Merge pull request #505 from OiAnthony/f_dotenv

feat: 添加.env配置文件和初始化环境变量

.env.example

@@ -0,0 +1,71 @@
+# 端口号
+# PORT=3000
+# 前端基础URL
+# FRONTEND_BASE_URL=https://your-frontend-url.com
+
+
+# 调试相关配置
+# 启用pprof
+# ENABLE_PPROF=true
+
+# 数据库相关配置
+# 数据库连接字符串
+# SQL_DSN=mysql://user:password@tcp(127.0.0.1:3306)/dbname?parseTime=true
+# 日志数据库连接字符串
+# LOG_SQL_DSN=mysql://user:password@tcp(127.0.0.1:3306)/logdb?parseTime=true
+# SQLite数据库路径
+# SQLITE_PATH=/path/to/sqlite.db
+# 数据库最大空闲连接数
+# SQL_MAX_IDLE_CONNS=100
+# 数据库最大打开连接数
+# SQL_MAX_OPEN_CONNS=1000
+# 数据库连接最大生命周期（秒）
+# SQL_MAX_LIFETIME=60
+
+
+# 缓存相关配置
+# Redis连接字符串
+# REDIS_CONN_STRING=redis://user:password@localhost:6379/0
+# 同步频率（单位：秒）
+# SYNC_FREQUENCY=60
+# 内存缓存启用
+# MEMORY_CACHE_ENABLED=true
+# 渠道更新频率（单位：秒）
+# CHANNEL_UPDATE_FREQUENCY=30
+# 批量更新启用
+# BATCH_UPDATE_ENABLED=true
+# 批量更新间隔（单位：秒）
+# BATCH_UPDATE_INTERVAL=5
+
+# 任务和功能配置
+# 更新任务启用
+# UPDATE_TASK=true
+
+
+# 会话密钥
+# SESSION_SECRET=random_string
+
+# 其他配置
+# 渠道测试频率（单位：秒）
+# CHANNEL_TEST_FREQUENCY=10
+# 生成默认token
+# GENERATE_DEFAULT_TOKEN=false
+# Gemini 安全设置
+# GEMINI_SAFETY_SETTING=BLOCK_NONE
+# Gemini版本设置
+# GEMINI_MODEL_MAP=gemini-1.5-pro-latest:v1beta,gemini-1.5-pro-001:v1beta
+# Cohere 安全设置
+# COHERE_SAFETY_SETTING=NONE
+# 是否统计图片token
+# GET_MEDIA_TOKEN=true
+# 是否在非流（stream=false）情况下统计图片token
+# GET_MEDIA_TOKEN_NOT_STREAM=true
+# 设置 Dify 渠道是否输出工作流和节点信息到客户端
+# DIFY_DEBUG=true
+# 设置流式一次回复的超时时间
+# STREAMING_TIMEOUT=90
+
+
+# 节点类型
+# 如果是主节点则为master
+# NODE_TYPE=master

.gitignore

@@ -6,4 +6,5 @@ upload
 build
 *.db-journal
 logs
-web/dist
+web/dist
+.env

README.md

@@ -48,6 +48,7 @@
     4. Telegram Bot 名称是bot username 去掉@后的字符串
 13. 添加 [Suno API](https://github.com/Suno-API/Suno-API)接口的支持，[对接文档](Suno.md)
 14. 支持Rerank模型，目前仅兼容Cohere和Jina，可接入Dify，[对接文档](Rerank.md)
+15. **[OpenAI Realtime API](https://platform.openai.com/docs/guides/realtime/integration)** - 支持OpenAI的Realtime API，支持Azure渠道。
 
 ## 模型支持
 此版本额外支持以下模型：

common/constants.go

@@ -41,6 +41,7 @@ var PasswordLoginEnabled = true
 var PasswordRegisterEnabled = true
 var EmailVerificationEnabled = false
 var GitHubOAuthEnabled = false
+var LinuxDOOAuthEnabled = false
 var WeChatAuthEnabled = false
 var TelegramOAuthEnabled = false
 var TurnstileCheckEnabled = false
@@ -75,6 +76,9 @@ var SMTPToken = ""
 var GitHubClientId = ""
 var GitHubClientSecret = ""
 
+var LinuxDOClientId = ""
+var LinuxDOClientSecret = ""
+
 var WeChatServerAddress = ""
 var WeChatServerToken = ""
 var WeChatAccountQRCodeImageURL = ""
@@ -222,6 +226,7 @@ const (
 	ChannelCloudflare         = 39
 	ChannelTypeSiliconFlow    = 40
 	ChannelTypeVertexAi       = 41
+	ChannelTypeMistral        = 42
 
 	ChannelTypeDummy // this one is only for count, do not add any channel after this
 
@@ -270,4 +275,5 @@ var ChannelBaseURLs = []string{
 	"https://api.cloudflare.com",                //39
 	"https://api.siliconflow.cn",                //40
 	"",                                          //41
+	"https://api.mistral.ai",                    //42
 }

common/model-ratio.go

@@ -32,26 +32,29 @@ var defaultModelRatio = map[string]float64{
 	"gpt-4-0613": 15,
 	"gpt-4-32k":  30,
 	//"gpt-4-32k-0314":               30, //deprecated
-	"gpt-4-32k-0613":            30,
-	"gpt-4-1106-preview":        5,    // $0.01 / 1K tokens
-	"gpt-4-0125-preview":        5,    // $0.01 / 1K tokens
-	"gpt-4-turbo-preview":       5,    // $0.01 / 1K tokens
-	"gpt-4-vision-preview":      5,    // $0.01 / 1K tokens
-	"gpt-4-1106-vision-preview": 5,    // $0.01 / 1K tokens
-	"chatgpt-4o-latest":         2.5,  // $0.01 / 1K tokens
-	"gpt-4o":                    2.5,  // $0.01 / 1K tokens
-	"gpt-4o-2024-05-13":         2.5,  // $0.01 / 1K tokens
-	"gpt-4o-2024-08-06":         1.25, // $0.01 / 1K tokens
-	"o1-preview":                7.5,
-	"o1-preview-2024-09-12":     7.5,
-	"o1-mini":                   1.5,
-	"o1-mini-2024-09-12":        1.5,
-	"gpt-4o-mini":               0.075,
-	"gpt-4o-mini-2024-07-18":    0.075,
-	"gpt-4-turbo":               5,    // $0.01 / 1K tokens
-	"gpt-4-turbo-2024-04-09":    5,    // $0.01 / 1K tokens
-	"gpt-3.5-turbo":             0.25, // $0.0015 / 1K tokens
+	"gpt-4-32k-0613":                  30,
+	"gpt-4-1106-preview":              5,    // $0.01 / 1K tokens
+	"gpt-4-0125-preview":              5,    // $0.01 / 1K tokens
+	"gpt-4-turbo-preview":             5,    // $0.01 / 1K tokens
+	"gpt-4-vision-preview":            5,    // $0.01 / 1K tokens
+	"gpt-4-1106-vision-preview":       5,    // $0.01 / 1K tokens
+	"chatgpt-4o-latest":               2.5,  // $0.01 / 1K tokens
+	"gpt-4o":                          1.25, // $0.01 / 1K tokens
+	"gpt-4o-audio-preview":            1.25, // $0.0015 / 1K tokens
+	"gpt-4o-audio-preview-2024-10-01": 1.25, // $0.0015 / 1K tokens
+	"gpt-4o-2024-08-06":               1.25, // $0.01 / 1K tokens
+	"gpt-4o-2024-05-13":               2.5,
+	"gpt-4o-realtime-preview":         2.5,
+	"o1-preview":                      7.5,
+	"o1-preview-2024-09-12":           7.5,
+	"o1-mini":                         1.5,
+	"o1-mini-2024-09-12":              1.5,
+	"gpt-4o-mini":                     0.075,
+	"gpt-4o-mini-2024-07-18":          0.075,
+	"gpt-4-turbo":                     5, // $0.01 / 1K tokens
+	"gpt-4-turbo-2024-04-09":          5, // $0.01 / 1K tokens
 	//"gpt-3.5-turbo-0301":           0.75, //deprecated
+	"gpt-3.5-turbo":          0.25,
 	"gpt-3.5-turbo-0613":     0.75,
 	"gpt-3.5-turbo-16k":      1.5, // $0.003 / 1K tokens
 	"gpt-3.5-turbo-16k-0613": 1.5,
@@ -86,6 +89,7 @@ var defaultModelRatio = map[string]float64{
 	"claude-2.0":                     4,     // $8 / 1M tokens
 	"claude-2.1":                     4,     // $8 / 1M tokens
 	"claude-3-haiku-20240307":        0.125, // $0.25 / 1M tokens
+	"claude-3-5-haiku-20241022":      0.5,   // $1 / 1M tokens
 	"claude-3-sonnet-20240229":       1.5,   // $3 / 1M tokens
 	"claude-3-5-sonnet-20240620":     1.5,
 	"claude-3-5-sonnet-20241022":     1.5,
@@ -337,15 +341,15 @@ func GetCompletionRatio(name string) float64 {
 		name = "gpt-4o-gizmo-*"
 	}
 	if strings.HasPrefix(name, "gpt-4") && !strings.HasSuffix(name, "-all") && !strings.HasSuffix(name, "-gizmo-*") {
-		if strings.HasPrefix(name, "gpt-4-turbo") || strings.HasSuffix(name, "preview") {
-			return 3
-		}
 		if strings.HasPrefix(name, "gpt-4o") {
 			if name == "gpt-4o-2024-05-13" {
 				return 3
 			}
 			return 4
 		}
+		if strings.HasPrefix(name, "gpt-4-turbo") || strings.HasSuffix(name, "preview") {
+			return 3
+		}
 		return 2
 	}
 	if strings.HasPrefix(name, "o1-") {
@@ -418,6 +422,36 @@ func GetCompletionRatio(name string) float64 {
 	return 1
 }
 
+func GetAudioRatio(name string) float64 {
+	if strings.HasPrefix(name, "gpt-4o-realtime") {
+		return 20
+	} else if strings.HasPrefix(name, "gpt-4o-audio") {
+		return 40
+	}
+	return 20
+}
+
+func GetAudioCompletionRatio(name string) float64 {
+	if strings.HasPrefix(name, "gpt-4o-realtime") {
+		return 2
+	}
+	return 2
+}
+
+//func GetAudioPricePerMinute(name string) float64 {
+//	if strings.HasPrefix(name, "gpt-4o-realtime") {
+//		return 0.06
+//	}
+//	return 0.06
+//}
+//
+//func GetAudioCompletionPricePerMinute(name string) float64 {
+//	if strings.HasPrefix(name, "gpt-4o-realtime") {
+//		return 0.24
+//	}
+//	return 0.24
+//}
+
 func GetCompletionRatioMap() map[string]float64 {
 	if CompletionRatio == nil {
 		CompletionRatio = defaultCompletionRatio

controller/channel-test.go

@@ -102,17 +102,22 @@ func testChannel(channel *model.Channel, testModel string) (err error, openAIErr
 	if err != nil {
 		return err, nil
 	}
-	if resp != nil && resp.StatusCode != http.StatusOK {
-		err := service.RelayErrorHandler(resp)
-		return fmt.Errorf("status code %d: %s", resp.StatusCode, err.Error.Message), err
+	var httpResp *http.Response
+	if resp != nil {
+		httpResp = resp.(*http.Response)
+		if httpResp.StatusCode != http.StatusOK {
+			err := service.RelayErrorHandler(httpResp)
+			return fmt.Errorf("status code %d: %s", httpResp.StatusCode, err.Error.Message), err
+		}
 	}
-	usage, respErr := adaptor.DoResponse(c, resp, meta)
+	usageA, respErr := adaptor.DoResponse(c, httpResp, meta)
 	if respErr != nil {
 		return fmt.Errorf("%s", respErr.Error.Message), respErr
 	}
-	if usage == nil {
+	if usageA == nil {
 		return errors.New("usage is nil"), nil
 	}
+	usage := usageA.(*dto.Usage)
 	result := w.Result()
 	respBody, err := io.ReadAll(result.Body)
 	if err != nil {

controller/linuxdo.go

@@ -0,0 +1,265 @@
+package controller
+
+import (
+	"encoding/base64"
+	"encoding/json"
+	"errors"
+	"fmt"
+	"net/http"
+	"net/url"
+	"one-api/common"
+	"one-api/model"
+	"strconv"
+	"strings"
+	"time"
+
+	"github.com/gin-contrib/sessions"
+	"github.com/gin-gonic/gin"
+)
+
+type LinuxdoUser struct {
+	Id         int    `json:"id"`
+	Username   string `json:"username"`
+	Name       string `json:"name"`
+	Active     bool   `json:"active"`
+	TrustLevel int    `json:"trust_level"`
+	Silenced   bool   `json:"silenced"`
+}
+
+func LinuxDoBind(c *gin.Context) {
+	if !common.LinuxDOOAuthEnabled {
+		c.JSON(http.StatusOK, gin.H{
+			"success": false,
+			"message": "管理员未开启通过 Linux DO 登录以及注册",
+		})
+		return
+	}
+
+	code := c.Query("code")
+	linuxdoUser, err := getLinuxdoUserInfoByCode(code, c)
+	if err != nil {
+		c.JSON(http.StatusOK, gin.H{
+			"success": false,
+			"message": err.Error(),
+		})
+		return
+	}
+
+	user := model.User{
+		LinuxDOId: strconv.Itoa(linuxdoUser.Id),
+	}
+
+	if model.IsLinuxDOIdAlreadyTaken(user.LinuxDOId) {
+		c.JSON(http.StatusOK, gin.H{
+			"success": false,
+			"message": "该 Linux DO 账户已被绑定",
+		})
+		return
+	}
+
+	session := sessions.Default(c)
+	id := session.Get("id")
+	user.Id = id.(int)
+
+	err = user.FillUserById()
+	if err != nil {
+		c.JSON(http.StatusOK, gin.H{
+			"success": false,
+			"message": err.Error(),
+		})
+		return
+	}
+
+	user.LinuxDOId = strconv.Itoa(linuxdoUser.Id)
+	err = user.Update(false)
+	if err != nil {
+		c.JSON(http.StatusOK, gin.H{
+			"success": false,
+			"message": err.Error(),
+		})
+		return
+	}
+
+	c.JSON(http.StatusOK, gin.H{
+		"success": true,
+		"message": "bind",
+	})
+}
+
+func getLinuxdoUserInfoByCode(code string, c *gin.Context) (*LinuxdoUser, error) {
+	if code == "" {
+		return nil, errors.New("invalid code")
+	}
+
+	// Get access token using Basic auth
+	tokenEndpoint := "https://connect.linux.do/oauth2/token"
+	credentials := common.LinuxDOClientId + ":" + common.LinuxDOClientSecret
+	basicAuth := "Basic " + base64.StdEncoding.EncodeToString([]byte(credentials))
+
+	// Get redirect URI from request
+	scheme := "http"
+	if c.Request.TLS != nil {
+		scheme = "https"
+	}
+	redirectURI := fmt.Sprintf("%s://%s/api/oauth/linuxdo", scheme, c.Request.Host)
+
+	data := url.Values{}
+	data.Set("grant_type", "authorization_code")
+	data.Set("code", code)
+	data.Set("redirect_uri", redirectURI)
+
+	req, err := http.NewRequest("POST", tokenEndpoint, strings.NewReader(data.Encode()))
+	if err != nil {
+		return nil, err
+	}
+
+	req.Header.Set("Authorization", basicAuth)
+	req.Header.Set("Content-Type", "application/x-www-form-urlencoded")
+	req.Header.Set("Accept", "application/json")
+
+	client := http.Client{Timeout: 5 * time.Second}
+	res, err := client.Do(req)
+	if err != nil {
+		return nil, errors.New("failed to connect to Linux DO server")
+	}
+	defer res.Body.Close()
+
+	var tokenRes struct {
+		AccessToken string `json:"access_token"`
+		Message     string `json:"message"`
+	}
+	if err := json.NewDecoder(res.Body).Decode(&tokenRes); err != nil {
+		return nil, err
+	}
+
+	if tokenRes.AccessToken == "" {
+		return nil, fmt.Errorf("failed to get access token: %s", tokenRes.Message)
+	}
+
+	// Get user info
+	userEndpoint := "https://connect.linux.do/api/user"
+	req, err = http.NewRequest("GET", userEndpoint, nil)
+	if err != nil {
+		return nil, err
+	}
+	req.Header.Set("Authorization", "Bearer "+tokenRes.AccessToken)
+	req.Header.Set("Accept", "application/json")
+
+	res2, err := client.Do(req)
+	if err != nil {
+		return nil, errors.New("failed to get user info from Linux DO")
+	}
+	defer res2.Body.Close()
+
+	var linuxdoUser LinuxdoUser
+	if err := json.NewDecoder(res2.Body).Decode(&linuxdoUser); err != nil {
+		return nil, err
+	}
+
+	if linuxdoUser.Id == 0 {
+		return nil, errors.New("invalid user info returned")
+	}
+
+	return &linuxdoUser, nil
+}
+
+func LinuxdoOAuth(c *gin.Context) {
+	session := sessions.Default(c)
+
+	errorCode := c.Query("error")
+	if errorCode != "" {
+		errorDescription := c.Query("error_description")
+		c.JSON(http.StatusOK, gin.H{
+			"success": false,
+			"message": errorDescription,
+		})
+		return
+	}
+
+	state := c.Query("state")
+	if state == "" || session.Get("oauth_state") == nil || state != session.Get("oauth_state").(string) {
+		c.JSON(http.StatusForbidden, gin.H{
+			"success": false,
+			"message": "state is empty or not same",
+		})
+		return
+	}
+
+	username := session.Get("username")
+	if username != nil {
+		LinuxDoBind(c)
+		return
+	}
+
+	if !common.LinuxDOOAuthEnabled {
+		c.JSON(http.StatusOK, gin.H{
+			"success": false,
+			"message": "管理员未开启通过 Linux DO 登录以及注册",
+		})
+		return
+	}
+
+	code := c.Query("code")
+	linuxdoUser, err := getLinuxdoUserInfoByCode(code, c)
+	if err != nil {
+		c.JSON(http.StatusOK, gin.H{
+			"success": false,
+			"message": err.Error(),
+		})
+		return
+	}
+
+	user := model.User{
+		LinuxDOId: strconv.Itoa(linuxdoUser.Id),
+	}
+
+	// Check if user exists
+	if model.IsLinuxDOIdAlreadyTaken(user.LinuxDOId) {
+		err := user.FillUserByLinuxDOId()
+		if err != nil {
+			c.JSON(http.StatusOK, gin.H{
+				"success": false,
+				"message": err.Error(),
+			})
+			return
+		}
+		if user.Id == 0 {
+			c.JSON(http.StatusOK, gin.H{
+				"success": false,
+				"message": "用户已注销",
+			})
+			return
+		}
+	} else {
+		if common.RegisterEnabled {
+			user.Username = "linuxdo_" + strconv.Itoa(model.GetMaxUserId()+1)
+			user.DisplayName = linuxdoUser.Name
+			user.Role = common.RoleCommonUser
+			user.Status = common.UserStatusEnabled
+
+			if err := user.Insert(0); err != nil {
+				c.JSON(http.StatusOK, gin.H{
+					"success": false,
+					"message": err.Error(),
+				})
+				return
+			}
+		} else {
+			c.JSON(http.StatusOK, gin.H{
+				"success": false,
+				"message": "管理员关闭了新用户注册",
+			})
+			return
+		}
+	}
+
+	if user.Status != common.UserStatusEnabled {
+		c.JSON(http.StatusOK, gin.H{
+			"message": "用户已被封禁",
+			"success": false,
+		})
+		return
+	}
+
+	setupLogin(&user, c)
+}

controller/misc.go

@@ -38,6 +38,8 @@ func GetStatus(c *gin.Context) {
 			"email_verification":       common.EmailVerificationEnabled,
 			"github_oauth":             common.GitHubOAuthEnabled,
 			"github_client_id":         common.GitHubClientId,
+			"linuxdo_oauth":            common.LinuxDOOAuthEnabled,
+			"linuxdo_client_id":        common.LinuxDOClientId,
 			"telegram_oauth":           common.TelegramOAuthEnabled,
 			"telegram_bot_name":        common.TelegramBotName,
 			"system_name":              common.SystemName,

controller/option.go

@@ -50,6 +50,14 @@ func UpdateOption(c *gin.Context) {
 			})
 			return
 		}
+	case "LinuxDOOAuthEnabled":
+		if option.Value == "true" && common.LinuxDOClientId == "" {
+			c.JSON(http.StatusOK, gin.H{
+				"success": false,
+				"message": "无法启用 LinuxDO OAuth，请先填入 LinuxDO Client Id 以及 LinuxDO Client Secret！",
+			})
+			return
+		}
 	case "EmailDomainRestrictionEnabled":
 		if option.Value == "true" && len(common.EmailDomainWhitelist) == 0 {
 			c.JSON(http.StatusOK, gin.H{

controller/relay.go

@@ -5,6 +5,7 @@ import (
 	"errors"
 	"fmt"
 	"github.com/gin-gonic/gin"
+	"github.com/gorilla/websocket"
 	"io"
 	"log"
 	"net/http"
@@ -38,6 +39,15 @@ func relayHandler(c *gin.Context, relayMode int) *dto.OpenAIErrorWithStatusCode
 	return err
 }
 
+func wsHandler(c *gin.Context, ws *websocket.Conn, relayMode int) *dto.OpenAIErrorWithStatusCode {
+	var err *dto.OpenAIErrorWithStatusCode
+	switch relayMode {
+	default:
+		err = relay.TextHelper(c)
+	}
+	return err
+}
+
 func Playground(c *gin.Context) {
 	var openaiErr *dto.OpenAIErrorWithStatusCode
 
