merge upstream

Signed-off-by: wozulong <>

.gitignore

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

README.md

@@ -2,15 +2,21 @@
 # New API
 
 > [!NOTE]
-> 本项目为开源项目，在[One API](https://github.com/songquanpeng/one-api)的基础上进行二次开发，感谢原作者的无私奉献。 
+> 本项目为开源项目，在[One API](https://github.com/songquanpeng/one-api)的基础上进行二次开发
-> 使用者必须在遵循 OpenAI 的[使用条款](https://openai.com/policies/terms-of-use)以及**法律法规**的情况下使用，不得用于非法用途。
 
-> 本项目为个人学习使用，不保证稳定性，且不提供任何技术支持，使用者必须在遵循 OpenAI 的使用条款以及法律法规的情况下使用，不得用于非法用途。  
+> [!IMPORTANT]
+> 使用者必须在遵循 OpenAI 的[使用条款](https://openai.com/policies/terms-of-use)以及**法律法规**的情况下使用，不得用于非法用途。
+> 本项目仅供个人学习使用，不保证稳定性，且不提供任何技术支持。
 > 根据[《生成式人工智能服务管理暂行办法》](http://www.cac.gov.cn/2023-07/13/c_1690898327029107.htm)的要求，请勿对中国地区公众提供一切未经备案的生成式人工智能服务。
 
-> [!NOTE]
+> [!TIP]
-> 最新版Docker镜像 calciumion/new-api:latest  
+> 最新版Docker镜像：`calciumion/new-api:latest`  
-> 更新指令 docker run --rm -v /var/run/docker.sock:/var/run/docker.sock containrrr/watchtower -cR
+> 默认账号root 密码123456  
+> 更新指令：
+> ```
+> docker run --rm -v /var/run/docker.sock:/var/run/docker.sock containrrr/watchtower -cR
+> ```
+
 
 ## 主要变更
 此分叉版本的主要变更如下：
@@ -18,9 +24,9 @@
 1. 全新的UI界面（部分界面还待更新）
 2. 添加[Midjourney-Proxy(Plus)](https://github.com/novicezk/midjourney-proxy)接口的支持，[对接文档](Midjourney.md)
 3. 支持在线充值功能，可在系统设置中设置，当前支持的支付接口：
-   + [x] 易支付
+    + [x] 易支付
 4. 支持用key查询使用额度:
-   + 配合项目[neko-api-key-tool](https://github.com/Calcium-Ion/neko-api-key-tool)可实现用key查询使用
+    + 配合项目[neko-api-key-tool](https://github.com/Calcium-Ion/neko-api-key-tool)可实现用key查询使用
 5. 渠道显示已使用额度，支持指定组织访问
 6. 分页支持选择每页显示数量
 7. 兼容原版One API的数据库，可直接使用原版数据库（one-api.db）
@@ -48,32 +54,19 @@
 8. [Suno API](https://github.com/Suno-API/Suno-API) 接口，[对接文档](Suno.md)
 9. Rerank模型，目前支持[Cohere](https://cohere.ai/)和[Jina](https://jina.ai/)，[对接文档](Rerank.md)
 10. Dify
+11. Vertex AI，目前兼容Claude，Gemini，Llama3.1
 
