fix: fix token create

统一文心计费模型名称

common/conv/any.go

@@ -0,0 +1,6 @@
+package conv
+
+func AsString(v any) string {
+	str, _ := v.(string)
+	return str
+}

common/model-ratio.go

@@ -75,7 +75,7 @@ var ModelRatio = map[string]float64{
 	"ERNIE-Bot":       0.8572,     // ￥0.012 / 1k tokens
 	"ERNIE-Bot-turbo": 0.5715,     // ￥0.008 / 1k tokens
 	"ERNIE-Bot-4":     0.12 * RMB, // ￥0.12 / 1k tokens
-	"ERNIE-Bot-8k":    0.024 * RMB,
+	"ERNIE-Bot-8K":    0.024 * RMB,
 	"Embedding-V1":    0.1429, // ￥0.002 / 1k tokens
 	"bge-large-zh":    0.002 * RMB,
 	"bge-large-en":    0.002 * RMB,

controller/model.go

@@ -4,12 +4,14 @@ import (
 	"fmt"
 	"github.com/gin-gonic/gin"
 	"github.com/songquanpeng/one-api/common"
+	"github.com/songquanpeng/one-api/model"
 	"github.com/songquanpeng/one-api/relay/channel/openai"
 	"github.com/songquanpeng/one-api/relay/constant"
 	"github.com/songquanpeng/one-api/relay/helper"
 	relaymodel "github.com/songquanpeng/one-api/relay/model"
 	"github.com/songquanpeng/one-api/relay/util"
 	"net/http"
+	"strings"
 )
 
 // https://platform.openai.com/docs/api-reference/models/list
@@ -120,9 +122,41 @@ func DashboardListModels(c *gin.Context) {
 }
 
 func ListModels(c *gin.Context) {
+	ctx := c.Request.Context()
+	var availableModels []string
+	if c.GetString("available_models") != "" {
+		availableModels = strings.Split(c.GetString("available_models"), ",")
+	} else {
+		userId := c.GetInt("id")
+		userGroup, _ := model.CacheGetUserGroup(userId)
+		availableModels, _ = model.CacheGetGroupModels(ctx, userGroup)
+	}
+	modelSet := make(map[string]bool)
+	for _, availableModel := range availableModels {
+		modelSet[availableModel] = true
+	}
+	var availableOpenAIModels []OpenAIModels
+	for _, model := range openAIModels {
+		if _, ok := modelSet[model.Id]; ok {
+			modelSet[model.Id] = false
+			availableOpenAIModels = append(availableOpenAIModels, model)
+		}
+	}
+	for modelName, ok := range modelSet {
+		if ok {
+			availableOpenAIModels = append(availableOpenAIModels, OpenAIModels{
+				Id:      modelName,
+				Object:  "model",
+				Created: 1626777600,
+				OwnedBy: "custom",
+				Root:    modelName,
+				Parent:  nil,
+			})
+		}
+	}
 	c.JSON(200, gin.H{
 		"object": "list",
-		"data":   openAIModels,
+		"data":   availableOpenAIModels,
 	})
 }
 
@@ -142,3 +176,30 @@ func RetrieveModel(c *gin.Context) {
 		})
 	}
 }
+
+func GetUserAvailableModels(c *gin.Context) {
+	ctx := c.Request.Context()
+	id := c.GetInt("id")
+	userGroup, err := model.CacheGetUserGroup(id)
+	if err != nil {
+		c.JSON(http.StatusOK, gin.H{
+			"success": false,
+			"message": err.Error(),
+		})
+		return
+	}
+	models, err := model.CacheGetGroupModels(ctx, userGroup)
+	if err != nil {
+		c.JSON(http.StatusOK, gin.H{
+			"success": false,
+			"message": err.Error(),
+		})
+		return
+	}
+	c.JSON(http.StatusOK, gin.H{
+		"success": true,
+		"message": "",
+		"data":    models,
+	})
+	return
+}

controller/token.go

@@ -130,6 +130,7 @@ func AddToken(c *gin.Context) {
 		ExpiredTime:    token.ExpiredTime,
 		RemainQuota:    token.RemainQuota,
 		UnlimitedQuota: token.UnlimitedQuota,
+		Models:         token.Models,
 	}
 	err = cleanToken.Insert()
 	if err != nil {
@@ -216,6 +217,7 @@ func UpdateToken(c *gin.Context) {
 		cleanToken.ExpiredTime = token.ExpiredTime
 		cleanToken.RemainQuota = token.RemainQuota
 		cleanToken.UnlimitedQuota = token.UnlimitedQuota
+		cleanToken.Models = token.Models
 	}
 	err = cleanToken.Update()
 	if err != nil {