@@ -134,6 +144,67 @@ func Relay(c *gin.Context) {
 	}
 }
 
+var upgrader = websocket.Upgrader{
+	Subprotocols: []string{"realtime"}, // WS 握手支持的协议，如果有使用 Sec-WebSocket-Protocol，则必须在此声明对应的 Protocol TODO add other protocol
+	CheckOrigin: func(r *http.Request) bool {
+		return true // 允许跨域
+	},
+}
+
+func WssRelay(c *gin.Context) {
+	// 将 HTTP 连接升级为 WebSocket 连接
+
+	ws, err := upgrader.Upgrade(c.Writer, c.Request, nil)
+	defer ws.Close()
+
+	if err != nil {
+		openaiErr := service.OpenAIErrorWrapper(err, "get_channel_failed", http.StatusInternalServerError)
+		service.WssError(c, ws, openaiErr.Error)
+		return
+	}
+
+	relayMode := constant.Path2RelayMode(c.Request.URL.Path)
+	requestId := c.GetString(common.RequestIdKey)
+	group := c.GetString("group")
+	//wss://api.openai.com/v1/realtime?model=gpt-4o-realtime-preview-2024-10-01
+	originalModel := c.GetString("original_model")
+	var openaiErr *dto.OpenAIErrorWithStatusCode
+
+	for i := 0; i <= common.RetryTimes; i++ {
+		channel, err := getChannel(c, group, originalModel, i)
+		if err != nil {
+			common.LogError(c, err.Error())
+			openaiErr = service.OpenAIErrorWrapperLocal(err, "get_channel_failed", http.StatusInternalServerError)
+			break
+		}
+
+		openaiErr = wssRequest(c, ws, relayMode, channel)
+
+		if openaiErr == nil {
+			return // 成功处理请求，直接返回
+		}
+
+		go processChannelError(c, channel.Id, channel.Type, channel.Name, channel.GetAutoBan(), openaiErr)
+
+		if !shouldRetry(c, openaiErr, common.RetryTimes-i) {
+			break
+		}
+	}
+	useChannel := c.GetStringSlice("use_channel")
+	if len(useChannel) > 1 {
+		retryLogStr := fmt.Sprintf("重试：%s", strings.Trim(strings.Join(strings.Fields(fmt.Sprint(useChannel)), "->"), "[]"))
+		common.LogInfo(c, retryLogStr)
+	}
+
+	if openaiErr != nil {
+		if openaiErr.StatusCode == http.StatusTooManyRequests {
+			openaiErr.Error.Message = "当前分组上游负载已饱和，请稍后再试"
+		}
+		openaiErr.Error.Message = common.MessageWithRequestId(openaiErr.Error.Message, requestId)
+		service.WssError(c, ws, openaiErr.Error)
+	}
+}
+
 func relayRequest(c *gin.Context, relayMode int, channel *model.Channel) *dto.OpenAIErrorWithStatusCode {
 	addUsedChannel(c, channel.Id)
 	requestBody, _ := common.GetRequestBody(c)
@@ -141,6 +212,13 @@ func relayRequest(c *gin.Context, relayMode int, channel *model.Channel) *dto.Op
 	return relayHandler(c, relayMode)
 }
 
+func wssRequest(c *gin.Context, ws *websocket.Conn, relayMode int, channel *model.Channel) *dto.OpenAIErrorWithStatusCode {
+	addUsedChannel(c, channel.Id)
+	requestBody, _ := common.GetRequestBody(c)
+	c.Request.Body = io.NopCloser(bytes.NewBuffer(requestBody))
+	return relay.WssHelper(c, ws)
+}
+
 func addUsedChannel(c *gin.Context, channelId int) {
 	useChannel := c.GetStringSlice("use_channel")
 	useChannel = append(useChannel, fmt.Sprintf("%d", channelId))

dto/openai_request.go

@@ -34,6 +34,8 @@ type GeneralOpenAIRequest struct {
 	LogProbs            bool           `json:"logprobs,omitempty"`
 	TopLogProbs         int            `json:"top_logprobs,omitempty"`
 	Dimensions          int            `json:"dimensions,omitempty"`
+	Modalities          any            `json:"modalities,omitempty"`
+	Audio               any            `json:"audio,omitempty"`
 }
 
 type OpenAITools struct {
@@ -83,9 +85,10 @@ type Message struct {
 }
 
 type MediaMessage struct {
-	Type     string `json:"type"`
-	Text     string `json:"text"`
-	ImageUrl any    `json:"image_url,omitempty"`
+	Type       string `json:"type"`
+	Text       string `json:"text"`
+	ImageUrl   any    `json:"image_url,omitempty"`
+	InputAudio any    `json:"input_audio,omitempty"`
 }
 
 type MessageImageUrl struct {
@@ -93,9 +96,15 @@ type MessageImageUrl struct {
 	Detail string `json:"detail"`
 }
 
+type MessageInputAudio struct {
+	Data   string `json:"data"` //base64
+	Format string `json:"format"`
+}
+
 const (
-	ContentTypeText     = "text"
-	ContentTypeImageURL = "image_url"
+	ContentTypeText       = "text"
+	ContentTypeImageURL   = "image_url"
+	ContentTypeInputAudio = "input_audio"
 )
 
 func (m Message) StringContent() string {
@@ -168,11 +177,19 @@ func (m Message) ParseContent() []MediaMessage {
 						},
 					})
 				}
-
+			case ContentTypeInputAudio:
+				if subObj, ok := contentMap["input_audio"].(map[string]any); ok {
+					contentList = append(contentList, MediaMessage{
+						Type: ContentTypeInputAudio,
+						InputAudio: MessageInputAudio{
+							Data:   subObj["data"].(string),
+							Format: subObj["format"].(string),
+						},
+					})
+				}
 			}
 		}
 		return contentList
 	}
-
 	return nil
 }

dto/openai_response.go

@@ -128,7 +128,9 @@ type CompletionsStreamResponse struct {
 }
 
 type Usage struct {
-	PromptTokens     int `json:"prompt_tokens"`
-	CompletionTokens int `json:"completion_tokens"`
-	TotalTokens      int `json:"total_tokens"`
+	PromptTokens           int                `json:"prompt_tokens"`
+	CompletionTokens       int                `json:"completion_tokens"`
+	TotalTokens            int                `json:"total_tokens"`
+	PromptTokensDetails    InputTokenDetails  `json:"prompt_tokens_details"`
+	CompletionTokenDetails OutputTokenDetails `json:"completion_tokens_details"`
 }

dto/realtime.go

@@ -0,0 +1,98 @@
+package dto
+
+const (
+	RealtimeEventTypeError              = "error"
+	RealtimeEventTypeSessionUpdate      = "session.update"
+	RealtimeEventTypeConversationCreate = "conversation.item.create"
+	RealtimeEventTypeResponseCreate     = "response.create"
+	RealtimeEventInputAudioBufferAppend = "input_audio_buffer.append"
+)
+
+const (
+	RealtimeEventTypeResponseDone                   = "response.done"
+	RealtimeEventTypeSessionUpdated                 = "session.updated"
+	RealtimeEventTypeSessionCreated                 = "session.created"
+	RealtimeEventResponseAudioDelta                 = "response.audio.delta"
+	RealtimeEventResponseAudioTranscriptionDelta    = "response.audio_transcript.delta"
+	RealtimeEventResponseFunctionCallArgumentsDelta = "response.function_call_arguments.delta"
+	RealtimeEventResponseFunctionCallArgumentsDone  = "response.function_call_arguments.done"
+	RealtimeEventConversationItemCreated            = "conversation.item.created"
+)
+
+type RealtimeEvent struct {
+	EventId string `json:"event_id"`
+	Type    string `json:"type"`
+	//PreviousItemId string `json:"previous_item_id"`
+	Session  *RealtimeSession  `json:"session,omitempty"`
+	Item     *RealtimeItem     `json:"item,omitempty"`
+	Error    *OpenAIError      `json:"error,omitempty"`
+	Response *RealtimeResponse `json:"response,omitempty"`
+	Delta    string            `json:"delta,omitempty"`
+	Audio    string            `json:"audio,omitempty"`
+}
+
+type RealtimeResponse struct {
+	Usage *RealtimeUsage `json:"usage"`
+}
+
+type RealtimeUsage struct {
+	TotalTokens        int                `json:"total_tokens"`
+	InputTokens        int                `json:"input_tokens"`
+	OutputTokens       int                `json:"output_tokens"`
+	InputTokenDetails  InputTokenDetails  `json:"input_token_details"`
+	OutputTokenDetails OutputTokenDetails `json:"output_token_details"`
+}
+
+type InputTokenDetails struct {
+	CachedTokens int `json:"cached_tokens"`
+	TextTokens   int `json:"text_tokens"`
+	AudioTokens  int `json:"audio_tokens"`
+	ImageTokens  int `json:"image_tokens"`
+}
+
+type OutputTokenDetails struct {
+	TextTokens  int `json:"text_tokens"`
+	AudioTokens int `json:"audio_tokens"`
+}
+
+type RealtimeSession struct {
+	Modalities              []string                `json:"modalities"`
+	Instructions            string                  `json:"instructions"`
+	Voice                   string                  `json:"voice"`
+	InputAudioFormat        string                  `json:"input_audio_format"`
+	OutputAudioFormat       string                  `json:"output_audio_format"`
+	InputAudioTranscription InputAudioTranscription `json:"input_audio_transcription"`
+	TurnDetection           interface{}             `json:"turn_detection"`
+	Tools                   []RealTimeTool          `json:"tools"`
+	ToolChoice              string                  `json:"tool_choice"`
+	Temperature             float64                 `json:"temperature"`
+	//MaxResponseOutputTokens int                     `json:"max_response_output_tokens"`
+}
+
+type InputAudioTranscription struct {
+	Model string `json:"model"`
+}
+
+type RealTimeTool struct {
+	Type        string `json:"type"`
+	Name        string `json:"name"`
+	Description string `json:"description"`
+	Parameters  any    `json:"parameters"`
+}
+
+type RealtimeItem struct {
+	Id        string            `json:"id"`
+	Type      string            `json:"type"`
+	Status    string            `json:"status"`
+	Role      string            `json:"role"`
+	Content   []RealtimeContent `json:"content"`
+	Name      *string           `json:"name,omitempty"`
+	ToolCalls any               `json:"tool_calls,omitempty"`
+	CallId    string            `json:"call_id,omitempty"`
+}
+type RealtimeContent struct {
+	Type       string `json:"type"`
+	Text       string `json:"text,omitempty"`
+	Audio      string `json:"audio,omitempty"` // Base64-encoded audio bytes.
+	Transcript string `json:"transcript,omitempty"`
+}

go.mod

@@ -63,6 +63,7 @@ require (
 	github.com/jackc/puddle/v2 v2.2.1 // indirect
 	github.com/jinzhu/inflection v1.0.0 // indirect
 	github.com/jinzhu/now v1.1.5 // indirect
+	github.com/joho/godotenv v1.5.1 // indirect
 	github.com/json-iterator/go v1.1.12 // indirect
 	github.com/klauspost/cpuid/v2 v2.2.4 // indirect
 	github.com/leodido/go-urn v1.4.0 // indirect

go.sum

@@ -111,6 +111,8 @@ github.com/jinzhu/inflection v1.0.0/go.mod h1:h+uFLlag+Qp1Va5pdKtLDYj+kHp5pxUVkr
 github.com/jinzhu/now v1.1.4/go.mod h1:d3SSVoowX0Lcu0IBviAWJpolVfI5UJVZZ7cO71lE/z8=
 github.com/jinzhu/now v1.1.5 h1:/o9tlHleP7gOFmsnYNz3RGnqzefHA47wQpKrrdTIwXQ=
 github.com/jinzhu/now v1.1.5/go.mod h1:d3SSVoowX0Lcu0IBviAWJpolVfI5UJVZZ7cO71lE/z8=
+github.com/joho/godotenv v1.5.1 h1:7eLL/+HRGLY0ldzfGMeQkb7vMd0as4CfYvUVzLqw0N0=
+github.com/joho/godotenv v1.5.1/go.mod h1:f4LDr5Voq0i2e/R5DDNOoa2zzDfwtkZa6DnEwAbqwq4=
 github.com/json-iterator/go v1.1.9/go.mod h1:KdQUCv79m/52Kvf8AW2vK1V8akMuk1QjK/uOdHXbAo4=
 github.com/json-iterator/go v1.1.12 h1:PV8peI4a0ysnczrg+LtxykD8LfKY9ML6u2jnxaEnrnM=
 github.com/json-iterator/go v1.1.12/go.mod h1:e30LSqwooZae/UwlEbR2852Gd8hjQvJoHmT4TnhNGBo=

main.go

@@ -3,10 +3,6 @@ package main
 import (
 	"embed"
 	"fmt"
-	"github.com/bytedance/gopkg/util/gopool"
-	"github.com/gin-contrib/sessions"
-	"github.com/gin-contrib/sessions/cookie"
-	"github.com/gin-gonic/gin"
 	"log"
 	"net/http"
 	"one-api/common"
@@ -19,6 +15,12 @@ import (
 	"os"
 	"strconv"
 
+	"github.com/bytedance/gopkg/util/gopool"
+	"github.com/gin-contrib/sessions"
+	"github.com/gin-contrib/sessions/cookie"
+	"github.com/gin-gonic/gin"
+	"github.com/joho/godotenv"
+
 	_ "net/http/pprof"
 )
 
@@ -29,6 +31,11 @@ var buildFS embed.FS
 var indexPage []byte
 
 func main() {
+	err := godotenv.Load(".env")
+	if err != nil {
+		common.SysLog("Can't load .env file")
+	}
+
 	common.SetupLogger()
 	common.SysLog("New API " + common.Version + " started")
 	if os.Getenv("GIN_MODE") != "debug" {
@@ -38,7 +45,7 @@ func main() {
 		common.SysLog("running in debug mode")
 	}
 	// Initialize SQL Database
-	err := model.InitDB()
+	err = model.InitDB()
 	if err != nil {
 		common.FatalLog("failed to initialize database: " + err.Error())
 	}

middleware/auth.go

@@ -155,8 +155,27 @@ func RootAuth() func(c *gin.Context) {
 	}
 }
 
+func WssAuth(c *gin.Context) {
+
+}
+
 func TokenAuth() func(c *gin.Context) {
 	return func(c *gin.Context) {
+		// 先检测是否为ws
+		if c.Request.Header.Get("Sec-WebSocket-Protocol") != "" {
+			// Sec-WebSocket-Protocol: realtime, openai-insecure-api-key.sk-xxx, openai-beta.realtime-v1
+			// read sk from Sec-WebSocket-Protocol
+			key := c.Request.Header.Get("Sec-WebSocket-Protocol")
+			parts := strings.Split(key, ",")
+			for _, part := range parts {
+				part = strings.TrimSpace(part)
+				if strings.HasPrefix(part, "openai-insecure-api-key") {
+					key = strings.TrimPrefix(part, "openai-insecure-api-key.")
+					break
+				}
+			}
+			c.Request.Header.Set("Authorization", "Bearer "+key)
+		}
 		key := c.Request.Header.Get("Authorization")
 		parts := make([]string, 0)
 		key = strings.TrimPrefix(key, "Bearer ")

middleware/distributor.go

@@ -170,6 +170,10 @@ func getModelRequest(c *gin.Context) (*ModelRequest, bool, error) {
 		abortWithOpenAiMessage(c, http.StatusBadRequest, "无效的请求, "+err.Error())
 		return nil, false, errors.New("无效的请求, " + err.Error())
 	}
+	if strings.HasPrefix(c.Request.URL.Path, "/v1/realtime") {
+		//wss://api.openai.com/v1/realtime?model=gpt-4o-realtime-preview-2024-10-01
+		modelRequest.Model = c.Query("model")
+	}
 	if strings.HasPrefix(c.Request.URL.Path, "/v1/moderations") {
 		if modelRequest.Model == "" {
 			modelRequest.Model = "text-moderation-stable"

model/option.go

@@ -31,6 +31,7 @@ func InitOptionMap() {
 	common.OptionMap["PasswordRegisterEnabled"] = strconv.FormatBool(common.PasswordRegisterEnabled)
 	common.OptionMap["EmailVerificationEnabled"] = strconv.FormatBool(common.EmailVerificationEnabled)
 	common.OptionMap["GitHubOAuthEnabled"] = strconv.FormatBool(common.GitHubOAuthEnabled)
+	common.OptionMap["LinuxDOOAuthEnabled"] = strconv.FormatBool(common.LinuxDOOAuthEnabled)
 	common.OptionMap["TelegramOAuthEnabled"] = strconv.FormatBool(common.TelegramOAuthEnabled)
 	common.OptionMap["WeChatAuthEnabled"] = strconv.FormatBool(common.WeChatAuthEnabled)
 	common.OptionMap["TurnstileCheckEnabled"] = strconv.FormatBool(common.TurnstileCheckEnabled)
@@ -175,6 +176,8 @@ func updateOptionMap(key string, value string) (err error) {
 			common.EmailVerificationEnabled = boolValue
 		case "GitHubOAuthEnabled":
 			common.GitHubOAuthEnabled = boolValue
+		case "LinuxDOOAuthEnabled":
+			common.LinuxDOOAuthEnabled = boolValue
 		case "WeChatAuthEnabled":
 			common.WeChatAuthEnabled = boolValue
 		case "TelegramOAuthEnabled":
@@ -267,6 +270,10 @@ func updateOptionMap(key string, value string) (err error) {
 		common.GitHubClientId = value
 	case "GitHubClientSecret":
 		common.GitHubClientSecret = value
+	case "LinuxDOClientId":
+		common.LinuxDOClientId = value
+	case "LinuxDOClientSecret":
+		common.LinuxDOClientSecret = value
 	case "Footer":
 		common.Footer = value
 	case "SystemName":

model/user.go

@@ -36,6 +36,7 @@ type User struct {
 	AffHistoryQuota  int            `json:"aff_history_quota" gorm:"type:int;default:0;column:aff_history"` // 邀请历史额度
 	InviterId        int            `json:"inviter_id" gorm:"type:int;column:inviter_id;index"`
 	DeletedAt        gorm.DeletedAt `gorm:"index"`
+	LinuxDOId        string         `json:"linux_do_id" gorm:"column:linux_do_id;index"`
 }
 
 func (user *User) GetAccessToken() string {
@@ -537,3 +538,17 @@ func GetUsernameById(id int) (username string, err error) {
 	err = DB.Model(&User{}).Where("id = ?", id).Select("username").Find(&username).Error
 	return username, err
 }
+
+func IsLinuxDOIdAlreadyTaken(linuxDOId string) bool {
+	var user User
+	err := DB.Unscoped().Where("linux_do_id = ?", linuxDOId).First(&user).Error
+	return !errors.Is(err, gorm.ErrRecordNotFound)
+}
+
+func (u *User) FillUserByLinuxDOId() error {
+	if u.LinuxDOId == "" {
+		return errors.New("linux do id is empty")
+	}
+	err := DB.Where("linux_do_id = ?", u.LinuxDOId).First(u).Error
+	return err
+}

relay/channel/adapter.go

@@ -12,13 +12,13 @@ type Adaptor interface {
 	// Init IsStream bool
 	Init(info *relaycommon.RelayInfo)
 	GetRequestURL(info *relaycommon.RelayInfo) (string, error)
-	SetupRequestHeader(c *gin.Context, req *http.Request, info *relaycommon.RelayInfo) error
+	SetupRequestHeader(c *gin.Context, req *http.Header, info *relaycommon.RelayInfo) error
 	ConvertRequest(c *gin.Context, info *relaycommon.RelayInfo, request *dto.GeneralOpenAIRequest) (any, error)
 	ConvertRerankRequest(c *gin.Context, relayMode int, request dto.RerankRequest) (any, error)
 	ConvertAudioRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.AudioRequest) (io.Reader, error)
 	ConvertImageRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.ImageRequest) (any, error)
-	DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (*http.Response, error)
-	DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage *dto.Usage, err *dto.OpenAIErrorWithStatusCode)
+	DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error)
+	DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage any, err *dto.OpenAIErrorWithStatusCode)
 	GetModelList() []string
 	GetChannelName() string
 }

relay/channel/ali/adaptor.go

@@ -32,14 +32,14 @@ func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {
 	return fullRequestURL, nil
 }
 
-func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, info *relaycommon.RelayInfo) error {
+func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Header, info *relaycommon.RelayInfo) error {
 	channel.SetupApiRequestHeader(info, c, req)
-	req.Header.Set("Authorization", "Bearer "+info.ApiKey)
+	req.Set("Authorization", "Bearer "+info.ApiKey)
 	if info.IsStream {
-		req.Header.Set("X-DashScope-SSE", "enable")
+		req.Set("X-DashScope-SSE", "enable")
 	}
 	if c.GetString("plugin") != "" {
-		req.Header.Set("X-DashScope-Plugin", c.GetString("plugin"))
+		req.Set("X-DashScope-Plugin", c.GetString("plugin"))
 	}
 	return nil
 }
@@ -72,11 +72,11 @@ func (a *Adaptor) ConvertAudioRequest(c *gin.Context, info *relaycommon.RelayInf
 	return nil, errors.New("not implemented")
 }
 
-func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (*http.Response, error) {
+func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	return channel.DoApiRequest(a, c, info, requestBody)
 }
 