 您可以在渠道中添加自定义模型gpt-4-gizmo-*或g-*，此模型并非OpenAI官方模型，而是第三方模型，使用官方key无法调用。
 
-## 渠道重试
-渠道重试功能已经实现，可以在`设置->运营设置->通用设置`设置重试次数，**建议开启缓存**功能。  
-如果开启了重试功能，第一次重试使用同优先级，第二次重试使用下一个优先级，以此类推。  
-### 缓存设置方法
-1. `REDIS_CONN_STRING`：设置之后将使用 Redis 作为缓存使用。
-    + 例子：`REDIS_CONN_STRING=redis://default:redispw@localhost:49153`
-2. `MEMORY_CACHE_ENABLED`：启用内存缓存（如果设置了`REDIS_CONN_STRING`，则无需手动设置），会导致用户额度的更新存在一定的延迟，可选值为 `true` 和 `false`，未设置则默认为 `false`。
-    + 例子：`MEMORY_CACHE_ENABLED=true`
-### 为什么有的时候没有重试
-这些错误码不会重试：400，504，524
-### 我想让400也重试
-在`渠道->编辑`中，将`状态码复写`改为
-```json
-{
-  "400": "500"
-}
-```
-可以实现400错误转为500错误，从而重试
-
 ## 比原版One API多出的配置
-- `STREAMING_TIMEOUT`：设置流式一次回复的超时时间，默认为 30 秒
+- `STREAMING_TIMEOUT`：设置流式一次回复的超时时间，默认为 30 秒。
-- `DIFY_DEBUG`：设置 Dify 渠道是否输出工作流和节点信息到客户端，默认为 `true`， 可选值为 `true` 和 `false`
+- `DIFY_DEBUG`：设置 Dify 渠道是否输出工作流和节点信息到客户端，默认为 `true`。
-- `FORCE_STREAM_OPTION`：覆盖客户端stream_options参数，请求上游返回流模式usage，目前仅支持 `OpenAI` 渠道类型
+- `FORCE_STREAM_OPTION`：是否覆盖客户端stream_options参数，请求上游返回流模式usage，默认为 `true`，建议开启，不影响客户端传入stream_options参数返回结果。
+- `GET_MEDIA_TOKEN`：是统计图片token，默认为 `true`，关闭后将不再在本地计算图片token，可能会导致和上游计费不同，此项覆盖 `GET_MEDIA_TOKEN_NOT_STREAM` 选项作用。
+- `GET_MEDIA_TOKEN_NOT_STREAM`：是否在非流（`stream=false`）情况下统计图片token，默认为 `true`。
+- `UPDATE_TASK`：是否更新异步任务（Midjourney、Suno），默认为 `true`，关闭后将不会更新任务进度。
+- `GEMINI_MODEL_MAP`：Gemini模型指定版本(v1/v1beta)，使用“模型:版本”指定，","分隔，例如：-e GEMINI_MODEL_MAP="gemini-1.5-pro-latest:v1beta,gemini-1.5-pro-001:v1beta"，为空则使用默认配置
+
 ## 部署
 ### 部署要求
 - 本地数据库（默认）：SQLite（Docker 部署默认使用 SQLite，必须挂载 `/data` 目录到宿主机）
@@ -96,8 +89,25 @@ docker run --name new-api -d --restart always -p 3000:3000 -e TZ=Asia/Shanghai -
 docker run --name new-api -d --restart always -p 3000:3000 -e SQL_DSN="root:123456@tcp(宝塔的服务器地址:宝塔数据库端口)/宝塔数据库名称" -e TZ=Asia/Shanghai -v /www/wwwroot/new-api:/data calciumion/new-api:latest
 # 注意：数据库要开启远程访问，并且只允许服务器IP访问
 ```
-### 默认账号密码
+
-默认账号root 密码123456
+## 渠道重试
+渠道重试功能已经实现，可以在`设置->运营设置->通用设置`设置重试次数，**建议开启缓存**功能。  
+如果开启了重试功能，第一次重试使用同优先级，第二次重试使用下一个优先级，以此类推。
+### 缓存设置方法
+1. `REDIS_CONN_STRING`：设置之后将使用 Redis 作为缓存使用。
+    + 例子：`REDIS_CONN_STRING=redis://default:redispw@localhost:49153`
+2. `MEMORY_CACHE_ENABLED`：启用内存缓存（如果设置了`REDIS_CONN_STRING`，则无需手动设置），会导致用户额度的更新存在一定的延迟，可选值为 `true` 和 `false`，未设置则默认为 `false`。
+    + 例子：`MEMORY_CACHE_ENABLED=true`
+### 为什么有的时候没有重试
+这些错误码不会重试：400，504，524
+### 我想让400也重试
+在`渠道->编辑`中，将`状态码复写`改为
+```json
+{
+  "400": "500"
+}
+```
+可以实现400错误转为500错误，从而重试
 
 ## Midjourney接口设置文档
 [对接文档](Midjourney.md)

common/constants.go

@@ -235,6 +235,9 @@ const (
 	ChannelTypeSunoAPI        = 36
 	ChannelTypeDify           = 37
 	ChannelTypeJina           = 38
+	ChannelCloudflare         = 39
+	ChannelTypeSiliconFlow    = 40
+	ChannelTypeVertexAi       = 41
 
 	ChannelTypeDummy // this one is only for count, do not add any channel after this
 
@@ -280,4 +283,7 @@ var ChannelBaseURLs = []string{
 	"",                                          //36
 	"",                                          //37
 	"https://api.jina.ai",                       //38
+	"https://api.cloudflare.com",                //39
+	"https://api.siliconflow.cn",                //40
