrideMessage
+ // 与 MaskSensitiveErrorWithStatusCode 保持一致：无状态码时不写 status_code=0
+ if err.StatusCode != 0 {
+ logContent = fmt.Sprintf("status_code=%d, %s", err.StatusCode, logContent)
+ }
+ }
other["admin_info"] = adminInfo
startTime := common.GetContextKeyTime(c, constant.ContextKeyRequestStartTime)
if startTime.IsZero() {
startTime = time.Now()
}
useTimeSeconds := int(time.Since(startTime).Seconds())
- model.RecordErrorLog(c, userId, channelId, modelName, tokenName, err.MaskSensitiveErrorWithStatusCode(), tokenId, useTimeSeconds, common.GetContextKeyBool(c, constant.ContextKeyIsStream), userGroup, other)
+ model.RecordErrorLog(c, userId, channelId, modelName, tokenName, logContent, tokenId, useTimeSeconds, common.GetContextKeyBool(c, constant.ContextKeyIsStream), userGroup, other)
}
}
@@ -475,13 +495,23 @@ func RelayMidjourney(c *gin.Context) {
mjErr.Result = "当前分组负载已饱和，请稍后再试，或升级账户以提升服务质量。"
statusCode = http.StatusTooManyRequests
}
+ description := fmt.Sprintf("%s %s", mjErr.Description, mjErr.Result)
+ channelId := c.GetInt("channel_id")
+ logger.LogError(c, fmt.Sprintf("relay error (channel #%d, status code %d): %s", channelId, statusCode, description))
+ // 这里一律不覆写：能走到 mjErr 的 MidjourneyResponse 全部是本站自产的（参数校验、
+ // 额度不足、DB/IO 失败）。mj-proxy 各 handler 拿到上游响应后是把上游 body 原样
+ // io.Copy 给客户端再 return nil 的，上游错误文案根本不经过这个分支。
+ // 之前在此处按关键词覆写会把本站的 quota_not_enough（mjproxy_handler.go 的
+ // RelaySwapFace / RelayMidjourneySubmit）误伤成 Service Unavailable，
+ // 让用户看不到自己额度不足的真实原因。
+ // MJ 上游文案的出口是被代理的响应体本身与任务的 FailReason，前者要改写就得重写
+ // 上游 JSON，会破坏 mj-proxy 协议兼容性，故不在本功能范围内；后者已在
+ // TaskModel2Dto / coverMidjourneyTaskDto 的读取边界处理。
c.JSON(statusCode, gin.H{
- "description": fmt.Sprintf("%s %s", mjErr.Description, mjErr.Result),
+ "description": description,
"type": "upstream_error",
"code": mjErr.Code,
})
- channelId := c.GetInt("channel_id")
- logger.LogError(c, fmt.Sprintf("relay error (channel #%d, status code %d): %s", channelId, statusCode, fmt.Sprintf("%s %s", mjErr.Description, mjErr.Result)))
}
}
@@ -617,7 +647,7 @@ func RelayTask(c *gin.Context) {
processChannelError(c,
*types.NewChannelError(channel.Id, channel.Type, channel.Name, channel.ChannelInfo.IsMultiKey,
common.GetContextKeyString(c, constant.ContextKeyChannelKey), channel.GetAutoBan()),
- types.NewOpenAIError(taskErr.Error, types.ErrorCodeBadResponseStatusCode, taskErr.StatusCode))
+ service.APIErrorFromTaskError(taskErr))
}
if taskFailoverEnabled {
@@ -674,10 +704,24 @@ func respondTaskError(c *gin.Context, taskErr *taskdto.TaskError) {
if taskErr.StatusCode == http.StatusTooManyRequests {
taskErr.Message = "当前分组上游负载已饱和，请稍后再试"
}
+ // 仅覆写文案确实取自上游任务平台的错误。这里必须用正向标记 FromUpstream 判定：
+ // 本站自产错误（读 body 失败、解析失败、预扣费额度不足）默认 LocalError == false，
+ // 用 !LocalError 反推会把它们一并掩盖。
+ if taskErr.FromUpstream {
+ if overridden := operation_setting.OverrideUpstreamMessage(taskErr.Message); overridden != taskErr.Message {
+ // 覆写前先记录原始上游文案，后台日志与排障始终可见全文
+ logger.LogError(c, fmt.Sprintf("task upstream error overridden: %s", common.LocalLogPreview(taskErr.Message)))
+ taskErr.Message = overridden
+ // Data 带 json:"data" 会一并返回客户端。当前 task 适配器不往里塞上游 body，
+ // 但覆写掉 Message 却留着一个可能承载上游原文的字段是自相矛盾的，
+ // 与 NewAPIError.ReplaceMessage 清 Metadata/Param 保持一致。
+ taskErr.Data = nil
+ }
+ }
c.JSON(taskErr.StatusCode, taskErr)
}
-func shouldRetryTaskRelay(c *gin.Context, channelId int, taskErr *taskdto.TaskError, retryTimes int) bool {
+func shouldRetryTaskRelay(c *gin.Context, channelId int, taskErr *taskdto.TaskError, retryTimes int, failoverEnabled bool) bool {
if taskErr == nil {
return false
}
diff --git a/controller/task.go b/controller/task.go
index a80f1a687aab..feec5fa42678 100644
--- a/controller/task.go
+++ b/controller/task.go
@@ -32,7 +32,8 @@ func GetAllTask(c *gin.Context) {
items := model.TaskGetAllTasks(pageInfo.GetStartIdx(), pageInfo.GetPageSize(), queryParams)
total := model.TaskCountAllTasks(queryParams)
pageInfo.SetTotal(int(total))
- pageInfo.SetItems(tasksToDto(items, true))
+ // 管理员视图不掩盖上游失败原因，排障需要原文
+ pageInfo.SetItems(tasksToDto(items, true, false))
common.ApiSuccess(c, pageInfo)
}
@@ -56,11 +57,13 @@ func GetUserTask(c *gin.Context) {
items := model.TaskGetAllUserTask(userId, pageInfo.GetStartIdx(), pageInfo.GetPageSize(), queryParams)
total := model.TaskCountAllUserTask(userId, queryParams)
pageInfo.SetTotal(int(total))
- pageInfo.SetItems(tasksToDto(items, false))
+ pageInfo.SetItems(tasksToDto(items, false, true))
common.ApiSuccess(c, pageInfo)
}
-func tasksToDto(tasks []*model.Task, fillUser bool) []*dto.TaskDto {
+// tasksToDto 转换任务列表。maskUpstreamFailReason 单独传参而不复用 fillUser：填充用户名
+// 与是否掩盖上游失败原因是两个无关维度，绑在一起会让后续加入新调用方时选错默认值。
+func tasksToDto(tasks []*model.Task, fillUser bool, maskUpstreamFailReason bool) []*dto.TaskDto {
var userIdMap map[int]*model.UserBase
if fillUser {
userIdMap = make(map[int]*model.UserBase)
@@ -82,7 +85,7 @@ func tasksToDto(tasks []*model.Task, fillUser bool) []*dto.TaskDto {
task.Username = user.Username
}
}
- result[i] = relay.TaskModel2Dto(task)
+ result[i] = relay.TaskModel2Dto(task, maskUpstreamFailReason)
}
return result
}
diff --git a/controller/task_error_override_test.go b/controller/task_error_override_test.go
new file mode 100644
index 000000000000..f31a2f8813fa
--- /dev/null
+++ b/controller/task_error_override_test.go
@@ -0,0 +1,152 @@
+package controller
+
+import (
+ "errors"
+ "net/http"
+ "net/http/httptest"
+ "testing"
+
+ taskdto "github.com/QuantumNous/new-api/dto"
+ "github.com/QuantumNous/new-api/relaykit/types"
+ "github.com/QuantumNous/new-api/service"
+ "github.com/QuantumNous/new-api/setting/operation_setting"
+
+ "github.com/gin-gonic/gin"
+ "github.com/stretchr/testify/assert"
+ "github.com/stretchr/testify/require"
+)
+
+// enableTaskErrorOverride turns the override on with the shipped default keywords and
+// restores the previous package state afterwards.
+func enableTaskErrorOverride(t *testing.T) {
+ t.Helper()
+ origEnabled := operation_setting.ErrorOverrideEnabled
+ origKeywords := operation_setting.ErrorOverrideKeywords
+ t.Cleanup(func() {
+ operation_setting.ErrorOverrideEnabled = origEnabled
+ operation_setting.ErrorOverrideKeywords = origKeywords
+ })
+ operation_setting.ErrorOverrideEnabled = true
+ operation_setting.ErrorOverrideKeywords = []string{"no available", "quota", "credits", "top-up"}
+}
+
+// respondTaskError must key the override off the positive FromUpstream marker. Local
+// errors default to LocalError == false, so gating on !LocalError would mask this
+// site's own billing and infrastructure failures whenever their text happens to
+// contain an override keyword.
+func TestRespondTaskError_OverrideGate(t *testing.T) {
+ enableTaskErrorOverride(t)
+ gin.SetMode(gin.TestMode)
+
+ cases := []struct {
+ name string
+ taskErr *taskdto.TaskError
+ wantMessage string
+ }{
+ {
+ // Regression: PreConsumeBilling produces this locally. It contains "quota",
+ // so an inverted gate would hide the real reason the request failed.
+ name: "local subscription quota error is not overridden",
+ taskErr: service.TaskErrorFromAPIError(types.NewErrorWithStatusCode(
+ errors.New("订阅额度不足或未配置订阅: subscription quota insufficient"),
+ types.ErrorCodeInsufficientUserQuota, http.StatusForbidden)),
+ wantMessage: "订阅额度不足或未配置订阅: subscription quota insufficient",
+ },
+ {
+ name: "local user quota error is not overridden",
+ taskErr: service.TaskErrorFromAPIError(types.NewErrorWithStatusCode(
+ errors.New("用户额度不足, 剩余额度: $0.00"),
+ types.ErrorCodeInsufficientUserQuota, http.StatusForbidden)),
+ wantMessage: "用户额度不足, 剩余额度: $0.00",
+ },
+ {
+ name: "local no-available-key error is not overridden",
+ taskErr: service.TaskErrorWrapper(
+ errors.New("no available channel for model under group default"),
+ "channel_no_available_key", http.StatusServiceUnavailable),
+ wantMessage: "no available channel for model under group default",
+ },
+ {
+ name: "upstream keyword match is overridden",
+ taskErr: service.TaskErrorWrapperUpstream(
+ errors.New("insufficient credits, please top up your account"),
+ "upstream_error", http.StatusPaymentRequired),
+ wantMessage: operation_setting.ErrorOverrideMessage,
+ },
+ {
+ name: "upstream without keyword keeps its message",
+ taskErr: service.TaskErrorWrapperUpstream(
+ errors.New("invalid prompt"), "upstream_error", http.StatusBadRequest),
+ wantMessage: "invalid prompt",
+ },
+ }
+
+ for _, tc := range cases {
+ t.Run(tc.name, func(t *testing.T) {
+ recorder := httptest.NewRecorder()
+ c, _ := gin.CreateTestContext(recorder)
+ c.Request = httptest.NewRequest(http.MethodPost, "/v1/video/generations", nil)
+
+ respondTaskError(c, tc.taskErr)
+
+ assert.Equal(t, tc.wantMessage, tc.taskErr.Message)
+ })
+ }
+}
+
+// Data carries json:"data" and reaches the client. Overriding Message while leaving a
+// field that may hold the upstream payload would defeat the override, so an overridden
+// task error must clear it — and a non-overridden one must keep it.
+func TestRespondTaskError_ClearsDataOnOverride(t *testing.T) {
+ enableTaskErrorOverride(t)
+ gin.SetMode(gin.TestMode)
+
+ t.Run("overridden error drops Data", func(t *testing.T) {
+ taskErr := service.TaskErrorWrapperUpstream(
+ errors.New("insufficient credits, please top up"),
+ "upstream_error", http.StatusPaymentRequired)
+ taskErr.Data = map[string]string{"raw": "upstream-billing-detail"}
+
+ recorder := httptest.NewRecorder()
+ c, _ := gin.CreateTestContext(recorder)
+ c.Request = httptest.NewRequest(http.MethodPost, "/v1/video/generations", nil)
+
+ respondTaskError(c, taskErr)
+
+ require.Equal(t, operation_setting.ErrorOverrideMessage, taskErr.Message)
+ assert.Nil(t, taskErr.Data)
+ })
+
+ t.Run("untouched error keeps Data", func(t *testing.T) {
+ taskErr := service.TaskErrorWrapperUpstream(
+ errors.New("invalid prompt"), "upstream_error", http.StatusBadRequest)
+ taskErr.Data = map[string]string{"field": "prompt"}
+
+ recorder := httptest.NewRecorder()
+ c, _ := gin.CreateTestContext(recorder)
+ c.Request = httptest.NewRequest(http.MethodPost, "/v1/video/generations", nil)
+
+ respondTaskError(c, taskErr)
+
+ require.Equal(t, "invalid prompt", taskErr.Message)
+ assert.NotNil(t, taskErr.Data)
+ })
+}
+
+// A 429 rewrite happens before the override and its text is this site's own, so it
+// must survive regardless of the upstream marker.