-func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage *dto.Usage, err *dto.OpenAIErrorWithStatusCode) {
+func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage any, err *dto.OpenAIErrorWithStatusCode) {
 	switch info.RelayMode {
 	case constant.RelayModeImagesGenerations:
 		err, usage = aliImageHandler(c, resp, info)

relay/channel/api_request.go

@@ -4,6 +4,7 @@ import (
 	"errors"
 	"fmt"
 	"github.com/gin-gonic/gin"
+	"github.com/gorilla/websocket"
 	"io"
 	"net/http"
 	"one-api/relay/common"
@@ -11,14 +12,16 @@ import (
 	"one-api/service"
 )
 
-func SetupApiRequestHeader(info *common.RelayInfo, c *gin.Context, req *http.Request) {
+func SetupApiRequestHeader(info *common.RelayInfo, c *gin.Context, req *http.Header) {
 	if info.RelayMode == constant.RelayModeAudioTranscription || info.RelayMode == constant.RelayModeAudioTranslation {
 		// multipart/form-data
+	} else if info.RelayMode == constant.RelayModeRealtime {
+		// websocket
 	} else {
-		req.Header.Set("Content-Type", c.Request.Header.Get("Content-Type"))
-		req.Header.Set("Accept", c.Request.Header.Get("Accept"))
+		req.Set("Content-Type", c.Request.Header.Get("Content-Type"))
+		req.Set("Accept", c.Request.Header.Get("Accept"))
 		if info.IsStream && c.Request.Header.Get("Accept") == "" {
-			req.Header.Set("Accept", "text/event-stream")
+			req.Set("Accept", "text/event-stream")
 		}
 	}
 }
@@ -32,7 +35,7 @@ func DoApiRequest(a Adaptor, c *gin.Context, info *common.RelayInfo, requestBody
 	if err != nil {
 		return nil, fmt.Errorf("new request failed: %w", err)
 	}
-	err = a.SetupRequestHeader(c, req, info)
+	err = a.SetupRequestHeader(c, &req.Header, info)
 	if err != nil {
 		return nil, fmt.Errorf("setup request header failed: %w", err)
 	}
@@ -55,7 +58,7 @@ func DoFormRequest(a Adaptor, c *gin.Context, info *common.RelayInfo, requestBod
 	// set form data
 	req.Header.Set("Content-Type", c.Request.Header.Get("Content-Type"))
 
-	err = a.SetupRequestHeader(c, req, info)
+	err = a.SetupRequestHeader(c, &req.Header, info)
 	if err != nil {
 		return nil, fmt.Errorf("setup request header failed: %w", err)
 	}
@@ -66,6 +69,27 @@ func DoFormRequest(a Adaptor, c *gin.Context, info *common.RelayInfo, requestBod
 	return resp, nil
 }
 
+func DoWssRequest(a Adaptor, c *gin.Context, info *common.RelayInfo, requestBody io.Reader) (*websocket.Conn, error) {
+	fullRequestURL, err := a.GetRequestURL(info)
+	if err != nil {
+		return nil, fmt.Errorf("get request url failed: %w", err)
+	}
+	targetHeader := http.Header{}
+	err = a.SetupRequestHeader(c, &targetHeader, info)
+	if err != nil {
+		return nil, fmt.Errorf("setup request header failed: %w", err)
+	}
+	targetHeader.Set("Content-Type", c.Request.Header.Get("Content-Type"))
+	targetConn, _, err := websocket.DefaultDialer.Dial(fullRequestURL, targetHeader)
+	if err != nil {
+		return nil, fmt.Errorf("dial failed to %s: %w", fullRequestURL, err)
+	}
+	// send request body
+	//all, err := io.ReadAll(requestBody)
+	//err = service.WssString(c, targetConn, string(all))
+	return targetConn, nil
+}
+
 func doRequest(c *gin.Context, req *http.Request) (*http.Response, error) {
 	resp, err := service.GetHttpClient().Do(req)
 	if err != nil {

relay/channel/aws/adaptor.go

@@ -37,7 +37,7 @@ func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {
 	return "", nil
 }
 
-func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, info *relaycommon.RelayInfo) error {
+func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Header, info *relaycommon.RelayInfo) error {
 	return nil
 }
 
@@ -59,11 +59,11 @@ func (a *Adaptor) ConvertRerankRequest(c *gin.Context, relayMode int, request dt
 	return nil, nil
 }
 
-func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (*http.Response, error) {
+func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	return nil, nil
 }
 
-func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage *dto.Usage, err *dto.OpenAIErrorWithStatusCode) {
+func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage any, err *dto.OpenAIErrorWithStatusCode) {
 	if info.IsStream {
 		err, usage = awsStreamHandler(c, resp, info, a.RequestMode)
 	} else {

relay/channel/baidu/adaptor.go

@@ -98,9 +98,9 @@ func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {
 	return fullRequestURL, nil
 }
 
-func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, info *relaycommon.RelayInfo) error {
+func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Header, info *relaycommon.RelayInfo) error {
 	channel.SetupApiRequestHeader(info, c, req)
-	req.Header.Set("Authorization", "Bearer "+info.ApiKey)
+	req.Set("Authorization", "Bearer "+info.ApiKey)
 	return nil
 }
 
@@ -122,11 +122,11 @@ func (a *Adaptor) ConvertRerankRequest(c *gin.Context, relayMode int, request dt
 	return nil, nil
 }
 
-func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (*http.Response, error) {
+func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	return channel.DoApiRequest(a, c, info, requestBody)
 }
 
-func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage *dto.Usage, err *dto.OpenAIErrorWithStatusCode) {
+func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage any, err *dto.OpenAIErrorWithStatusCode) {
 	if info.IsStream {
 		err, usage = baiduStreamHandler(c, resp)
 	} else {

relay/channel/claude/adaptor.go

@@ -47,14 +47,14 @@ func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {
 	}
 }
 
-func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, info *relaycommon.RelayInfo) error {
+func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Header, info *relaycommon.RelayInfo) error {
 	channel.SetupApiRequestHeader(info, c, req)
-	req.Header.Set("x-api-key", info.ApiKey)
+	req.Set("x-api-key", info.ApiKey)
 	anthropicVersion := c.Request.Header.Get("anthropic-version")
 	if anthropicVersion == "" {
 		anthropicVersion = "2023-06-01"
 	}
-	req.Header.Set("anthropic-version", anthropicVersion)
+	req.Set("anthropic-version", anthropicVersion)
 	return nil
 }
 
@@ -73,11 +73,11 @@ func (a *Adaptor) ConvertRerankRequest(c *gin.Context, relayMode int, request dt
 	return nil, nil
 }
 
-func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (*http.Response, error) {
+func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	return channel.DoApiRequest(a, c, info, requestBody)
 }
 
-func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage *dto.Usage, err *dto.OpenAIErrorWithStatusCode) {
+func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage any, err *dto.OpenAIErrorWithStatusCode) {
 	if info.IsStream {
 		err, usage = ClaudeStreamHandler(c, resp, info, a.RequestMode)
 	} else {

relay/channel/claude/constants.go

@@ -8,6 +8,7 @@ var ModelList = []string{
 	"claude-3-sonnet-20240229",
 	"claude-3-opus-20240229",
 	"claude-3-haiku-20240307",
+	"claude-3-5-haiku-20241022",
 	"claude-3-5-sonnet-20240620",
 	"claude-3-5-sonnet-20241022",
 }

relay/channel/claude/relay-claude.go

@@ -509,7 +509,7 @@ func ClaudeHandler(c *gin.Context, resp *http.Response, requestMode int, info *r
 		}, nil
 	}
 	fullTextResponse := ResponseClaude2OpenAI(requestMode, &claudeResponse)
-	completionTokens, err := service.CountTokenText(claudeResponse.Completion, info.OriginModelName)
+	completionTokens, err := service.CountTextToken(claudeResponse.Completion, info.OriginModelName)
 	if err != nil {
 		return service.OpenAIErrorWrapper(err, "count_token_text_failed", http.StatusInternalServerError), nil
 	}

relay/channel/cloudflare/adaptor.go

@@ -30,9 +30,9 @@ func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {
 	}
 }
 
-func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, info *relaycommon.RelayInfo) error {
+func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Header, info *relaycommon.RelayInfo) error {
 	channel.SetupApiRequestHeader(info, c, req)
-	req.Header.Set("Authorization", fmt.Sprintf("Bearer %s", info.ApiKey))
+	req.Set("Authorization", fmt.Sprintf("Bearer %s", info.ApiKey))
 	return nil
 }
 
@@ -48,7 +48,7 @@ func (a *Adaptor) ConvertRequest(c *gin.Context, info *relaycommon.RelayInfo, re
 	}
 }
 
-func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (*http.Response, error) {
+func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	return channel.DoApiRequest(a, c, info, requestBody)
 }
 
@@ -78,7 +78,7 @@ func (a *Adaptor) ConvertImageRequest(c *gin.Context, info *relaycommon.RelayInf
 	return nil, errors.New("not implemented")
 }
 
-func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage *dto.Usage, err *dto.OpenAIErrorWithStatusCode) {
+func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage any, err *dto.OpenAIErrorWithStatusCode) {
 	switch info.RelayMode {
 	case constant.RelayModeEmbeddings:
 		fallthrough

relay/channel/cloudflare/relay_cloudflare.go

@@ -149,7 +149,7 @@ func cfSTTHandler(c *gin.Context, resp *http.Response, info *relaycommon.RelayIn
 
 	usage := &dto.Usage{}
 	usage.PromptTokens = info.PromptTokens
-	usage.CompletionTokens, _ = service.CountTokenText(cfResp.Result.Text, info.UpstreamModelName)
+	usage.CompletionTokens, _ = service.CountTextToken(cfResp.Result.Text, info.UpstreamModelName)
 	usage.TotalTokens = usage.PromptTokens + usage.CompletionTokens
 
 	return nil, usage

relay/channel/cohere/adaptor.go

@@ -36,9 +36,9 @@ func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {
 	}
 }
 
-func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, info *relaycommon.RelayInfo) error {
+func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Header, info *relaycommon.RelayInfo) error {
 	channel.SetupApiRequestHeader(info, c, req)
-	req.Header.Set("Authorization", fmt.Sprintf("Bearer %s", info.ApiKey))
+	req.Set("Authorization", fmt.Sprintf("Bearer %s", info.ApiKey))
 	return nil
 }
 
@@ -46,7 +46,7 @@ func (a *Adaptor) ConvertRequest(c *gin.Context, info *relaycommon.RelayInfo, re
 	return requestOpenAI2Cohere(*request), nil
 }
 
-func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (*http.Response, error) {
+func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	return channel.DoApiRequest(a, c, info, requestBody)
 }
 
@@ -54,7 +54,7 @@ func (a *Adaptor) ConvertRerankRequest(c *gin.Context, relayMode int, request dt
 	return requestConvertRerank2Cohere(request), nil
 }
 
-func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage *dto.Usage, err *dto.OpenAIErrorWithStatusCode) {
+func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage any, err *dto.OpenAIErrorWithStatusCode) {
 	if info.RelayMode == constant.RelayModeRerank {
 		err, usage = cohereRerankHandler(c, resp, info)
 	} else {

relay/channel/dify/adaptor.go

@@ -31,9 +31,9 @@ func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {
 	return fmt.Sprintf("%s/v1/chat-messages", info.BaseUrl), nil
 }
 
-func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, info *relaycommon.RelayInfo) error {
+func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Header, info *relaycommon.RelayInfo) error {
 	channel.SetupApiRequestHeader(info, c, req)
-	req.Header.Set("Authorization", "Bearer "+info.ApiKey)
+	req.Set("Authorization", "Bearer "+info.ApiKey)
 	return nil
 }
 
@@ -48,11 +48,11 @@ func (a *Adaptor) ConvertRerankRequest(c *gin.Context, relayMode int, request dt
 	return nil, nil
 }
 
-func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (*http.Response, error) {
+func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	return channel.DoApiRequest(a, c, info, requestBody)
 }
 
-func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage *dto.Usage, err *dto.OpenAIErrorWithStatusCode) {
+func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage any, err *dto.OpenAIErrorWithStatusCode) {
 	if info.IsStream {
 		err, usage = difyStreamHandler(c, resp, info)
 	} else {

relay/channel/dify/relay-dify.go

@@ -108,7 +108,7 @@ func difyStreamHandler(c *gin.Context, resp *http.Response, info *relaycommon.Re
 	}
 	if usage.TotalTokens == 0 {
 		usage.PromptTokens = info.PromptTokens
-		usage.CompletionTokens, _ = service.CountTokenText("gpt-3.5-turbo", responseText)
+		usage.CompletionTokens, _ = service.CountTextToken("gpt-3.5-turbo", responseText)
 		usage.TotalTokens = usage.PromptTokens + usage.CompletionTokens
 	}
 	return nil, usage

relay/channel/gemini/adaptor.go

@@ -47,9 +47,9 @@ func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {
 	return fmt.Sprintf("%s/%s/models/%s:%s", info.BaseUrl, version, info.UpstreamModelName, action), nil
 }
 
-func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, info *relaycommon.RelayInfo) error {
+func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Header, info *relaycommon.RelayInfo) error {
 	channel.SetupApiRequestHeader(info, c, req)
-	req.Header.Set("x-goog-api-key", info.ApiKey)
+	req.Set("x-goog-api-key", info.ApiKey)
 	return nil
 }
 
@@ -64,11 +64,11 @@ func (a *Adaptor) ConvertRerankRequest(c *gin.Context, relayMode int, request dt
 	return nil, nil
 }
 
-func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (*http.Response, error) {
+func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	return channel.DoApiRequest(a, c, info, requestBody)
 }
 
-func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage *dto.Usage, err *dto.OpenAIErrorWithStatusCode) {
+func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage any, err *dto.OpenAIErrorWithStatusCode) {
 	if info.IsStream {
 		err, usage = GeminiChatStreamHandler(c, resp, info)
 	} else {

relay/channel/jina/adaptor.go

@@ -37,9 +37,9 @@ func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {
 	return "", errors.New("invalid relay mode")
 }
 
-func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, info *relaycommon.RelayInfo) error {
+func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Header, info *relaycommon.RelayInfo) error {
 	channel.SetupApiRequestHeader(info, c, req)
-	req.Header.Set("Authorization", fmt.Sprintf("Bearer %s", info.ApiKey))
+	req.Set("Authorization", fmt.Sprintf("Bearer %s", info.ApiKey))
 	return nil
 }
 
@@ -47,7 +47,7 @@ func (a *Adaptor) ConvertRequest(c *gin.Context, info *relaycommon.RelayInfo, re
 	return request, nil
 }
 
-func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (*http.Response, error) {
+func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	return channel.DoApiRequest(a, c, info, requestBody)
 }
 
@@ -55,7 +55,7 @@ func (a *Adaptor) ConvertRerankRequest(c *gin.Context, relayMode int, request dt
 	return request, nil
 }
 
-func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage *dto.Usage, err *dto.OpenAIErrorWithStatusCode) {
+func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage any, err *dto.OpenAIErrorWithStatusCode) {
 	if info.RelayMode == constant.RelayModeRerank {
 		err, usage = jinaRerankHandler(c, resp)
 	} else if info.RelayMode == constant.RelayModeEmbeddings {

relay/channel/mistral/adaptor.go

@@ -0,0 +1,72 @@
+package mistral
+
+import (
+	"errors"
+	"github.com/gin-gonic/gin"
+	"io"
+	"net/http"
+	"one-api/dto"
+	"one-api/relay/channel"
+	"one-api/relay/channel/openai"
+	relaycommon "one-api/relay/common"
+)
+
+type Adaptor struct {
+}
+
+func (a *Adaptor) ConvertAudioRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.AudioRequest) (io.Reader, error) {
+	//TODO implement me
+	return nil, errors.New("not implemented")
+}
+
+func (a *Adaptor) ConvertImageRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.ImageRequest) (any, error) {
+	//TODO implement me
+	return nil, errors.New("not implemented")
+}
+
+func (a *Adaptor) Init(info *relaycommon.RelayInfo) {
+}
+
+func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {
+	return relaycommon.GetFullRequestURL(info.BaseUrl, info.RequestURLPath, info.ChannelType), nil
+}
+
+func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Header, info *relaycommon.RelayInfo) error {
+	channel.SetupApiRequestHeader(info, c, req)
+	req.Set("Authorization", "Bearer "+info.ApiKey)
+	return nil
+}
+
+func (a *Adaptor) ConvertRequest(c *gin.Context, info *relaycommon.RelayInfo, request *dto.GeneralOpenAIRequest) (any, error) {
+	if request == nil {
+		return nil, errors.New("request is nil")
+	}
+	mistralReq := requestOpenAI2Mistral(*request)
+	//common.LogJson(c, "body", mistralReq)
+	return mistralReq, nil
+}
+
+func (a *Adaptor) ConvertRerankRequest(c *gin.Context, relayMode int, request dto.RerankRequest) (any, error) {
+	return nil, nil
+}
+
+func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
+	return channel.DoApiRequest(a, c, info, requestBody)
+}
+
+func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage any, err *dto.OpenAIErrorWithStatusCode) {
+	if info.IsStream {
+		err, usage = openai.OaiStreamHandler(c, resp, info)
+	} else {
+		err, usage = openai.OpenaiHandler(c, resp, info.PromptTokens, info.UpstreamModelName)
+	}
+	return
+}
+
+func (a *Adaptor) GetModelList() []string {
+	return ModelList
+}
+
+func (a *Adaptor) GetChannelName() string {
+	return ChannelName
+}

relay/channel/mistral/constants.go

@@ -0,0 +1,12 @@
+package mistral
+
+var ModelList = []string{
+	"open-mistral-7b",
+	"open-mixtral-8x7b",
+	"mistral-small-latest",
+	"mistral-medium-latest",
+	"mistral-large-latest",
+	"mistral-embed",
+}
+
+var ChannelName = "mistral"

relay/channel/mistral/text.go

@@ -0,0 +1,40 @@
+package mistral
+
+import (
+	"encoding/json"
+	"one-api/dto"
+)
+
+func requestOpenAI2Mistral(request dto.GeneralOpenAIRequest) *dto.GeneralOpenAIRequest {
+	messages := make([]dto.Message, 0, len(request.Messages))
+	for _, message := range request.Messages {
+		if !message.IsStringContent() {
+			mediaMessages := message.ParseContent()
+			for j, mediaMessage := range mediaMessages {
+				if mediaMessage.Type == dto.ContentTypeImageURL {
+					imageUrl := mediaMessage.ImageUrl.(dto.MessageImageUrl)
+					mediaMessage.ImageUrl = imageUrl.Url
+					mediaMessages[j] = mediaMessage
+				}
+			}
+			messageRaw, _ := json.Marshal(mediaMessages)
+			message.Content = messageRaw
+		}
+		messages = append(messages, dto.Message{
+			Role:       message.Role,
+			Content:    message.Content,
+			ToolCalls:  message.ToolCalls,
+			ToolCallId: message.ToolCallId,
+		})
+	}
+	return &dto.GeneralOpenAIRequest{
+		Model:       request.Model,
+		Stream:      request.Stream,
+		Messages:    messages,
+		Temperature: request.Temperature,
+		TopP:        request.TopP,
+		MaxTokens:   request.MaxTokens,
+		Tools:       request.Tools,
+		ToolChoice:  request.ToolChoice,
+	}
+}

relay/channel/ollama/adaptor.go

@@ -31,13 +31,13 @@ func (a *Adaptor) Init(info *relaycommon.RelayInfo) {
 func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {
 	switch info.RelayMode {
 	case relayconstant.RelayModeEmbeddings:
-		return info.BaseUrl + "/api/embeddings", nil
+		return info.BaseUrl + "/api/embed", nil
 	default:
 		return relaycommon.GetFullRequestURL(info.BaseUrl, info.RequestURLPath, info.ChannelType), nil
 	}
 }
 
-func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, info *relaycommon.RelayInfo) error {
+func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Header, info *relaycommon.RelayInfo) error {
 	channel.SetupApiRequestHeader(info, c, req)
 	return nil
 }
@@ -58,11 +58,11 @@ func (a *Adaptor) ConvertRerankRequest(c *gin.Context, relayMode int, request dt
 	return nil, nil
 }
 
-func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (*http.Response, error) {
+func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	return channel.DoApiRequest(a, c, info, requestBody)
 }
 