middleware/auth.go

@@ -1,6 +1,7 @@
 package middleware
 
 import (
+	"fmt"
 	"github.com/gin-contrib/sessions"
 	"github.com/gin-gonic/gin"
 	"github.com/songquanpeng/one-api/common"
@@ -107,6 +108,19 @@ func TokenAuth() func(c *gin.Context) {
 			abortWithMessage(c, http.StatusForbidden, "用户已被封禁")
 			return
 		}
+		requestModel, err := getRequestModel(c)
+		if err != nil {
+			abortWithMessage(c, http.StatusBadRequest, err.Error())
+			return
+		}
+		c.Set("request_model", requestModel)
+		if token.Models != nil && *token.Models != "" {
+			c.Set("available_models", *token.Models)
+			if requestModel != "" && !isModelInList(requestModel, *token.Models) {
+				abortWithMessage(c, http.StatusForbidden, fmt.Sprintf("该令牌无权使用模型：%s", requestModel))
+				return
+			}
+		}
 		c.Set("id", token.UserId)
 		c.Set("token_id", token.Id)
 		c.Set("token_name", token.Name)

middleware/distributor.go

@@ -2,14 +2,12 @@ package middleware
 
 import (
 	"fmt"
+	"github.com/gin-gonic/gin"
 	"github.com/songquanpeng/one-api/common"
 	"github.com/songquanpeng/one-api/common/logger"
 	"github.com/songquanpeng/one-api/model"
 	"net/http"
 	"strconv"
-	"strings"
-
-	"github.com/gin-gonic/gin"
 )
 
 type ModelRequest struct {
@@ -40,37 +38,11 @@ func Distribute() func(c *gin.Context) {
 				return
 			}
 		} else {
-			// Select a channel for the user
-			var modelRequest ModelRequest
-			err := common.UnmarshalBodyReusable(c, &modelRequest)
+			requestModel := c.GetString("request_model")
+			var err error
+			channel, err = model.CacheGetRandomSatisfiedChannel(userGroup, requestModel, false)
 			if err != nil {
-				abortWithMessage(c, http.StatusBadRequest, "无效的请求")
-				return
-			}
-			if strings.HasPrefix(c.Request.URL.Path, "/v1/moderations") {
-				if modelRequest.Model == "" {
-					modelRequest.Model = "text-moderation-stable"
-				}
-			}
-			if strings.HasSuffix(c.Request.URL.Path, "embeddings") {
-				if modelRequest.Model == "" {
-					modelRequest.Model = c.Param("model")
-				}
-			}
-			if strings.HasPrefix(c.Request.URL.Path, "/v1/images/generations") {
-				if modelRequest.Model == "" {
-					modelRequest.Model = "dall-e-2"
-				}
-			}
-			if strings.HasPrefix(c.Request.URL.Path, "/v1/audio/transcriptions") || strings.HasPrefix(c.Request.URL.Path, "/v1/audio/translations") {
-				if modelRequest.Model == "" {
-					modelRequest.Model = "whisper-1"
-				}
-			}
-			requestModel = modelRequest.Model
-			channel, err = model.CacheGetRandomSatisfiedChannel(userGroup, modelRequest.Model, false)
-			if err != nil {
-				message := fmt.Sprintf("当前分组 %s 下对于模型 %s 无可用渠道", userGroup, modelRequest.Model)
+				message := fmt.Sprintf("当前分组 %s 下对于模型 %s 无可用渠道", userGroup, requestModel)
 				if channel != nil {
 					logger.SysError(fmt.Sprintf("渠道不存在：%d", channel.Id))
 					message = "数据库一致性已被破坏，请联系管理员"

middleware/utils.go

@@ -1,9 +1,12 @@
 package middleware
 
 import (
+	"fmt"
 	"github.com/gin-gonic/gin"
+	"github.com/songquanpeng/one-api/common"
 	"github.com/songquanpeng/one-api/common/helper"
 	"github.com/songquanpeng/one-api/common/logger"
+	"strings"
 )
 
 func abortWithMessage(c *gin.Context, statusCode int, message string) {
@@ -16,3 +19,42 @@ func abortWithMessage(c *gin.Context, statusCode int, message string) {
 	c.Abort()
 	logger.Error(c.Request.Context(), message)
 }
+
+func getRequestModel(c *gin.Context) (string, error) {
+	var modelRequest ModelRequest
+	err := common.UnmarshalBodyReusable(c, &modelRequest)
+	if err != nil {
+		return "", fmt.Errorf("common.UnmarshalBodyReusable failed: %w", err)
+	}
+	if strings.HasPrefix(c.Request.URL.Path, "/v1/moderations") {
+		if modelRequest.Model == "" {
+			modelRequest.Model = "text-moderation-stable"
+		}
+	}
+	if strings.HasSuffix(c.Request.URL.Path, "embeddings") {
+		if modelRequest.Model == "" {
+			modelRequest.Model = c.Param("model")
+		}
+	}
+	if strings.HasPrefix(c.Request.URL.Path, "/v1/images/generations") {
+		if modelRequest.Model == "" {
+			modelRequest.Model = "dall-e-2"
+		}
+	}
+	if strings.HasPrefix(c.Request.URL.Path, "/v1/audio/transcriptions") || strings.HasPrefix(c.Request.URL.Path, "/v1/audio/translations") {
+		if modelRequest.Model == "" {
+			modelRequest.Model = "whisper-1"
+		}
+	}
+	return modelRequest.Model, nil
+}
+
+func isModelInList(modelName string, models string) bool {
+	modelList := strings.Split(models, ",")
+	for _, model := range modelList {
+		if modelName == model {
+			return true
+		}
+	}
+	return false
+}