+func TestRespondTaskError_RateLimitMessagePreserved(t *testing.T) {
+ enableTaskErrorOverride(t)
+ gin.SetMode(gin.TestMode)
+
+ taskErr := service.TaskErrorWrapperUpstream(
+ errors.New("upstream quota exhausted"), "upstream_error", http.StatusTooManyRequests)
+
+ recorder := httptest.NewRecorder()
+ c, _ := gin.CreateTestContext(recorder)
+ c.Request = httptest.NewRequest(http.MethodPost, "/v1/video/generations", nil)
+
+ respondTaskError(c, taskErr)
+
+ require.Equal(t, "当前分组上游负载已饱和，请稍后再试", taskErr.Message)
+}
diff --git a/dto/task.go b/dto/task.go
index 4a9a8e2e6d18..1fe309b3542d 100644
--- a/dto/task.go
+++ b/dto/task.go
@@ -10,7 +10,11 @@ type TaskError struct {
Data any `json:"data"`
StatusCode int `json:"-"`
LocalError bool `json:"-"`
- Error error `json:"-"`
+ // FromUpstream 表示 Message 的文案取自上游任务平台的响应体，而非本站生成。
+ // 错误信息覆写只作用于该标记为 true 的错误；LocalError 是重试判定用的独立维度，
+ // 不能反过来当作「来自上游」的依据（本站自产错误默认 LocalError == false）。
+ FromUpstream bool `json:"-"`
+ Error error `json:"-"`
}
type TaskData interface {
diff --git a/middleware/distributor.go b/middleware/distributor.go
index 2b8a794776ff..67fa9d365f34 100644
--- a/middleware/distributor.go
+++ b/middleware/distributor.go
@@ -52,7 +52,7 @@ func Distribute() func(c *gin.Context) {
}
if aliasApplied {
common.SetContextKey(c, constant.ContextKeyRequestedModelAlias, modelRequest.Model)
- rewriteRequestBodyModel(c, strings.TrimSuffix(resolvedModel, ratio_setting.CompactModelSuffix))
+ rewriteRequestBodyModel(c, resolvedModel)
modelRequest.Model = resolvedModel
}
}
diff --git a/model/channel_cache_failover_test.go b/model/channel_cache_failover_test.go
index 80e6e311a865..ec2ef51f5e09 100644
--- a/model/channel_cache_failover_test.go
+++ b/model/channel_cache_failover_test.go
@@ -4,7 +4,7 @@ import (
"testing"
"github.com/QuantumNous/new-api/common"
- "github.com/QuantumNous/new-api/dto"
+ relaykitdto "github.com/QuantumNous/new-api/relaykit/dto"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
)
@@ -31,7 +31,7 @@ func setupFailoverChannelCache(t *testing.T, channels []*Channel, group string,
oldAdvanced := channel2advancedCustomConfig
channelsIDM = newIDM
group2model2channels = newG2M
- channel2advancedCustomConfig = make(map[int]*dto.AdvancedCustomConfig)
+ channel2advancedCustomConfig = make(map[int]*relaykitdto.AdvancedCustomConfig)
channelSyncLock.Unlock()
t.Cleanup(func() {
diff --git a/model/option.go b/model/option.go
index 89eb56f2a427..f7e81c28b94b 100644
--- a/model/option.go
+++ b/model/option.go
@@ -175,6 +175,8 @@ func InitOptionMap() {
common.OptionMap["SensitiveWords"] = setting.SensitiveWordsToString()
common.OptionMap["StreamCacheQueueLength"] = strconv.Itoa(setting.StreamCacheQueueLength)
common.OptionMap["AutomaticDisableKeywords"] = operation_setting.AutomaticDisableKeywordsToString()
+ common.OptionMap["ErrorOverrideEnabled"] = strconv.FormatBool(operation_setting.ErrorOverrideEnabled)
+ common.OptionMap["ErrorOverrideKeywords"] = operation_setting.ErrorOverrideKeywordsToString()
common.OptionMap["AutomaticDisableStatusCodes"] = operation_setting.AutomaticDisableStatusCodesToString()
common.OptionMap["AutomaticRetryStatusCodes"] = operation_setting.AutomaticRetryStatusCodesToString()
common.OptionMap["ExposeRatioEnabled"] = strconv.FormatBool(ratio_setting.IsExposeRatioEnabled())
@@ -373,6 +375,8 @@ func updateOptionMap(key string, value string) (err error) {
operation_setting.DemoSiteEnabled = boolValue
case "SelfUseModeEnabled":
operation_setting.SelfUseModeEnabled = boolValue
+ case "ErrorOverrideEnabled":
+ operation_setting.ErrorOverrideEnabled = boolValue
case "ChannelFailoverEnabled":
operation_setting.ChannelFailoverEnabled = boolValue
case "CheckSensitiveOnPromptEnabled":
@@ -593,6 +597,8 @@ func updateOptionMap(key string, value string) (err error) {
setting.SensitiveWordsFromString(value)
case "AutomaticDisableKeywords":
operation_setting.AutomaticDisableKeywordsFromString(value)
+ case "ErrorOverrideKeywords":
+ operation_setting.ErrorOverrideKeywordsFromString(value)
case "AutomaticDisableStatusCodes":
err = operation_setting.AutomaticDisableStatusCodesFromString(value)
case "AutomaticRetryStatusCodes":
diff --git a/relay/channel/ali/image.go b/relay/channel/ali/image.go
index 6913fa346aa6..cf46725e487e 100644
--- a/relay/channel/ali/image.go
+++ b/relay/channel/ali/image.go
@@ -317,7 +317,7 @@ func aliImageHandler(a *Adaptor, c *gin.Context, resp *http.Response, info *rela
return types.NewError(err, types.ErrorCodeBadResponse), nil
}
if aliResponse.Output.TaskStatus != "SUCCEEDED" {
- return types.WithOpenAIError(types.OpenAIError{
+ return types.WithUpstreamOpenAIError(types.OpenAIError{
Message: aliResponse.Output.Message,
Type: "ali_error",
Param: "",
diff --git a/relay/channel/ali/rerank.go b/relay/channel/ali/rerank.go
index ac2afbd3d3db..7bc3e358a2f6 100644
--- a/relay/channel/ali/rerank.go
+++ b/relay/channel/ali/rerank.go
@@ -46,7 +46,7 @@ func RerankHandler(c *gin.Context, resp *http.Response, info *relaycommon.RelayI
}
if aliResponse.Code != "" {
- return types.WithOpenAIError(types.OpenAIError{
+ return types.WithUpstreamOpenAIError(types.OpenAIError{
Message: aliResponse.Message,
Type: aliResponse.Code,
Param: aliResponse.RequestId,
diff --git a/relay/channel/claude/relay-claude.go b/relay/channel/claude/relay-claude.go
index 2f424b32abdd..b3615ab70b33 100644
--- a/relay/channel/claude/relay-claude.go
+++ b/relay/channel/claude/relay-claude.go
@@ -91,7 +91,7 @@ func HandleStreamResponseData(c *gin.Context, info *relaycommon.RelayInfo, claud
return types.NewError(err, types.ErrorCodeBadResponseBody)
}
if claudeError := claudeResponse.GetClaudeError(); claudeError != nil && claudeError.Type != "" {
- return types.WithClaudeError(*claudeError, http.StatusInternalServerError)
+ return types.WithUpstreamClaudeError(*claudeError, http.StatusInternalServerError)
}
if claudeResponse.StopReason != "" {
maybeMarkClaudeRefusal(c, claudeResponse.StopReason)
@@ -221,7 +221,7 @@ func HandleClaudeResponseData(c *gin.Context, info *relaycommon.RelayInfo, claud
return types.NewError(err, types.ErrorCodeBadResponseBody)
}
if claudeError := claudeResponse.GetClaudeError(); claudeError != nil && claudeError.Type != "" {
- return types.WithClaudeError(*claudeError, http.StatusInternalServerError)
+ return types.WithUpstreamClaudeError(*claudeError, http.StatusInternalServerError)
}
maybeMarkClaudeRefusal(c, claudeResponse.StopReason)
if claudeInfo.Usage == nil {
diff --git a/relay/channel/jimeng/image.go b/relay/channel/jimeng/image.go
index 888531bdd99d..689e5ed1be6d 100644
--- a/relay/channel/jimeng/image.go
+++ b/relay/channel/jimeng/image.go
@@ -64,7 +64,7 @@ func jimengImageHandler(c *gin.Context, resp *http.Response, info *relaycommon.R
// Check if the response indicates an error
if jimengResponse.Code != 10000 {
- return nil, types.WithOpenAIError(types.OpenAIError{
+ return nil, types.WithUpstreamOpenAIError(types.OpenAIError{
Message: jimengResponse.Message,
Type: "jimeng_error",
Param: "",
diff --git a/relay/channel/minimax/image.go b/relay/channel/minimax/image.go
index c86cd47dba72..c7349ae28e9a 100644
--- a/relay/channel/minimax/image.go
+++ b/relay/channel/minimax/image.go
@@ -187,7 +187,7 @@ func miniMaxImageHandler(c *gin.Context, resp *http.Response, info *relaycommon.
return nil, types.NewOpenAIError(err, types.ErrorCodeBadResponseBody, http.StatusInternalServerError)
}
if minimaxResponse.BaseResp.StatusCode != 0 {
- return nil, types.WithOpenAIError(types.OpenAIError{
+ return nil, types.WithUpstreamOpenAIError(types.OpenAIError{
Message: minimaxResponse.BaseResp.StatusMsg,
Type: "minimax_image_error",
Code: fmt.Sprintf("%d", minimaxResponse.BaseResp.StatusCode),
diff --git a/relay/channel/openai/chat_via_responses.go b/relay/channel/openai/chat_via_responses.go
index 25caeb5854bd..a6dfa071481b 100644
--- a/relay/channel/openai/chat_via_responses.go
+++ b/relay/channel/openai/chat_via_responses.go
@@ -38,7 +38,7 @@ func OaiResponsesToChatHandler(c *gin.Context, info *relaycommon.RelayInfo, resp
}
if oaiError := responsesResp.GetOpenAIError(); oaiError != nil && oaiError.Type != "" {
- return nil, types.WithOpenAIError(*oaiError, resp.StatusCode)
+ return nil, types.WithUpstreamOpenAIError(*oaiError, resp.StatusCode)
}
chatResult, err := relayconvert.ConvertResponse(c, info, types.RelayFormatOpenAI, &responsesResp)
@@ -124,7 +124,7 @@ func OaiResponsesToChatBufferedStreamHandler(c *gin.Context, info *relaycommon.R
case "response.failed", "response.error":
if streamResp.Response != nil {
if oaiErr := streamResp.Response.GetOpenAIError(); oaiErr != nil && oaiErr.Type != "" {
- streamErr = types.WithOpenAIError(*oaiErr, http.StatusInternalServerError)
+ streamErr = types.WithUpstreamOpenAIError(*oaiErr, http.StatusInternalServerError)
break
}
}
@@ -283,7 +283,7 @@ func OaiResponsesToChatStreamHandler(c *gin.Context, info *relaycommon.RelayInfo
if streamResp.Type == "response.error" || streamResp.Type == "response.failed" {
if streamResp.Response != nil {
if oaiErr := streamResp.Response.GetOpenAIError(); oaiErr != nil && oaiErr.Type != "" {
- streamErr = types.WithOpenAIError(*oaiErr, http.StatusInternalServerError)
+ streamErr = types.WithUpstreamOpenAIError(*oaiErr, http.StatusInternalServerError)
sr.Stop(streamErr)
return
}
diff --git a/relay/channel/openai/relay-openai.go b/relay/channel/openai/relay-openai.go
index 9a0619eb27f5..3f9ceaff70d1 100644
--- a/relay/channel/openai/relay-openai.go
+++ b/relay/channel/openai/relay-openai.go
@@ -250,7 +250,7 @@ func OpenaiHandler(c *gin.Context, info *relaycommon.RelayInfo, resp *http.Respo
}
if oaiError := simpleResponse.GetOpenAIError(); oaiError != nil && oaiError.Type != "" {
- return nil, types.WithOpenAIError(*oaiError, resp.StatusCode)
+ return nil, types.WithUpstreamOpenAIError(*oaiError, resp.StatusCode)
}
for _, choice := range simpleResponse.Choices {
diff --git a/relay/channel/openai/relay_image.go b/relay/channel/openai/relay_image.go
index 1e6be0dd4cbe..60682cc8cbaf 100644
--- a/relay/channel/openai/relay_image.go
+++ b/relay/channel/openai/relay_image.go
@@ -46,7 +46,7 @@ func OpenaiImageHandler(c *gin.Context, info *relaycommon.RelayInfo, resp *http.
}
if oaiError := usageResp.GetOpenAIError(); oaiError != nil && oaiError.Type != "" {
- return nil, types.WithOpenAIError(*oaiError, resp.StatusCode)
+ return nil, types.WithUpstreamOpenAIError(*oaiError, resp.StatusCode)
}
updateOpenAIImageCount(info, gjson.GetBytes(responseBody, "data.#").Int())
@@ -247,7 +247,7 @@ func openaiImageJSONAsStreamHandler(c *gin.Context, info *relaycommon.RelayInfo,
return nil, types.NewOpenAIError(err, types.ErrorCodeBadResponseBody, http.StatusInternalServerError)
}
if oaiError := usageResp.GetOpenAIError(); oaiError != nil && oaiError.Type != "" {
- return nil, types.WithOpenAIError(*oaiError, resp.StatusCode)
+ return nil, types.WithUpstreamOpenAIError(*oaiError, resp.StatusCode)
}
normalizeOpenAIUsage(&usageResp.Usage)
applyUsagePostProcessing(info, &usageResp.Usage, responseBody)
diff --git a/relay/channel/openai/relay_responses.go b/relay/channel/openai/relay_responses.go
index ceca1af3b381..3fa6f5f7d6a9 100644
--- a/relay/channel/openai/relay_responses.go
+++ b/relay/channel/openai/relay_responses.go
@@ -31,7 +31,7 @@ func OaiResponsesHandler(c *gin.Context, info *relaycommon.RelayInfo, resp *http
return nil, types.NewOpenAIError(err, types.ErrorCodeBadResponseBody, http.StatusInternalServerError)
}
if oaiError := responsesResponse.GetOpenAIError(); oaiError != nil && oaiError.Type != "" {
- return nil, types.WithOpenAIError(*oaiError, resp.StatusCode)
+ return nil, types.WithUpstreamOpenAIError(*oaiError, resp.StatusCode)
}
// 写入新的 response body
diff --git a/relay/channel/openai/relay_responses_compact.go b/relay/channel/openai/relay_responses_compact.go
index ff30d36e417b..4bbceb11a514 100644
--- a/relay/channel/openai/relay_responses_compact.go
+++ b/relay/channel/openai/relay_responses_compact.go
@@ -25,7 +25,7 @@ func OaiResponsesCompactionHandler(c *gin.Context, resp *http.Response) (*dto.Us
return nil, types.NewOpenAIError(err, types.ErrorCodeBadResponseBody, http.StatusInternalServerError)
}
if oaiError := compactResp.GetOpenAIError(); oaiError != nil && oaiError.Type != "" {
- return nil, types.WithOpenAIError(*oaiError, resp.StatusCode)
+ return nil, types.WithUpstreamOpenAIError(*oaiError, resp.StatusCode)
}
service.IOCopyBytesGracefully(c, resp, responseBody)
diff --git a/relay/channel/openai/responses_via_chat.go b/relay/channel/openai/responses_via_chat.go
index 53b9d33cbc0e..47880d0f2cea 100644
--- a/relay/channel/openai/responses_via_chat.go
+++ b/relay/channel/openai/responses_via_chat.go
@@ -32,7 +32,7 @@ func OaiChatToResponsesHandler(c *gin.Context, info *relaycommon.RelayInfo, resp
return nil, types.NewOpenAIError(err, types.ErrorCodeBadResponseBody, http.StatusInternalServerError)
}
if oaiError := chatResp.GetOpenAIError(); oaiError != nil && oaiError.Type != "" {
- return nil, types.WithOpenAIError(*oaiError, resp.StatusCode)
+ return nil, types.WithUpstreamOpenAIError(*oaiError, resp.StatusCode)
}
if responseID := helper.GetResponseID(c); responseID != "" {
@@ -97,7 +97,7 @@ func OaiChatToResponsesStreamHandler(c *gin.Context, info *relaycommon.RelayInfo
var errorResp dto.OpenAITextResponse
if err := common.UnmarshalJsonStr(data, &errorResp); err == nil {
if oaiError := errorResp.GetOpenAIError(); oaiError != nil && oaiError.Type != "" {
- streamErr = types.WithOpenAIError(*oaiError, resp.StatusCode)
+ streamErr = types.WithUpstreamOpenAIError(*oaiError, resp.StatusCode)
sr.Stop(streamErr)
return
}
diff --git a/relay/channel/palm/relay-palm.go b/relay/channel/palm/relay-palm.go
index 89ac14c7d010..fc8a24dc0f10 100644
--- a/relay/channel/palm/relay-palm.go
+++ b/relay/channel/palm/relay-palm.go
@@ -113,7 +113,7 @@ func palmHandler(c *gin.Context, info *relaycommon.RelayInfo, resp *http.Respons
return nil, types.NewOpenAIError(err, types.ErrorCodeBadResponseBody, http.StatusInternalServerError)
}
if palmResponse.Error.Code != 0 || len(palmResponse.Candidates) == 0 {
- return nil, types.WithOpenAIError(types.OpenAIError{
+ return nil, types.WithUpstreamOpenAIError(types.OpenAIError{
Message: palmResponse.Error.Message,
Type: palmResponse.Error.Status,
Param: "",
diff --git a/relay/channel/task/ali/adaptor.go b/relay/channel/task/ali/adaptor.go
index 7452c614dbdb..2ec9558ffb70 100644
--- a/relay/channel/task/ali/adaptor.go
+++ b/relay/channel/task/ali/adaptor.go