-func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage *dto.Usage, err *dto.OpenAIErrorWithStatusCode) {
+func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage any, err *dto.OpenAIErrorWithStatusCode) {
 	if info.IsStream {
 		err, usage = openai.OaiStreamHandler(c, resp, info)
 	} else {

relay/channel/ollama/dto.go

@@ -37,5 +37,5 @@ type OllamaEmbeddingRequest struct {
 type OllamaEmbeddingResponse struct {
 	Error     string    `json:"error,omitempty"`
 	Model     string    `json:"model"`
-	Embedding []float64 `json:"embedding,omitempty"`
+	Embedding [][]float64 `json:"embeddings,omitempty"`
 }

relay/channel/ollama/relay-ollama.go

@@ -73,9 +73,10 @@ func ollamaEmbeddingHandler(c *gin.Context, resp *http.Response, promptTokens in
 	if ollamaEmbeddingResponse.Error != "" {
 		return service.OpenAIErrorWrapper(err, "ollama_error", resp.StatusCode), nil
 	}
+	flattenedEmbeddings := flattenEmbeddings(ollamaEmbeddingResponse.Embedding)
 	data := make([]dto.OpenAIEmbeddingResponseItem, 0, 1)
 	data = append(data, dto.OpenAIEmbeddingResponseItem{
-		Embedding: ollamaEmbeddingResponse.Embedding,
+		Embedding: flattenedEmbeddings,
 		Object:    "embedding",
 	})
 	usage := &dto.Usage{
@@ -120,3 +121,11 @@ func ollamaEmbeddingHandler(c *gin.Context, resp *http.Response, promptTokens in
 	}
 	return nil, usage
 }
+
+func flattenEmbeddings(embeddings [][]float64) []float64 {
+flattened := []float64{}
+for _, row := range embeddings {
+	flattened = append(flattened, row...)
+}
+return flattened
+}

relay/channel/openai/adaptor.go

@@ -31,6 +31,13 @@ func (a *Adaptor) Init(info *relaycommon.RelayInfo) {
 }
 
 func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {
+	if info.RelayMode == constant.RelayModeRealtime {
+		// trim https
+		baseUrl := strings.TrimPrefix(info.BaseUrl, "https://")
+		baseUrl = strings.TrimPrefix(baseUrl, "http://")
+		baseUrl = "wss://" + baseUrl
+		info.BaseUrl = baseUrl
+	}
 	switch info.ChannelType {
 	case common.ChannelTypeAzure:
 		// https://learn.microsoft.com/en-us/azure/cognitive-services/openai/chatgpt-quickstart?pivots=rest-api&tabs=command-line#rest-api
@@ -40,8 +47,10 @@ func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {
 		model_ := info.UpstreamModelName
 		model_ = strings.Replace(model_, ".", "", -1)
 		// https://github.com/songquanpeng/one-api/issues/67
-
 		requestURL = fmt.Sprintf("/openai/deployments/%s/%s", model_, task)
+		if info.RelayMode == constant.RelayModeRealtime {
+			requestURL = fmt.Sprintf("/openai/realtime?deployment=%s&api-version=%s", model_, info.ApiVersion)
+		}
 		return relaycommon.GetFullRequestURL(info.BaseUrl, requestURL, info.ChannelType), nil
 	case common.ChannelTypeMiniMax:
 		return minimax.GetRequestURL(info)
@@ -54,16 +63,34 @@ func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {
 	}
 }
 
-func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, info *relaycommon.RelayInfo) error {
-	channel.SetupApiRequestHeader(info, c, req)
+func (a *Adaptor) SetupRequestHeader(c *gin.Context, header *http.Header, info *relaycommon.RelayInfo) error {
+	channel.SetupApiRequestHeader(info, c, header)
 	if info.ChannelType == common.ChannelTypeAzure {
-		req.Header.Set("api-key", info.ApiKey)
+		header.Set("api-key", info.ApiKey)
 		return nil
 	}
 	if info.ChannelType == common.ChannelTypeOpenAI && "" != info.Organization {
-		req.Header.Set("OpenAI-Organization", info.Organization)
+		header.Set("OpenAI-Organization", info.Organization)
+	}
+	if info.RelayMode == constant.RelayModeRealtime {
+		swp := c.Request.Header.Get("Sec-WebSocket-Protocol")
+		if swp != "" {
+			items := []string{
+				"realtime",
+				"openai-insecure-api-key." + info.ApiKey,
+				"openai-beta.realtime-v1",
+			}
+			header.Set("Sec-WebSocket-Protocol", strings.Join(items, ","))
+			//req.Header.Set("Sec-WebSocket-Key", c.Request.Header.Get("Sec-WebSocket-Key"))
+			//req.Header.Set("Sec-Websocket-Extensions", c.Request.Header.Get("Sec-Websocket-Extensions"))
+			//req.Header.Set("Sec-Websocket-Version", c.Request.Header.Get("Sec-Websocket-Version"))
+		} else {
+			header.Set("openai-beta", "realtime=v1")
+			header.Set("Authorization", "Bearer "+info.ApiKey)
+		}
+	} else {
+		header.Set("Authorization", "Bearer "+info.ApiKey)
 	}
-	req.Header.Set("Authorization", "Bearer "+info.ApiKey)
 	//if info.ChannelType == common.ChannelTypeOpenRouter {
 	//	req.Header.Set("HTTP-Referer", "https://github.com/songquanpeng/one-api")
 	//	req.Header.Set("X-Title", "One API")
@@ -131,16 +158,20 @@ func (a *Adaptor) ConvertImageRequest(c *gin.Context, info *relaycommon.RelayInf
 	return request, nil
 }
 
-func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (*http.Response, error) {
+func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	if info.RelayMode == constant.RelayModeAudioTranscription || info.RelayMode == constant.RelayModeAudioTranslation {
 		return channel.DoFormRequest(a, c, info, requestBody)
+	} else if info.RelayMode == constant.RelayModeRealtime {
+		return channel.DoWssRequest(a, c, info, requestBody)
 	} else {
 		return channel.DoApiRequest(a, c, info, requestBody)
 	}
 }
 
-func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage *dto.Usage, err *dto.OpenAIErrorWithStatusCode) {
+func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage any, err *dto.OpenAIErrorWithStatusCode) {
 	switch info.RelayMode {
+	case constant.RelayModeRealtime:
+		err, usage = OpenaiRealtimeHandler(c, info)
 	case constant.RelayModeAudioSpeech:
 		err, usage = OpenaiTTSHandler(c, resp, info)
 	case constant.RelayModeAudioTranslation:

relay/channel/openai/constant.go

@@ -13,6 +13,8 @@ var ModelList = []string{
 	"gpt-4o-mini", "gpt-4o-mini-2024-07-18",
 	"o1-preview", "o1-preview-2024-09-12",
 	"o1-mini", "o1-mini-2024-09-12",
+	"gpt-4o-audio-preview", "gpt-4o-audio-preview-2024-10-01",
+	"gpt-4o-realtime-preview", "gpt-4o-realtime-preview-2024-10-01",
 	"text-embedding-ada-002", "text-embedding-3-small", "text-embedding-3-large",
 	"text-curie-001", "text-babbage-001", "text-ada-001",
 	"text-moderation-latest", "text-moderation-stable",

relay/channel/openai/relay-openai.go

@@ -7,6 +7,7 @@ import (
 	"fmt"
 	"github.com/bytedance/gopkg/util/gopool"
 	"github.com/gin-gonic/gin"
+	"github.com/gorilla/websocket"
 	"io"
 	"net/http"
 	"one-api/common"
@@ -231,7 +232,7 @@ func OpenaiHandler(c *gin.Context, resp *http.Response, promptTokens int, model
 	if simpleResponse.Usage.TotalTokens == 0 || (simpleResponse.Usage.PromptTokens == 0 && simpleResponse.Usage.CompletionTokens == 0) {
 		completionTokens := 0
 		for _, choice := range simpleResponse.Choices {
-			ctkm, _ := service.CountTokenText(string(choice.Message.Content), model)
+			ctkm, _ := service.CountTextToken(string(choice.Message.Content), model)
 			completionTokens += ctkm
 		}
 		simpleResponse.Usage = dto.Usage{
@@ -324,7 +325,7 @@ func OpenaiSTTHandler(c *gin.Context, resp *http.Response, info *relaycommon.Rel
 
 	usage := &dto.Usage{}
 	usage.PromptTokens = info.PromptTokens
-	usage.CompletionTokens, _ = service.CountTokenText(text, info.UpstreamModelName)
+	usage.CompletionTokens, _ = service.CountTextToken(text, info.UpstreamModelName)
 	usage.TotalTokens = usage.PromptTokens + usage.CompletionTokens
 	return nil, usage
 }
@@ -373,3 +374,210 @@ func getTextFromJSON(body []byte) (string, error) {
 	}
 	return whisperResponse.Text, nil
 }
+
+func OpenaiRealtimeHandler(c *gin.Context, info *relaycommon.RelayInfo) (*dto.OpenAIErrorWithStatusCode, *dto.RealtimeUsage) {
+	info.IsStream = true
+	clientConn := info.ClientWs
+	targetConn := info.TargetWs
+
+	clientClosed := make(chan struct{})
+	targetClosed := make(chan struct{})
+	sendChan := make(chan []byte, 100)
+	receiveChan := make(chan []byte, 100)
+	errChan := make(chan error, 2)
+
+	usage := &dto.RealtimeUsage{}
+	localUsage := &dto.RealtimeUsage{}
+	sumUsage := &dto.RealtimeUsage{}
+
+	gopool.Go(func() {
+		for {
+			select {
+			case <-c.Done():
+				return
+			default:
+				_, message, err := clientConn.ReadMessage()
+				if err != nil {
+					if !websocket.IsCloseError(err, websocket.CloseNormalClosure, websocket.CloseGoingAway) {
+						errChan <- fmt.Errorf("error reading from client: %v", err)
+					}
+					close(clientClosed)
+					return
+				}
+
+				realtimeEvent := &dto.RealtimeEvent{}
+				err = json.Unmarshal(message, realtimeEvent)
+				if err != nil {
+					errChan <- fmt.Errorf("error unmarshalling message: %v", err)
+					return
+				}
+
+				if realtimeEvent.Type == dto.RealtimeEventTypeSessionUpdate {
+					if realtimeEvent.Session != nil {
+						if realtimeEvent.Session.Tools != nil {
+							info.RealtimeTools = realtimeEvent.Session.Tools
+						}
+					}
+				}
+
+				textToken, audioToken, err := service.CountTokenRealtime(info, *realtimeEvent, info.UpstreamModelName)
+				if err != nil {
+					errChan <- fmt.Errorf("error counting text token: %v", err)
+					return
+				}
+				common.LogInfo(c, fmt.Sprintf("type: %s, textToken: %d, audioToken: %d", realtimeEvent.Type, textToken, audioToken))
+				localUsage.TotalTokens += textToken + audioToken
+				localUsage.InputTokens += textToken + audioToken
+				localUsage.InputTokenDetails.TextTokens += textToken
+				localUsage.InputTokenDetails.AudioTokens += audioToken
+
+				err = service.WssString(c, targetConn, string(message))
+				if err != nil {
+					errChan <- fmt.Errorf("error writing to target: %v", err)
+					return
+				}
+
+				select {
+				case sendChan <- message:
+				default:
+				}
+			}
+		}
+	})
+
+	gopool.Go(func() {
+		for {
+			select {
+			case <-c.Done():
+				return
+			default:
+				_, message, err := targetConn.ReadMessage()
+				if err != nil {
+					if !websocket.IsCloseError(err, websocket.CloseNormalClosure, websocket.CloseGoingAway) {
+						errChan <- fmt.Errorf("error reading from target: %v", err)
+					}
+					close(targetClosed)
+					return
+				}
+				info.SetFirstResponseTime()
+				realtimeEvent := &dto.RealtimeEvent{}
+				err = json.Unmarshal(message, realtimeEvent)
+				if err != nil {
+					errChan <- fmt.Errorf("error unmarshalling message: %v", err)
+					return
+				}
+
+				if realtimeEvent.Type == dto.RealtimeEventTypeResponseDone {
+					realtimeUsage := realtimeEvent.Response.Usage
+					if realtimeUsage != nil {
+						usage.TotalTokens += realtimeUsage.TotalTokens
+						usage.InputTokens += realtimeUsage.InputTokens
+						usage.OutputTokens += realtimeUsage.OutputTokens
+						usage.InputTokenDetails.AudioTokens += realtimeUsage.InputTokenDetails.AudioTokens
+						usage.InputTokenDetails.CachedTokens += realtimeUsage.InputTokenDetails.CachedTokens
+						usage.InputTokenDetails.TextTokens += realtimeUsage.InputTokenDetails.TextTokens
+						usage.OutputTokenDetails.AudioTokens += realtimeUsage.OutputTokenDetails.AudioTokens
+						usage.OutputTokenDetails.TextTokens += realtimeUsage.OutputTokenDetails.TextTokens
+						err := preConsumeUsage(c, info, usage, sumUsage)
+						if err != nil {
+							errChan <- fmt.Errorf("error consume usage: %v", err)
+							return
+						}
+						// 本次计费完成，清除
+						usage = &dto.RealtimeUsage{}
+
+						localUsage = &dto.RealtimeUsage{}
+					} else {
+						textToken, audioToken, err := service.CountTokenRealtime(info, *realtimeEvent, info.UpstreamModelName)
+						if err != nil {
+							errChan <- fmt.Errorf("error counting text token: %v", err)
+							return
+						}
+						common.LogInfo(c, fmt.Sprintf("type: %s, textToken: %d, audioToken: %d", realtimeEvent.Type, textToken, audioToken))
+						localUsage.TotalTokens += textToken + audioToken
+						info.IsFirstRequest = false
+						localUsage.InputTokens += textToken + audioToken
+						localUsage.InputTokenDetails.TextTokens += textToken
+						localUsage.InputTokenDetails.AudioTokens += audioToken
+						err = preConsumeUsage(c, info, localUsage, sumUsage)
+						if err != nil {
+							errChan <- fmt.Errorf("error consume usage: %v", err)
+							return
+						}
+						// 本次计费完成，清除
+						localUsage = &dto.RealtimeUsage{}
+						// print now usage
+					}
+					//common.LogInfo(c, fmt.Sprintf("realtime streaming sumUsage: %v", sumUsage))
+					//common.LogInfo(c, fmt.Sprintf("realtime streaming localUsage: %v", localUsage))
+					//common.LogInfo(c, fmt.Sprintf("realtime streaming localUsage: %v", localUsage))
+
+				} else if realtimeEvent.Type == dto.RealtimeEventTypeSessionUpdated || realtimeEvent.Type == dto.RealtimeEventTypeSessionCreated {
+					realtimeSession := realtimeEvent.Session
+					if realtimeSession != nil {
+						// update audio format
+						info.InputAudioFormat = common.GetStringIfEmpty(realtimeSession.InputAudioFormat, info.InputAudioFormat)
+						info.OutputAudioFormat = common.GetStringIfEmpty(realtimeSession.OutputAudioFormat, info.OutputAudioFormat)
+					}
+				} else {
+					textToken, audioToken, err := service.CountTokenRealtime(info, *realtimeEvent, info.UpstreamModelName)
+					if err != nil {
+						errChan <- fmt.Errorf("error counting text token: %v", err)
+						return
+					}
+					common.LogInfo(c, fmt.Sprintf("type: %s, textToken: %d, audioToken: %d", realtimeEvent.Type, textToken, audioToken))
+					localUsage.TotalTokens += textToken + audioToken
+					localUsage.OutputTokens += textToken + audioToken
+					localUsage.OutputTokenDetails.TextTokens += textToken
+					localUsage.OutputTokenDetails.AudioTokens += audioToken
+				}
+
+				err = service.WssString(c, clientConn, string(message))
+				if err != nil {
+					errChan <- fmt.Errorf("error writing to client: %v", err)
+					return
+				}
+
+				select {
+				case receiveChan <- message:
+				default:
+				}
+			}
+		}
+	})
+
+	select {
+	case <-clientClosed:
+	case <-targetClosed:
+	case err := <-errChan:
+		//return service.OpenAIErrorWrapper(err, "realtime_error", http.StatusInternalServerError), nil
+		common.LogError(c, "realtime error: "+err.Error())
+	case <-c.Done():
+	}
+
+	if usage.TotalTokens != 0 {
+		_ = preConsumeUsage(c, info, usage, sumUsage)
+	}
+
+	if localUsage.TotalTokens != 0 {
+		_ = preConsumeUsage(c, info, localUsage, sumUsage)
+	}
+
+	// check usage total tokens, if 0, use local usage
+
+	return nil, sumUsage
+}
+
+func preConsumeUsage(ctx *gin.Context, info *relaycommon.RelayInfo, usage *dto.RealtimeUsage, totalUsage *dto.RealtimeUsage) error {
+	totalUsage.TotalTokens += usage.TotalTokens
+	totalUsage.InputTokens += usage.InputTokens
+	totalUsage.OutputTokens += usage.OutputTokens
+	totalUsage.InputTokenDetails.CachedTokens += usage.InputTokenDetails.CachedTokens
+	totalUsage.InputTokenDetails.TextTokens += usage.InputTokenDetails.TextTokens
+	totalUsage.InputTokenDetails.AudioTokens += usage.InputTokenDetails.AudioTokens
+	totalUsage.OutputTokenDetails.TextTokens += usage.OutputTokenDetails.TextTokens
+	totalUsage.OutputTokenDetails.AudioTokens += usage.OutputTokenDetails.AudioTokens
+	// clear usage
+	err := service.PreWssConsumeQuota(ctx, info, usage)
+	return err
+}

relay/channel/palm/adaptor.go

@@ -32,9 +32,9 @@ func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {
 	return fmt.Sprintf("%s/v1beta2/models/chat-bison-001:generateMessage", info.BaseUrl), nil
 }
 
-func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, info *relaycommon.RelayInfo) error {
+func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Header, info *relaycommon.RelayInfo) error {
 	channel.SetupApiRequestHeader(info, c, req)
-	req.Header.Set("x-goog-api-key", info.ApiKey)
+	req.Set("x-goog-api-key", info.ApiKey)
 	return nil
 }
 
@@ -49,11 +49,11 @@ func (a *Adaptor) ConvertRerankRequest(c *gin.Context, relayMode int, request dt
 	return nil, nil
 }
 
-func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (*http.Response, error) {
+func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	return channel.DoApiRequest(a, c, info, requestBody)
 }
 
-func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage *dto.Usage, err *dto.OpenAIErrorWithStatusCode) {
+func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage any, err *dto.OpenAIErrorWithStatusCode) {
 	if info.IsStream {
 		var responseText string
 		err, responseText = palmStreamHandler(c, resp)

relay/channel/palm/relay-palm.go

@@ -156,7 +156,7 @@ func palmHandler(c *gin.Context, resp *http.Response, promptTokens int, model st
 		}, nil
 	}
 	fullTextResponse := responsePaLM2OpenAI(&palmResponse)
-	completionTokens, _ := service.CountTokenText(palmResponse.Candidates[0].Content, model)
+	completionTokens, _ := service.CountTextToken(palmResponse.Candidates[0].Content, model)
 	usage := dto.Usage{
 		PromptTokens:     promptTokens,
 		CompletionTokens: completionTokens,

relay/channel/perplexity/adaptor.go

@@ -32,9 +32,9 @@ func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {
 	return fmt.Sprintf("%s/chat/completions", info.BaseUrl), nil
 }
 
-func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, info *relaycommon.RelayInfo) error {
+func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Header, info *relaycommon.RelayInfo) error {
 	channel.SetupApiRequestHeader(info, c, req)
-	req.Header.Set("Authorization", "Bearer "+info.ApiKey)
+	req.Set("Authorization", "Bearer "+info.ApiKey)
 	return nil
 }
 
@@ -52,11 +52,11 @@ func (a *Adaptor) ConvertRerankRequest(c *gin.Context, relayMode int, request dt
 	return nil, nil
 }
 
-func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (*http.Response, error) {
+func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	return channel.DoApiRequest(a, c, info, requestBody)
 }
 
-func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage *dto.Usage, err *dto.OpenAIErrorWithStatusCode) {
+func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage any, err *dto.OpenAIErrorWithStatusCode) {
 	if info.IsStream {
 		err, usage = openai.OaiStreamHandler(c, resp, info)
 	} else {

relay/channel/siliconflow/adaptor.go

@@ -40,9 +40,9 @@ func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {
 	return "", errors.New("invalid relay mode")
 }
 
-func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, info *relaycommon.RelayInfo) error {
+func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Header, info *relaycommon.RelayInfo) error {
 	channel.SetupApiRequestHeader(info, c, req)
-	req.Header.Set("Authorization", fmt.Sprintf("Bearer %s", info.ApiKey))
+	req.Set("Authorization", fmt.Sprintf("Bearer %s", info.ApiKey))
 	return nil
 }
 
@@ -50,7 +50,7 @@ func (a *Adaptor) ConvertRequest(c *gin.Context, info *relaycommon.RelayInfo, re
 	return request, nil
 }
 
-func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (*http.Response, error) {
+func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	return channel.DoApiRequest(a, c, info, requestBody)
 }
 
@@ -58,7 +58,7 @@ func (a *Adaptor) ConvertRerankRequest(c *gin.Context, relayMode int, request dt
 	return request, nil
 }
 
-func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage *dto.Usage, err *dto.OpenAIErrorWithStatusCode) {
+func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage any, err *dto.OpenAIErrorWithStatusCode) {
 	switch info.RelayMode {
 	case constant.RelayModeRerank:
 		err, usage = siliconflowRerankHandler(c, resp)

relay/channel/tencent/adaptor.go

@@ -43,12 +43,12 @@ func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {
 	return fmt.Sprintf("%s/", info.BaseUrl), nil
 }
 
-func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, info *relaycommon.RelayInfo) error {
+func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Header, info *relaycommon.RelayInfo) error {
 	channel.SetupApiRequestHeader(info, c, req)
-	req.Header.Set("Authorization", a.Sign)
-	req.Header.Set("X-TC-Action", a.Action)
-	req.Header.Set("X-TC-Version", a.Version)
-	req.Header.Set("X-TC-Timestamp", strconv.FormatInt(a.Timestamp, 10))
+	req.Set("Authorization", a.Sign)
+	req.Set("X-TC-Action", a.Action)
+	req.Set("X-TC-Version", a.Version)
+	req.Set("X-TC-Timestamp", strconv.FormatInt(a.Timestamp, 10))
 	return nil
 }
 
@@ -73,11 +73,11 @@ func (a *Adaptor) ConvertRerankRequest(c *gin.Context, relayMode int, request dt
 	return nil, nil
 }
 
-func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (*http.Response, error) {
+func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	return channel.DoApiRequest(a, c, info, requestBody)
 }
 
-func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage *dto.Usage, err *dto.OpenAIErrorWithStatusCode) {
+func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage any, err *dto.OpenAIErrorWithStatusCode) {
 	if info.IsStream {
 		var responseText string
 		err, responseText = tencentStreamHandler(c, resp)

relay/channel/vertex/adaptor.go

@@ -107,13 +107,13 @@ func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {
 	return "", errors.New("unsupported request mode")
 }
 