+	"",                                          //41
 }

common/email-outlook-auth.go

@@ -0,0 +1,40 @@
+package common
+
+import (
+	"errors"
+	"net/smtp"
+	"strings"
+)
+
+type outlookAuth struct {
+	username, password string
+}
+
+func LoginAuth(username, password string) smtp.Auth {
+	return &outlookAuth{username, password}
+}
+
+func (a *outlookAuth) Start(_ *smtp.ServerInfo) (string, []byte, error) {
+	return "LOGIN", []byte{}, nil
+}
+
+func (a *outlookAuth) Next(fromServer []byte, more bool) ([]byte, error) {
+	if more {
+		switch string(fromServer) {
+		case "Username:":
+			return []byte(a.username), nil
+		case "Password:":
+			return []byte(a.password), nil
+		default:
+			return nil, errors.New("unknown fromServer")
+		}
+	}
+	return nil, nil
+}
+
+func isOutlookServer(server string) bool {
+	// 兼容多地区的outlook邮箱和ofb邮箱
+	// 其实应该加一个Option来区分是否用LOGIN的方式登录
+	// 先临时兼容一下
+	return strings.Contains(server, "outlook") || strings.Contains(server, "onmicrosoft")
+}

common/email.go

@@ -9,17 +9,26 @@ import (
 	"time"
 )
 
+func generateMessageID() string {
+	domain := strings.Split(SMTPAccount, "@")[1]
+	return fmt.Sprintf("<%d.%s@%s>", time.Now().UnixNano(), GetRandomString(12), domain)
+}
+
 func SendEmail(subject string, receiver string, content string) error {
 	if SMTPFrom == "" { // for compatibility
 		SMTPFrom = SMTPAccount
 	}
+	if SMTPServer == "" && SMTPAccount == "" {
+		return fmt.Errorf("SMTP 服务器未配置")
+	}
 	encodedSubject := fmt.Sprintf("=?UTF-8?B?%s?=", base64.StdEncoding.EncodeToString([]byte(subject)))
 	mail := []byte(fmt.Sprintf("To: %s\r\n"+
 		"From: %s<%s>\r\n"+
 		"Subject: %s\r\n"+
 		"Date: %s\r\n"+
+		"Message-ID: %s\r\n"+ // 添加 Message-ID 头
 		"Content-Type: text/html; charset=UTF-8\r\n\r\n%s\r\n",
-		receiver, SystemName, SMTPFrom, encodedSubject, time.Now().Format(time.RFC1123Z), content))
+		receiver, SystemName, SMTPFrom, encodedSubject, time.Now().Format(time.RFC1123Z), generateMessageID(), content))
 	auth := smtp.PlainAuth("", SMTPAccount, SMTPToken, SMTPServer)
 	addr := fmt.Sprintf("%s:%d", SMTPServer, SMTPPort)
 	to := strings.Split(receiver, ";")
@@ -62,6 +71,9 @@ func SendEmail(subject string, receiver string, content string) error {
 		if err != nil {
 			return err
 		}
+	} else if isOutlookServer(SMTPAccount) {
+		auth = LoginAuth(SMTPAccount, SMTPToken)
+		err = smtp.SendMail(addr, auth, SMTPAccount, to, mail)
 	} else {
 		err = smtp.SendMail(addr, auth, SMTPAccount, to, mail)
 	}

common/model-ratio.go

@@ -3,6 +3,7 @@ package common
 import (
 	"encoding/json"
 	"strings"
+	"sync"
 )
 
 // from songquanpeng/one-api
@@ -30,8 +31,12 @@ var defaultModelRatio = map[string]float64{
 	"gpt-4-32k":                      30,
 	"gpt-4-32k-0314":                 30,
 	"gpt-4-32k-0613":                 30,
+	"gpt-4o-mini":                    0.075, // $0.00015 / 1K tokens
+	"gpt-4o-mini-2024-07-18":         0.075,
+	"chatgpt-4o-latest":              2.5,  // $0.01 / 1K tokens
 	"gpt-4o":                         2.5,  // $0.005 / 1K tokens
 	"gpt-4o-2024-05-13":              2.5,  // $0.005 / 1K tokens