@@ -489,13 +489,13 @@ func (a *TaskAdaptor) DoResponse(c *gin.Context, resp *http.Response, info *rela
// 解析阿里响应
var aliResp AliVideoResponse
if err := common.Unmarshal(responseBody, &aliResp); err != nil {
- taskErr = service.TaskErrorWrapper(errors.Wrapf(err, "body: %s", responseBody), "unmarshal_response_body_failed", http.StatusInternalServerError)
+ taskErr = service.TaskErrorWrapperUpstream(errors.Wrapf(err, "body: %s", responseBody), "unmarshal_response_body_failed", http.StatusInternalServerError)
return
}
// 检查错误
if aliResp.Code != "" {
- taskErr = service.TaskErrorWrapper(fmt.Errorf("%s: %s", aliResp.Code, aliResp.Message), "ali_api_error", resp.StatusCode)
+ taskErr = service.TaskErrorWrapperUpstream(fmt.Errorf("%s: %s", aliResp.Code, aliResp.Message), "ali_api_error", resp.StatusCode)
return
}
diff --git a/relay/channel/task/doubao/adaptor.go b/relay/channel/task/doubao/adaptor.go
index 69302a676290..eade387ee2b9 100644
--- a/relay/channel/task/doubao/adaptor.go
+++ b/relay/channel/task/doubao/adaptor.go
@@ -219,7 +219,7 @@ func (a *TaskAdaptor) DoResponse(c *gin.Context, resp *http.Response, info *rela
// Parse Doubao response
var dResp responsePayload
if err := common.Unmarshal(responseBody, &dResp); err != nil {
- taskErr = service.TaskErrorWrapper(errors.Wrapf(err, "body: %s", responseBody), "unmarshal_response_body_failed", http.StatusInternalServerError)
+ taskErr = service.TaskErrorWrapperUpstream(errors.Wrapf(err, "body: %s", responseBody), "unmarshal_response_body_failed", http.StatusInternalServerError)
return
}
diff --git a/relay/channel/task/hailuo/adaptor.go b/relay/channel/task/hailuo/adaptor.go
index af9f5c57c52a..899fd0c1324c 100644
--- a/relay/channel/task/hailuo/adaptor.go
+++ b/relay/channel/task/hailuo/adaptor.go
@@ -89,12 +89,12 @@ func (a *TaskAdaptor) DoResponse(c *gin.Context, resp *http.Response, info *rela
var hResp VideoResponse
if err := common.Unmarshal(responseBody, &hResp); err != nil {
- taskErr = service.TaskErrorWrapper(errors.Wrapf(err, "body: %s", responseBody), "unmarshal_response_body_failed", http.StatusInternalServerError)
+ taskErr = service.TaskErrorWrapperUpstream(errors.Wrapf(err, "body: %s", responseBody), "unmarshal_response_body_failed", http.StatusInternalServerError)
return
}
if hResp.BaseResp.StatusCode != StatusSuccess {
- taskErr = service.TaskErrorWrapper(
+ taskErr = service.TaskErrorWrapperUpstream(
fmt.Errorf("hailuo api error: %s", hResp.BaseResp.StatusMsg),
strconv.Itoa(hResp.BaseResp.StatusCode),
http.StatusBadRequest,
diff --git a/relay/channel/task/jimeng/adaptor.go b/relay/channel/task/jimeng/adaptor.go
index 5e788d3415b5..197fcee1cd36 100644
--- a/relay/channel/task/jimeng/adaptor.go
+++ b/relay/channel/task/jimeng/adaptor.go
@@ -194,12 +194,12 @@ func (a *TaskAdaptor) DoResponse(c *gin.Context, resp *http.Response, info *rela
// Parse Jimeng response
var jResp responsePayload
if err := common.Unmarshal(responseBody, &jResp); err != nil {
- taskErr = service.TaskErrorWrapper(errors.Wrapf(err, "body: %s", responseBody), "unmarshal_response_body_failed", http.StatusInternalServerError)
+ taskErr = service.TaskErrorWrapperUpstream(errors.Wrapf(err, "body: %s", responseBody), "unmarshal_response_body_failed", http.StatusInternalServerError)
return
}
if jResp.Code != 10000 {
- taskErr = service.TaskErrorWrapper(fmt.Errorf("%s", jResp.Message), fmt.Sprintf("%d", jResp.Code), http.StatusInternalServerError)
+ taskErr = service.TaskErrorWrapperUpstream(fmt.Errorf("%s", jResp.Message), fmt.Sprintf("%d", jResp.Code), http.StatusInternalServerError)
return
}
diff --git a/relay/channel/task/kling/adaptor.go b/relay/channel/task/kling/adaptor.go
index 200c3c6829ee..41c88671f7aa 100644
--- a/relay/channel/task/kling/adaptor.go
+++ b/relay/channel/task/kling/adaptor.go
@@ -203,7 +203,9 @@ func (a *TaskAdaptor) DoResponse(c *gin.Context, resp *http.Response, info *rela
return
}
if kResp.Code != 0 {
+ // LocalError 保持 true：该错误不重试。文案取自上游，需同时标记以便错误信息覆写生效。
taskErr = service.TaskErrorWrapperLocal(fmt.Errorf("%s", kResp.Message), "task_failed", http.StatusBadRequest)
+ taskErr.FromUpstream = true
return
}
ov := dto.NewOpenAIVideo()
diff --git a/relay/channel/task/sora/adaptor.go b/relay/channel/task/sora/adaptor.go
index 7f81e5335ebb..9b6e8d4a42d8 100644
--- a/relay/channel/task/sora/adaptor.go
+++ b/relay/channel/task/sora/adaptor.go
@@ -236,7 +236,7 @@ func (a *TaskAdaptor) DoResponse(c *gin.Context, resp *http.Response, info *rela
// Parse Sora response
var dResp responseTask
if err := common.Unmarshal(responseBody, &dResp); err != nil {
- taskErr = service.TaskErrorWrapper(errors.Wrapf(err, "body: %s", responseBody), "unmarshal_response_body_failed", http.StatusInternalServerError)
+ taskErr = service.TaskErrorWrapperUpstream(errors.Wrapf(err, "body: %s", responseBody), "unmarshal_response_body_failed", http.StatusInternalServerError)
return
}
diff --git a/relay/channel/task/suno/adaptor.go b/relay/channel/task/suno/adaptor.go
index 35b5e423b7ff..14e0b49b7314 100644
--- a/relay/channel/task/suno/adaptor.go
+++ b/relay/channel/task/suno/adaptor.go
@@ -105,7 +105,7 @@ func (a *TaskAdaptor) DoResponse(c *gin.Context, resp *http.Response, info *rela
return
}
if !sunoResponse.IsSuccess() {
- taskErr = service.TaskErrorWrapper(fmt.Errorf("%s", sunoResponse.Message), sunoResponse.Code, http.StatusInternalServerError)
+ taskErr = service.TaskErrorWrapperUpstream(fmt.Errorf("%s", sunoResponse.Message), sunoResponse.Code, http.StatusInternalServerError)
return
}
diff --git a/relay/channel/task/vidu/adaptor.go b/relay/channel/task/vidu/adaptor.go
index 62e029bbe88e..dd03dc0d69df 100644
--- a/relay/channel/task/vidu/adaptor.go
+++ b/relay/channel/task/vidu/adaptor.go
@@ -172,7 +172,7 @@ func (a *TaskAdaptor) DoResponse(c *gin.Context, resp *http.Response, info *rela
var vResp responsePayload
err = common.Unmarshal(responseBody, &vResp)
if err != nil {
- taskErr = service.TaskErrorWrapper(errors.Wrap(err, fmt.Sprintf("%s", responseBody)), "unmarshal_response_failed", http.StatusInternalServerError)
+ taskErr = service.TaskErrorWrapperUpstream(errors.Wrap(err, fmt.Sprintf("%s", responseBody)), "unmarshal_response_failed", http.StatusInternalServerError)
return
}
diff --git a/relay/channel/tencent/relay-tencent.go b/relay/channel/tencent/relay-tencent.go
index 611c186defa7..34e39d1fefaa 100644
--- a/relay/channel/tencent/relay-tencent.go
+++ b/relay/channel/tencent/relay-tencent.go
@@ -145,7 +145,7 @@ func tencentHandler(c *gin.Context, info *relaycommon.RelayInfo, resp *http.Resp
return nil, types.NewOpenAIError(err, types.ErrorCodeBadResponseBody, http.StatusInternalServerError)
}
if tencentSb.Response.Error.Code != 0 {
- return nil, types.WithOpenAIError(types.OpenAIError{
+ return nil, types.WithUpstreamOpenAIError(types.OpenAIError{
Message: tencentSb.Response.Error.Message,
Code: tencentSb.Response.Error.Code,
}, resp.StatusCode)
diff --git a/relay/channel/zhipu/relay-zhipu.go b/relay/channel/zhipu/relay-zhipu.go
index 0c280e2b705d..fae2f2f2332a 100644
--- a/relay/channel/zhipu/relay-zhipu.go
+++ b/relay/channel/zhipu/relay-zhipu.go
@@ -234,7 +234,7 @@ func zhipuHandler(c *gin.Context, info *relaycommon.RelayInfo, resp *http.Respon
return nil, types.NewOpenAIError(err, types.ErrorCodeBadResponseBody, http.StatusInternalServerError)
}
if !zhipuResponse.Success {
- return nil, types.WithOpenAIError(types.OpenAIError{
+ return nil, types.WithUpstreamOpenAIError(types.OpenAIError{
Message: zhipuResponse.Msg,
Code: zhipuResponse.Code,
}, resp.StatusCode)
diff --git a/relay/channel/zhipu_4v/image.go b/relay/channel/zhipu_4v/image.go
index cdb35a82c8ad..bd97804c5dfd 100644
--- a/relay/channel/zhipu_4v/image.go
+++ b/relay/channel/zhipu_4v/image.go
@@ -67,7 +67,7 @@ func zhipu4vImageHandler(c *gin.Context, resp *http.Response, info *relaycommon.
}
if zhipuResp.Error != nil && zhipuResp.Error.Message != "" {
- return nil, types.WithOpenAIError(types.OpenAIError{
+ return nil, types.WithUpstreamOpenAIError(types.OpenAIError{
Message: zhipuResp.Error.Message,
Type: "zhipu_image_error",
Code: zhipuResp.Error.Code,
diff --git a/relay/mjproxy_handler.go b/relay/mjproxy_handler.go
index daa0585a0797..f64dc9e98df3 100644
--- a/relay/mjproxy_handler.go
+++ b/relay/mjproxy_handler.go
@@ -21,6 +21,7 @@ import (
"github.com/QuantumNous/new-api/relay/helper"
"github.com/QuantumNous/new-api/service"
"github.com/QuantumNous/new-api/setting"
+ "github.com/QuantumNous/new-api/setting/operation_setting"
"github.com/QuantumNous/new-api/setting/system_setting"
"github.com/gin-gonic/gin"
@@ -161,7 +162,9 @@ func coverMidjourneyTaskDto(c *gin.Context, originTask *model.Midjourney) (midjo
midjourneyTask.VideoUrl = originTask.VideoUrl
}
midjourneyTask.Status = originTask.Status
- midjourneyTask.FailReason = originTask.FailReason
+ // FailReason 落库时保留上游原文（排障与管理员视图需要），在这个面向用户的读取边界
+ // 应用覆写。coverMidjourneyTaskDto 只服务 /mj/task/... 查询，没有管理员调用方。
+ midjourneyTask.FailReason = operation_setting.OverrideUpstreamMessage(originTask.FailReason)
midjourneyTask.Action = originTask.Action
midjourneyTask.Description = originTask.Description
midjourneyTask.Prompt = originTask.Prompt
diff --git a/relay/relay_task.go b/relay/relay_task.go
index fb384d18937a..ebd9f0a95892 100644
--- a/relay/relay_task.go
+++ b/relay/relay_task.go
@@ -19,7 +19,10 @@ import (
relayconstant "github.com/QuantumNous/new-api/relay/constant"
"github.com/QuantumNous/new-api/relay/helper"
"github.com/QuantumNous/new-api/service"
+ "github.com/QuantumNous/new-api/setting/operation_setting"
"github.com/gin-gonic/gin"
+ "github.com/tidwall/gjson"
+ "github.com/tidwall/sjson"
)
type TaskSubmitResult struct {
@@ -223,7 +226,7 @@ func RelayTaskSubmit(c *gin.Context, info *relaycommon.RelayInfo) (*TaskSubmitRe
}
if resp != nil && resp.StatusCode != http.StatusOK {
responseBody, _ := io.ReadAll(resp.Body)
- return nil, service.TaskErrorWrapper(fmt.Errorf("%s", string(responseBody)), "fail_to_fetch_task", resp.StatusCode)
+ return nil, service.TaskErrorWrapperUpstream(fmt.Errorf("%s", string(responseBody)), "fail_to_fetch_task", resp.StatusCode)
}
// 10. 返回 OtherRatios 给下游（header 必须在 DoResponse 写 body 之前设置）
@@ -335,7 +338,7 @@ func sunoFetchRespBodyBuilder(c *gin.Context) (respBody []byte, taskResp *dto.Ta
return
}
for _, task := range taskModels {
- tasks = append(tasks, TaskModel2Dto(task))
+ tasks = append(tasks, TaskModel2Dto(task, true))
}
} else {
tasks = make([]any, 0)
@@ -363,7 +366,7 @@ func sunoFetchByIDRespBodyBuilder(c *gin.Context) (respBody []byte, taskResp *dt
respBody, err = common.Marshal(dto.TaskResponse[any]{
Code: "success",
- Data: TaskModel2Dto(originTask),
+ Data: TaskModel2Dto(originTask, true),
})
return
}
@@ -406,7 +409,7 @@ func videoFetchByIDRespBodyBuilder(c *gin.Context) (respBody []byte, taskResp *d
taskResp = service.TaskErrorWrapper(err, "convert_to_openai_video_failed", http.StatusInternalServerError)
return
}
- respBody = openAIVideoData
+ respBody = overrideOpenAIVideoUpstreamError(openAIVideoData)
return
}
taskResp = service.TaskErrorWrapperLocal(fmt.Errorf("not_implemented:%s", originTask.Platform), "not_implemented", http.StatusNotImplemented)
@@ -416,7 +419,7 @@ func videoFetchByIDRespBodyBuilder(c *gin.Context) (respBody []byte, taskResp *d
// 通用 TaskDto 格式
respBody, err = common.Marshal(dto.TaskResponse[any]{
Code: "success",
- Data: TaskModel2Dto(originTask),
+ Data: TaskModel2Dto(originTask, true),
})
if err != nil {
taskResp = service.TaskErrorWrapper(err, "marshal_response_failed", http.StatusInternalServerError)
@@ -547,7 +550,44 @@ func mapTaskStatusToSimple(status model.TaskStatus) string {
}
}
-func TaskModel2Dto(task *model.Task) *dto.TaskDto {
+// overrideOpenAIVideoUpstreamError 覆写 OpenAI Video 响应体里的上游错误文案。
+// /v1/videos/{id} 走各 adaptor 的 ConvertToOpenAIVideo，响应体由落库的上游原始数据构建
+// （service/task_polling.go 的 task.Data），不经过 TaskModel2Dto，是上游失败文案的另一个
+// 用户可见出口。
+//
+// 直接改写 JSON 而不是反序列化成 dto.OpenAIVideo 再序列化：sora 的 converter 把上游对象整体
+// 透传，结构体往返会丢掉上游多返回的字段。与 relay 链路一致，只替换 message，error.code 保留。
+func overrideOpenAIVideoUpstreamError(respBody []byte) []byte {
+ message := gjson.GetBytes(respBody, "error.message")
+ if message.Type != gjson.String || message.String() == "" {
+ return respBody
+ }
+ overridden := operation_setting.OverrideUpstreamMessage(message.String())
+ if overridden == message.String() {
+ return respBody
+ }
+ masked, err := sjson.SetBytes(respBody, "error.message", overridden)
+ if err != nil {
+ // 改写失败时宁可不返回上游原文
+ return []byte(fmt.Sprintf(`{"error":{"message":%q}}`, overridden))
+ }
+ return masked
+}
+
+// TaskModel2Dto 把任务模型转为对外 DTO。maskUpstreamFailReason 为 true 时对 FailReason
+// 应用错误信息覆写：异步任务把上游失败原因落库（service/task_polling.go 的
+// task.FailReason = taskResult.Reason），用户随后通过任务查询接口读到它，这是同一份上游
+// 文案的另一个出口。管理员视图传 false，始终看原文。
+//
+// 与 relay/task 链路不同，这里只能按关键词门控：Task 表没有来源标记列，补一列要跨三种
+// 数据库做迁移，代价与收益不匹配。本站自产的 FailReason 是「任务超时（%d分钟）」这类
+// 中文文案（sweepTimedOutTasks）和 upstream returned error 这类固定串（FailTaskInfo），
+// 都不含默认关键词；但管理员配置过宽的关键词时本站文案仍可能被误伤。
+func TaskModel2Dto(task *model.Task, maskUpstreamFailReason bool) *dto.TaskDto {
+ failReason := task.FailReason
+ if maskUpstreamFailReason {
+ failReason = operation_setting.OverrideUpstreamMessage(failReason)
+ }
return &dto.TaskDto{
ID: task.ID,
CreatedAt: task.CreatedAt,
@@ -560,7 +600,7 @@ func TaskModel2Dto(task *model.Task) *dto.TaskDto {
Quota: task.Quota,
Action: task.Action,
Status: string(task.Status),
- FailReason: task.FailReason,
+ FailReason: failReason,
ResultURL: task.GetResultURL(),
SubmitTime: task.SubmitTime,
StartTime: task.StartTime,
diff --git a/relay/task_fail_reason_override_test.go b/relay/task_fail_reason_override_test.go
new file mode 100644
index 000000000000..0af1b7adf10e
--- /dev/null
+++ b/relay/task_fail_reason_override_test.go
@@ -0,0 +1,119 @@
+package relay
+
+import (
+ "testing"
+
+ "github.com/QuantumNous/new-api/model"
+ "github.com/QuantumNous/new-api/setting/operation_setting"
+
+ "github.com/stretchr/testify/assert"
+ "github.com/stretchr/testify/require"
+ "github.com/tidwall/gjson"
+)
+
+// Async tasks persist the upstream failure reason (service/task_polling.go sets
+// task.FailReason = taskResult.Reason) and users read it back through the task query
+// endpoints. That is a second outlet for the same upstream text the relay response
+// override hides, so TaskModel2Dto must mask it for user-facing callers while the admin
+// view keeps the original for diagnosis.