model/ability.go

@@ -1,7 +1,10 @@
 package model
 
 import (
+	"context"
 	"github.com/songquanpeng/one-api/common"
+	"gorm.io/gorm"
+	"sort"
 	"strings"
 )
 
@@ -13,7 +16,7 @@ type Ability struct {
 	Priority  *int64 `json:"priority" gorm:"bigint;default:0;index"`
 }
 
-func GetRandomSatisfiedChannel(group string, model string) (*Channel, error) {
+func GetRandomSatisfiedChannel(group string, model string, ignoreFirstPriority bool) (*Channel, error) {
 	ability := Ability{}
 	groupCol := "`group`"
 	trueVal := "1"
@@ -23,8 +26,13 @@ func GetRandomSatisfiedChannel(group string, model string) (*Channel, error) {
 	}
 
 	var err error = nil
-	maxPrioritySubQuery := DB.Model(&Ability{}).Select("MAX(priority)").Where(groupCol+" = ? and model = ? and enabled = "+trueVal, group, model)
-	channelQuery := DB.Where(groupCol+" = ? and model = ? and enabled = "+trueVal+" and priority = (?)", group, model, maxPrioritySubQuery)
+	var channelQuery *gorm.DB
+	if ignoreFirstPriority {
+		channelQuery = DB.Where(groupCol+" = ? and model = ? and enabled = "+trueVal, group, model)
+	} else {
+		maxPrioritySubQuery := DB.Model(&Ability{}).Select("MAX(priority)").Where(groupCol+" = ? and model = ? and enabled = "+trueVal, group, model)
+		channelQuery = DB.Where(groupCol+" = ? and model = ? and enabled = "+trueVal+" and priority = (?)", group, model, maxPrioritySubQuery)
+	}
 	if common.UsingSQLite || common.UsingPostgreSQL {
 		err = channelQuery.Order("RANDOM()").First(&ability).Error
 	} else {
@@ -82,3 +90,19 @@ func (channel *Channel) UpdateAbilities() error {
 func UpdateAbilityStatus(channelId int, status bool) error {
 	return DB.Model(&Ability{}).Where("channel_id = ?", channelId).Select("enabled").Update("enabled", status).Error
 }
+
+func GetGroupModels(ctx context.Context, group string) ([]string, error) {
+	groupCol := "`group`"
+	trueVal := "1"
+	if common.UsingPostgreSQL {
+		groupCol = `"group"`
+		trueVal = "true"
+	}
+	var models []string
+	err := DB.Model(&Ability{}).Distinct("model").Where(groupCol+" = ? and enabled = "+trueVal, group).Pluck("model", &models).Error
+	if err != nil {
+		return nil, err
+	}
+	sort.Strings(models)
+	return models, err
+}

model/cache.go

@@ -21,6 +21,7 @@ var (
 	UserId2GroupCacheSeconds  = config.SyncFrequency
 	UserId2QuotaCacheSeconds  = config.SyncFrequency
 	UserId2StatusCacheSeconds = config.SyncFrequency
+	GroupModelsCacheSeconds   = config.SyncFrequency
 )
 
 func CacheGetTokenByKey(key string) (*Token, error) {
@@ -146,6 +147,25 @@ func CacheIsUserEnabled(userId int) (bool, error) {
 	return userEnabled, err
 }
 
+func CacheGetGroupModels(ctx context.Context, group string) ([]string, error) {
+	if !common.RedisEnabled {
+		return GetGroupModels(ctx, group)
+	}
+	modelsStr, err := common.RedisGet(fmt.Sprintf("group_models:%s", group))
+	if err == nil {
+		return strings.Split(modelsStr, ","), nil
+	}
+	models, err := GetGroupModels(ctx, group)
+	if err != nil {
+		return nil, err
+	}
+	err = common.RedisSet(fmt.Sprintf("group_models:%s", group), strings.Join(models, ","), time.Duration(GroupModelsCacheSeconds)*time.Second)
+	if err != nil {
+		logger.SysError("Redis set group models error: " + err.Error())
+	}
+	return models, nil
+}
+
 var group2model2channels map[string]map[string][]*Channel
 var channelSyncLock sync.RWMutex
 