+	"gpt-4o-2024-08-06":              1.25, // $0.01 / 1K tokens
 	"gpt-4-turbo":                    5,    // $0.01 / 1K tokens
 	"gpt-4-turbo-2024-04-09":         5,    // $0.01 / 1K tokens
 	"gpt-4-1106-preview":             5,    // $0.01 / 1K tokens
@@ -71,13 +76,13 @@ var defaultModelRatio = map[string]float64{
 	"text-search-ada-doc-001":        10,
 	"text-moderation-stable":         0.1,
 	"text-moderation-latest":         0.1,
-	"claude-instant-1":               0.4,    // $0.8 / 1M tokens
+	"claude-instant-1":               0.4,   // $0.8 / 1M tokens
-	"claude-2.0":                     4,      // $8 / 1M tokens
+	"claude-2.0":                     4,     // $8 / 1M tokens
-	"claude-2.1":                     4,      // $8 / 1M tokens
+	"claude-2.1":                     4,     // $8 / 1M tokens
-	"claude-3-haiku-20240307":        0.125,  // $0.25 / 1M tokens
+	"claude-3-haiku-20240307":        0.125, // $0.25 / 1M tokens
-	"claude-3-5-sonnet-20240620":     1.5,    // $3 / 1M tokens
+	"claude-3-5-sonnet-20240620":     1.5,   // $3 / 1M tokens
-	"claude-3-sonnet-20240229":       1.5,    // $3 / 1M tokens
+	"claude-3-sonnet-20240229":       1.5,   // $3 / 1M tokens
-	"claude-3-opus-20240229":         7.5,    // $15 / 1M tokens
+	"claude-3-opus-20240229":         7.5,   // $15 / 1M tokens
 	"ERNIE-4.0-8K":                   0.120 * RMB,
 	"ERNIE-3.5-8K":                   0.012 * RMB,
 	"ERNIE-3.5-8K-0205":              0.024 * RMB,
@@ -104,13 +109,21 @@ var defaultModelRatio = map[string]float64{
 	"gemini-1.0-pro-latest":          1,
 	"gemini-1.0-pro-vision-latest":   1,
 	"gemini-ultra":                   1,
-	"chatglm_turbo":                  0.3572, // ￥0.005 / 1k tokens
+	"chatglm_turbo":                  0.3572,     // ￥0.005 / 1k tokens
-	"chatglm_pro":                    0.7143, // ￥0.01 / 1k tokens
+	"chatglm_pro":                    0.7143,     // ￥0.01 / 1k tokens
-	"chatglm_std":                    0.3572, // ￥0.005 / 1k tokens
+	"chatglm_std":                    0.3572,     // ￥0.005 / 1k tokens
-	"chatglm_lite":                   0.1429, // ￥0.002 / 1k tokens
+	"chatglm_lite":                   0.1429,     // ￥0.002 / 1k tokens
-	"glm-4":                          7.143,  // ￥0.1 / 1k tokens
+	"glm-4":                          7.143,      // ￥0.1 / 1k tokens
-	"glm-4v":                         7.143,  // ￥0.1 / 1k tokens
+	"glm-4v":                         0.05 * RMB, // ￥0.05 / 1k tokens
+	"glm-4-alltools":                 0.1 * RMB,  // ￥0.1 / 1k tokens
 	"glm-3-turbo":                    0.3572,
+	"glm-4-plus":                     0.05 * RMB,
+	"glm-4-0520":                     0.1 * RMB,
+	"glm-4-air":                      0.001 * RMB,
+	"glm-4-airx":                     0.01 * RMB,
+	"glm-4-long":                     0.001 * RMB,
+	"glm-4-flash":                    0,
+	"glm-4v-plus":                    0.01 * RMB,
 	"qwen-turbo":                     0.8572, // ￥0.012 / 1k tokens
 	"qwen-plus":                      10,     // ￥0.14 / 1k tokens
 	"text-embedding-v1":              0.05,   // ￥0.0007 / 1k tokens