-func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, info *relaycommon.RelayInfo) error {
+func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Header, info *relaycommon.RelayInfo) error {
 	channel.SetupApiRequestHeader(info, c, req)
 	accessToken, err := getAccessToken(a, info)
 	if err != nil {
 		return err
 	}
-	req.Header.Set("Authorization", "Bearer "+accessToken)
+	req.Set("Authorization", "Bearer "+accessToken)
 	return nil
 }
 
@@ -148,11 +148,11 @@ func (a *Adaptor) ConvertRerankRequest(c *gin.Context, relayMode int, request dt
 	return nil, nil
 }
 
-func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (*http.Response, error) {
+func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	return channel.DoApiRequest(a, c, info, requestBody)
 }
 
-func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage *dto.Usage, err *dto.OpenAIErrorWithStatusCode) {
+func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage any, err *dto.OpenAIErrorWithStatusCode) {
 	if info.IsStream {
 		switch a.RequestMode {
 		case RequestModeClaude:

relay/channel/xunfei/adaptor.go

@@ -33,7 +33,7 @@ func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {
 	return "", nil
 }
 
-func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, info *relaycommon.RelayInfo) error {
+func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Header, info *relaycommon.RelayInfo) error {
 	channel.SetupApiRequestHeader(info, c, req)
 	return nil
 }
@@ -50,14 +50,14 @@ func (a *Adaptor) ConvertRerankRequest(c *gin.Context, relayMode int, request dt
 	return nil, nil
 }
 
-func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (*http.Response, error) {
+func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	// xunfei's request is not http request, so we don't need to do anything here
 	dummyResp := &http.Response{}
 	dummyResp.StatusCode = http.StatusOK
 	return dummyResp, nil
 }
 
-func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage *dto.Usage, err *dto.OpenAIErrorWithStatusCode) {
+func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage any, err *dto.OpenAIErrorWithStatusCode) {
 	splits := strings.Split(info.ApiKey, "|")
 	if len(splits) != 3 {
 		return nil, service.OpenAIErrorWrapper(errors.New("invalid auth"), "invalid_auth", http.StatusBadRequest)

relay/channel/zhipu/adaptor.go

@@ -35,10 +35,10 @@ func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {
 	return fmt.Sprintf("%s/api/paas/v3/model-api/%s/%s", info.BaseUrl, info.UpstreamModelName, method), nil
 }
 
-func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, info *relaycommon.RelayInfo) error {
+func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Header, info *relaycommon.RelayInfo) error {
 	channel.SetupApiRequestHeader(info, c, req)
 	token := getZhipuToken(info.ApiKey)
-	req.Header.Set("Authorization", token)
+	req.Set("Authorization", token)
 	return nil
 }
 
@@ -56,11 +56,11 @@ func (a *Adaptor) ConvertRerankRequest(c *gin.Context, relayMode int, request dt
 	return nil, nil
 }
 
-func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (*http.Response, error) {
+func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	return channel.DoApiRequest(a, c, info, requestBody)
 }
 
-func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage *dto.Usage, err *dto.OpenAIErrorWithStatusCode) {
+func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage any, err *dto.OpenAIErrorWithStatusCode) {
 	if info.IsStream {
 		err, usage = zhipuStreamHandler(c, resp)
 	} else {

relay/channel/zhipu_4v/adaptor.go

@@ -32,10 +32,10 @@ func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {
 	return fmt.Sprintf("%s/api/paas/v4/chat/completions", info.BaseUrl), nil
 }
 
-func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, info *relaycommon.RelayInfo) error {
+func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Header, info *relaycommon.RelayInfo) error {
 	channel.SetupApiRequestHeader(info, c, req)
 	token := getZhipuToken(info.ApiKey)
-	req.Header.Set("Authorization", token)
+	req.Set("Authorization", token)
 	return nil
 }
 
@@ -53,11 +53,11 @@ func (a *Adaptor) ConvertRerankRequest(c *gin.Context, relayMode int, request dt
 	return nil, nil
 }
 
-func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (*http.Response, error) {
+func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	return channel.DoApiRequest(a, c, info, requestBody)
 }
 
-func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage *dto.Usage, err *dto.OpenAIErrorWithStatusCode) {
+func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage any, err *dto.OpenAIErrorWithStatusCode) {
 	if info.IsStream {
 		err, usage = openai.OaiStreamHandler(c, resp, info)
 	} else {

relay/common/relay_info.go

@@ -2,7 +2,9 @@ package common
 
 import (
 	"github.com/gin-gonic/gin"
+	"github.com/gorilla/websocket"
 	"one-api/common"
+	"one-api/dto"
 	"one-api/relay/constant"
 	"strings"
 	"time"
@@ -21,6 +23,7 @@ type RelayInfo struct {
 	ApiType              int
 	IsStream             bool
 	IsPlayground         bool
+	UsePrice             bool
 	RelayMode            int
 	UpstreamModelName    string
 	OriginModelName      string
@@ -32,6 +35,22 @@ type RelayInfo struct {
 	BaseUrl              string
 	SupportStreamOptions bool
 	ShouldIncludeUsage   bool
+	ClientWs             *websocket.Conn
+	TargetWs             *websocket.Conn
+	InputAudioFormat     string
+	OutputAudioFormat    string
+	RealtimeTools        []dto.RealTimeTool
+	IsFirstRequest       bool
+	AudioUsage           bool
+}
+
+func GenRelayInfoWs(c *gin.Context, ws *websocket.Conn) *RelayInfo {
+	info := GenRelayInfo(c)
+	info.ClientWs = ws
+	info.InputAudioFormat = "pcm16"
+	info.OutputAudioFormat = "pcm16"
+	info.IsFirstRequest = true
+	return info
 }
 
 func GenRelayInfo(c *gin.Context) *RelayInfo {

relay/constant/api_type.go

@@ -25,6 +25,7 @@ const (
 	APITypeCloudflare
 	APITypeSiliconFlow
 	APITypeVertexAi
+	APITypeMistral
 
 	APITypeDummy // this one is only for count, do not add any channel after this
 )
@@ -72,6 +73,8 @@ func ChannelType2APIType(channelType int) (int, bool) {
 		apiType = APITypeSiliconFlow
 	case common.ChannelTypeVertexAi:
 		apiType = APITypeVertexAi
+	case common.ChannelTypeMistral:
+		apiType = APITypeMistral
 	}
 	if apiType == -1 {
 		return APITypeOpenAI, false

relay/constant/relay_mode.go

@@ -38,6 +38,8 @@ const (
 	RelayModeSunoSubmit
 
 	RelayModeRerank
+
+	RelayModeRealtime
 )
 
 func Path2RelayMode(path string) int {
@@ -64,6 +66,8 @@ func Path2RelayMode(path string) int {
 		relayMode = RelayModeAudioTranslation
 	} else if strings.HasPrefix(path, "/v1/rerank") {
 		relayMode = RelayModeRerank
+	} else if strings.HasPrefix(path, "/v1/realtime") {
+		relayMode = RelayModeRealtime
 	}
 	return relayMode
 }

relay/relay-audio.go

@@ -46,7 +46,7 @@ func getAndValidAudioRequest(c *gin.Context, info *relaycommon.RelayInfo) (*dto.
 	return audioRequest, nil
 }
 
-func AudioHelper(c *gin.Context) *dto.OpenAIErrorWithStatusCode {
+func AudioHelper(c *gin.Context) (openaiErr *dto.OpenAIErrorWithStatusCode) {
 	relayInfo := relaycommon.GenRelayInfo(c)
 	audioRequest, err := getAndValidAudioRequest(c, relayInfo)
 
@@ -58,7 +58,7 @@ func AudioHelper(c *gin.Context) *dto.OpenAIErrorWithStatusCode {
 	promptTokens := 0
 	preConsumedTokens := common.PreConsumedQuota
 	if relayInfo.RelayMode == relayconstant.RelayModeAudioSpeech {
-		promptTokens, err = service.CountAudioToken(audioRequest.Input, audioRequest.Model)
+		promptTokens, err = service.CountTTSToken(audioRequest.Input, audioRequest.Model)
 		if err != nil {
 			return service.OpenAIErrorWrapper(err, "count_audio_token_failed", http.StatusInternalServerError)
 		}
@@ -92,6 +92,11 @@ func AudioHelper(c *gin.Context) *dto.OpenAIErrorWithStatusCode {
 			return service.OpenAIErrorWrapperLocal(err, "pre_consume_token_quota_failed", http.StatusForbidden)
 		}
 	}
+	defer func() {
+		if openaiErr != nil {
+			returnPreConsumedQuota(c, relayInfo, userQuota, preConsumedQuota)
+		}
+	}()
 
 	// map model name
 	modelMapping := c.GetString("model_mapping")
@@ -122,27 +127,27 @@ func AudioHelper(c *gin.Context) *dto.OpenAIErrorWithStatusCode {
 	if err != nil {
 		return service.OpenAIErrorWrapper(err, "do_request_failed", http.StatusInternalServerError)
 	}
-
 	statusCodeMappingStr := c.GetString("status_code_mapping")
+
+	var httpResp *http.Response
 	if resp != nil {
-		if resp.StatusCode != http.StatusOK {
-			returnPreConsumedQuota(c, relayInfo, userQuota, preConsumedQuota)
-			openaiErr := service.RelayErrorHandler(resp)
+		httpResp = resp.(*http.Response)
+		if httpResp.StatusCode != http.StatusOK {
+			openaiErr = service.RelayErrorHandler(httpResp)
 			// reset status code 重置状态码
 			service.ResetStatusCode(openaiErr, statusCodeMappingStr)
 			return openaiErr
 		}
 	}
 
-	usage, openaiErr := adaptor.DoResponse(c, resp, relayInfo)
+	usage, openaiErr := adaptor.DoResponse(c, httpResp, relayInfo)
 	if openaiErr != nil {
-		returnPreConsumedQuota(c, relayInfo, userQuota, preConsumedQuota)
 		// reset status code 重置状态码
 		service.ResetStatusCode(openaiErr, statusCodeMappingStr)
 		return openaiErr
 	}
 
-	postConsumeQuota(c, relayInfo, audioRequest.Model, usage, ratio, preConsumedQuota, userQuota, modelRatio, groupRatio, 0, false, "")
+	postConsumeQuota(c, relayInfo, audioRequest.Model, usage.(*dto.Usage), ratio, preConsumedQuota, userQuota, modelRatio, groupRatio, 0, false, "")
 
 	return nil
 }

relay/relay-image.go

@@ -149,22 +149,24 @@ func ImageHelper(c *gin.Context, relayMode int) *dto.OpenAIErrorWithStatusCode {
 	requestBody = bytes.NewBuffer(jsonData)
 
 	statusCodeMappingStr := c.GetString("status_code_mapping")
+
 	resp, err := adaptor.DoRequest(c, relayInfo, requestBody)
 	if err != nil {
 		return service.OpenAIErrorWrapper(err, "do_request_failed", http.StatusInternalServerError)
 	}
-
+	var httpResp *http.Response
 	if resp != nil {
-		relayInfo.IsStream = relayInfo.IsStream || strings.HasPrefix(resp.Header.Get("Content-Type"), "text/event-stream")
-		if resp.StatusCode != http.StatusOK {
-			openaiErr := service.RelayErrorHandler(resp)
+		httpResp = resp.(*http.Response)
+		relayInfo.IsStream = relayInfo.IsStream || strings.HasPrefix(httpResp.Header.Get("Content-Type"), "text/event-stream")
+		if httpResp.StatusCode != http.StatusOK {
+			openaiErr := service.RelayErrorHandler(httpResp)
 			// reset status code 重置状态码
 			service.ResetStatusCode(openaiErr, statusCodeMappingStr)
 			return openaiErr
 		}
 	}
 
-	_, openaiErr := adaptor.DoResponse(c, resp, relayInfo)
+	_, openaiErr := adaptor.DoResponse(c, httpResp, relayInfo)
 	if openaiErr != nil {
 		// reset status code 重置状态码
 		service.ResetStatusCode(openaiErr, statusCodeMappingStr)

relay/relay-text.go

@@ -64,7 +64,7 @@ func getAndValidateTextRequest(c *gin.Context, relayInfo *relaycommon.RelayInfo)
 	return textRequest, nil
 }
 
-func TextHelper(c *gin.Context) *dto.OpenAIErrorWithStatusCode {
+func TextHelper(c *gin.Context) (openaiErr *dto.OpenAIErrorWithStatusCode) {
 
 	relayInfo := relaycommon.GenRelayInfo(c)
 
@@ -131,7 +131,11 @@ func TextHelper(c *gin.Context) *dto.OpenAIErrorWithStatusCode {
 	if openaiErr != nil {
 		return openaiErr
 	}
-
+	defer func() {
+		if openaiErr != nil {
+			returnPreConsumedQuota(c, relayInfo, userQuota, preConsumedQuota)
+		}
+	}()
 	includeUsage := false
 	// 判断用户是否需要返回使用情况
 	if textRequest.StreamOptions != nil && textRequest.StreamOptions.IncludeUsage {
@@ -180,30 +184,35 @@ func TextHelper(c *gin.Context) *dto.OpenAIErrorWithStatusCode {
 	}
 
 	statusCodeMappingStr := c.GetString("status_code_mapping")
+	var httpResp *http.Response
 	resp, err := adaptor.DoRequest(c, relayInfo, requestBody)
 	if err != nil {
 		return service.OpenAIErrorWrapper(err, "do_request_failed", http.StatusInternalServerError)
 	}
 
 	if resp != nil {
-		relayInfo.IsStream = relayInfo.IsStream || strings.HasPrefix(resp.Header.Get("Content-Type"), "text/event-stream")
-		if resp.StatusCode != http.StatusOK {
-			returnPreConsumedQuota(c, relayInfo, userQuota, preConsumedQuota)
-			openaiErr := service.RelayErrorHandler(resp)
+		httpResp = resp.(*http.Response)
+		relayInfo.IsStream = relayInfo.IsStream || strings.HasPrefix(httpResp.Header.Get("Content-Type"), "text/event-stream")
+		if httpResp.StatusCode != http.StatusOK {
+			openaiErr = service.RelayErrorHandler(httpResp)
 			// reset status code 重置状态码
 			service.ResetStatusCode(openaiErr, statusCodeMappingStr)
 			return openaiErr
 		}
 	}
 
-	usage, openaiErr := adaptor.DoResponse(c, resp, relayInfo)
+	usage, openaiErr := adaptor.DoResponse(c, httpResp, relayInfo)
 	if openaiErr != nil {
-		returnPreConsumedQuota(c, relayInfo, userQuota, preConsumedQuota)
 		// reset status code 重置状态码
 		service.ResetStatusCode(openaiErr, statusCodeMappingStr)
 		return openaiErr
 	}
-	postConsumeQuota(c, relayInfo, textRequest.Model, usage, ratio, preConsumedQuota, userQuota, modelRatio, groupRatio, modelPrice, getModelPriceSuccess, "")
+
+	if strings.HasPrefix(relayInfo.UpstreamModelName, "gpt-4o-audio") {
+		service.PostAudioConsumeQuota(c, relayInfo, usage.(*dto.Usage), ratio, preConsumedQuota, userQuota, modelRatio, groupRatio, modelPrice, getModelPriceSuccess, "")
+	} else {
+		postConsumeQuota(c, relayInfo, textRequest.Model, usage.(*dto.Usage), ratio, preConsumedQuota, userQuota, modelRatio, groupRatio, modelPrice, getModelPriceSuccess, "")
+	}
 	return nil
 }
 

relay/relay_adaptor.go

@@ -12,6 +12,7 @@ import (
 	"one-api/relay/channel/dify"
 	"one-api/relay/channel/gemini"
 	"one-api/relay/channel/jina"
+	"one-api/relay/channel/mistral"
 	"one-api/relay/channel/ollama"
 	"one-api/relay/channel/openai"
 	"one-api/relay/channel/palm"
@@ -68,6 +69,8 @@ func GetAdaptor(apiType int) channel.Adaptor {
 		return &siliconflow.Adaptor{}
 	case constant.APITypeVertexAi:
 		return &vertex.Adaptor{}
+	case constant.APITypeMistral:
+		return &mistral.Adaptor{}
 	}
 	return nil
 }

relay/relay_rerank.go

@@ -23,7 +23,7 @@ func getRerankPromptToken(rerankRequest dto.RerankRequest) int {
 	return token
 }
 
-func RerankHelper(c *gin.Context, relayMode int) *dto.OpenAIErrorWithStatusCode {
+func RerankHelper(c *gin.Context, relayMode int) (openaiErr *dto.OpenAIErrorWithStatusCode) {
 	relayInfo := relaycommon.GenRelayInfo(c)
 
 	var rerankRequest *dto.RerankRequest
@@ -79,6 +79,12 @@ func RerankHelper(c *gin.Context, relayMode int) *dto.OpenAIErrorWithStatusCode
 	if openaiErr != nil {
 		return openaiErr
 	}
+	defer func() {
+		if openaiErr != nil {
+			returnPreConsumedQuota(c, relayInfo, userQuota, preConsumedQuota)
+		}
+	}()
+
 	adaptor := GetAdaptor(relayInfo.ApiType)
 	if adaptor == nil {
 		return service.OpenAIErrorWrapperLocal(fmt.Errorf("invalid api type: %d", relayInfo.ApiType), "invalid_api_type", http.StatusBadRequest)
@@ -99,23 +105,24 @@ func RerankHelper(c *gin.Context, relayMode int) *dto.OpenAIErrorWithStatusCode
 	if err != nil {
 		return service.OpenAIErrorWrapper(err, "do_request_failed", http.StatusInternalServerError)
 	}
+
+	var httpResp *http.Response
 	if resp != nil {
-		if resp.StatusCode != http.StatusOK {
-			returnPreConsumedQuota(c, relayInfo, userQuota, preConsumedQuota)
-			openaiErr := service.RelayErrorHandler(resp)
+		httpResp = resp.(*http.Response)
+		if httpResp.StatusCode != http.StatusOK {
+			openaiErr = service.RelayErrorHandler(httpResp)
 			// reset status code 重置状态码
 			service.ResetStatusCode(openaiErr, statusCodeMappingStr)
 			return openaiErr
 		}
 	}
 
-	usage, openaiErr := adaptor.DoResponse(c, resp, relayInfo)
+	usage, openaiErr := adaptor.DoResponse(c, httpResp, relayInfo)
 	if openaiErr != nil {
-		returnPreConsumedQuota(c, relayInfo, userQuota, preConsumedQuota)
 		// reset status code 重置状态码
 		service.ResetStatusCode(openaiErr, statusCodeMappingStr)
 		return openaiErr
 	}
-	postConsumeQuota(c, relayInfo, rerankRequest.Model, usage, ratio, preConsumedQuota, userQuota, modelRatio, groupRatio, modelPrice, success, "")
+	postConsumeQuota(c, relayInfo, rerankRequest.Model, usage.(*dto.Usage), ratio, preConsumedQuota, userQuota, modelRatio, groupRatio, modelPrice, success, "")
 	return nil
 }