+func TestTaskModel2Dto_FailReasonOverride(t *testing.T) {
+ origEnabled := operation_setting.ErrorOverrideEnabled
+ origKeywords := operation_setting.ErrorOverrideKeywords
+ t.Cleanup(func() {
+ operation_setting.ErrorOverrideEnabled = origEnabled
+ operation_setting.ErrorOverrideKeywords = origKeywords
+ })
+ operation_setting.ErrorOverrideEnabled = true
+ operation_setting.ErrorOverrideKeywords = []string{"no available", "quota", "credits", "top-up"}
+
+ upstreamReason := "insufficient credits, please top up your account"
+
+ t.Run("user view masks an upstream reason", func(t *testing.T) {
+ task := &model.Task{TaskID: "t-1", FailReason: upstreamReason}
+ assert.Equal(t, operation_setting.ErrorOverrideMessage,
+ TaskModel2Dto(task, true).FailReason)
+ })
+
+ t.Run("admin view keeps the original reason", func(t *testing.T) {
+ task := &model.Task{TaskID: "t-1", FailReason: upstreamReason}
+ assert.Equal(t, upstreamReason, TaskModel2Dto(task, false).FailReason)
+ })
+
+ // Local reasons produced by this site must survive the user-facing path. These are
+ // the strings sweepTimedOutTasks and FailTaskInfo actually write.
+ t.Run("local reasons are not masked", func(t *testing.T) {
+ for _, reason := range []string{
+ "任务超时（30分钟）",
+ "任务超时（旧系统遗留任务，不进行退款，请联系管理员）",
+ "upstream returned error",
+ "upstream returned unrecognized message",
+ } {
+ task := &model.Task{TaskID: "t-2", FailReason: reason}
+ assert.Equal(t, reason, TaskModel2Dto(task, true).FailReason, reason)
+ }
+ })
+
+ t.Run("empty reason stays empty", func(t *testing.T) {
+ task := &model.Task{TaskID: "t-3"}
+ require.Empty(t, TaskModel2Dto(task, true).FailReason)
+ })
+
+ t.Run("disabled override keeps the upstream reason", func(t *testing.T) {
+ operation_setting.ErrorOverrideEnabled = false
+ defer func() { operation_setting.ErrorOverrideEnabled = true }()
+
+ task := &model.Task{TaskID: "t-4", FailReason: upstreamReason}
+ assert.Equal(t, upstreamReason, TaskModel2Dto(task, true).FailReason)
+ })
+}
+
+// /v1/videos/{id} builds its response from the stored upstream payload through each
+// adaptor's ConvertToOpenAIVideo, bypassing TaskModel2Dto entirely. It is therefore a
+// third outlet for the same upstream failure text and must be masked at the JSON level:
+// the sora converter passes the whole upstream object through, so a struct round-trip
+// would silently drop fields the upstream returned.
+func TestOverrideOpenAIVideoUpstreamError(t *testing.T) {
+ origEnabled := operation_setting.ErrorOverrideEnabled
+ origKeywords := operation_setting.ErrorOverrideKeywords
+ t.Cleanup(func() {
+ operation_setting.ErrorOverrideEnabled = origEnabled
+ operation_setting.ErrorOverrideKeywords = origKeywords
+ })
+ operation_setting.ErrorOverrideEnabled = true
+ operation_setting.ErrorOverrideKeywords = []string{"no available", "quota", "credits", "top-up"}
+
+ t.Run("masks the message and keeps every other upstream field", func(t *testing.T) {
+ body := []byte(`{"id":"video_1","status":"failed","seconds":"8","error":{"code":"billing_hard_limit_reached","message":"You have insufficient credits, please top-up","upstream_only":"keep me"}}`)
+
+ got := overrideOpenAIVideoUpstreamError(body)
+
+ assert.Equal(t, operation_setting.ErrorOverrideMessage, gjson.GetBytes(got, "error.message").String())
+ // Status code equivalents and any extra upstream fields survive the rewrite.
+ assert.Equal(t, "billing_hard_limit_reached", gjson.GetBytes(got, "error.code").String())
+ assert.Equal(t, "keep me", gjson.GetBytes(got, "error.upstream_only").String())
+ assert.Equal(t, "video_1", gjson.GetBytes(got, "id").String())
+ assert.Equal(t, "failed", gjson.GetBytes(got, "status").String())
+ assert.Equal(t, "8", gjson.GetBytes(got, "seconds").String())
+ })
+
+ t.Run("leaves bodies without a matching message byte-identical", func(t *testing.T) {
+ for _, body := range []string{
+ `{"id":"video_2","status":"completed"}`,
+ `{"id":"video_3","status":"failed","error":{"code":"moderation_blocked","message":"content policy violation"}}`,
+ `{"id":"video_4","error":null}`,
+ `{"id":"video_5","error":"insufficient credits"}`,
+ `not json at all, please top-up`,
+ ``,
+ } {
+ assert.Equal(t, body, string(overrideOpenAIVideoUpstreamError([]byte(body))), body)
+ }
+ })
+
+ t.Run("disabled override keeps the upstream message", func(t *testing.T) {
+ operation_setting.ErrorOverrideEnabled = false
+ defer func() { operation_setting.ErrorOverrideEnabled = true }()
+
+ body := `{"error":{"message":"You have insufficient credits, please top-up"}}`
+ assert.Equal(t, body, string(overrideOpenAIVideoUpstreamError([]byte(body))))
+ })
+}
diff --git a/relaykit/types/error.go b/relaykit/types/error.go
index 387fdad76948..0691143e7108 100644
--- a/relaykit/types/error.go
+++ b/relaykit/types/error.go
@@ -92,6 +92,7 @@ type NewAPIError struct {
RelayError any
skipRetry bool
recordErrorLog *bool
+ fromUpstream bool
errorType ErrorType
errorCode ErrorCode
StatusCode int
@@ -177,6 +178,44 @@ func (e *NewAPIError) SetMessage(message string) {
e.Err = errors.New(message)
}
+// MarkUpstreamOrigin records that this error's message text came from an upstream
+// provider response rather than being produced locally by the gateway. Callers that
+// decode an upstream error body must set this so the message can be treated as
+// upstream-controlled text downstream.
+func (e *NewAPIError) MarkUpstreamOrigin() {
+ if e == nil {
+ return
+ }
+ e.fromUpstream = true
+}
+
+// ReplaceMessage rewrites the user-facing message on every representation of the
+// error while preserving StatusCode, type and code. Unlike SetMessage it also
+// updates RelayError, because ToOpenAIError/ToClaudeError return the provider
+// payload verbatim for upstream errors and never read Err.
+//
+// Metadata and Param are cleared alongside the message: metadata carries the raw
+// provider error body, and Param carries provider-supplied detail such as an
+// upstream request id (see the Ali rerank adaptor). Leaving either in place would
+// keep leaking the upstream trail that replacing the message is meant to hide.
+func (e *NewAPIError) ReplaceMessage(message string) {
+ if e == nil {
+ return
+ }
+ e.Err = errors.New(message)
+ e.Metadata = nil
+ switch relayError := e.RelayError.(type) {
+ case OpenAIError:
+ relayError.Message = message
+ relayError.Metadata = nil
+ relayError.Param = ""
+ e.RelayError = relayError
+ case ClaudeError:
+ relayError.Message = message
+ e.RelayError = relayError
+ }
+}
+
func (e *NewAPIError) ToOpenAIError() OpenAIError {
var result OpenAIError
switch e.errorType {
@@ -363,6 +402,23 @@ func WithClaudeError(claudeError ClaudeError, statusCode int, ops ...NewAPIError
return e
}
+// WithUpstreamOpenAIError builds an error from an OpenAI-shaped payload decoded out
+// of an upstream provider response, recording that the message text is upstream
+// controlled. Channel adaptors that surface a provider error must use this instead of
+// WithOpenAIError, which is also used for locally constructed errors.
+func WithUpstreamOpenAIError(openAIError OpenAIError, statusCode int, ops ...NewAPIErrorOptions) *NewAPIError {
+ e := WithOpenAIError(openAIError, statusCode, ops...)
+ e.MarkUpstreamOrigin()
+ return e
+}
+
+// WithUpstreamClaudeError is the Claude-shaped counterpart of WithUpstreamOpenAIError.
+func WithUpstreamClaudeError(claudeError ClaudeError, statusCode int, ops ...NewAPIErrorOptions) *NewAPIError {
+ e := WithClaudeError(claudeError, statusCode, ops...)
+ e.MarkUpstreamOrigin()
+ return e
+}
+
func IsChannelError(err *NewAPIError) bool {
if err == nil {
return false
@@ -378,6 +434,16 @@ func IsSkipRetryError(err *NewAPIError) bool {
return err.skipRetry
}
+// IsFromUpstreamError reports whether the error message originated from an upstream
+// provider response. Errors produced by the gateway itself return false.
+func IsFromUpstreamError(err *NewAPIError) bool {
+ if err == nil {
+ return false
+ }
+
+ return err.fromUpstream
+}
+
func ErrOptionWithSkipRetry() NewAPIErrorOptions {
return func(e *NewAPIError) {
e.skipRetry = true
diff --git a/relaykit/types/error_test.go b/relaykit/types/error_test.go
new file mode 100644
index 000000000000..87749dcc978c
--- /dev/null
+++ b/relaykit/types/error_test.go
@@ -0,0 +1,143 @@
+package types
+
+import (
+ "encoding/json"
+ "errors"
+ "net/http"
+ "testing"
+
+ "github.com/stretchr/testify/assert"
+ "github.com/stretchr/testify/require"
+)
+
+func TestMarkUpstreamOrigin_NilSafe(t *testing.T) {
+ var e *NewAPIError
+ require.NotPanics(t, func() { e.MarkUpstreamOrigin() })
+ assert.False(t, IsFromUpstreamError(e))
+}
+
+func TestReplaceMessage_NilSafe(t *testing.T) {
+ var e *NewAPIError
+ require.NotPanics(t, func() { e.ReplaceMessage("x") })
+}
+
+func TestIsFromUpstreamError_Nil(t *testing.T) {
+ assert.False(t, IsFromUpstreamError(nil))
+}
+
+func TestWithUpstreamOpenAIError_MarksOrigin(t *testing.T) {
+ e := WithUpstreamOpenAIError(OpenAIError{
+ Message: "Your credit balance is too low, please top up",
+ Type: "upstream_error",
+ Code: "credits_exhausted",
+ }, http.StatusPaymentRequired)
+
+ require.True(t, IsFromUpstreamError(e))
+ assert.Equal(t, http.StatusPaymentRequired, e.StatusCode)
+ assert.Equal(t, ErrorTypeOpenAIError, e.GetErrorType())
+ assert.Equal(t, ErrorCode("credits_exhausted"), e.GetErrorCode())
+}
+
+func TestWithOpenAIError_DoesNotMarkOrigin(t *testing.T) {
+ e := WithOpenAIError(OpenAIError{
+ Message: "some local message",
+ Type: "upstream_error",
+ }, http.StatusBadRequest)
+ assert.False(t, IsFromUpstreamError(e))
+}
+
+func TestWithUpstreamClaudeError_MarksOrigin(t *testing.T) {
+ e := WithUpstreamClaudeError(ClaudeError{
+ Type: "error",
+ Message: "You exceeded your current quota",
+ }, http.StatusTooManyRequests)
+
+ require.True(t, IsFromUpstreamError(e))
+ assert.Equal(t, ErrorTypeClaudeError, e.GetErrorType())
+ assert.Equal(t, http.StatusTooManyRequests, e.StatusCode)
+}
+
+func TestWithClaudeError_DoesNotMarkOrigin(t *testing.T) {
+ e := WithClaudeError(ClaudeError{
+ Type: "error",
+ Message: "local",
+ }, http.StatusBadRequest)
+ assert.False(t, IsFromUpstreamError(e))
+}
+
+// ReplaceMessage must rewrite every user-facing representation while preserving
+// StatusCode, type, code, and clearing upstream metadata.