@@ -205,7 +225,7 @@ func SyncChannelCache(frequency int) {
 
 func CacheGetRandomSatisfiedChannel(group string, model string, ignoreFirstPriority bool) (*Channel, error) {
 	if !config.MemoryCacheEnabled {
-		return GetRandomSatisfiedChannel(group, model)
+		return GetRandomSatisfiedChannel(group, model, ignoreFirstPriority)
 	}
 	channelSyncLock.RLock()
 	defer channelSyncLock.RUnlock()

model/token.go

@@ -12,24 +12,25 @@ import (
 )
 
 type Token struct {
-	Id             int    `json:"id"`
-	UserId         int    `json:"user_id"`
-	Key            string `json:"key" gorm:"type:char(48);uniqueIndex"`
-	Status         int    `json:"status" gorm:"default:1"`
-	Name           string `json:"name" gorm:"index" `
-	CreatedTime    int64  `json:"created_time" gorm:"bigint"`
-	AccessedTime   int64  `json:"accessed_time" gorm:"bigint"`
-	ExpiredTime    int64  `json:"expired_time" gorm:"bigint;default:-1"` // -1 means never expired
-	RemainQuota    int64  `json:"remain_quota" gorm:"bigint;default:0"`
-	UnlimitedQuota bool   `json:"unlimited_quota" gorm:"default:false"`
-	UsedQuota      int64  `json:"used_quota" gorm:"bigint;default:0"` // used quota
+	Id             int     `json:"id"`
+	UserId         int     `json:"user_id"`
+	Key            string  `json:"key" gorm:"type:char(48);uniqueIndex"`
+	Status         int     `json:"status" gorm:"default:1"`
+	Name           string  `json:"name" gorm:"index" `
+	CreatedTime    int64   `json:"created_time" gorm:"bigint"`
+	AccessedTime   int64   `json:"accessed_time" gorm:"bigint"`
+	ExpiredTime    int64   `json:"expired_time" gorm:"bigint;default:-1"` // -1 means never expired
+	RemainQuota    int64   `json:"remain_quota" gorm:"bigint;default:0"`
+	UnlimitedQuota bool    `json:"unlimited_quota" gorm:"default:false"`
+	UsedQuota      int64   `json:"used_quota" gorm:"bigint;default:0"` // used quota
+	Models         *string `json:"models" gorm:"default:''"`
 }
 
 func GetAllUserTokens(userId int, startIdx int, num int, order string) ([]*Token, error) {
 	var tokens []*Token
 	var err error
 	query := DB.Where("user_id = ?", userId)
-	
+
 	switch order {
 	case "remain_quota":
 		query = query.Order("unlimited_quota desc, remain_quota desc")
@@ -38,7 +39,7 @@ func GetAllUserTokens(userId int, startIdx int, num int, order string) ([]*Token
 	default:
 		query = query.Order("id desc")
 	}
-	
+
 	err = query.Limit(num).Offset(startIdx).Find(&tokens).Error
 	return tokens, err
 }
@@ -121,7 +122,7 @@ func (token *Token) Insert() error {
 // Update Make sure your token's fields is completed, because this will update non-zero values
 func (token *Token) Update() error {
 	var err error
-	err = DB.Model(token).Select("name", "status", "expired_time", "remain_quota", "unlimited_quota").Updates(token).Error
+	err = DB.Model(token).Select("name", "status", "expired_time", "remain_quota", "unlimited_quota", "models").Updates(token).Error
 	return err
 }
 

relay/channel/ali/main.go

@@ -48,6 +48,9 @@ func ConvertRequest(request model.GeneralOpenAIRequest) *ChatRequest {
 			MaxTokens:         request.MaxTokens,
 			Temperature:       request.Temperature,
 			TopP:              request.TopP,
+			TopK:              request.TopK,
+			ResultFormat:      "message",
+			Tools:             request.Tools,
 		},
 	}
 }
@@ -117,19 +120,11 @@ func embeddingResponseAli2OpenAI(response *EmbeddingResponse) *openai.EmbeddingR
 }
 
 func responseAli2OpenAI(response *ChatResponse) *openai.TextResponse {
-	choice := openai.TextResponseChoice{
-		Index: 0,
-		Message: model.Message{
-			Role:    "assistant",
-			Content: response.Output.Text,
-		},
-		FinishReason: response.Output.FinishReason,
-	}
 	fullTextResponse := openai.TextResponse{
 		Id:      response.RequestId,
 		Object:  "chat.completion",
 		Created: helper.GetTimestamp(),
-		Choices: []openai.TextResponseChoice{choice},
+		Choices: response.Output.Choices,
 		Usage: model.Usage{
 			PromptTokens:     response.Usage.InputTokens,
 			CompletionTokens: response.Usage.OutputTokens,
@@ -140,10 +135,14 @@ func responseAli2OpenAI(response *ChatResponse) *openai.TextResponse {
 }
 