relay/websocket.go

@@ -0,0 +1,159 @@
+package relay
+
+import (
+	"encoding/json"
+	"fmt"
+	"github.com/gin-gonic/gin"
+	"github.com/gorilla/websocket"
+	"net/http"
+	"one-api/common"
+	"one-api/dto"
+	relaycommon "one-api/relay/common"
+	"one-api/service"
+)
+
+//func getAndValidateWssRequest(c *gin.Context, ws *websocket.Conn) (*dto.RealtimeEvent, error) {
+//	_, p, err := ws.ReadMessage()
+//	if err != nil {
+//		return nil, err
+//	}
+//	realtimeEvent := &dto.RealtimeEvent{}
+//	err = json.Unmarshal(p, realtimeEvent)
+//	if err != nil {
+//		return nil, err
+//	}
+//	// save the original request
+//	if realtimeEvent.Session == nil {
+//		return nil, errors.New("session object is nil")
+//	}
+//	c.Set("first_wss_request", p)
+//	return realtimeEvent, nil
+//}
+
+func WssHelper(c *gin.Context, ws *websocket.Conn) (openaiErr *dto.OpenAIErrorWithStatusCode) {
+	relayInfo := relaycommon.GenRelayInfoWs(c, ws)
+
+	// get & validate textRequest 获取并验证文本请求
+	//realtimeEvent, err := getAndValidateWssRequest(c, ws)
+	//if err != nil {
+	//	common.LogError(c, fmt.Sprintf("getAndValidateWssRequest failed: %s", err.Error()))
+	//	return service.OpenAIErrorWrapperLocal(err, "invalid_text_request", http.StatusBadRequest)
+	//}
+
+	// map model name
+	modelMapping := c.GetString("model_mapping")
+	//isModelMapped := false
+	if modelMapping != "" && modelMapping != "{}" {
+		modelMap := make(map[string]string)
+		err := json.Unmarshal([]byte(modelMapping), &modelMap)
+		if err != nil {
+			return service.OpenAIErrorWrapperLocal(err, "unmarshal_model_mapping_failed", http.StatusInternalServerError)
+		}
+		if modelMap[relayInfo.OriginModelName] != "" {
+			relayInfo.UpstreamModelName = modelMap[relayInfo.OriginModelName]
+			// set upstream model name
+			//isModelMapped = true
+		}
+	}
+	//relayInfo.UpstreamModelName = textRequest.Model
+	modelPrice, getModelPriceSuccess := common.GetModelPrice(relayInfo.UpstreamModelName, false)
+	groupRatio := common.GetGroupRatio(relayInfo.Group)
+
+	var preConsumedQuota int
+	var ratio float64
+	var modelRatio float64
+	//err := service.SensitiveWordsCheck(textRequest)
+
+	//if constant.ShouldCheckPromptSensitive() {
+	//	err = checkRequestSensitive(textRequest, relayInfo)
+	//	if err != nil {
+	//		return service.OpenAIErrorWrapperLocal(err, "sensitive_words_detected", http.StatusBadRequest)
+	//	}
+	//}
+
+	//promptTokens, err := getWssPromptTokens(realtimeEvent, relayInfo)
+	//// count messages token error 计算promptTokens错误
+	//if err != nil {
+	//	return service.OpenAIErrorWrapper(err, "count_token_messages_failed", http.StatusInternalServerError)
+	//}
+	//
+	if !getModelPriceSuccess {
+		preConsumedTokens := common.PreConsumedQuota
+		//if realtimeEvent.Session.MaxResponseOutputTokens != 0 {
+		//	preConsumedTokens = promptTokens + int(realtimeEvent.Session.MaxResponseOutputTokens)
+		//}
+		modelRatio = common.GetModelRatio(relayInfo.UpstreamModelName)
+		ratio = modelRatio * groupRatio
+		preConsumedQuota = int(float64(preConsumedTokens) * ratio)
+	} else {
+		preConsumedQuota = int(modelPrice * common.QuotaPerUnit * groupRatio)
+		relayInfo.UsePrice = true
+	}
+
+	// pre-consume quota 预消耗配额
+	preConsumedQuota, userQuota, openaiErr := preConsumeQuota(c, preConsumedQuota, relayInfo)
+	if openaiErr != nil {
+		return openaiErr
+	}
+
+	defer func() {
+		if openaiErr != nil {
+			returnPreConsumedQuota(c, relayInfo, userQuota, preConsumedQuota)
+		}
+	}()
+
+	adaptor := GetAdaptor(relayInfo.ApiType)
+	if adaptor == nil {
+		return service.OpenAIErrorWrapperLocal(fmt.Errorf("invalid api type: %d", relayInfo.ApiType), "invalid_api_type", http.StatusBadRequest)
+	}
+	adaptor.Init(relayInfo)
+	//var requestBody io.Reader
+	//firstWssRequest, _ := c.Get("first_wss_request")
+	//requestBody = bytes.NewBuffer(firstWssRequest.([]byte))
+
+	statusCodeMappingStr := c.GetString("status_code_mapping")
+	resp, err := adaptor.DoRequest(c, relayInfo, nil)
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "do_request_failed", http.StatusInternalServerError)
+	}
+
+	if resp != nil {
+		relayInfo.TargetWs = resp.(*websocket.Conn)
+		defer relayInfo.TargetWs.Close()
+	}
+
+	usage, openaiErr := adaptor.DoResponse(c, nil, relayInfo)
+	if openaiErr != nil {
+		// reset status code 重置状态码
+		service.ResetStatusCode(openaiErr, statusCodeMappingStr)
+		return openaiErr
+	}
+	service.PostWssConsumeQuota(c, relayInfo, relayInfo.UpstreamModelName, usage.(*dto.RealtimeUsage), ratio, preConsumedQuota, userQuota, modelRatio, groupRatio, modelPrice, getModelPriceSuccess, "")
+	return nil
+}
+
+//func getWssPromptTokens(textRequest *dto.RealtimeEvent, info *relaycommon.RelayInfo) (int, error) {
+//	var promptTokens int
+//	var err error
+//	switch info.RelayMode {
+//	default:
+//		promptTokens, err = service.CountTokenRealtime(*textRequest, info.UpstreamModelName)
+//	}
+//	info.PromptTokens = promptTokens
+//	return promptTokens, err
+//}
+
+//func checkWssRequestSensitive(textRequest *dto.GeneralOpenAIRequest, info *relaycommon.RelayInfo) error {
+//	var err error
+//	switch info.RelayMode {
+//	case relayconstant.RelayModeChatCompletions:
+//		err = service.CheckSensitiveMessages(textRequest.Messages)
+//	case relayconstant.RelayModeCompletions:
+//		err = service.CheckSensitiveInput(textRequest.Prompt)
+//	case relayconstant.RelayModeModerations:
+//		err = service.CheckSensitiveInput(textRequest.Input)
+//	case relayconstant.RelayModeEmbeddings:
+//		err = service.CheckSensitiveInput(textRequest.Input)
+//	}
+//	return err
+//}

router/api-router.go

@@ -25,6 +25,7 @@ func SetApiRouter(router *gin.Engine) {
 		apiRouter.GET("/reset_password", middleware.CriticalRateLimit(), middleware.TurnstileCheck(), controller.SendPasswordResetEmail)
 		apiRouter.POST("/user/reset", middleware.CriticalRateLimit(), controller.ResetPassword)
 		apiRouter.GET("/oauth/github", middleware.CriticalRateLimit(), controller.GitHubOAuth)
+		apiRouter.GET("/oauth/linuxdo", middleware.CriticalRateLimit(), controller.LinuxdoOAuth)
 		apiRouter.GET("/oauth/state", middleware.CriticalRateLimit(), controller.GenerateOAuthCode)
 		apiRouter.GET("/oauth/wechat", middleware.CriticalRateLimit(), controller.WeChatAuth)
 		apiRouter.GET("/oauth/wechat/bind", middleware.CriticalRateLimit(), middleware.UserAuth(), controller.WeChatBind)

router/relay-router.go

@@ -22,32 +22,41 @@ func SetRelayRouter(router *gin.Engine) {
 		playgroundRouter.POST("/chat/completions", controller.Playground)
 	}
 	relayV1Router := router.Group("/v1")
-	relayV1Router.Use(middleware.TokenAuth(), middleware.Distribute())
+	relayV1Router.Use(middleware.TokenAuth())
 	{
-		relayV1Router.POST("/completions", controller.Relay)
-		relayV1Router.POST("/chat/completions", controller.Relay)
-		relayV1Router.POST("/edits", controller.Relay)
-		relayV1Router.POST("/images/generations", controller.Relay)
-		relayV1Router.POST("/images/edits", controller.RelayNotImplemented)
-		relayV1Router.POST("/images/variations", controller.RelayNotImplemented)
-		relayV1Router.POST("/embeddings", controller.Relay)
-		relayV1Router.POST("/engines/:model/embeddings", controller.Relay)
-		relayV1Router.POST("/audio/transcriptions", controller.Relay)
-		relayV1Router.POST("/audio/translations", controller.Relay)
-		relayV1Router.POST("/audio/speech", controller.Relay)
-		relayV1Router.GET("/files", controller.RelayNotImplemented)
-		relayV1Router.POST("/files", controller.RelayNotImplemented)
-		relayV1Router.DELETE("/files/:id", controller.RelayNotImplemented)
-		relayV1Router.GET("/files/:id", controller.RelayNotImplemented)
-		relayV1Router.GET("/files/:id/content", controller.RelayNotImplemented)
-		relayV1Router.POST("/fine-tunes", controller.RelayNotImplemented)
-		relayV1Router.GET("/fine-tunes", controller.RelayNotImplemented)
-		relayV1Router.GET("/fine-tunes/:id", controller.RelayNotImplemented)
-		relayV1Router.POST("/fine-tunes/:id/cancel", controller.RelayNotImplemented)
-		relayV1Router.GET("/fine-tunes/:id/events", controller.RelayNotImplemented)
-		relayV1Router.DELETE("/models/:model", controller.RelayNotImplemented)
-		relayV1Router.POST("/moderations", controller.Relay)
-		relayV1Router.POST("/rerank", controller.Relay)
+		// WebSocket 路由
+		wsRouter := relayV1Router.Group("")
+		wsRouter.Use(middleware.Distribute())
+		wsRouter.GET("/realtime", controller.WssRelay)
+	}
+	{
+		//http router
+		httpRouter := relayV1Router.Group("")
+		httpRouter.Use(middleware.Distribute())
+		httpRouter.POST("/completions", controller.Relay)
+		httpRouter.POST("/chat/completions", controller.Relay)
+		httpRouter.POST("/edits", controller.Relay)
+		httpRouter.POST("/images/generations", controller.Relay)
+		httpRouter.POST("/images/edits", controller.RelayNotImplemented)
+		httpRouter.POST("/images/variations", controller.RelayNotImplemented)
+		httpRouter.POST("/embeddings", controller.Relay)
+		httpRouter.POST("/engines/:model/embeddings", controller.Relay)
+		httpRouter.POST("/audio/transcriptions", controller.Relay)
+		httpRouter.POST("/audio/translations", controller.Relay)
+		httpRouter.POST("/audio/speech", controller.Relay)
+		httpRouter.GET("/files", controller.RelayNotImplemented)
+		httpRouter.POST("/files", controller.RelayNotImplemented)
+		httpRouter.DELETE("/files/:id", controller.RelayNotImplemented)
+		httpRouter.GET("/files/:id", controller.RelayNotImplemented)
+		httpRouter.GET("/files/:id/content", controller.RelayNotImplemented)
+		httpRouter.POST("/fine-tunes", controller.RelayNotImplemented)
+		httpRouter.GET("/fine-tunes", controller.RelayNotImplemented)
+		httpRouter.GET("/fine-tunes/:id", controller.RelayNotImplemented)
+		httpRouter.POST("/fine-tunes/:id/cancel", controller.RelayNotImplemented)
+		httpRouter.GET("/fine-tunes/:id/events", controller.RelayNotImplemented)
+		httpRouter.DELETE("/models/:model", controller.RelayNotImplemented)
+		httpRouter.POST("/moderations", controller.Relay)
+		httpRouter.POST("/rerank", controller.Relay)
 	}
 
 	relayMjRouter := router.Group("/mj")

service/audio.go

@@ -0,0 +1,31 @@
+package service
+
+import (
+	"encoding/base64"
+	"fmt"
+)
+
+func parseAudio(audioBase64 string, format string) (duration float64, err error) {
+	audioData, err := base64.StdEncoding.DecodeString(audioBase64)
+	if err != nil {
+		return 0, fmt.Errorf("base64 decode error: %v", err)
+	}
+
+	var samplesCount int
+	var sampleRate int
+
+	switch format {
+	case "pcm16":
+		samplesCount = len(audioData) / 2 // 16位 = 2字节每样本
+		sampleRate = 24000                // 24kHz
+	case "g711_ulaw", "g711_alaw":
+		samplesCount = len(audioData) // 8位 = 1字节每样本
+		sampleRate = 8000             // 8kHz
+	default:
+		samplesCount = len(audioData) // 8位 = 1字节每样本
+		sampleRate = 8000             // 8kHz
+	}
+
+	duration = float64(samplesCount) / float64(sampleRate)
+	return duration, nil
+}

service/log.go

@@ -2,6 +2,7 @@ package service
 
 import (
 	"github.com/gin-gonic/gin"
+	"one-api/dto"
 	relaycommon "one-api/relay/common"
 )
 
@@ -17,3 +18,27 @@ func GenerateTextOtherInfo(ctx *gin.Context, relayInfo *relaycommon.RelayInfo, m
 	other["admin_info"] = adminInfo
 	return other
 }
+
+func GenerateWssOtherInfo(ctx *gin.Context, relayInfo *relaycommon.RelayInfo, usage *dto.RealtimeUsage, modelRatio, groupRatio, completionRatio, audioRatio, audioCompletionRatio, modelPrice float64) map[string]interface{} {
+	info := GenerateTextOtherInfo(ctx, relayInfo, modelRatio, groupRatio, completionRatio, modelPrice)
+	info["ws"] = true
+	info["audio_input"] = usage.InputTokenDetails.AudioTokens
+	info["audio_output"] = usage.OutputTokenDetails.AudioTokens
+	info["text_input"] = usage.InputTokenDetails.TextTokens
+	info["text_output"] = usage.OutputTokenDetails.TextTokens
+	info["audio_ratio"] = audioRatio
+	info["audio_completion_ratio"] = audioCompletionRatio
+	return info
+}
+
+func GenerateAudioOtherInfo(ctx *gin.Context, relayInfo *relaycommon.RelayInfo, usage *dto.Usage, modelRatio, groupRatio, completionRatio, audioRatio, audioCompletionRatio, modelPrice float64) map[string]interface{} {
+	info := GenerateTextOtherInfo(ctx, relayInfo, modelRatio, groupRatio, completionRatio, modelPrice)
+	info["audio"] = true
+	info["audio_input"] = usage.PromptTokensDetails.AudioTokens
+	info["audio_output"] = usage.CompletionTokenDetails.AudioTokens
+	info["text_input"] = usage.PromptTokensDetails.TextTokens
+	info["text_output"] = usage.CompletionTokenDetails.TextTokens
+	info["audio_ratio"] = audioRatio
+	info["audio_completion_ratio"] = audioCompletionRatio
+	return info
+}

service/quota.go

@@ -0,0 +1,212 @@
+package service
+
+import (
+	"errors"
+	"fmt"
+	"github.com/gin-gonic/gin"
+	"math"
+	"one-api/common"
+	"one-api/dto"
+	"one-api/model"
+	relaycommon "one-api/relay/common"
+	"strings"
+	"time"
+)
+
+func PreWssConsumeQuota(ctx *gin.Context, relayInfo *relaycommon.RelayInfo, usage *dto.RealtimeUsage) error {
+	if relayInfo.UsePrice {
+		return nil
+	}
+	userQuota, err := model.GetUserQuota(relayInfo.UserId)
+	if err != nil {
+		return err
+	}
+
+	token, err := model.CacheGetTokenByKey(strings.TrimLeft(relayInfo.ApiKey, "sk-"))
+	if err != nil {
+		return err
+	}
+
+	modelName := relayInfo.UpstreamModelName
+	textInputTokens := usage.InputTokenDetails.TextTokens
+	textOutTokens := usage.OutputTokenDetails.TextTokens
+	audioInputTokens := usage.InputTokenDetails.AudioTokens
+	audioOutTokens := usage.OutputTokenDetails.AudioTokens
+
+	completionRatio := common.GetCompletionRatio(modelName)
+	audioRatio := common.GetAudioRatio(relayInfo.UpstreamModelName)
+	audioCompletionRatio := common.GetAudioCompletionRatio(modelName)
+	groupRatio := common.GetGroupRatio(relayInfo.Group)
+	modelRatio := common.GetModelRatio(modelName)
+
+	ratio := groupRatio * modelRatio
+
+	quota := textInputTokens + int(math.Round(float64(textOutTokens)*completionRatio))
+	quota += int(math.Round(float64(audioInputTokens)*audioRatio)) + int(math.Round(float64(audioOutTokens)*audioRatio*audioCompletionRatio))
+
+	quota = int(math.Round(float64(quota) * ratio))
+	if ratio != 0 && quota <= 0 {
+		quota = 1
+	}
+
+	if userQuota < quota {
+		return errors.New(fmt.Sprintf("用户额度不足，剩余额度为 %d", userQuota))
+	}
+
+	if token.RemainQuota < quota {
+		return errors.New(fmt.Sprintf("令牌额度不足，剩余额度为 %d", token.RemainQuota))
+	}
+
+	err = model.PostConsumeTokenQuota(relayInfo, 0, quota, 0, false)
+	if err != nil {
+		return err
+	}
+	common.LogInfo(ctx, "realtime streaming consume quota success, quota: "+fmt.Sprintf("%d", quota))
+	err = model.CacheUpdateUserQuota(relayInfo.UserId)
+	if err != nil {
+		return err
+	}
+	return nil
+}
+
+func PostWssConsumeQuota(ctx *gin.Context, relayInfo *relaycommon.RelayInfo, modelName string,
+	usage *dto.RealtimeUsage, ratio float64, preConsumedQuota int, userQuota int, modelRatio float64,
+	groupRatio float64,
+	modelPrice float64, usePrice bool, extraContent string) {
+
+	useTimeSeconds := time.Now().Unix() - relayInfo.StartTime.Unix()
+	textInputTokens := usage.InputTokenDetails.TextTokens
+	textOutTokens := usage.OutputTokenDetails.TextTokens
+
+	audioInputTokens := usage.InputTokenDetails.AudioTokens
+	audioOutTokens := usage.OutputTokenDetails.AudioTokens
+
+	tokenName := ctx.GetString("token_name")
+	completionRatio := common.GetCompletionRatio(modelName)
+	audioRatio := common.GetAudioRatio(relayInfo.UpstreamModelName)
+	audioCompletionRatio := common.GetAudioCompletionRatio(modelName)
+
+	quota := 0
+	if !usePrice {
+		quota = int(math.Round(float64(textInputTokens) + float64(textOutTokens)*completionRatio))
+		quota += int(math.Round(float64(audioInputTokens)*audioRatio + float64(audioOutTokens)*audioRatio*audioCompletionRatio))
+		quota = int(math.Round(float64(quota) * ratio))
+		if ratio != 0 && quota <= 0 {
+			quota = 1
+		}
+	} else {
+		quota = int(modelPrice * common.QuotaPerUnit * groupRatio)
+	}
+	totalTokens := usage.TotalTokens
+	var logContent string
+	if !usePrice {
+		logContent = fmt.Sprintf("模型倍率 %.2f，补全倍率 %.2f，音频倍率 %.2f，音频补全倍率 %.2f，分组倍率 %.2f", modelRatio, completionRatio, audioRatio, audioCompletionRatio, groupRatio)
+	} else {
+		logContent = fmt.Sprintf("模型价格 %.2f，分组倍率 %.2f", modelPrice, groupRatio)
+	}
+
+	// record all the consume log even if quota is 0
+	if totalTokens == 0 {
+		// in this case, must be some error happened
+		// we cannot just return, because we may have to return the pre-consumed quota
+		quota = 0
+		logContent += fmt.Sprintf("（可能是上游超时）")
+		common.LogError(ctx, fmt.Sprintf("total tokens is 0, cannot consume quota, userId %d, channelId %d, "+
+			"tokenId %d, model %s， pre-consumed quota %d", relayInfo.UserId, relayInfo.ChannelId, relayInfo.TokenId, modelName, preConsumedQuota))
+	} else {
+		//if sensitiveResp != nil {
+		//	logContent += fmt.Sprintf("，敏感词：%s", strings.Join(sensitiveResp.SensitiveWords, ", "))
+		//}
+		//quotaDelta := quota - preConsumedQuota
+		//if quotaDelta != 0 {
+		//	err := model.PostConsumeTokenQuota(relayInfo, userQuota, quotaDelta, preConsumedQuota, true)
+		//	if err != nil {
+		//		common.LogError(ctx, "error consuming token remain quota: "+err.Error())
+		//	}
+		//}
+
+		//err := model.CacheUpdateUserQuota(relayInfo.UserId)
+		//if err != nil {
+		//	common.LogError(ctx, "error update user quota cache: "+err.Error())
+		//}
+		model.UpdateUserUsedQuotaAndRequestCount(relayInfo.UserId, quota)
+		model.UpdateChannelUsedQuota(relayInfo.ChannelId, quota)
+	}
+
+	logModel := modelName
+	if extraContent != "" {
+		logContent += ", " + extraContent
+	}
+	other := GenerateWssOtherInfo(ctx, relayInfo, usage, modelRatio, groupRatio, completionRatio, audioRatio, audioCompletionRatio, modelPrice)
+	model.RecordConsumeLog(ctx, relayInfo.UserId, relayInfo.ChannelId, usage.InputTokens, usage.OutputTokens, logModel,
+		tokenName, quota, logContent, relayInfo.TokenId, userQuota, int(useTimeSeconds), relayInfo.IsStream, other)
+}
+
+func PostAudioConsumeQuota(ctx *gin.Context, relayInfo *relaycommon.RelayInfo,
+	usage *dto.Usage, ratio float64, preConsumedQuota int, userQuota int, modelRatio float64,
+	groupRatio float64,
+	modelPrice float64, usePrice bool, extraContent string) {
+
+	useTimeSeconds := time.Now().Unix() - relayInfo.StartTime.Unix()
+	textInputTokens := usage.PromptTokensDetails.TextTokens
+	textOutTokens := usage.CompletionTokenDetails.TextTokens
+
+	audioInputTokens := usage.PromptTokensDetails.AudioTokens
+	audioOutTokens := usage.CompletionTokenDetails.AudioTokens
+
+	tokenName := ctx.GetString("token_name")
+	completionRatio := common.GetCompletionRatio(relayInfo.UpstreamModelName)
+	audioRatio := common.GetAudioRatio(relayInfo.UpstreamModelName)
+	audioCompletionRatio := common.GetAudioCompletionRatio(relayInfo.UpstreamModelName)
+
+	quota := 0
+	if !usePrice {
+		quota = int(math.Round(float64(textInputTokens) + float64(textOutTokens)*completionRatio))
+		quota += int(math.Round(float64(audioInputTokens)*audioRatio + float64(audioOutTokens)*audioRatio*audioCompletionRatio))
+		quota = int(math.Round(float64(quota) * ratio))
+		if ratio != 0 && quota <= 0 {
+			quota = 1
+		}
+	} else {
+		quota = int(modelPrice * common.QuotaPerUnit * groupRatio)
+	}
+	totalTokens := usage.TotalTokens
+	var logContent string
+	if !usePrice {
+		logContent = fmt.Sprintf("模型倍率 %.2f，补全倍率 %.2f，音频倍率 %.2f，音频补全倍率 %.2f，分组倍率 %.2f", modelRatio, completionRatio, audioRatio, audioCompletionRatio, groupRatio)
+	} else {
+		logContent = fmt.Sprintf("模型价格 %.2f，分组倍率 %.2f", modelPrice, groupRatio)
+	}
+
+	// record all the consume log even if quota is 0
+	if totalTokens == 0 {
+		// in this case, must be some error happened
+		// we cannot just return, because we may have to return the pre-consumed quota
+		quota = 0
+		logContent += fmt.Sprintf("（可能是上游超时）")
+		common.LogError(ctx, fmt.Sprintf("total tokens is 0, cannot consume quota, userId %d, channelId %d, "+
+			"tokenId %d, model %s， pre-consumed quota %d", relayInfo.UserId, relayInfo.ChannelId, relayInfo.TokenId, relayInfo.UpstreamModelName, preConsumedQuota))
+	} else {
+		quotaDelta := quota - preConsumedQuota
+		if quotaDelta != 0 {
+			err := model.PostConsumeTokenQuota(relayInfo, userQuota, quotaDelta, preConsumedQuota, true)
+			if err != nil {
+				common.LogError(ctx, "error consuming token remain quota: "+err.Error())
+			}
+		}
+		err := model.CacheUpdateUserQuota(relayInfo.UserId)
+		if err != nil {
+			common.LogError(ctx, "error update user quota cache: "+err.Error())
+		}
+		model.UpdateUserUsedQuotaAndRequestCount(relayInfo.UserId, quota)
+		model.UpdateChannelUsedQuota(relayInfo.ChannelId, quota)
+	}
+
+	logModel := relayInfo.UpstreamModelName
+	if extraContent != "" {
+		logContent += ", " + extraContent
+	}
+	other := GenerateAudioOtherInfo(ctx, relayInfo, usage, modelRatio, groupRatio, completionRatio, audioRatio, audioCompletionRatio, modelPrice)
+	model.RecordConsumeLog(ctx, relayInfo.UserId, relayInfo.ChannelId, usage.PromptTokens, usage.CompletionTokens, logModel,
+		tokenName, quota, logContent, relayInfo.TokenId, userQuota, int(useTimeSeconds), relayInfo.IsStream, other)
+}

service/relay.go

@@ -5,6 +5,7 @@ import (
 	"errors"
 	"fmt"
 	"github.com/gin-gonic/gin"
+	"github.com/gorilla/websocket"
 	"net/http"
 	"one-api/common"
 	"one-api/dto"
@@ -42,11 +43,47 @@ func Done(c *gin.Context) {
 	_ = StringData(c, "[DONE]")
 }
 