+func TestReplaceMessage_OpenAIError(t *testing.T) {
+ metadata := json.RawMessage(`{"raw":"upstream-secret"}`)
+ e := WithUpstreamOpenAIError(OpenAIError{
+ Message: "No available channel for model gpt-4o under group default",
+ Type: "upstream_error",
+ Code: "no_available_channel",
+ Param: "upstream-request-id-123",
+ Metadata: metadata,
+ }, http.StatusServiceUnavailable)
+ e.Metadata = metadata
+
+ e.ReplaceMessage("Service Unavailable")
+
+ // Err mirrors the new message.
+ assert.Equal(t, "Service Unavailable", e.Error())
+
+ // ToOpenAIError returns the RelayError payload verbatim, so it must reflect the new message.
+ openAI := e.ToOpenAIError()
+ assert.Equal(t, "Service Unavailable", openAI.Message)
+ assert.Nil(t, openAI.Metadata)
+ // Param carries provider detail such as an upstream request id and must not survive.
+ assert.Equal(t, "", openAI.Param)
+
+ // Preserved fields.
+ assert.Equal(t, http.StatusServiceUnavailable, e.StatusCode)
+ assert.Equal(t, ErrorTypeOpenAIError, e.GetErrorType())
+ assert.Equal(t, ErrorCode("no_available_channel"), e.GetErrorCode())
+ assert.Nil(t, e.Metadata)
+}
+
+func TestReplaceMessage_ClaudeError(t *testing.T) {
+ e := WithUpstreamClaudeError(ClaudeError{
+ Type: "error",
+ Message: "You exceeded your current quota",
+ }, http.StatusTooManyRequests)
+
+ e.ReplaceMessage("Service Unavailable")
+
+ assert.Equal(t, "Service Unavailable", e.Error())
+ claude := e.ToClaudeError()
+ assert.Equal(t, "Service Unavailable", claude.Message)
+ assert.Equal(t, "error", claude.Type)
+ assert.Equal(t, http.StatusTooManyRequests, e.StatusCode)
+ assert.Equal(t, ErrorTypeClaudeError, e.GetErrorType())
+}
+
+func TestReplaceMessage_DefaultErrorType(t *testing.T) {
+ e := NewErrorWithStatusCode(errors.New("upstream quota exhausted"),
+ ErrorCodeBadResponseStatusCode, http.StatusBadGateway)
+ e.MarkUpstreamOrigin()
+ require.True(t, IsFromUpstreamError(e))
+
+ e.ReplaceMessage("Service Unavailable")
+
+ assert.Equal(t, "Service Unavailable", e.Error())
+ openAI := e.ToOpenAIError()
+ assert.Equal(t, "Service Unavailable", openAI.Message)
+ assert.Equal(t, http.StatusBadGateway, e.StatusCode)
+ assert.Equal(t, ErrorTypeNewAPIError, e.GetErrorType())
+}
+
+// NewError preserves a deeply wrapped *NewAPIError, so the upstream marker must
+// survive wrapping carried by errors.As.
+func TestUpstreamMarker_SurvivesNewErrorWrap(t *testing.T) {
+ upstream := WithUpstreamOpenAIError(OpenAIError{
+ Message: "Please top up your credits",
+ Type: "upstream_error",
+ }, http.StatusPaymentRequired)
+
+ wrapped := NewError(upstream, ErrorCodeBadResponseStatusCode)
+
+ require.True(t, IsFromUpstreamError(wrapped))
+ assert.Equal(t, http.StatusPaymentRequired, wrapped.StatusCode)
+}
diff --git a/service/error.go b/service/error.go
index f14f1bbab660..1f21155994f2 100644
--- a/service/error.go
+++ b/service/error.go
@@ -89,8 +89,10 @@ func RelayErrorHandler(ctx context.Context, resp *http.Response, showBodyWhenFai
responseBody, err := io.ReadAll(resp.Body)
if err != nil {
+ // 读取上游响应体失败是本站基础设施错误，文案并非来自上游，不应标记上游来源
return
}
+
CloseResponseBodyGracefully(resp)
var errResponse dto.GeneralErrorResponse
responseBodyText := string(responseBody)
@@ -105,8 +107,12 @@ func RelayErrorHandler(ctx context.Context, resp *http.Response, showBodyWhenFai
err = common.Unmarshal(responseBody, &errResponse)
if err != nil {
if showBodyWhenFail {
+ // 文案内嵌上游响应体原文，属于上游来源
newApiErr.Err = buildErrWithBody("")
+ newApiErr.MarkUpstreamOrigin()
} else {
+ // 响应体解析失败且不回显 body 时，文案完全由本站生成（只含状态码），
+ // 不标记上游来源，避免覆写关键词误伤本站文案
logger.LogError(ctx, fmt.Sprintf("bad response status code %d, body: %s", resp.StatusCode, responseBodyPreview))
newApiErr.Err = fmt.Errorf("bad response status code %d", resp.StatusCode)
}
@@ -117,7 +123,7 @@ func RelayErrorHandler(ctx context.Context, resp *http.Response, showBodyWhenFai
// General format error (OpenAI, Anthropic, Gemini, etc.)
oaiError := errResponse.TryToOpenAIError()
if oaiError != nil {
- newApiErr = types.WithOpenAIError(*oaiError, resp.StatusCode)
+ newApiErr = types.WithUpstreamOpenAIError(*oaiError, resp.StatusCode)
if showBodyWhenFail {
newApiErr.Err = buildErrWithBody(newApiErr.Error())
}
@@ -131,6 +137,8 @@ func RelayErrorHandler(ctx context.Context, resp *http.Response, showBodyWhenFai
logger.LogError(ctx, fmt.Sprintf("bad response status code %d with empty error message, body: %s", resp.StatusCode, responseBodyPreview))
}
newApiErr = types.NewOpenAIError(errors.New(message), types.ErrorCodeBadResponseStatusCode, resp.StatusCode)
+ // message 取自上游响应体（errResponse.ToMessage()），标记为上游来源
+ newApiErr.MarkUpstreamOrigin()
if showBodyWhenFail {
newApiErr.Err = buildErrWithBody(newApiErr.Error())
}
@@ -197,6 +205,15 @@ func TaskErrorWrapperLocal(err error, code string, statusCode int) *taskdto.Task
return openaiErr
}
+// TaskErrorWrapperUpstream 包装文案取自上游任务平台响应体的错误，标记后可被错误信息覆写识别。
+// 适配器把上游返回的 message/code 直接塞进错误时必须用它，而不是 TaskErrorWrapper —— 后者
+// 同样服务于本站自产的读取/解析失败。
+func TaskErrorWrapperUpstream(err error, code string, statusCode int) *taskdto.TaskError {
+ taskErr := TaskErrorWrapper(err, code, statusCode)
+ taskErr.FromUpstream = true
+ return taskErr
+}
+
func TaskErrorWrapper(err error, code string, statusCode int) *taskdto.TaskError {
text := err.Error()
lowerText := strings.ToLower(text)
@@ -217,14 +234,38 @@ func TaskErrorWrapper(err error, code string, statusCode int) *taskdto.TaskError
}
// TaskErrorFromAPIError 将 PreConsumeBilling 返回的 NewAPIError 转换为 TaskError。
+// 预扣费错误（额度不足、订阅未配置等）由本站产生，FromUpstream 保持 false，
+// 文案不会被错误信息覆写掩盖，用户仍能看到真实的额度原因。
func TaskErrorFromAPIError(apiErr *types.NewAPIError) *taskdto.TaskError {
if apiErr == nil {
return nil
}
return &taskdto.TaskError{
- Code: string(apiErr.GetErrorCode()),
- Message: apiErr.Err.Error(),
- StatusCode: apiErr.StatusCode,
- Error: apiErr.Err,
+ Code: string(apiErr.GetErrorCode()),
+ Message: apiErr.Err.Error(),
+ StatusCode: apiErr.StatusCode,
+ FromUpstream: types.IsFromUpstreamError(apiErr),
+ Error: apiErr.Err,
+ }
+}
+
+// APIErrorFromTaskError 是 TaskErrorFromAPIError 的反向转换，供任务链路复用 relay 的渠道
+// 错误处理（processChannelError：渠道自动禁用判定与用户可见的错误日志）。
+//
+// FromUpstream 必须一并带过去：错误日志的 Content 会通过 /api/log/self 回显给用户，
+// 是和响应体同级的对外出口。丢掉标记会让任务链路的日志始终写入上游原文，客户端在响应里
+// 看到 Service Unavailable，转头在日志页仍能读到上游账务细节。
+func APIErrorFromTaskError(taskErr *taskdto.TaskError) *types.NewAPIError {
+ if taskErr == nil {
+ return nil
+ }
+ err := taskErr.Error
+ if err == nil {
+ err = errors.New(taskErr.Message)
+ }
+ apiErr := types.NewOpenAIError(err, types.ErrorCodeBadResponseStatusCode, taskErr.StatusCode)
+ if taskErr.FromUpstream {
+ apiErr.MarkUpstreamOrigin()
}
+ return apiErr
}
diff --git a/service/error_test.go b/service/error_test.go
index 266d2a875560..0efba762cb2d 100644
--- a/service/error_test.go
+++ b/service/error_test.go
@@ -3,6 +3,7 @@ package service
import (
"bytes"
"context"
+ "errors"
"fmt"
"io"
"net/http"
@@ -10,6 +11,7 @@ import (
"testing"
"github.com/QuantumNous/new-api/common"
+ taskdto "github.com/QuantumNous/new-api/dto"
"github.com/QuantumNous/new-api/relaykit/types"
"github.com/gin-gonic/gin"
"github.com/stretchr/testify/require"
@@ -150,6 +152,141 @@ func TestRelayErrorHandlerKeepsInvalidJSONBodyInDebugLog(t *testing.T) {
require.Contains(t, logBuffer.String(), body)
}
+// TestRelayErrorHandlerIOErrorNotMarkedUpstream verifies that a local I/O failure
+// while reading the upstream body is not marked as upstream-origin. The defer-based
+// marker must only cover paths whose message text is sourced from the response body.
+func TestRelayErrorHandlerIOErrorNotMarkedUpstream(t *testing.T) {
+ t.Parallel()
+
+ // A reader that always errors on Read, so io.ReadAll fails before any
+ // upstream-derived message is constructed.
+ resp := &http.Response{
+ StatusCode: http.StatusBadGateway,
+ Body: io.NopCloser(errReader{}),
+ }
+
+ newAPIError := RelayErrorHandler(context.Background(), resp, false)
+
+ require.NotNil(t, newAPIError)
+ require.False(t, types.IsFromUpstreamError(newAPIError),
+ "I/O failure reading upstream body is a local error and must not be marked upstream")
+}
+
+// The upstream marker must follow the text, not merely the position in the function:
+// a body that fails to parse yields a message this site builds from the status code
+// alone, so it stays local unless the raw body is echoed back into it.
+func TestRelayErrorHandlerMarksUpstreamPerPath(t *testing.T) {
+ cases := []struct {
+ name string
+ body string
+ showBodyWhenFail bool
+ wantUpstream bool
+ }{
+ {
+ name: "unparseable body without echo stays local",
+ body: "502 Bad Gateway",
+ showBodyWhenFail: false,
+ wantUpstream: false,
+ },
+ {
+ name: "unparseable body echoed into message is upstream",
+ body: "502 Bad Gateway",
+ showBodyWhenFail: true,
+ wantUpstream: true,
+ },
+ {
+ name: "structured provider error is upstream",
+ body: `{"error":{"message":"You exceeded your current quota","type":"insufficient_quota"}}`,
+ showBodyWhenFail: false,
+ wantUpstream: true,
+ },
+ {
+ name: "plain message body is upstream",
+ body: `{"message":"Please top up your credits"}`,
+ showBodyWhenFail: false,
+ wantUpstream: true,
+ },
+ }
+
+ for _, tc := range cases {
+ t.Run(tc.name, func(t *testing.T) {
+ withDebugEnabled(t, false)
+
+ resp := &http.Response{
+ StatusCode: http.StatusBadGateway,
+ Body: io.NopCloser(strings.NewReader(tc.body)),
+ }
+
+ newAPIError := RelayErrorHandler(context.Background(), resp, tc.showBodyWhenFail)
+
+ require.NotNil(t, newAPIError)
+ require.Equal(t, tc.wantUpstream, types.IsFromUpstreamError(newAPIError))
+ })
+ }
+}
+
+// The task relay loop reuses processChannelError for channel auto-disable and for the
+// user-visible error log, so it converts TaskError back into NewAPIError. The upstream
+// marker has to survive that conversion: without it the log always records the raw
+// upstream text, and a user who sees "Service Unavailable" in the response can still
+// read the upstream billing details on the log page.
+func TestAPIErrorFromTaskError(t *testing.T) {
+ t.Parallel()
+
+ t.Run("nil task error converts to nil", func(t *testing.T) {
+ t.Parallel()
+ require.Nil(t, APIErrorFromTaskError(nil))
+ })
+
+ t.Run("upstream marker and status code survive the conversion", func(t *testing.T) {
+ t.Parallel()
+
+ taskErr := TaskErrorWrapperUpstream(errors.New("insufficient credits"), "fail_to_fetch_task", http.StatusPaymentRequired)
+ apiErr := APIErrorFromTaskError(taskErr)
+
+ require.NotNil(t, apiErr)
+ require.True(t, types.IsFromUpstreamError(apiErr))
+ require.Equal(t, http.StatusPaymentRequired, apiErr.StatusCode)
+ require.Equal(t, "insufficient credits", apiErr.Error())
+ })
+
+ t.Run("locally produced task errors stay local", func(t *testing.T) {
+ t.Parallel()
+
+ taskErr := TaskErrorWrapperLocal(errors.New("video_id is required"), "invalid_request", http.StatusBadRequest)
+ apiErr := APIErrorFromTaskError(taskErr)
+
+ require.NotNil(t, apiErr)
+ require.False(t, types.IsFromUpstreamError(apiErr))
+ require.Equal(t, "video_id is required", apiErr.Error())
+ })
+
+ // TaskErrorFromAPIError leaves Error unset when the source carried no wrapped error,
+ // so the message is the only text available.
+ t.Run("falls back to Message when Error is nil", func(t *testing.T) {
+ t.Parallel()
+
+ apiErr := APIErrorFromTaskError(&taskdto.TaskError{
+ Message: "upstream rejected the request",
+ StatusCode: http.StatusBadGateway,
+ FromUpstream: true,
+ })
+
+ require.NotNil(t, apiErr)
+ require.Equal(t, "upstream rejected the request", apiErr.Error())
+ require.True(t, types.IsFromUpstreamError(apiErr))
+ })
+}
+
+// errReader is an io.ReadCloser whose Read always returns an error.