@@ -129,26 +142,28 @@ var defaultModelRatio = map[string]float64{
 	"hunyuan":                        7.143,  // ¥0.1 / 1k tokens  // https://cloud.tencent.com/document/product/1729/97731#e0e6be58-60c8-469f-bdeb-6c264ce3b4d0
 	// https://platform.lingyiwanwu.com/docs#-计费单元
 	// 已经按照 7.2 来换算美元价格
-	"yi-34b-chat-0205":      0.18,
+	"yi-34b-chat-0205":       0.18,
-	"yi-34b-chat-200k":      0.864,
+	"yi-34b-chat-200k":       0.864,
-	"yi-vl-plus":            0.432,
+	"yi-vl-plus":             0.432,
-	"yi-large":              20.0 / 1000 * RMB,
+	"yi-large":               20.0 / 1000 * RMB,
-	"yi-medium":             2.5 / 1000 * RMB,
+	"yi-medium":              2.5 / 1000 * RMB,
-	"yi-vision":             6.0 / 1000 * RMB,
+	"yi-vision":              6.0 / 1000 * RMB,
-	"yi-medium-200k":        12.0 / 1000 * RMB,
+	"yi-medium-200k":         12.0 / 1000 * RMB,
-	"yi-spark":              1.0 / 1000 * RMB,
+	"yi-spark":               1.0 / 1000 * RMB,
-	"yi-large-rag":          25.0 / 1000 * RMB,
+	"yi-large-rag":           25.0 / 1000 * RMB,
-	"yi-large-turbo":        12.0 / 1000 * RMB,
+	"yi-large-turbo":         12.0 / 1000 * RMB,
-	"yi-large-preview":      20.0 / 1000 * RMB,
+	"yi-large-preview":       20.0 / 1000 * RMB,
-	"yi-large-rag-preview":  25.0 / 1000 * RMB,
+	"yi-large-rag-preview":   25.0 / 1000 * RMB,
-	"command":               0.5,
+	"command":                0.5,
-	"command-nightly":       0.5,
+	"command-nightly":        0.5,
-	"command-light":         0.5,
+	"command-light":          0.5,
-	"command-light-nightly": 0.5,
+	"command-light-nightly":  0.5,
-	"command-r":             0.25,
+	"command-r":              0.25,
-	"command-r-plus	":       1.5,
+	"command-r-plus":         1.5,
-	"deepseek-chat":         0.07,
+	"command-r-08-2024":      0.075,
-	"deepseek-coder":        0.07,
+	"command-r-plus-08-2024": 1.25,
+	"deepseek-chat":          0.07,
+	"deepseek-coder":         0.07,
 	// Perplexity online 模型对搜索额外收费，有需要应自行调整，此处不计入搜索费用
 	"llama-3-sonar-small-32k-chat":   0.2 / 1000 * USD,
 	"llama-3-sonar-small-32k-online": 0.2 / 1000 * USD,
@@ -157,6 +172,8 @@ var defaultModelRatio = map[string]float64{
 }
 
 var defaultModelPrice = map[string]float64{
+	"suno_music":        0.1,
+	"suno_lyrics":       0.01,
 	"dall-e-2":          0.02,
 	"dall-e-3":          0.04,
 	"gpt-4-gizmo-*":     0.1,
@@ -176,10 +193,17 @@ var defaultModelPrice = map[string]float64{
 	"mj_describe":       0.05,
 	"mj_upscale":        0.05,
 	"swap_face":         0.05,
+	"mj_upload":         0.05,
 }
 
-var modelPrice map[string]float64 = nil
+var (
-var modelRatio map[string]float64 = nil
+	modelPriceMap      map[string]float64 = nil
+	modelPriceMapMutex                    = sync.RWMutex{}
+)
+var (
+	modelRatioMap      map[string]float64 = nil
+	modelRatioMapMutex                    = sync.RWMutex{}
+)
 
 var CompletionRatio map[string]float64 = nil
 var defaultCompletionRatio = map[string]float64{
@@ -189,11 +213,18 @@ var defaultCompletionRatio = map[string]float64{
 	"gpt-4o-all":    2,
 }
 
-func ModelPrice2JSONString() string {
+func GetModelPriceMap() map[string]float64 {