 func streamResponseAli2OpenAI(aliResponse *ChatResponse) *openai.ChatCompletionsStreamResponse {
+	if len(aliResponse.Output.Choices) == 0 {
+		return nil
+	}
+	aliChoice := aliResponse.Output.Choices[0]
 	var choice openai.ChatCompletionsStreamResponseChoice
-	choice.Delta.Content = aliResponse.Output.Text
-	if aliResponse.Output.FinishReason != "null" {
-		finishReason := aliResponse.Output.FinishReason
+	choice.Delta = aliChoice.Message
+	if aliChoice.FinishReason != "null" {
+		finishReason := aliChoice.FinishReason
 		choice.FinishReason = &finishReason
 	}
 	response := openai.ChatCompletionsStreamResponse{
@@ -204,6 +203,9 @@ func StreamHandler(c *gin.Context, resp *http.Response) (*model.ErrorWithStatusC
 				usage.TotalTokens = aliResponse.Usage.InputTokens + aliResponse.Usage.OutputTokens
 			}
 			response := streamResponseAli2OpenAI(&aliResponse)
+			if response == nil {
+				return true
+			}
 			//response.Choices[0].Delta.Content = strings.TrimPrefix(response.Choices[0].Delta.Content, lastResponseText)
 			//lastResponseText = aliResponse.Output.Text
 			jsonResponse, err := json.Marshal(response)
@@ -226,6 +228,7 @@ func StreamHandler(c *gin.Context, resp *http.Response) (*model.ErrorWithStatusC
 }
 
 func Handler(c *gin.Context, resp *http.Response) (*model.ErrorWithStatusCode, *model.Usage) {
+	ctx := c.Request.Context()
 	var aliResponse ChatResponse
 	responseBody, err := io.ReadAll(resp.Body)
 	if err != nil {
@@ -235,6 +238,7 @@ func Handler(c *gin.Context, resp *http.Response) (*model.ErrorWithStatusCode, *
 	if err != nil {
 		return openai.ErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), nil
 	}
+	logger.Debugf(ctx, "response body: %s\n", responseBody)
 	err = json.Unmarshal(responseBody, &aliResponse)
 	if err != nil {
 		return openai.ErrorWrapper(err, "unmarshal_response_body_failed", http.StatusInternalServerError), nil

relay/channel/ali/model.go

@@ -1,5 +1,10 @@
 package ali
 
+import (
+	"github.com/songquanpeng/one-api/relay/channel/openai"
+	"github.com/songquanpeng/one-api/relay/model"
+)
+
 type Message struct {
 	Content string `json:"content"`
 	Role    string `json:"role"`
@@ -11,13 +16,15 @@ type Input struct {
 }
 
 type Parameters struct {
-	TopP              float64 `json:"top_p,omitempty"`
-	TopK              int     `json:"top_k,omitempty"`
-	Seed              uint64  `json:"seed,omitempty"`
-	EnableSearch      bool    `json:"enable_search,omitempty"`
-	IncrementalOutput bool    `json:"incremental_output,omitempty"`
-	MaxTokens         int     `json:"max_tokens,omitempty"`
-	Temperature       float64 `json:"temperature,omitempty"`
+	TopP              float64      `json:"top_p,omitempty"`
+	TopK              int          `json:"top_k,omitempty"`
+	Seed              uint64       `json:"seed,omitempty"`
+	EnableSearch      bool         `json:"enable_search,omitempty"`
+	IncrementalOutput bool         `json:"incremental_output,omitempty"`
+	MaxTokens         int          `json:"max_tokens,omitempty"`
+	Temperature       float64      `json:"temperature,omitempty"`
+	ResultFormat      string       `json:"result_format,omitempty"`
+	Tools             []model.Tool `json:"tools,omitempty"`
 }
 
 type ChatRequest struct {
@@ -62,8 +69,9 @@ type Usage struct {
 }
 
 type Output struct {
-	Text         string `json:"text"`
-	FinishReason string `json:"finish_reason"`
+	//Text         string                      `json:"text"`
+	//FinishReason string                      `json:"finish_reason"`
+	Choices []openai.TextResponseChoice `json:"choices"`
 }
 
 type ChatResponse struct {

relay/channel/anthropic/main.go

@@ -38,6 +38,7 @@ func ConvertRequest(textRequest model.GeneralOpenAIRequest) *Request {
 		MaxTokens:   textRequest.MaxTokens,
 		Temperature: textRequest.Temperature,
 		TopP:        textRequest.TopP,
+		TopK:        textRequest.TopK,
 		Stream:      textRequest.Stream,
 	}
 	if claudeRequest.MaxTokens == 0 {