+type errReader struct{}
+
+func (errReader) Read(p []byte) (int, error) {
+ return 0, io.ErrUnexpectedEOF
+}
+
+func (errReader) Close() error { return nil }
+
func withDebugEnabled(t *testing.T, enabled bool) {
t.Helper()
diff --git a/service/task_billing.go b/service/task_billing.go
index f310a8230c63..329c1af2859d 100644
--- a/service/task_billing.go
+++ b/service/task_billing.go
@@ -10,6 +10,7 @@ import (
"github.com/QuantumNous/new-api/logger"
"github.com/QuantumNous/new-api/model"
relaycommon "github.com/QuantumNous/new-api/relay/common"
+ "github.com/QuantumNous/new-api/setting/operation_setting"
"github.com/QuantumNous/new-api/setting/ratio_setting"
"github.com/QuantumNous/new-api/types"
"github.com/gin-gonic/gin"
@@ -184,6 +185,13 @@ func RefundTaskQuota(ctx context.Context, task *model.Task, reason string) bool
// 3. 记录日志
other := taskBillingOther(task)
other["task_id"] = task.TaskID
+ // 退款日志会通过 /api/log/self 回显给用户，reason 与任务 FailReason 是同一份上游文案，
+ // 因此和 relay 的错误日志一样在写库前覆写，原文改记到 admin_info
+ // （model.formatUserLogs 会为普通用户剥离整个 admin_info）。
+ if masked := operation_setting.OverrideUpstreamMessage(reason); masked != reason {
+ other["admin_info"] = map[string]interface{}{"original_reason": reason}
+ reason = masked
+ }
other["reason"] = reason
model.RecordTaskBillingLog(model.RecordTaskBillingLogParams{
UserId: task.UserId,
diff --git a/service/task_billing_test.go b/service/task_billing_test.go
index 53e3f680d01c..a7a0739c5c77 100644
--- a/service/task_billing_test.go
+++ b/service/task_billing_test.go
@@ -12,6 +12,7 @@ import (
"github.com/QuantumNous/new-api/common"
"github.com/QuantumNous/new-api/model"
relaycommon "github.com/QuantumNous/new-api/relay/common"
+ "github.com/QuantumNous/new-api/setting/operation_setting"
"github.com/QuantumNous/new-api/types"
"github.com/glebarez/sqlite"
"github.com/shopspring/decimal"
@@ -422,6 +423,73 @@ func TestRefundTaskQuota_FundingFailureKeepsPendingMarker(t *testing.T) {
assert.Equal(t, int64(0), countLogs(t))
}
+// The refund reason is the same upstream text the task query endpoints mask, and the
+// refund log is echoed to the user through /api/log/self. It therefore has to be
+// overridden before the row is written, with the original kept under admin_info —
+// model.formatUserLogs strips that whole key for non-admin views.
+func TestRefundTaskQuota_MasksUpstreamReason(t *testing.T) {
+ origEnabled := operation_setting.ErrorOverrideEnabled
+ origKeywords := operation_setting.ErrorOverrideKeywords
+ t.Cleanup(func() {
+ operation_setting.ErrorOverrideEnabled = origEnabled
+ operation_setting.ErrorOverrideKeywords = origKeywords
+ })
+ operation_setting.ErrorOverrideEnabled = true
+ operation_setting.ErrorOverrideKeywords = []string{"no available", "quota", "credits", "top-up"}
+
+ const upstreamReason = "insufficient credits, please top-up your account"
+ const localReason = "任务超时（30分钟）"
+
+ cases := []struct {
+ name string
+ userID int
+ reason string
+ wantReason string
+ wantOriginal string
+ }{
+ {
+ name: "upstream reason is masked and the original moves to admin_info",
+ userID: 11,
+ reason: upstreamReason,
+ wantReason: operation_setting.ErrorOverrideMessage,
+ wantOriginal: upstreamReason,
+ },
+ {
+ name: "local reason is written verbatim without admin_info",
+ userID: 12,
+ reason: localReason,
+ wantReason: localReason,
+ },
+ }
+
+ for _, tc := range cases {
+ t.Run(tc.name, func(t *testing.T) {
+ truncate(t)
+ const channelID, preConsumed = 11, 700
+ seedUser(t, tc.userID, 10000)
+ seedChannel(t, channelID)
+
+ task := makeTask(tc.userID, channelID, preConsumed, 0, BillingSourceWallet, 0)
+ require.NoError(t, model.DB.Create(task).Error)
+ require.True(t, RefundTaskQuota(context.Background(), task, tc.reason))
+
+ log := getLastLog(t)
+ require.NotNil(t, log)
+ other, err := common.StrToMap(log.Other)
+ require.NoError(t, err)
+ assert.Equal(t, tc.wantReason, other["reason"])
+
+ if tc.wantOriginal == "" {
+ assert.NotContains(t, other, "admin_info")
+ return
+ }
+ adminInfo, ok := other["admin_info"].(map[string]interface{})
+ require.True(t, ok, "admin_info should carry the original reason")
+ assert.Equal(t, tc.wantOriginal, adminInfo["original_reason"])
+ })
+ }
+}
+
// ===========================================================================
// RecalculateTaskQuota tests
// ===========================================================================
diff --git a/service/violation_fee.go b/service/violation_fee.go
index f51533629d76..26c34bfb66d5 100644
--- a/service/violation_fee.go
+++ b/service/violation_fee.go
@@ -45,7 +45,12 @@ func WrapAsViolationFeeGrokCSAM(err *types.NewAPIError) *types.NewAPIError {
oai := err.ToOpenAIError()
oai.Type = string(types.ErrorCodeViolationFeeGrokCSAM)
oai.Code = string(types.ErrorCodeViolationFeeGrokCSAM)
- return types.WithOpenAIError(oai, err.StatusCode, types.ErrOptionWithSkipRetry())
+ wrapped := types.WithOpenAIError(oai, err.StatusCode, types.ErrOptionWithSkipRetry())
+ // 重新包装不应改变错误来源，否则上游错误会被误判为本站错误
+ if types.IsFromUpstreamError(err) {
+ wrapped.MarkUpstreamOrigin()
+ }
+ return wrapped
}
// NormalizeViolationFeeError ensures:
@@ -64,7 +69,11 @@ func NormalizeViolationFeeError(err *types.NewAPIError) *types.NewAPIError {
if IsViolationFeeCode(err.GetErrorCode()) {
oai := err.ToOpenAIError()
- return types.WithOpenAIError(oai, err.StatusCode, types.ErrOptionWithSkipRetry())
+ wrapped := types.WithOpenAIError(oai, err.StatusCode, types.ErrOptionWithSkipRetry())
+ if types.IsFromUpstreamError(err) {
+ wrapped.MarkUpstreamOrigin()
+ }
+ return wrapped
}
return err
diff --git a/setting/model_setting/model_alias.go b/setting/model_setting/model_alias.go
index 078690c01e5b..0d7f4e5ae4e9 100644
--- a/setting/model_setting/model_alias.go
+++ b/setting/model_setting/model_alias.go
@@ -7,7 +7,6 @@ import (
"github.com/QuantumNous/new-api/common"
"github.com/QuantumNous/new-api/setting/config"
- "github.com/QuantumNous/new-api/setting/ratio_setting"
"github.com/QuantumNous/new-api/types"
)
@@ -27,18 +26,13 @@ func init() {
}
// ResolveModelAlias 把全局别名解析为真实模型名，支持链式映射（a->b->c）并检测环。
-// 自映射（a->a）视为未配置别名。带 compact 后缀的模型名先剥离后缀参与解析，结果再补回后缀。
+// 自映射（a->a）视为未配置别名。
// 返回（解析后的模型名, 是否发生了别名替换, error）。
func ResolveModelAlias(requested string) (string, bool, error) {
if requested == "" || modelAliasSetting.Mapping.Len() == 0 {
return requested, false, nil
}
- baseName := requested
- hasCompactSuffix := strings.HasSuffix(requested, ratio_setting.CompactModelSuffix)
- if hasCompactSuffix {
- baseName = strings.TrimSuffix(requested, ratio_setting.CompactModelSuffix)
- }
- current := baseName
+ current := requested
visited := map[string]bool{current: true}
applied := false
for {
@@ -56,9 +50,6 @@ func ResolveModelAlias(requested string) (string, bool, error) {
if !applied {
return requested, false, nil
}
- if hasCompactSuffix {
- current = ratio_setting.WithCompactModelSuffix(current)
- }
return current, true, nil
}
diff --git a/setting/model_setting/model_alias_test.go b/setting/model_setting/model_alias_test.go
index 4d04a53745fe..6dcfefbc815f 100644
--- a/setting/model_setting/model_alias_test.go
+++ b/setting/model_setting/model_alias_test.go
@@ -74,12 +74,22 @@ func TestResolveModelAlias(t *testing.T) {
wantApplied: false,
},
{
- name: "compact suffix stripped and restored",
- mapping: map[string]string{"cinax": "cinax-pro"},
+ // 别名按完整模型名整体匹配，不对名字做任何分段拆解。
+ name: "alias matches the full model name verbatim",
+ mapping: map[string]string{"cinax-openai-compact": "cinax-pro-openai-compact"},
requested: "cinax-openai-compact",
wantResolved: "cinax-pro-openai-compact",
wantApplied: true,
},
+ {
+ // 基名映射不会命中更长的模型名：解析不再剥离任何后缀，
+ // 带后缀的模型必须显式配置自己的映射。
+ name: "base-name mapping does not match a longer model name",
+ mapping: map[string]string{"cinax": "cinax-pro"},
+ requested: "cinax-openai-compact",
+ wantResolved: "cinax-openai-compact",
+ wantApplied: false,
+ },
{
name: "empty requested model returns as-is",
mapping: map[string]string{"cinax": "cinax-pro"},
diff --git a/setting/operation_setting/error_override_setting.go b/setting/operation_setting/error_override_setting.go
new file mode 100644
index 000000000000..60e41185e8c4
--- /dev/null
+++ b/setting/operation_setting/error_override_setting.go
@@ -0,0 +1,75 @@
+package operation_setting
+
+import (
+ "strings"
+
+ "github.com/QuantumNous/new-api/relaykit/types"
+)
+
+// ErrorOverrideMessage 覆写后统一返回给客户端的错误文案
+const ErrorOverrideMessage = "Service Unavailable"
+
+// 错误信息覆写：命中关键词的上游错误统一替换文案，避免向客户端暴露上游账务状态与转售链路。
+// 仅作用于上游传递到本站的错误；本站自身产生的错误（用户额度不足、无可用渠道等）不受影响。
+var ErrorOverrideEnabled = false
+
+// 关键词统一以小写存储，匹配时对错误文案做小写化后比对
+var ErrorOverrideKeywords = []string{"no available", "quota", "credits", "top-up"}
+
+func ErrorOverrideKeywordsToString() string {
+ return strings.Join(ErrorOverrideKeywords, "\n")
+}
+
+func ErrorOverrideKeywordsFromString(s string) {
+ ErrorOverrideKeywords = []string{}
+ ak := strings.Split(s, "\n")
+ for _, k := range ak {
+ k = strings.TrimSpace(k)
+ k = strings.ToLower(k)
+ if k != "" {
+ ErrorOverrideKeywords = append(ErrorOverrideKeywords, k)
+ }
+ }
+}
+
+// ShouldOverrideUpstreamError 报告该错误写给客户端时是否会被覆写，但不修改错误本身。
+// 供需要在覆写前分叉行为的调用方使用（例如错误日志要区分记录原文还是覆写后的文案）。
+func ShouldOverrideUpstreamError(err *types.NewAPIError) bool {
+ if !types.IsFromUpstreamError(err) {
+ return false
+ }
+ return shouldOverrideErrorMessage(err.Error())
+}
+
+// OverrideUpstreamError 在错误写给客户端前覆写其文案，返回是否已覆写。必须在错误日志记录
+// 之后调用，后台日志与渠道自动禁用判定始终使用原始上游文案。状态码、error.type、error.code
+// 保持不变。
+func OverrideUpstreamError(err *types.NewAPIError) bool {
+ if !ShouldOverrideUpstreamError(err) {
+ return false
+ }
+ err.ReplaceMessage(ErrorOverrideMessage)
+ return true
+}
+
+// OverrideUpstreamMessage 供不经过 NewAPIError 的链路使用（任务平台、Midjourney）。
+// 调用方需自行确认文案来自上游。
+func OverrideUpstreamMessage(message string) string {
+ if !shouldOverrideErrorMessage(message) {
+ return message
+ }
+ return ErrorOverrideMessage
+}
+
+func shouldOverrideErrorMessage(message string) bool {
+ if !ErrorOverrideEnabled || message == "" {
+ return false
+ }
+ lowerMessage := strings.ToLower(message)
+ for _, keyword := range ErrorOverrideKeywords {
+ if strings.Contains(lowerMessage, keyword) {
+ return true
+ }
+ }
+ return false
+}
diff --git a/setting/operation_setting/error_override_setting_test.go b/setting/operation_setting/error_override_setting_test.go
new file mode 100644
index 000000000000..de0ceaf8e475
--- /dev/null
+++ b/setting/operation_setting/error_override_setting_test.go
@@ -0,0 +1,240 @@
+package operation_setting
+
+import (
+ "errors"
+ "net/http"
+ "testing"
+
+ "github.com/stretchr/testify/assert"
+ "github.com/stretchr/testify/require"
+
+ "github.com/QuantumNous/new-api/relaykit/types"
+)
+
+// restoreOverrideState resets the package-level override settings to the defaults used
+// by each test and reinstalls them after the test runs, so tests stay isolated.
+func restoreOverrideState(t *testing.T) {
+ t.Helper()
+ origEnabled := ErrorOverrideEnabled
+ origKeywords := ErrorOverrideKeywords
+ t.Cleanup(func() {
+ ErrorOverrideEnabled = origEnabled
+ ErrorOverrideKeywords = origKeywords
+ })
+ ErrorOverrideEnabled = true
+ ErrorOverrideKeywords = []string{"no available", "quota", "credits", "top-up"}
+}
+
+func TestOverrideUpstreamMessage_DisabledKeepsMessage(t *testing.T) {
+ orig := ErrorOverrideEnabled
+ t.Cleanup(func() { ErrorOverrideEnabled = orig })
+ ErrorOverrideEnabled = false
+
+ msg := "Your credit balance is too low, please top up"
+ assert.Equal(t, msg, OverrideUpstreamMessage(msg))
+ // Empty message is never overridden even when enabled.