+func WssString(c *gin.Context, ws *websocket.Conn, str string) error {
+	if ws == nil {
+		common.LogError(c, "websocket connection is nil")
+		return errors.New("websocket connection is nil")
+	}
+	//common.LogInfo(c, fmt.Sprintf("sending message: %s", str))
+	return ws.WriteMessage(1, []byte(str))
+}
+
+func WssObject(c *gin.Context, ws *websocket.Conn, object interface{}) error {
+	jsonData, err := json.Marshal(object)
+	if err != nil {
+		return fmt.Errorf("error marshalling object: %w", err)
+	}
+	if ws == nil {
+		common.LogError(c, "websocket connection is nil")
+		return errors.New("websocket connection is nil")
+	}
+	//common.LogInfo(c, fmt.Sprintf("sending message: %s", jsonData))
+	return ws.WriteMessage(1, jsonData)
+}
+
+func WssError(c *gin.Context, ws *websocket.Conn, openaiError dto.OpenAIError) {
+	errorObj := &dto.RealtimeEvent{
+		Type:    "error",
+		EventId: GetLocalRealtimeID(c),
+		Error:   &openaiError,
+	}
+	_ = WssObject(c, ws, errorObj)
+}
+
 func GetResponseID(c *gin.Context) string {
-	logID := c.GetString("X-Oneapi-Request-Id")
+	logID := c.GetString(common.RequestIdKey)
 	return fmt.Sprintf("chatcmpl-%s", logID)
 }
 
+func GetLocalRealtimeID(c *gin.Context) string {
+	logID := c.GetString(common.RequestIdKey)
+	return fmt.Sprintf("evt_%s", logID)
+}
+
 func GenerateStopResponse(id string, createAt int64, model string, finishReason string) *dto.ChatCompletionsStreamResponse {
 	return &dto.ChatCompletionsStreamResponse{
 		Id:                id,

service/token_counter.go

@@ -11,6 +11,7 @@ import (
 	"one-api/common"
 	"one-api/constant"
 	"one-api/dto"
+	relaycommon "one-api/relay/common"
 	"strings"
 	"unicode/utf8"
 )
@@ -191,6 +192,72 @@ func CountTokenChatRequest(request dto.GeneralOpenAIRequest, model string) (int,
 	return tkm, nil
 }
 
+func CountTokenRealtime(info *relaycommon.RelayInfo, request dto.RealtimeEvent, model string) (int, int, error) {
+	audioToken := 0
+	textToken := 0
+	switch request.Type {
+	case dto.RealtimeEventTypeSessionUpdate:
+		if request.Session != nil {
+			msgTokens, err := CountTextToken(request.Session.Instructions, model)
+			if err != nil {
+				return 0, 0, err
+			}
+			textToken += msgTokens
+		}
+	case dto.RealtimeEventResponseAudioDelta:
+		// count audio token
+		atk, err := CountAudioTokenOutput(request.Delta, info.OutputAudioFormat)
+		if err != nil {
+			return 0, 0, fmt.Errorf("error counting audio token: %v", err)
+		}
+		audioToken += atk
+	case dto.RealtimeEventResponseAudioTranscriptionDelta, dto.RealtimeEventResponseFunctionCallArgumentsDelta:
+		// count text token
+		tkm, err := CountTextToken(request.Delta, model)
+		if err != nil {
+			return 0, 0, fmt.Errorf("error counting text token: %v", err)
+		}
+		textToken += tkm
+	case dto.RealtimeEventInputAudioBufferAppend:
+		// count audio token
+		atk, err := CountAudioTokenInput(request.Audio, info.InputAudioFormat)
+		if err != nil {
+			return 0, 0, fmt.Errorf("error counting audio token: %v", err)
+		}
+		audioToken += atk
+	case dto.RealtimeEventConversationItemCreated:
+		if request.Item != nil {
+			switch request.Item.Type {
+			case "message":
+				for _, content := range request.Item.Content {
+					if content.Type == "input_text" {
+						tokens, err := CountTextToken(content.Text, model)
+						if err != nil {
+							return 0, 0, err
+						}
+						textToken += tokens
+					}
+				}
+			}
+		}
+	case dto.RealtimeEventTypeResponseDone:
+		// count tools token
+		if !info.IsFirstRequest {
+			if info.RealtimeTools != nil && len(info.RealtimeTools) > 0 {
+				for _, tool := range info.RealtimeTools {
+					toolTokens, err := CountTokenInput(tool, model)
+					if err != nil {
+						return 0, 0, err
+					}
+					textToken += 8
+					textToken += toolTokens
+				}
+			}
+		}
+	}
+	return textToken, audioToken, nil
+}
+
 func CountTokenMessages(messages []dto.Message, model string, stream bool) (int, error) {
 	//recover when panic
 	tokenEncoder := getTokenEncoder(model)
@@ -223,7 +290,7 @@ func CountTokenMessages(messages []dto.Message, model string, stream bool) (int,
 			} else {
 				arrayContent := message.ParseContent()
 				for _, m := range arrayContent {
-					if m.Type == "image_url" {
+					if m.Type == dto.ContentTypeImageURL {
 						imageUrl := m.ImageUrl.(dto.MessageImageUrl)
 						imageTokenNum, err := getImageToken(&imageUrl, model, stream)
 						if err != nil {
@@ -231,6 +298,9 @@ func CountTokenMessages(messages []dto.Message, model string, stream bool) (int,
 						}
 						tokenNum += imageTokenNum
 						log.Printf("image token num: %d", imageTokenNum)
+					} else if m.Type == dto.ContentTypeInputAudio {
+						// TODO: 音频token数量计算
+						tokenNum += 100
 					} else {
 						tokenNum += getTokenNum(tokenEncoder, m.Text)
 					}
@@ -245,13 +315,13 @@ func CountTokenMessages(messages []dto.Message, model string, stream bool) (int,
 func CountTokenInput(input any, model string) (int, error) {
 	switch v := input.(type) {
 	case string:
-		return CountTokenText(v, model)
+		return CountTextToken(v, model)
 	case []string:
 		text := ""
 		for _, s := range v {
 			text += s
 		}
-		return CountTokenText(text, model)
+		return CountTextToken(text, model)
 	}
 	return CountTokenInput(fmt.Sprintf("%v", input), model)
 }
@@ -273,16 +343,44 @@ func CountTokenStreamChoices(messages []dto.ChatCompletionsStreamResponseChoice,
 	return tokens
 }
 
-func CountAudioToken(text string, model string) (int, error) {
+func CountTTSToken(text string, model string) (int, error) {
 	if strings.HasPrefix(model, "tts") {
 		return utf8.RuneCountInString(text), nil
 	} else {
-		return CountTokenText(text, model)
+		return CountTextToken(text, model)
 	}
 }
 
-// CountTokenText 统计文本的token数量，仅当文本包含敏感词，返回错误，同时返回token数量
-func CountTokenText(text string, model string) (int, error) {
+func CountAudioTokenInput(audioBase64 string, audioFormat string) (int, error) {
+	if audioBase64 == "" {
+		return 0, nil
+	}
+	duration, err := parseAudio(audioBase64, audioFormat)
+	if err != nil {
+		return 0, err
+	}
+	return int(duration / 60 * 100 / 0.06), nil
+}
+
+func CountAudioTokenOutput(audioBase64 string, audioFormat string) (int, error) {
+	if audioBase64 == "" {
+		return 0, nil
+	}
+	duration, err := parseAudio(audioBase64, audioFormat)
+	if err != nil {
+		return 0, err
+	}
+	return int(duration / 60 * 200 / 0.24), nil
+}
+
+//func CountAudioToken(sec float64, audioType string) {
+//	if audioType == "input" {
+//
+//	}
+//}
+
+// CountTextToken 统计文本的token数量，仅当文本包含敏感词，返回错误，同时返回token数量
+func CountTextToken(text string, model string) (int, error) {
 	var err error
 	tokenEncoder := getTokenEncoder(model)
 	return getTokenNum(tokenEncoder, text), err

service/usage_helpr.go

@@ -19,7 +19,7 @@ import (
 func ResponseText2Usage(responseText string, modeName string, promptTokens int) (*dto.Usage, error) {
 	usage := &dto.Usage{}
 	usage.PromptTokens = promptTokens
-	ctkm, err := CountTokenText(responseText, modeName)
+	ctkm, err := CountTextToken(responseText, modeName)
 	usage.CompletionTokens = ctkm
 	usage.TotalTokens = usage.PromptTokens + usage.CompletionTokens
 	return usage, err

web/src/App.js

@@ -26,6 +26,7 @@ import Midjourney from './pages/Midjourney';
 import Pricing from './pages/Pricing/index.js';
 import Task from "./pages/Task/index.js";
 import Playground from './components/Playground.js';
+import LinuxDoOAuth from './components/LinuxDoOAuth.js';
 
 const Home = lazy(() => import('./pages/Home'));
 const Detail = lazy(() => import('./pages/Detail'));
@@ -181,6 +182,14 @@ function App() {
             </Suspense>
           }
         />
+        <Route
+          path='/oauth/linuxdo'
+          element={
+            <Suspense fallback={<Loading></Loading>}>
+              <LinuxDoOAuth />
+            </Suspense>
+          }
+        />
         <Route
           path='/setting'
           element={

web/src/components/LinuxDoIcon.js

@@ -0,0 +1,27 @@
+import React from 'react';
+import { Icon } from '@douyinfe/semi-ui';
+
+const LinuxDoIcon = (props) => {
+  function CustomIcon() {
+    return (
+      <svg
+        className='icon'
+        viewBox='0 0 24 24'
+        version='1.1'
+        xmlns='http://www.w3.org/2000/svg'
+        width='1em'
+        height='1em'
+        {...props}
+      >
+        <path
+          d='M19.7,17.6c-0.1-0.2-0.2-0.4-0.2-0.6c0-0.4-0.2-0.7-0.5-1c-0.1-0.1-0.3-0.2-0.4-0.2c0.6-1.8-0.3-3.6-1.3-4.9c0,0,0,0,0,0c-0.8-1.2-2-2.1-1.9-3.7c0-1.9,0.2-5.4-3.3-5.1C8.5,2.3,9.5,6,9.4,7.3c0,1.1-0.5,2.2-1.3,3.1c-0.2,0.2-0.4,0.5-0.5,0.7c-1,1.2-1.5,2.8-1.5,4.3c-0.2,0.2-0.4,0.4-0.5,0.6c-0.1,0.1-0.2,0.2-0.2,0.3c-0.1,0.1-0.3,0.2-0.5,0.3c-0.4,0.1-0.7,0.3-0.9,0.7c-0.1,0.3-0.2,0.7-0.1,1.1c0.1,0.2,0.1,0.4,0,0.7c-0.2,0.4-0.2,0.9,0,1.4c0.3,0.4,0.8,0.5,1.5,0.6c0.5,0,1.1,0.2,1.6,0.4l0,0c0.5,0.3,1.1,0.5,1.7,0.5c0.3,0,0.7-0.1,1-0.2c0.3-0.2,0.5-0.4,0.6-0.7c0.4,0,1-0.2,1.7-0.2c0.6,0,1.2,0.2,2,0.1c0,0.1,0,0.2,0.1,0.3c0.2,0.5,0.7,0.9,1.3,1c0.1,0,0.1,0,0.2,0c0.8-0.1,1.6-0.5,2.1-1.1l0,0c0.4-0.4,0.9-0.7,1.4-0.9c0.6-0.3,1-0.5,1.1-1C20.3,18.6,20.1,18.2,19.7,17.6z M12.8,4.8c0.6,0.1,1.1,0.6,1,1.2c0,0.3-0.1,0.6-0.3,0.9c0,0,0,0-0.1,0c-0.2-0.1-0.3-0.1-0.4-0.2c0.1-0.1,0.1-0.3,0.2-0.5c0-0.4-0.2-0.7-0.4-0.7c-0.3,0-0.5,0.3-0.5,0.7c0,0,0,0.1,0,0.1c-0.1-0.1-0.3-0.1-0.4-0.2c0,0,0-0.1,0-0.1C11.8,5.5,12.2,4.9,12.8,4.8z M12.5,6.8c0.1,0.1,0.3,0.2,0.4,0.2c0.1,0,0.3,0.1,0.4,0.2c0.2,0.1,0.4,0.2,0.4,0.5c0,0.3-0.3,0.6-0.9,0.8c-0.2,0.1-0.3,0.1-0.4,0.2c-0.3,0.2-0.6,0.3-1,0.3c-0.3,0-0.6-0.2-0.8-0.4c-0.1-0.1-0.2-0.2-0.4-0.3C10.1,8.2,9.9,8,9.8,7.7c0-0.1,0.1-0.2,0.2-0.3c0.3-0.2,0.4-0.3,0.5-0.4l0.1-0.1c0.2-0.3,0.6-0.5,1-0.5C11.9,6.5,12.2,6.6,12.5,6.8z M10.4,5c0.4,0,0.7,0.4,0.8,1.1c0,0.1,0,0.1,0,0.2c-0.1,0-0.3,0.1-0.4,0.2c0,0,0-0.1,0-0.2c0-0.3-0.2-0.6-0.4-0.5c-0.2,0-0.3,0.3-0.3,0.6c0,0.2,0.1,0.3,0.2,0.4l0,0c0,0-0.1,0.1-0.2,0.1C9.9,6.7,9.7,6.4,9.7,6.1C9.7,5.5,10,5,10.4,5z M9.4,21.1c-0.7,0.3-1.6,0.2-2.2-0.2c-0.6-0.3-1.1-0.4-1.8-0.4c-0.5-0.1-1-0.1-1.1-0.3c-0.1-0.2-0.1-0.5,0.1-1c0.1-0.3,0.1-0.6,0-0.9c-0.1-0.3-0.1-0.5,0-0.8C4.5,17.2,4.7,17.1,5,17c0.3-0.1,0.5-0.2,0.7-0.4c0.1-0.1,0.2-0.2,0.3-0.4c0.3-0.4,0.5-0.6,0.8-0.6c0.6,0.1,1.1,1,1.5,1.9c0.2,0.3,0.4,0.7,0.7,1c0.4,0.5,0.9,1.2,0.9,1.6C9.9,20.6,9.7,20.9,9.4,21.1z M14.3,18.9c0,0.1,0,0.1-0.1,0.2c-1.2,0.9-2.8,1-4.1,0.3c-0.2-0.3-0.4-0.6-0.6-0.9c0.9-0.1,0.7-1.3-1.2-2.5c-2-1.3-0.6-3.7,0.1-4.8c0.1-0.1,0.1,0-0.3,0.8c-0.3,0.6-0.9,2.1-0.1,3.2c0-0.8,0.2-1.6,0.5-2.4c0.7-1.3,1.2-2.8,1.5-4.3c0.1,0.1,0.1,0.1,0.2,0.1c0.1,0.1,0.2,0.2,0.3,0.2c0.2,0.3,0.6,0.4,0.9,0.4c0,0,0.1,0,0.1,0c0.4,0,0.8-0.1,1.1-0.4c0.1-0.1,0.2-0.2,0.4-0.2c0.3-0.1,0.6-0.3,0.9-0.6c0.4,1.3,0.8,2.5,1.4,3.6c0.4,0.8,0.7,1.6,0.9,2.5c0.3,0,0.7,0.1,1,0.3c0.8,0.4,1.1,0.7,1,1.2c-0.1,0-0.1,0-0.2,0c0-0.3-0.2-0.6-0.9-0.9c-0.7-0.3-1.3-0.3-1.5,0.4c-0.1,0-0.2,0.1-0.3,0.1c-0.8,0.4-0.8,1.5-0.9,2.6C14.5,18.2,14.4,18.5,14.3,18.9z M18.9,19.5c-0.6,0.2-1.1,0.6-1.5,1.1c-0.4,0.6-1.1,1-1.9,0.9c-0.4,0-0.8-0.3-0.9-0.7c-0.1-0.6-0.1-1.2,0.2-1.8c0.1-0.4,0.2-0.7,0.3-1.1c0.1-1.2,0.1-1.9,0.6-2.2h0c0,0.5,0.3,0.8,0.7,1c0.5,0,1-0.1,1.4-0.5c0.1,0,0.1,0,0.2,0c0.3,0,0.5,0,0.7,0.2c0.2,0.2,0.3,0.5,0.3,0.7c0,0.3,0.2,0.6,0.3,0.9c0.5,0.5,0.5,0.8,0.5,0.9C19.7,19.1,19.3,19.3,18.9,19.5z M9.9,7.5c-0.1,0-0.1,0-0.1,0.1c0,0,0,0.1,0.1,0.1c0,0,0,0,0,0c0.1,0,0.1,0.1,0.1,0.1c0.3,0.4,0.8,0.6,1.4,0.7c0.5-0.1,1-0.2,1.5-0.6c0.2-0.1,0.4-0.2,0.6-0.3c0.1,0,0.1-0.1,0.1-0.1c0-0.1,0-0.1-0.1-0.1l0,0c-0.2,0.1-0.5,0.2-0.7,0.3c-0.4,0.3-0.9,0.5-1.4,0.5c-0.5,0-0.9-0.3-1.2-0.6C10.1,7.6,10,7.5,9.9,7.5z'
+          fill='currentColor'
+        />
+      </svg>
+    );
+  }
+
+  return <Icon svg={<CustomIcon />} />;
+};
+
+export default LinuxDoIcon;

web/src/components/LinuxDoOAuth.js

@@ -0,0 +1,61 @@
+import React, { useContext, useEffect, useState } from 'react';
+import { Dimmer, Loader, Segment } from 'semantic-ui-react';
+import { useNavigate, useSearchParams } from 'react-router-dom';
+import { API, showError, showSuccess, updateAPI } from '../helpers';
+import { UserContext } from '../context/User';
+import { setUserData } from '../helpers/data.js';
+
+const LinuxDoOAuth = () => {
+  const [searchParams, setSearchParams] = useSearchParams();
+
+  const [userState, userDispatch] = useContext(UserContext);
+  const [prompt, setPrompt] = useState('处理中...');
+  const [processing, setProcessing] = useState(true);
+
+  let navigate = useNavigate();
+
+  const sendCode = async (code, state, count) => {
+    const res = await API.get(`/api/oauth/linuxdo?code=${code}&state=${state}`);
+    const { success, message, data } = res.data;
+    if (success) {
+      if (message === 'bind') {
+        showSuccess('绑定成功！');
+        navigate('/setting');
+      } else {
+        userDispatch({ type: 'login', payload: data });
+        localStorage.setItem('user', JSON.stringify(data));
+        setUserData(data);
+        updateAPI()
+        showSuccess('登录成功！');
+        navigate('/token');
+      }
+    } else {
+      showError(message);
+      if (count === 0) {
+        setPrompt(`操作失败，重定向至登录界面中...`);
+        navigate('/setting'); // in case this is failed to bind GitHub
+        return;
+      }
+      count++;
+      setPrompt(`出现错误，第 ${count} 次重试中...`);
+      await new Promise((resolve) => setTimeout(resolve, count * 2000));
+      await sendCode(code, state, count);
+    }
+  };
+
+  useEffect(() => {
+    let code = searchParams.get('code');
+    let state = searchParams.get('state');
+    sendCode(code, state, 0).then();
+  }, []);
+
+  return (
+    <Segment style={{ minHeight: '300px' }}>
+      <Dimmer active inverted>
+        <Loader size='large'>{prompt}</Loader>
+      </Dimmer>
+    </Segment>
+  );
+};
+
+export default LinuxDoOAuth;

web/src/components/LoginForm.js

@@ -1,8 +1,15 @@
 import React, { useContext, useEffect, useState } from 'react';
 import { Link, useNavigate, useSearchParams } from 'react-router-dom';
 import { UserContext } from '../context/User';
-import { API, getLogo, showError, showInfo, showSuccess, updateAPI } from '../helpers';
-import { onGitHubOAuthClicked } from './utils';
+import {
+  API,
+  getLogo,
+  showError,
+  showInfo,
+  showSuccess,
+  updateAPI,
+} from '../helpers';
+import { onGitHubOAuthClicked, onLinuxDOOAuthClicked } from './utils';
 import Turnstile from 'react-turnstile';
 import {
   Button,
@@ -17,9 +24,10 @@ import Title from '@douyinfe/semi-ui/lib/es/typography/title';
 import Text from '@douyinfe/semi-ui/lib/es/typography/text';
 import TelegramLoginButton from 'react-telegram-login';
 