relay/channel/openai/adaptor.go

@@ -70,8 +70,10 @@ func (a *Adaptor) DoRequest(c *gin.Context, meta *util.RelayMeta, requestBody io
 func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, meta *util.RelayMeta) (usage *model.Usage, err *model.ErrorWithStatusCode) {
 	if meta.IsStream {
 		var responseText string
-		err, responseText, _ = StreamHandler(c, resp, meta.Mode)
-		usage = ResponseText2Usage(responseText, meta.ActualModelName, meta.PromptTokens)
+		err, responseText, usage = StreamHandler(c, resp, meta.Mode)
+		if usage == nil {
+			usage = ResponseText2Usage(responseText, meta.ActualModelName, meta.PromptTokens)
+		}
 	} else {
 		err, usage = Handler(c, resp, meta.PromptTokens, meta.ActualModelName)
 	}

relay/channel/openai/main.go

@@ -6,6 +6,7 @@ import (
 	"encoding/json"
 	"github.com/gin-gonic/gin"
 	"github.com/songquanpeng/one-api/common"
+	"github.com/songquanpeng/one-api/common/conv"
 	"github.com/songquanpeng/one-api/common/logger"
 	"github.com/songquanpeng/one-api/relay/constant"
 	"github.com/songquanpeng/one-api/relay/model"
@@ -53,7 +54,7 @@ func StreamHandler(c *gin.Context, resp *http.Response, relayMode int) (*model.E
 						continue // just ignore the error
 					}
 					for _, choice := range streamResponse.Choices {
-						responseText += choice.Delta.Content
+						responseText += conv.AsString(choice.Delta.Content)
 					}
 					if streamResponse.Usage != nil {
 						usage = streamResponse.Usage

relay/channel/openai/model.go

@@ -118,12 +118,9 @@ type ImageResponse struct {
 }
 
 type ChatCompletionsStreamResponseChoice struct {
-	Index int `json:"index"`
-	Delta struct {
-		Content string `json:"content"`
-		Role    string `json:"role,omitempty"`
-	} `json:"delta"`
-	FinishReason *string `json:"finish_reason,omitempty"`
+	Index        int           `json:"index"`
+	Delta        model.Message `json:"delta"`
+	FinishReason *string       `json:"finish_reason,omitempty"`
 }
 
 type ChatCompletionsStreamResponse struct {

relay/channel/tencent/main.go

@@ -10,6 +10,7 @@ import (
 	"fmt"
 	"github.com/gin-gonic/gin"
 	"github.com/songquanpeng/one-api/common"
+	"github.com/songquanpeng/one-api/common/conv"
 	"github.com/songquanpeng/one-api/common/helper"
 	"github.com/songquanpeng/one-api/common/logger"
 	"github.com/songquanpeng/one-api/relay/channel/openai"
@@ -129,7 +130,7 @@ func StreamHandler(c *gin.Context, resp *http.Response) (*model.ErrorWithStatusC
 			}
 			response := streamResponseTencent2OpenAI(&TencentResponse)
 			if len(response.Choices) != 0 {
-				responseText += response.Choices[0].Delta.Content
+				responseText += conv.AsString(response.Choices[0].Delta.Content)
 			}
 			jsonResponse, err := json.Marshal(response)
 			if err != nil {

relay/channel/xunfei/main.go

@@ -26,7 +26,11 @@ import (
 
 func requestOpenAI2Xunfei(request model.GeneralOpenAIRequest, xunfeiAppId string, domain string) *ChatRequest {
 	messages := make([]Message, 0, len(request.Messages))
+	var lastToolCalls []model.Tool
 	for _, message := range request.Messages {
+		if message.ToolCalls != nil {
+			lastToolCalls = message.ToolCalls
+		}
 		messages = append(messages, Message{
 			Role:    message.Role,
 			Content: message.StringContent(),
@@ -39,9 +43,33 @@ func requestOpenAI2Xunfei(request model.GeneralOpenAIRequest, xunfeiAppId string
 	xunfeiRequest.Parameter.Chat.TopK = request.N
 	xunfeiRequest.Parameter.Chat.MaxTokens = request.MaxTokens
 	xunfeiRequest.Payload.Message.Text = messages
+	if len(lastToolCalls) != 0 {
+		for _, toolCall := range lastToolCalls {
+			xunfeiRequest.Payload.Functions.Text = append(xunfeiRequest.Payload.Functions.Text, toolCall.Function)
+		}
+	}
+
 	return &xunfeiRequest
 }
 