+ assert.Equal(t, "", OverrideUpstreamMessage(""))
+}
+
+func TestOverrideUpstreamMessage_MatchesKeywords(t *testing.T) {
+ restoreOverrideState(t)
+
+ cases := []struct {
+ name string
+ message string
+ want string
+ }{
+ {"credits", "insufficient credits", ErrorOverrideMessage},
+ {"top-up", "Please top-up your account", ErrorOverrideMessage},
+ {"quota", "You exceeded your current quota", ErrorOverrideMessage},
+ {"no available", "No available channel for model gpt-4o under group default", ErrorOverrideMessage},
+ {"case-insensitive credits", "CREDITS exhausted", ErrorOverrideMessage},
+ {"case-insensitive quota", "QUOTA exceeded", ErrorOverrideMessage},
+ {"no match", "Internal Server Error", "Internal Server Error"},
+ {"empty", "", ""},
+ }
+ for _, tc := range cases {
+ t.Run(tc.name, func(t *testing.T) {
+ assert.Equal(t, tc.want, OverrideUpstreamMessage(tc.message))
+ })
+ }
+}
+
+func TestOverrideUpstreamError_LocalErrorsNotOverridden(t *testing.T) {
+ restoreOverrideState(t)
+
+ // 本地错误：用户额度不足（由本站产生，未标记上游来源）。
+ localQuota := types.NewErrorWithStatusCode(
+ errors.New("用户额度不足, 剩余额度: $0.00"),
+ types.ErrorCodeInsufficientUserQuota, http.StatusPaymentRequired)
+ assert.False(t, OverrideUpstreamError(localQuota))
+ assert.Equal(t, "用户额度不足, 剩余额度: $0.00", localQuota.Error())
+
+ // 本地错误：distributor 的「无可用渠道」文案虽然命中 "no available" 关键词，
+ // 但因未标记上游来源，不应被覆写。
+ localNoChannel := types.NewErrorWithStatusCode(
+ errors.New("No available channel for model gpt-4o under group default (distributor)"),
+ types.ErrorCodeBadResponseStatusCode, http.StatusServiceUnavailable)
+ assert.False(t, OverrideUpstreamError(localNoChannel))
+ assert.Contains(t, localNoChannel.Error(), "No available channel")
+}
+
+func TestOverrideUpstreamError_UpstreamErrorsOverridden(t *testing.T) {
+ restoreOverrideState(t)
+
+ cases := []struct {
+ name string
+ build func() *types.NewAPIError
+ message string
+ }{
+ {
+ name: "openai credits",
+ build: func() *types.NewAPIError {
+ return types.WithUpstreamOpenAIError(types.OpenAIError{
+ Message: "insufficient credits balance, please top up",
+ Type: "upstream_error",
+ Code: "credits_exhausted",
+ }, http.StatusPaymentRequired)
+ },
+ message: "insufficient credits balance, please top up",
+ },
+ {
+ name: "claude quota",
+ build: func() *types.NewAPIError {
+ return types.WithUpstreamClaudeError(types.ClaudeError{
+ Type: "error",
+ Message: "You exceeded your current quota",
+ }, http.StatusTooManyRequests)
+ },
+ message: "You exceeded your current quota",
+ },
+ {
+ name: "openai no available",
+ build: func() *types.NewAPIError {
+ return types.WithUpstreamOpenAIError(types.OpenAIError{
+ Message: "No available channel for model gpt-4o under group default",
+ Type: "upstream_error",
+ Code: "no_available_channel",
+ }, http.StatusServiceUnavailable)
+ },
+ message: "No available channel for model gpt-4o under group default",
+ },
+ }
+ for _, tc := range cases {
+ t.Run(tc.name, func(t *testing.T) {
+ err := tc.build()
+ require.True(t, types.IsFromUpstreamError(err))
+ require.True(t, OverrideUpstreamError(err))
+
+ // Every user-facing representation is rewritten.
+ assert.Equal(t, ErrorOverrideMessage, err.Error())
+ openAI := err.ToOpenAIError()
+ assert.Equal(t, ErrorOverrideMessage, openAI.Message)
+ assert.Nil(t, openAI.Metadata)
+ })
+ }
+}
+
+func TestOverrideUpstreamError_PreservesStatusCodeTypeCode(t *testing.T) {
+ restoreOverrideState(t)
+
+ err := types.WithUpstreamOpenAIError(types.OpenAIError{
+ Message: "Please top up your credits",
+ Type: "upstream_error",
+ Code: "credits_exhausted",
+ }, http.StatusPaymentRequired)
+ originalCode := err.GetErrorCode()
+ originalType := err.GetErrorType()
+
+ require.True(t, OverrideUpstreamError(err))
+
+ assert.Equal(t, http.StatusPaymentRequired, err.StatusCode)
+ assert.Equal(t, originalType, err.GetErrorType())
+ assert.Equal(t, originalCode, err.GetErrorCode())
+}
+
+func TestOverrideUpstreamError_DisabledDoesNothing(t *testing.T) {
+ orig := ErrorOverrideEnabled
+ t.Cleanup(func() { ErrorOverrideEnabled = orig })
+ ErrorOverrideEnabled = false
+
+ err := types.WithUpstreamOpenAIError(types.OpenAIError{
+ Message: "quota exhausted",
+ Type: "upstream_error",
+ }, http.StatusTooManyRequests)
+ assert.False(t, OverrideUpstreamError(err))
+ assert.Equal(t, "quota exhausted", err.Error())
+}
+
+func TestOverrideUpstreamError_NilSafe(t *testing.T) {
+ restoreOverrideState(t)
+ require.NotPanics(t, func() { OverrideUpstreamError(nil) })
+}
+
+// ShouldOverrideUpstreamError answers the same question as OverrideUpstreamError but
+// must not mutate the error. processChannelError relies on that to decide what to write
+// into the user-visible error log while the untouched error still drives retry and
+// channel-disable decisions.
+func TestShouldOverrideUpstreamError_DoesNotMutate(t *testing.T) {
+ restoreOverrideState(t)
+
+ upstream := types.WithUpstreamOpenAIError(types.OpenAIError{
+ Message: "insufficient credits, please top up",
+ Type: "upstream_error",
+ Code: "credits_exhausted",
+ }, http.StatusPaymentRequired)
+
+ require.True(t, ShouldOverrideUpstreamError(upstream))
+ // Asking twice must stay stable, and the message must survive untouched.
+ require.True(t, ShouldOverrideUpstreamError(upstream))
+ assert.Equal(t, "insufficient credits, please top up", upstream.Error())
+ assert.Equal(t, "insufficient credits, please top up", upstream.ToOpenAIError().Message)
+
+ local := types.NewErrorWithStatusCode(
+ errors.New("用户额度不足, 剩余额度: $0.00"),
+ types.ErrorCodeInsufficientUserQuota, http.StatusPaymentRequired)
+ assert.False(t, ShouldOverrideUpstreamError(local))
+ require.NotPanics(t, func() { ShouldOverrideUpstreamError(nil) })
+ assert.False(t, ShouldOverrideUpstreamError(nil))
+}
+
+func TestOverrideUpstreamMessage_NoKeywords(t *testing.T) {
+ origEnabled := ErrorOverrideEnabled
+ origKeywords := ErrorOverrideKeywords
+ t.Cleanup(func() {
+ ErrorOverrideEnabled = origEnabled
+ ErrorOverrideKeywords = origKeywords
+ })
+ ErrorOverrideEnabled = true
+ ErrorOverrideKeywords = nil
+
+ msg := "Your credits are too low"
+ assert.Equal(t, msg, OverrideUpstreamMessage(msg))
+}
+
+func TestErrorOverrideKeywordsFromString_ParsesAndNormalizes(t *testing.T) {
+ orig := ErrorOverrideKeywords
+ t.Cleanup(func() { ErrorOverrideKeywords = orig })
+
+ ErrorOverrideKeywordsFromString("Credits\n Quota \n\n\ntop-up\n ")
+
+ require.Equal(t, []string{"credits", "quota", "top-up"}, ErrorOverrideKeywords)
+}
+
+func TestErrorOverrideKeywordsFromString_EmptyStringClears(t *testing.T) {
+ orig := ErrorOverrideKeywords
+ t.Cleanup(func() { ErrorOverrideKeywords = orig })
+
+ ErrorOverrideKeywordsFromString("")
+ require.Empty(t, ErrorOverrideKeywords)
+}
+
+func TestErrorOverrideKeywordsToString_RoundTrip(t *testing.T) {
+ orig := ErrorOverrideKeywords
+ t.Cleanup(func() { ErrorOverrideKeywords = orig })
+
+ ErrorOverrideKeywords = []string{"no available", "quota", "credits", "top-up"}
+ s := ErrorOverrideKeywordsToString()
+ ErrorOverrideKeywordsFromString(s)
+ require.Equal(t, []string{"no available", "quota", "credits", "top-up"}, ErrorOverrideKeywords)
+}
diff --git a/web/src/features/keys/components/__tests__/api-key-group-cell.test.tsx b/web/src/features/keys/components/__tests__/api-key-group-cell.test.tsx
index 5cb64ae57f7b..6da76150afc0 100644
--- a/web/src/features/keys/components/__tests__/api-key-group-cell.test.tsx
+++ b/web/src/features/keys/components/__tests__/api-key-group-cell.test.tsx
@@ -100,7 +100,7 @@ describe('API key group table cell', () => {
domWindow.close()
})
- test('renders two unclipped rings and a localized Auto ratio when API data uses a nonlocalized string', async () => {
+ test('renders an unclipped ring and a localized Auto ratio when API data uses a nonlocalized string', async () => {
const container = document.createElement('div')
document.body.append(container)
const root = createRoot(container)
@@ -123,12 +123,14 @@ describe('API key group table cell', () => {
assert.equal(badgeCell.classList.contains('overflow-visible'), true)
assert.equal(badgeCell.classList.contains('overflow-hidden'), false)
+ // AutoGroupBadge 在 ApiKeyGroupCell 中被停用（由 Cross-group StatusBadge 承担该位置），
+ // 因此 auto 分组只剩 GroupRatioBadge 一个带流光边框的 frame。
const frames = container.querySelectorAll('[data-auto-group-frame]')
const movingRings = container.querySelectorAll(
'[data-auto-group-flow-border]'
)
- assert.equal(frames.length, 2)
- assert.equal(movingRings.length, 2)
+ assert.equal(frames.length, 1)
+ assert.equal(movingRings.length, 1)
for (const frame of frames) {
assert.equal(frame.classList.contains('relative'), true)
assert.equal(frame.classList.contains('overflow-visible'), true)
@@ -155,7 +157,7 @@ describe('API key group table cell', () => {
container.remove()
})
- test('keeps static Auto frames but omits both moving layers for reduced motion', async () => {
+ test('keeps the static Auto frame but omits the moving layer for reduced motion', async () => {
const container = document.createElement('div')
document.body.append(container)
const root = createRoot(container)
@@ -166,7 +168,7 @@ describe('API key group table cell', () => {
assert.equal(
container.querySelectorAll('[data-auto-group-frame]').length,
- 2
+ 1
)
assert.equal(
container.querySelectorAll('[data-auto-group-flow-border]').length,
@@ -177,7 +179,7 @@ describe('API key group table cell', () => {
container.remove()
})
- test('shows only the Auto badge when ratio data is unavailable', async () => {
+ test('shows only the Cross-group badge when ratio data is unavailable', async () => {
const container = document.createElement('div')
document.body.append(container)
const root = createRoot(container)
@@ -186,19 +188,21 @@ describe('API key group table cell', () => {
root.render()
)
+ // 无 ratio 时 GroupRatioBadge 返回 null，auto 分组不渲染任何 frame，
+ // 只保留 Cross-group 状态徽章。
assert.equal(
container.querySelectorAll('[data-auto-group-frame]').length,
- 1
+ 0
)
assert.equal(
container.querySelectorAll('[data-auto-group-flow-border]').length,
- 1
+ 0
)
assert.equal(
container.querySelector('[data-auto-group-effect="ratio"]'),
null
)
- assert.equal(container.textContent?.includes('Auto'), true)
+ assert.equal(container.textContent?.includes('Cross-group'), true)
assert.equal(container.textContent?.includes('Ratio'), false)
await act(async () => root.unmount())
diff --git a/web/src/features/system-settings/general/system-behavior-section.tsx b/web/src/features/system-settings/general/system-behavior-section.tsx
index 4413a2a02b1b..f59f2a100e30 100644
--- a/web/src/features/system-settings/general/system-behavior-section.tsx
+++ b/web/src/features/system-settings/general/system-behavior-section.tsx
@@ -26,9 +26,12 @@ import {
FormControl,
FormDescription,
FormField,
+ FormItem,
FormLabel,
+ FormMessage,
} from '@/components/ui/form'
import { Switch } from '@/components/ui/switch'
+import { Textarea } from '@/components/ui/textarea'
import {
SettingsForm,
@@ -44,6 +47,8 @@ const behaviorSchema = z.object({
DefaultCollapseSidebar: z.boolean(),
DemoSiteEnabled: z.boolean(),
SelfUseModeEnabled: z.boolean(),
+ ErrorOverrideEnabled: z.boolean(),
+ ErrorOverrideKeywords: z.string(),
ChannelFailoverEnabled: z.boolean(),
})
@@ -147,6 +152,54 @@ export function SystemBehaviorSection({
)}
/>
+ (
+
+
+ {t('Error Message Override')}
+
+ {t(
+ 'Replace upstream error messages that match a keyword with a generic message, so upstream billing state is not exposed. Errors produced by this site are never replaced.'
+ )}
+
+
+
+
+
+
+ )}
+ />
+
+ {form.watch('ErrorOverrideEnabled') && (
+ (
+
+ {t('Override keywords')}
+
+
+
+
+ {t(
+ 'One keyword per line, case-insensitive. Matching upstream errors return Service Unavailable with the original status code preserved.'