-import { IconGithubLogo } from '@douyinfe/semi-icons';
+import { IconGithubLogo, IconAlarm } from '@douyinfe/semi-icons';
 import WeChatIcon from './WeChatIcon';
 import { setUserData } from '../helpers/data.js';
+import LinuxDoIcon from './LinuxDoIcon.js';
 
 const LoginForm = () => {
   const [inputs, setInputs] = useState({
@@ -72,7 +80,7 @@ const LoginForm = () => {
       userDispatch({ type: 'login', payload: data });
       localStorage.setItem('user', JSON.stringify(data));
       setUserData(data);
-      updateAPI()
+      updateAPI();
       navigate('/');
       showSuccess('登录成功！');
       setShowWeChatLoginModal(false);
@@ -103,7 +111,7 @@ const LoginForm = () => {
       if (success) {
         userDispatch({ type: 'login', payload: data });
         setUserData(data);
-        updateAPI()
+        updateAPI();
         showSuccess('登录成功！');
         if (username === 'root' && password === '123456') {
           Modal.error({
@@ -146,7 +154,7 @@ const LoginForm = () => {
       localStorage.setItem('user', JSON.stringify(data));
       showSuccess('登录成功！');
       setUserData(data);
-      updateAPI()
+      updateAPI();
       navigate('/');
     } else {
       showError(message);
@@ -214,7 +222,8 @@ const LoginForm = () => {
                 </div>
                 {status.github_oauth ||
                 status.wechat_login ||
-                status.telegram_oauth ? (
+                status.telegram_oauth ||
+                status.linuxdo_oauth ? (
                   <>
                     <Divider margin='12px' align='center'>
                       第三方登录
@@ -237,6 +246,16 @@ const LoginForm = () => {
                       ) : (
                         <></>
                       )}
+                      {status.linuxdo_oauth ? (
+                        <Button
+                          icon={<LinuxDoIcon />}
+                          onClick={() =>
+                            onLinuxDOOAuthClicked(status.linuxdo_client_id)
+                          }
+                        />
+                      ) : (
+                        <></>
+                      )}
                       {status.wechat_login ? (
                         <Button
                           type='primary'

web/src/components/LogsTable.js

@@ -11,7 +11,7 @@ import {
 
 import {
   Avatar,
-  Button,
+  Button, Descriptions,
   Form,
   Layout,
   Modal,
@@ -20,14 +20,15 @@ import {
   Spin,
   Table,
   Tag,
-  Tooltip,
+  Tooltip
 } from '@douyinfe/semi-ui';
 import { ITEMS_PER_PAGE } from '../constants';
 import {
+  renderAudioModelPrice,
   renderModelPrice,
   renderNumber,
   renderQuota,
-  stringToColor,
+  stringToColor
 } from '../helpers/render';
 import Paragraph from '@douyinfe/semi-ui/lib/es/typography/paragraph';
 import { getLogOther } from '../helpers/other.js';
@@ -384,31 +385,31 @@ const LogsTable = () => {
             </Paragraph>
           );
         }
-        let content = renderModelPrice(
-          record.prompt_tokens,
-          record.completion_tokens,
-          other.model_ratio,
-          other.model_price,
-          other.completion_ratio,
-          other.group_ratio,
-        );
+
+        // let content = renderModelPrice(
+        //   record.prompt_tokens,
+        //   record.completion_tokens,
+        //   other.model_ratio,
+        //   other.model_price,
+        //   other.completion_ratio,
+        //   other.group_ratio,
+        // );
         return (
-          <Tooltip content={content}>
             <Paragraph
-              ellipsis={{
-                rows: 2,
-              }}
-              style={{ maxWidth: 240 }}
+                ellipsis={{
+                  rows: 2,
+                }}
+                style={{ maxWidth: 240 }}
             >
-              {text}
+              调用消费
             </Paragraph>
-          </Tooltip>
         );
       },
     },
   ];
 
   const [logs, setLogs] = useState([]);
+  const [expandData, setExpandData] = useState({});
   const [showStat, setShowStat] = useState(false);
   const [loading, setLoading] = useState(false);
   const [loadingStat, setLoadingStat] = useState(false);
@@ -512,10 +513,89 @@ const LogsTable = () => {
   };
 
   const setLogsFormat = (logs) => {
+    let expandDatesLocal = {};
     for (let i = 0; i < logs.length; i++) {
       logs[i].timestamp2string = timestamp2string(logs[i].created_at);
-      logs[i].key = '' + logs[i].id;
+      logs[i].key = i;
+      let other = getLogOther(logs[i].other);
+      let expandDataLocal = [];
+      if (isAdmin()) {
+        // let content = '渠道：' + logs[i].channel;
+        // if (other.admin_info !== undefined) {
+        //   if (
+        //     other.admin_info.use_channel !== null &&
+        //     other.admin_info.use_channel !== undefined &&
+        //     other.admin_info.use_channel !== ''
+        //   ) {
+        //     // channel id array
+        //     let useChannel = other.admin_info.use_channel;
+        //     let useChannelStr = useChannel.join('->');
+        //     content = `渠道：${useChannelStr}`;
+        //   }
+        // }
+        // expandDataLocal.push({
+        //   key: '渠道重试',
+        //   value: content,
+        // })
+      }
+      if (other?.ws || other?.audio) {
+        expandDataLocal.push({
+          key: '语音输入',
+          value: other.audio_input,
+        });
+        expandDataLocal.push({
+          key: '语音输出',
+          value: other.audio_output,
+        });
+        expandDataLocal.push({
+          key: '文字输入',
+          value: other.text_input,
+        });
+        expandDataLocal.push({
+          key: '文字输出',
+          value: other.text_output,
+        });
+      }
+      expandDataLocal.push({
+        key: '日志详情',
+        value: logs[i].content,
+      })
+      if (logs[i].type === 2) {
+        let content = '';
+        if (other?.ws || other?.audio) {
+          content = renderAudioModelPrice(
+            other.text_input,
+            other.text_output,
+            other.model_ratio,
+            other.model_price,
+            other.completion_ratio,
+            other.audio_input,
+            other.audio_output,
+            other?.audio_ratio,
+            other?.audio_completion_ratio,
+            other.group_ratio,
+          );
+        } else {
+          content = renderModelPrice(
+            logs[i].prompt_tokens,
+            logs[i].completion_tokens,
+            other.model_ratio,
+            other.model_price,
+            other.completion_ratio,
+            other.group_ratio,
+          );
+        }
+        expandDataLocal.push({
+          key: '计费过程',
+          value: content,
+        });
+      }
+
+      expandDatesLocal[logs[i].key] = expandDataLocal;
     }
+
+    setExpandData(expandDatesLocal);
+
     setLogs(logs);
   };
 
@@ -588,6 +668,10 @@ const LogsTable = () => {
     handleEyeClick();
   }, []);
 
+  const expandRowRender = (record, index) => {
+    return <Descriptions data={expandData[record.key]} />;
+  };
+
   return (
     <>
       <Layout>
@@ -686,7 +770,10 @@ const LogsTable = () => {
         <Table
           style={{ marginTop: 5 }}
           columns={columns}
+          expandedRowRender={expandRowRender}
+          expandRowByClick={true}
           dataSource={logs}
+          rowKey="key"
           pagination={{
             currentPage: activePage,
             pageSize: pageSize,

web/src/components/PersonalSetting.js

@@ -10,7 +10,7 @@ import {
 } from '../helpers';
 import Turnstile from 'react-turnstile';
 import { UserContext } from '../context/User';
-import { onGitHubOAuthClicked } from './utils';
+import { onGitHubOAuthClicked, onLinuxDOOAuthClicked } from './utils';
 import {
   Avatar,
   Banner,
@@ -519,7 +519,6 @@ const PersonalSetting = () => {
                   </div>
                 </div>
               </div>
-
               <div style={{ marginTop: 10 }}>
                 <Typography.Text strong>Telegram</Typography.Text>
                 <div
@@ -551,7 +550,36 @@ const PersonalSetting = () => {
                   </div>
                 </div>
               </div>
-
+              <div style={{ marginTop: 10 }}>
+                <Typography.Text strong>LinuxDO</Typography.Text>
+                <div
+                  style={{ display: 'flex', justifyContent: 'space-between' }}
+                >
+                  <div>
+                    <Input
+                      value={
+                        userState.user && userState.user.linux_do_id !== ''
+                          ? userState.user.linux_do_id
+                          : '未绑定'
+                      }
+                      readonly={true}
+                    ></Input>
+                  </div>
+                  <div>
+                    <Button
+                      onClick={() => {
+                        onLinuxDOOAuthClicked(status.linuxdo_client_id);
+                      }}
+                      disabled={
+                        (userState.user && userState.user.linux_do_id !== '') ||
+                        !status.linuxdo_oauth
+                      }
+                    >
+                      {status.linuxdo_oauth ? '绑定' : '未启用'}
+                    </Button>
+                  </div>
+                </div>
+              </div>
               <div style={{ marginTop: 10 }}>
                 <Space>
                   <Button onClick={generateAccessToken}>

web/src/components/Playground.js

@@ -84,6 +84,11 @@ const Playground = () => {
       // handleInputChange('group', localGroupOptions[0].value);
 
       if (localGroupOptions.length > 0) {
+        // set default group at first
+        localGroupOptions.unshift({
+          label: '用户分组',
+          value: '',
+        });
       } else {
         localGroupOptions = [{
           label: '用户分组',

web/src/components/SystemSetting.js

@@ -53,6 +53,9 @@ const SystemSetting = () => {
     TelegramOAuthEnabled: '',
     TelegramBotToken: '',
     TelegramBotName: '',
+    LinuxDOOAuthEnabled: '',
+    LinuxDOClientId: '',
+    LinuxDOClientSecret: '',
   });
   const [originInputs, setOriginInputs] = useState({});
   let [loading, setLoading] = useState(false);
@@ -103,6 +106,7 @@ const SystemSetting = () => {
       case 'PasswordRegisterEnabled':
       case 'EmailVerificationEnabled':
       case 'GitHubOAuthEnabled':
+      case 'LinuxDOOAuthEnabled':
       case 'WeChatAuthEnabled':
       case 'TelegramOAuthEnabled':
       case 'TurnstileCheckEnabled':
@@ -163,7 +167,9 @@ const SystemSetting = () => {
       name === 'EmailDomainWhitelist' ||
       name === 'TopupGroupRatio' ||
       name === 'TelegramBotToken' ||
-      name === 'TelegramBotName'
+      name === 'TelegramBotName' ||
+      name === 'LinuxDOClientId' ||
+      name === 'LinuxDOClientSecret'
     ) {
       setInputs((inputs) => ({ ...inputs, [name]: value }));
     } else {
@@ -182,7 +188,7 @@ const SystemSetting = () => {
     if (inputs.WorkerValidKey !== '') {
       await updateOption('WorkerValidKey', inputs.WorkerValidKey);
     }
-  }
+  };
 
   const submitPayAddress = async () => {
     if (inputs.ServerAddress === '') {
@@ -320,6 +326,18 @@ const SystemSetting = () => {
     }
   };
 
+  const submitLinuxDOOAuth = async () => {
+    if (originInputs['LinuxDOClientId'] !== inputs.LinuxDOClientId) {
+      await updateOption('LinuxDOClientId', inputs.LinuxDOClientId);
+    }
+    if (
+      originInputs['LinuxDOClientSecret'] !== inputs.LinuxDOClientSecret &&
+      inputs.LinuxDOClientSecret !== ''
+    ) {
+      await updateOption('LinuxDOClientSecret', inputs.LinuxDOClientSecret);
+    }
+  };
+
   return (
     <Grid columns={1}>
       <Grid.Column>
@@ -340,7 +358,15 @@ const SystemSetting = () => {
             更新服务器地址
           </Form.Button>
           <Header as='h3' inverted={isDark}>
-            代理设置（支持 <a href='https://github.com/Calcium-Ion/new-api-worker' target='_blank' rel='noreferrer'>new-api-worker</a>）
+            代理设置（支持{' '}
+            <a
+              href='https://github.com/Calcium-Ion/new-api-worker'
+              target='_blank'
+              rel='noreferrer'
+            >
+              new-api-worker
+            </a>
+            ）
           </Header>
           <Form.Group widths='equal'>
             <Form.Input
@@ -358,9 +384,7 @@ const SystemSetting = () => {
               onChange={handleInputChange}
             />
           </Form.Group>
-          <Form.Button onClick={submitWorker}>
-            更新Worker设置
-          </Form.Button>
+          <Form.Button onClick={submitWorker}>更新Worker设置</Form.Button>
           <Divider />
           <Header as='h3' inverted={isDark}>
             支付设置（当前仅支持易支付接口，默认使用上方服务器地址作为回调地址！）
@@ -483,6 +507,12 @@ const SystemSetting = () => {
               name='GitHubOAuthEnabled'
               onChange={handleInputChange}
             />
+            <Form.Checkbox
+              checked={inputs.LinuxDOOAuthEnabled === 'true'}
+              label='允许通过 LinuxDO 账户登录 & 注册'
+              name='LinuxDOOAuthEnabled'
+              onChange={handleInputChange}
+            />
             <Form.Checkbox
               checked={inputs.WeChatAuthEnabled === 'true'}
               label='允许通过微信登录 & 注册'
@@ -781,6 +811,48 @@ const SystemSetting = () => {
           <Form.Button onClick={submitTurnstile}>
             保存 Turnstile 设置
           </Form.Button>
+          <Divider />
+          <Header as='h3' inverted={isDark}>
+            配置 LinuxDO OAuth App
+            <Header.Subheader>
+              用以支持通过 LinuxDO 进行登录注册，
+              <a
+                href='https://connect.linux.do/'
+                target='_blank'
+                rel='noreferrer'
+              >
+                点击此处
+              </a>
+              管理你的 LinuxDO OAuth App
+            </Header.Subheader>
+          </Header>
+          <Message>
+            Homepage URL 填 <code>{inputs.ServerAddress}</code>
+            ，Authorization callback URL 填{' '}
+            <code>{`${inputs.ServerAddress}/oauth/linuxdo`}</code>
+          </Message>
+          <Form.Group widths={3}>
+            <Form.Input
+              label='LinuxDO Client ID'
+              name='LinuxDOClientId'
+              onChange={handleInputChange}
+              autoComplete='new-password'
+              value={inputs.LinuxDOClientId}
+              placeholder='输入你注册的 LinuxDO OAuth APP 的 ID'
+            />
+            <Form.Input
+              label='LinuxDO Client Secret'
+              name='LinuxDOClientSecret'
+              onChange={handleInputChange}
+              type='password'
+              autoComplete='new-password'
+              value={inputs.LinuxDOClientSecret}
+              placeholder='敏感信息不会发送到前端显示'
+            />
+          </Form.Group>
+          <Form.Button onClick={submitLinuxDOOAuth}>
+            保存 LinuxDO OAuth 设置
+          </Form.Button>
         </Form>
       </Grid.Column>
     </Grid>

web/src/components/TokensTable.js

@@ -393,8 +393,8 @@ const TokensTable = () => {
       serverAddress = window.location.origin;
     }
     let encodedServerAddress = encodeURIComponent(serverAddress);
-    url = url.replace('{address}', encodedServerAddress);
-    url = url.replace('{key}', 'sk-' + record.key);
+    url = url.replaceAll('{address}', encodedServerAddress);
+    url = url.replaceAll('{key}', 'sk-' + record.key);
     // console.log(url);
     // const chatLink = localStorage.getItem('chat_link');
     // const mjLink = localStorage.getItem('chat_link2');

web/src/components/utils.js

@@ -19,6 +19,14 @@ export async function onGitHubOAuthClicked(github_client_id) {
   );
 }
 
+export async function onLinuxDOOAuthClicked(linuxdo_client_id) {
+  const state = await getOAuthState();
+  if (!state) return;
+  window.open(
+    `https://connect.linux.do/oauth2/authorize?response_type=code&client_id=${linuxdo_client_id}&state=${state}`,
+  );
+}
+
 let channelModels = undefined;
 export async function loadChannelModels() {
   const res = await API.get('/api/models');

web/src/constants/channel.constants.js

@@ -109,6 +109,7 @@ export const CHANNEL_OPTIONS = [
   { key: 37, text: 'Dify', value: 37, color: 'teal', label: 'Dify' },
   { key: 38, text: 'Jina', value: 38, color: 'blue', label: 'Jina' },
   { key: 40, text: 'SiliconCloud', value: 40, color: 'purple', label: 'SiliconCloud' },
+  { key: 42, text: 'Mistral AI', value: 42, color: 'blue', label: 'Mistral AI' },
   { key: 8, text: '自定义渠道', value: 8, color: 'pink', label: '自定义渠道' },
   {
     key: 22,

web/src/helpers/render.js

@@ -173,6 +173,59 @@ export function renderModelPrice(
   }
 }
 
+export function renderAudioModelPrice(
+  inputTokens,
+  completionTokens,
+  modelRatio,
+  modelPrice = -1,
+  completionRatio,
+  audioInputTokens,
+  audioCompletionTokens,
+  audioRatio,
+  audioCompletionRatio,
+  groupRatio,
+) {
+  // 1 ratio = $0.002 / 1K tokens
+  if (modelPrice !== -1) {
+    return '模型价格：$' + modelPrice + ' * 分组倍率：' + groupRatio + ' = $' + modelPrice * groupRatio;
+  } else {
+    if (completionRatio === undefined) {
+      completionRatio = 0;
+    }
+    // 这里的 *2 是因为 1倍率=0.002刀，请勿删除
+    let inputRatioPrice = modelRatio * 2.0;
+    let completionRatioPrice = modelRatio * 2.0 * completionRatio;
+    let price =
+      (inputTokens / 1000000) * inputRatioPrice * groupRatio +
+      (completionTokens / 1000000) * completionRatioPrice * groupRatio +
+      (audioInputTokens / 1000000) * inputRatioPrice * audioRatio * groupRatio +
+      (audioCompletionTokens / 1000000) * inputRatioPrice * audioRatio * audioCompletionRatio * groupRatio;
+    return (
+      <>
+        <article>
+          <p>提示：${inputRatioPrice} * {groupRatio} = ${inputRatioPrice * groupRatio} / 1M tokens</p>
+          <p>补全：${completionRatioPrice} * {groupRatio} = ${completionRatioPrice * groupRatio} / 1M tokens</p>
+          <p>音频提示：${inputRatioPrice} * {groupRatio} * {audioRatio} = ${inputRatioPrice * audioRatio * groupRatio} / 1M tokens</p>
+          <p>音频补全：${inputRatioPrice} * {groupRatio} * {audioRatio} * {audioCompletionRatio} = ${inputRatioPrice * audioRatio * audioCompletionRatio * groupRatio} / 1M tokens</p>
+          <p></p>
+          <p>
+            提示 {inputTokens} tokens / 1M tokens * ${inputRatioPrice} + 补全{' '}
+            {completionTokens} tokens / 1M tokens * ${completionRatioPrice} +
+          </p>
+          <p>
+            音频提示 {audioInputTokens} tokens / 1M tokens * ${inputRatioPrice} * {audioRatio} + 音频补全 {audioCompletionTokens} tokens / 1M tokens * ${inputRatioPrice} * {audioRatio} * {audioCompletionRatio}
+          </p>
+          <p>
+            （文字 + 音频） * 分组 {groupRatio} =
+            ${price.toFixed(6)}
+          </p>
+          <p>仅供参考，以实际扣费为准</p>
+        </article>
+      </>
+    );
+  }
+}
+
 export function renderQuotaWithPrompt(quota, digits) {
   let displayInCurrency = localStorage.getItem('display_in_currency');
   displayInCurrency = displayInCurrency === 'true';

web/src/pages/Chat/index.js

@@ -20,8 +20,8 @@ const ChatPage = () => {
               if (Array.isArray(chats) && chats.length > 0) {
                   for (let k in chats[id]) {
                       link = chats[id][k];
-                      link = link.replace('{address}', encodeURIComponent(serverAddress));
-                      link = link.replace('{key}', 'sk-' + key);
+                      link = link.replaceAll('{address}', encodeURIComponent(serverAddress));
+                      link = link.replaceAll('{key}', 'sk-' + key);
                   }
               }
           }

web/src/pages/Home/index.js

@@ -150,6 +150,12 @@ const Home = () => {
                       ? '已启用'
                       : '未启用'}
                   </p>
+                  <p>
+                    Linux DO 身份验证：
+                    {statusState?.status?.linuxdo_oauth === true
+                      ? '已启用'
+                      : '未启用'}
+                  </p>
                 </Card>
               </Col>
             </Row>