+func getToolCalls(response *ChatResponse) []model.Tool {
+	var toolCalls []model.Tool
+	if len(response.Payload.Choices.Text) == 0 {
+		return toolCalls
+	}
+	item := response.Payload.Choices.Text[0]
+	if item.FunctionCall == nil {
+		return toolCalls
+	}
+	toolCall := model.Tool{
+		Id:       fmt.Sprintf("call_%s", helper.GetUUID()),
+		Type:     "function",
+		Function: *item.FunctionCall,
+	}
+	toolCalls = append(toolCalls, toolCall)
+	return toolCalls
+}
+
 func responseXunfei2OpenAI(response *ChatResponse) *openai.TextResponse {
 	if len(response.Payload.Choices.Text) == 0 {
 		response.Payload.Choices.Text = []ChatResponseTextItem{
@@ -53,8 +81,9 @@ func responseXunfei2OpenAI(response *ChatResponse) *openai.TextResponse {
 	choice := openai.TextResponseChoice{
 		Index: 0,
 		Message: model.Message{
-			Role:    "assistant",
-			Content: response.Payload.Choices.Text[0].Content,
+			Role:      "assistant",
+			Content:   response.Payload.Choices.Text[0].Content,
+			ToolCalls: getToolCalls(response),
 		},
 		FinishReason: constant.StopFinishReason,
 	}
@@ -78,6 +107,7 @@ func streamResponseXunfei2OpenAI(xunfeiResponse *ChatResponse) *openai.ChatCompl
 	}
 	var choice openai.ChatCompletionsStreamResponseChoice
 	choice.Delta.Content = xunfeiResponse.Payload.Choices.Text[0].Content
+	choice.Delta.ToolCalls = getToolCalls(xunfeiResponse)
 	if xunfeiResponse.Payload.Choices.Status == 2 {
 		choice.FinishReason = &constant.StopFinishReason
 	}

relay/channel/xunfei/model.go

@@ -26,13 +26,18 @@ type ChatRequest struct {
 		Message struct {
 			Text []Message `json:"text"`
 		} `json:"message"`
+		Functions struct {
+			Text []model.Function `json:"text,omitempty"`
+		} `json:"functions"`
 	} `json:"payload"`
 }
 
 type ChatResponseTextItem struct {
-	Content string `json:"content"`
-	Role    string `json:"role"`
-	Index   int    `json:"index"`
+	Content      string          `json:"content"`
+	Role         string          `json:"role"`
+	Index        int             `json:"index"`
+	ContentType  string          `json:"content_type"`
+	FunctionCall *model.Function `json:"function_call"`
 }
 
 type ChatResponse struct {

relay/model/general.go

@@ -5,25 +5,29 @@ type ResponseFormat struct {
 }
 
 type GeneralOpenAIRequest struct {
-	Model            string          `json:"model,omitempty"`
 	Messages         []Message       `json:"messages,omitempty"`
-	Prompt           any             `json:"prompt,omitempty"`
-	Stream           bool            `json:"stream,omitempty"`
-	MaxTokens        int             `json:"max_tokens,omitempty"`
-	Temperature      float64         `json:"temperature,omitempty"`
-	TopP             float64         `json:"top_p,omitempty"`
-	N                int             `json:"n,omitempty"`
-	Input            any             `json:"input,omitempty"`
-	Instruction      string          `json:"instruction,omitempty"`
-	Size             string          `json:"size,omitempty"`
-	Functions        any             `json:"functions,omitempty"`
+	Model            string          `json:"model,omitempty"`
 	FrequencyPenalty float64         `json:"frequency_penalty,omitempty"`
+	MaxTokens        int             `json:"max_tokens,omitempty"`
+	N                int             `json:"n,omitempty"`
 	PresencePenalty  float64         `json:"presence_penalty,omitempty"`
 	ResponseFormat   *ResponseFormat `json:"response_format,omitempty"`
 	Seed             float64         `json:"seed,omitempty"`
-	Tools            any             `json:"tools,omitempty"`
+	Stream           bool            `json:"stream,omitempty"`
+	Temperature      float64         `json:"temperature,omitempty"`
+	TopP             float64         `json:"top_p,omitempty"`
+	TopK             int             `json:"top_k,omitempty"`
+	Tools            []Tool          `json:"tools,omitempty"`
 	ToolChoice       any             `json:"tool_choice,omitempty"`
+	FunctionCall     any             `json:"function_call,omitempty"`
+	Functions        any             `json:"functions,omitempty"`
 	User             string          `json:"user,omitempty"`
+	Prompt           any             `json:"prompt,omitempty"`
+	Input            any             `json:"input,omitempty"`
+	EncodingFormat   string          `json:"encoding_format,omitempty"`
+	Dimensions       int             `json:"dimensions,omitempty"`
+	Instruction      string          `json:"instruction,omitempty"`
+	Size             string          `json:"size,omitempty"`
 }
 