+ )}
+
+
+
+ )}
+ />
+ )}
+
diff --git a/web/src/features/system-settings/types.ts b/web/src/features/system-settings/types.ts
index 927130266db2..33256ae9f060 100644
--- a/web/src/features/system-settings/types.ts
+++ b/web/src/features/system-settings/types.ts
@@ -341,6 +341,8 @@ export type OperationsSettings = {
DefaultCollapseSidebar: boolean
DemoSiteEnabled: boolean
SelfUseModeEnabled: boolean
+ ErrorOverrideEnabled: boolean
+ ErrorOverrideKeywords: string
ChannelFailoverEnabled: boolean
QuotaRemindThreshold: string
SMTPServer: string
diff --git a/web/src/i18n/locales/en.json b/web/src/i18n/locales/en.json
index 24b77adf0485..871e72319660 100644
--- a/web/src/i18n/locales/en.json
+++ b/web/src/i18n/locales/en.json
@@ -1744,6 +1744,7 @@
"Error establishing connection": "Error establishing connection",
"Error Message": "Error Message",
"Error Message (required)": "Error Message (required)",
+ "Error Message Override": "Error Message Override",
"Error parsing response data": "Error parsing response data",
"Error Type (optional)": "Error Type (optional)",
"Estimated cost": "Estimated cost",
@@ -3129,6 +3130,7 @@
"One domain per line (only used when domain restriction is enabled)": "One domain per line (only used when domain restriction is enabled)",
"One IP or CIDR range per line": "One IP or CIDR range per line",
"One IP per line (empty for no restriction)": "One IP per line (empty for no restriction)",
+ "One keyword per line, case-insensitive. Matching upstream errors return Service Unavailable with the original status code preserved.": "One keyword per line, case-insensitive. Matching upstream errors return Service Unavailable with the original status code preserved.",
"one keyword per line": "one keyword per line",
"online": "online",
"Online": "Online",
@@ -3238,6 +3240,7 @@
"override": "override",
"Override": "Override",
"Override auto-discovered endpoint": "Override auto-discovered endpoint",
+ "Override keywords": "Override keywords",
"Override matrix": "Override matrix",
"Override request headers": "Override request headers",
"Override request headers (JSON format)": "Override request headers (JSON format)",
@@ -3826,6 +3829,7 @@
"Replace all existing keys": "Replace all existing keys",
"Replace channel models": "Replace channel models",
"Replace mode: Will completely replace all existing keys": "Replace mode: Will completely replace all existing keys",
+ "Replace upstream error messages that match a keyword with a generic message, so upstream billing state is not exposed. Errors produced by this site are never replaced.": "Replace upstream error messages that match a keyword with a generic message, so upstream billing state is not exposed. Errors produced by this site are never replaced.",
"Replace With": "Replace With",
"replaced": "replaced",
"Replacement Model": "Replacement Model",
diff --git a/web/src/i18n/locales/fr.json b/web/src/i18n/locales/fr.json
index 1bf520865c2c..4a812d2cf2d7 100644
--- a/web/src/i18n/locales/fr.json
+++ b/web/src/i18n/locales/fr.json
@@ -1744,6 +1744,7 @@
"Error establishing connection": "Erreur lors de l’établissement de la connexion",
"Error Message": "Message d'erreur",
"Error Message (required)": "Message d'erreur (requis)",
+ "Error Message Override": "Remplacement des messages d'erreur",
"Error parsing response data": "Erreur lors de l’analyse des données de réponse",
"Error Type (optional)": "Type d'erreur (optionnel)",
"Estimated cost": "Coût estimé",
@@ -3129,6 +3130,7 @@
"One domain per line (only used when domain restriction is enabled)": "Un domaine par ligne (utilisé uniquement lorsque la restriction de domaine est activée)",
"One IP or CIDR range per line": "Une IP ou plage CIDR par ligne",
"One IP per line (empty for no restriction)": "Une IP par ligne (laisser vide pour aucune restriction)",
+ "One keyword per line, case-insensitive. Matching upstream errors return Service Unavailable with the original status code preserved.": "Un mot-clé par ligne, insensible à la casse. Les erreurs en amont correspondantes renvoient Service Unavailable en conservant le code de statut d'origine.",
"one keyword per line": "un mot-clé par ligne",
"online": "en ligne",
"Online": "En ligne",
@@ -3238,6 +3240,7 @@
"override": "remplacer",
"Override": "Remplacer",
"Override auto-discovered endpoint": "Remplacer le point de terminaison auto-découvert",
+ "Override keywords": "Mots-clés de remplacement",
"Override matrix": "Matrice de remplacement",
"Override request headers": "Remplacer les en-têtes de requête",
"Override request headers (JSON format)": "Surcharge des en-têtes de requête (format JSON)",
@@ -3826,6 +3829,7 @@
"Replace all existing keys": "Remplacer toutes les clés existantes",
"Replace channel models": "Remplacer les modèles du canal",
"Replace mode: Will completely replace all existing keys": "Mode remplacement : Remplacera complètement toutes les clés existantes",
+ "Replace upstream error messages that match a keyword with a generic message, so upstream billing state is not exposed. Errors produced by this site are never replaced.": "Remplace les messages d'erreur en amont correspondant à un mot-clé par un message générique, afin de ne pas exposer l'état de facturation en amont. Les erreurs générées par ce site ne sont jamais remplacées.",
"Replace With": "Remplacer par",
"replaced": "remplacé",
"Replacement Model": "Modèle de remplacement",
diff --git a/web/src/i18n/locales/ja.json b/web/src/i18n/locales/ja.json
index 668887e38155..0da1bd574a04 100644
--- a/web/src/i18n/locales/ja.json
+++ b/web/src/i18n/locales/ja.json
@@ -1744,6 +1744,7 @@
"Error establishing connection": "接続の確立に失敗しました",
"Error Message": "エラーメッセージ",
"Error Message (required)": "エラーメッセージ（必須）",
+ "Error Message Override": "エラーメッセージの上書き",
"Error parsing response data": "レスポンスデータの解析に失敗しました",
"Error Type (optional)": "エラータイプ（任意）",
"Estimated cost": "推定コスト",
@@ -3129,6 +3130,7 @@
"One domain per line (only used when domain restriction is enabled)": "1行に1つのドメイン (ドメイン制限が有効な場合のみ使用されます)",
"One IP or CIDR range per line": "1行に1つのIPまたはCIDR範囲",
"One IP per line (empty for no restriction)": "1行に1つのIP (制限なしの場合は空欄)",
+ "One keyword per line, case-insensitive. Matching upstream errors return Service Unavailable with the original status code preserved.": "1行に1つのキーワード、大文字と小文字を区別しません。一致した上流エラーは、元のステータスコードを保持したまま Service Unavailable を返します。",
"one keyword per line": "1行に1つのキーワード",
"online": "オンライン",
"Online": "オンライン",
@@ -3238,6 +3240,7 @@
"override": "上書き",
"Override": "上書き",
"Override auto-discovered endpoint": "自動検出されたエンドポイントを上書きする",
+ "Override keywords": "上書きキーワード",
"Override matrix": "上書きマトリクス",
"Override request headers": "リクエストヘッダーを上書きする",
"Override request headers (JSON format)": "リクエストヘッダーのオーバーライド (JSON 形式)",
@@ -3826,6 +3829,7 @@
"Replace all existing keys": "既存のすべてのキーを置き換える",
"Replace channel models": "チャネルモデルを置き換える",
"Replace mode: Will completely replace all existing keys": "置換モード: 既存のすべてのキーを完全に置き換えます",
+ "Replace upstream error messages that match a keyword with a generic message, so upstream billing state is not exposed. Errors produced by this site are never replaced.": "キーワードに一致する上流のエラーメッセージを汎用メッセージに置き換え、上流の課金状態が漏れないようにします。本サイトが生成したエラーは置き換えられません。",
"Replace With": "置換後",
"replaced": "置換済み",
"Replacement Model": "代替モデル",
diff --git a/web/src/i18n/locales/ru.json b/web/src/i18n/locales/ru.json
index 16e30d17b0ec..f976ab5940e0 100644
--- a/web/src/i18n/locales/ru.json
+++ b/web/src/i18n/locales/ru.json
@@ -1744,6 +1744,7 @@
"Error establishing connection": "Ошибка при установлении соединения",
"Error Message": "Сообщение об ошибке",
"Error Message (required)": "Сообщение об ошибке (обязательно)",
+ "Error Message Override": "Переопределение сообщений об ошибках",
"Error parsing response data": "Ошибка при разборе данных ответа",
"Error Type (optional)": "Тип ошибки (необязательно)",
"Estimated cost": "Примерная стоимость",
@@ -3129,6 +3130,7 @@
"One domain per line (only used when domain restriction is enabled)": "Один домен на строку (используется только при включении ограничения домена)",
"One IP or CIDR range per line": "Один IP или диапазон CIDR на строку",
"One IP per line (empty for no restriction)": "Один IP на строку (пусто для отсутствия ограничений)",
+ "One keyword per line, case-insensitive. Matching upstream errors return Service Unavailable with the original status code preserved.": "По одному ключевому слову на строку, без учёта регистра. Совпавшие ошибки вышестоящего провайдера возвращают Service Unavailable с сохранением исходного кода состояния.",
"one keyword per line": "одно ключевое слово на строку",
"online": "онлайн",
"Online": "Онлайн",
@@ -3238,6 +3240,7 @@
"override": "переопределить",
"Override": "Перезаписать",
"Override auto-discovered endpoint": "Переопределить автоматически обнаруженную конечную точку",
+ "Override keywords": "Ключевые слова переопределения",
"Override matrix": "Матрица переопределений",
"Override request headers": "Переопределить заголовки запроса",
"Override request headers (JSON format)": "Переопределение заголовков запроса (формат JSON)",
@@ -3826,6 +3829,7 @@
"Replace all existing keys": "Заменить все существующие ключи",
"Replace channel models": "Замена моделей каналов",
"Replace mode: Will completely replace all existing keys": "Режим замены: полностью заменит все существующие ключи",
+ "Replace upstream error messages that match a keyword with a generic message, so upstream billing state is not exposed. Errors produced by this site are never replaced.": "Сообщения об ошибках вышестоящего провайдера, совпадающие с ключевым словом, заменяются общим сообщением, чтобы не раскрывать состояние его биллинга. Ошибки, созданные этим сайтом, никогда не заменяются.",
"Replace With": "Заменить на",
"replaced": "заменено",
"Replacement Model": "Модель замены",
diff --git a/web/src/i18n/locales/vi.json b/web/src/i18n/locales/vi.json
index 633f21c24f7c..0e831e0a9447 100644
--- a/web/src/i18n/locales/vi.json
+++ b/web/src/i18n/locales/vi.json
@@ -1744,6 +1744,7 @@
"Error establishing connection": "Lỗi khi thiết lập kết nối",
"Error Message": "Thông báo lỗi",
"Error Message (required)": "Thông báo lỗi (bắt buộc)",
+ "Error Message Override": "Ghi đè thông báo lỗi",
"Error parsing response data": "Lỗi khi phân tích dữ liệu phản hồi",
"Error Type (optional)": "Loại lỗi (tùy chọn)",
"Estimated cost": "Chi phí ước tính",
@@ -3129,6 +3130,7 @@
"One domain per line (only used when domain restriction is enabled)": "Mỗi dòng một tên miền (chỉ được sử dụng khi hạn chế tên miền được bật)",
"One IP or CIDR range per line": "Một IP hoặc dải CIDR mỗi dòng",
"One IP per line (empty for no restriction)": "Mỗi IP một dòng (để trống nếu không giới hạn)",
+ "One keyword per line, case-insensitive. Matching upstream errors return Service Unavailable with the original status code preserved.": "Mỗi dòng một từ khóa, không phân biệt chữ hoa chữ thường. Lỗi thượng nguồn khớp sẽ trả về Service Unavailable và giữ nguyên mã trạng thái gốc.",
"one keyword per line": "Mỗi dòng một từ khóa",
"online": "trực tuyến",
"Online": "Trực tuyến",
@@ -3238,6 +3240,7 @@
"override": "ghi đè",
"Override": "Ghi đè",
"Override auto-discovered endpoint": "Ghi đè điểm cuối tự động phát hiện",
+ "Override keywords": "Từ khóa ghi đè",
"Override matrix": "Ma trận ghi đè",
"Override request headers": "Ghi đè tiêu đề yêu cầu",
"Override request headers (JSON format)": "Ghi đè tiêu đề yêu cầu (định dạng JSON)",
@@ -3826,6 +3829,7 @@
"Replace all existing keys": "Thay thế tất cả các khóa hiện có",
"Replace channel models": "Thay thế mô hình kênh",
"Replace mode: Will completely replace all existing keys": "Chế độ Thay thế: Sẽ thay thế hoàn toàn tất cả các khóa hiện có",
+ "Replace upstream error messages that match a keyword with a generic message, so upstream billing state is not exposed. Errors produced by this site are never replaced.": "Thay thế thông báo lỗi từ nhà cung cấp thượng nguồn khớp với từ khóa bằng một thông báo chung, để không làm lộ trạng thái thanh toán của thượng nguồn. Lỗi do trang này tạo ra sẽ không bao giờ bị thay thế.",
"Replace With": "Thay bằng",
"replaced": "thay thế",
"Replacement Model": "Replacement model",
diff --git a/web/src/i18n/locales/zh-TW.json b/web/src/i18n/locales/zh-TW.json
index d68bb320c9c0..66405dbe7656 100644
--- a/web/src/i18n/locales/zh-TW.json
+++ b/web/src/i18n/locales/zh-TW.json
@@ -1744,6 +1744,7 @@
"Error establishing connection": "建立連接失敗",
"Error Message": "錯誤訊息",
"Error Message (required)": "錯誤訊息（必填）",
+ "Error Message Override": "錯誤訊息覆寫",
"Error parsing response data": "解析回應數據失敗",
"Error Type (optional)": "錯誤類型（可選）",
"Estimated cost": "預計成本",
@@ -3129,6 +3130,7 @@
"One domain per line (only used when domain restriction is enabled)": "每行一個域名 (僅在啟用域名限制時使用)",
"One IP or CIDR range per line": "每行一個 IP 或 CIDR 範圍",
"One IP per line (empty for no restriction)": "每行一個 IP (留空表示無限制)",
+ "One keyword per line, case-insensitive. Matching upstream errors return Service Unavailable with the original status code preserved.": "每行一個關鍵詞，不區分大小寫。命中的上游錯誤回傳 Service Unavailable，HTTP 狀態碼保持原值。",
"one keyword per line": "每行一個關鍵詞",
"online": "在線",
"Online": "在線",
@@ -3238,6 +3240,7 @@
"override": "覆蓋",
"Override": "覆蓋",
"Override auto-discovered endpoint": "覆蓋自動發現的端點",
+ "Override keywords": "覆寫關鍵詞",
"Override matrix": "覆蓋矩陣",
"Override request headers": "覆蓋請求標頭",
"Override request headers (JSON format)": "覆蓋請求頭（JSON 格式）",
@@ -3826,6 +3829,7 @@
"Replace all existing keys": "替換所有現有金鑰",
"Replace channel models": "覆蓋渠道模型",
"Replace mode: Will completely replace all existing keys": "替換模式：將完全替換所有現有鍵",
+ "Replace upstream error messages that match a keyword with a generic message, so upstream billing state is not exposed. Errors produced by this site are never replaced.": "命中關鍵詞的上游錯誤訊息將被替換為通用提示，避免暴露上游帳務狀態。本站自身產生的錯誤不會被覆寫。",
"Replace With": "替換為",
"replaced": "已替換",
"Replacement Model": "替換模型",
diff --git a/web/src/i18n/locales/zh.json b/web/src/i18n/locales/zh.json
index 57c28403608d..97ef5f91d12f 100644
--- a/web/src/i18n/locales/zh.json
+++ b/web/src/i18n/locales/zh.json
@@ -1744,6 +1744,7 @@
"Error establishing connection": "建立连接失败",
"Error Message": "错误消息",
"Error Message (required)": "错误消息（必填）",
+ "Error Message Override": "错误信息覆写",
"Error parsing response data": "解析响应数据失败",
"Error Type (optional)": "错误类型（可选）",
"Estimated cost": "预计成本",
@@ -3129,6 +3130,7 @@
"One domain per line (only used when domain restriction is enabled)": "每行一个域名 (仅在启用域名限制时使用)",
"One IP or CIDR range per line": "每行一个 IP 或 CIDR 范围",
"One IP per line (empty for no restriction)": "每行一个 IP (留空表示无限制)",
+ "One keyword per line, case-insensitive. Matching upstream errors return Service Unavailable with the original status code preserved.": "每行一个关键词，不区分大小写。命中的上游错误返回 Service Unavailable，HTTP 状态码保持原值。",
"one keyword per line": "每行一个关键词",
"online": "在线",
"Online": "在线",
@@ -3238,6 +3240,7 @@
"override": "覆盖",
"Override": "覆盖",
"Override auto-discovered endpoint": "覆盖自动发现的端点",
+ "Override keywords": "覆写关键词",
"Override matrix": "覆盖矩阵",
"Override request headers": "覆盖请求标头",
"Override request headers (JSON format)": "覆盖请求头（JSON 格式）",
@@ -3826,6 +3829,7 @@
"Replace all existing keys": "替换所有现有密钥",
"Replace channel models": "覆盖渠道模型",
"Replace mode: Will completely replace all existing keys": "替换模式：将完全替换所有现有键",
+ "Replace upstream error messages that match a keyword with a generic message, so upstream billing state is not exposed. Errors produced by this site are never replaced.": "命中关键词的上游错误信息将被替换为通用提示，避免暴露上游账务状态。本站自身产生的错误不会被覆写。",
"Replace With": "替换为",
"replaced": "已替换",
"Replacement Model": "替换模型",