 func (r GeneralOpenAIRequest) ParseInput() []string {

relay/model/message.go

@@ -1,9 +1,10 @@
 package model
 
 type Message struct {
-	Role    string  `json:"role"`
-	Content any     `json:"content"`
-	Name    *string `json:"name,omitempty"`
+	Role      string  `json:"role,omitempty"`
+	Content   any     `json:"content,omitempty"`
+	Name      *string `json:"name,omitempty"`
+	ToolCalls []Tool  `json:"tool_calls,omitempty"`
 }
 
 func (m Message) IsStringContent() bool {

relay/model/tool.go

@@ -0,0 +1,14 @@
+package model
+
+type Tool struct {
+	Id       string   `json:"id,omitempty"`
+	Type     string   `json:"type"`
+	Function Function `json:"function"`
+}
+
+type Function struct {
+	Description string `json:"description,omitempty"`
+	Name        string `json:"name"`
+	Parameters  any    `json:"parameters,omitempty"` // request
+	Arguments   any    `json:"arguments,omitempty"`  // response
+}

router/api-router.go

@@ -43,6 +43,7 @@ func SetApiRouter(router *gin.Engine) {
 				selfRoute.GET("/token", controller.GenerateAccessToken)
 				selfRoute.GET("/aff", controller.GetAffCode)
 				selfRoute.POST("/topup", controller.TopUp)
+				selfRoute.GET("/available_models", controller.GetUserAvailableModels)
 			}
 
 			adminRoute := userRoute.Group("/")

web/default/src/pages/Token/EditToken.js

@@ -1,19 +1,21 @@
 import React, { useEffect, useState } from 'react';
 import { Button, Form, Header, Message, Segment } from 'semantic-ui-react';
-import { useParams, useNavigate } from 'react-router-dom';
-import { API, showError, showSuccess, timestamp2string } from '../../helpers';
-import { renderQuota, renderQuotaWithPrompt } from '../../helpers/render';
+import { useNavigate, useParams } from 'react-router-dom';
+import { API, copy, showError, showSuccess, timestamp2string } from '../../helpers';
+import { renderQuotaWithPrompt } from '../../helpers/render';
 
 const EditToken = () => {
   const params = useParams();
   const tokenId = params.id;
   const isEdit = tokenId !== undefined;
   const [loading, setLoading] = useState(isEdit);
+  const [modelOptions, setModelOptions] = useState([]);
   const originInputs = {
     name: '',
     remain_quota: isEdit ? 0 : 500000,
     expired_time: -1,
-    unlimited_quota: false
+    unlimited_quota: false,
+    models: []
   };
   const [inputs, setInputs] = useState(originInputs);
   const { name, remain_quota, expired_time, unlimited_quota } = inputs;
@@ -22,8 +24,8 @@ const EditToken = () => {
     setInputs((inputs) => ({ ...inputs, [name]: value }));
   };
   const handleCancel = () => {
-    navigate("/token");
-  }
+    navigate('/token');
+  };
   const setExpiredTime = (month, day, hour, minute) => {
     let now = new Date();
     let timestamp = now.getTime() / 1000;
@@ -50,6 +52,11 @@ const EditToken = () => {
       if (data.expired_time !== -1) {
         data.expired_time = timestamp2string(data.expired_time);
       }
+      if (data.models === '') {
+        data.models = [];
+      } else {
+        data.models = data.models.split(',');
+      }
       setInputs(data);
     } else {
       showError(message);
@@ -60,8 +67,26 @@ const EditToken = () => {
     if (isEdit) {
       loadToken().then();
     }
+    loadAvailableModels().then();
   }, []);
 
+  const loadAvailableModels = async () => {
+    let res = await API.get(`/api/user/available_models`);
+    const { success, message, data } = res.data;
+    if (success) {
+      let options = data.map((model) => {
+        return {
+          key: model,
+          text: model,
+          value: model
+        };
+      });
+      setModelOptions(options);
+    } else {
+      showError(message);
+    }
+  };
+
   const submit = async () => {
     if (!isEdit && inputs.name === '') return;
     let localInputs = inputs;
@@ -74,6 +99,7 @@ const EditToken = () => {
       }
       localInputs.expired_time = Math.ceil(time / 1000);
     }
+    localInputs.models = localInputs.models.join(',');
     let res;
     if (isEdit) {
       res = await API.put(`/api/token/`, { ...localInputs, id: parseInt(tokenId) });
@@ -109,6 +135,24 @@ const EditToken = () => {
               required={!isEdit}
             />
           </Form.Field>
+          <Form.Field>
+            <Form.Dropdown
+              label='模型范围'
+              placeholder={'请选择允许使用的模型，留空则不进行限制'}
+              name='models'
+              fluid
+              multiple
+              search
+              onLabelClick={(e, { value }) => {
+                copy(value).then();
+              }}
+              selection
+              onChange={handleInputChange}
+              value={inputs.models}
+              autoComplete='new-password'
+              options={modelOptions}
+            />
+          </Form.Field>
           <Form.Field>
             <Form.Input
               label='过期时间'