-	if modelPrice == nil {
+	modelPriceMapMutex.Lock()
-		modelPrice = defaultModelPrice
+	defer modelPriceMapMutex.Unlock()
+	if modelPriceMap == nil {
+		modelPriceMap = defaultModelPrice
 	}
-	jsonBytes, err := json.Marshal(modelPrice)
+	return modelPriceMap
+}
+
+func ModelPrice2JSONString() string {
+	GetModelPriceMap()
+	jsonBytes, err := json.Marshal(modelPriceMap)
 	if err != nil {
 		SysError("error marshalling model price: " + err.Error())
 	}
@@ -201,21 +232,21 @@ func ModelPrice2JSONString() string {
 }
 
 func UpdateModelPriceByJSONString(jsonStr string) error {
-	modelPrice = make(map[string]float64)
+	modelPriceMapMutex.Lock()
-	return json.Unmarshal([]byte(jsonStr), &modelPrice)
+	defer modelPriceMapMutex.Unlock()
+	modelPriceMap = make(map[string]float64)
+	return json.Unmarshal([]byte(jsonStr), &modelPriceMap)
 }
 
 // GetModelPrice 返回模型的价格，如果模型不存在则返回-1，false
 func GetModelPrice(name string, printErr bool) (float64, bool) {
-	if modelPrice == nil {
+	GetModelPriceMap()
-		modelPrice = defaultModelPrice
-	}
 	if strings.HasPrefix(name, "gpt-4-gizmo") {
 		name = "gpt-4-gizmo-*"
 	} else if strings.HasPrefix(name, "g-") {
 		name = "g-*"
 	}
-	price, ok := modelPrice[name]
+	price, ok := modelPriceMap[name]
 	if !ok {
 		if printErr {
 			SysError("model price not found: " + name)
@@ -225,18 +256,18 @@ func GetModelPrice(name string, printErr bool) (float64, bool) {
 	return price, true
 }
 
-func GetModelPriceMap() map[string]float64 {
+func GetModelRatioMap() map[string]float64 {
-	if modelPrice == nil {
+	modelRatioMapMutex.Lock()
-		modelPrice = defaultModelPrice
+	defer modelRatioMapMutex.Unlock()
+	if modelRatioMap == nil {
+		modelRatioMap = defaultModelRatio
 	}
-	return modelPrice
+	return modelRatioMap
 }
 
 func ModelRatio2JSONString() string {
-	if modelRatio == nil {
+	GetModelRatioMap()
-		modelRatio = defaultModelRatio
+	jsonBytes, err := json.Marshal(modelRatioMap)
-	}
-	jsonBytes, err := json.Marshal(modelRatio)
 	if err != nil {
 		SysError("error marshalling model ratio: " + err.Error())
 	}
@@ -244,20 +275,20 @@ func ModelRatio2JSONString() string {
 }
 
 func UpdateModelRatioByJSONString(jsonStr string) error {
-	modelRatio = make(map[string]float64)
+	modelRatioMapMutex.Lock()
-	return json.Unmarshal([]byte(jsonStr), &modelRatio)
+	defer modelRatioMapMutex.Unlock()
+	modelRatioMap = make(map[string]float64)
+	return json.Unmarshal([]byte(jsonStr), &modelRatioMap)
 }
 
 func GetModelRatio(name string) float64 {
-	if modelRatio == nil {
+	GetModelRatioMap()
-		modelRatio = defaultModelRatio
-	}
 	if strings.HasPrefix(name, "gpt-4-gizmo") {
 		name = "gpt-4-gizmo-*"
 	} else if strings.HasPrefix(name, "g-") {
 		name = "g-*"
 	}
-	ratio, ok := modelRatio[name]
+	ratio, ok := modelRatioMap[name]
 	if !ok {
 		SysError("model ratio not found: " + name)
 		return 30
@@ -313,11 +344,18 @@ func GetCompletionRatio(name string) float64 {
 		return 4.0 / 3.0
 	}
 	if strings.HasPrefix(name, "gpt-4") && name != "gpt-4-all" && name != "gpt-4-gizmo-*" {
+		if strings.HasPrefix(name, "gpt-4o-mini") || "gpt-4o-2024-08-06" == name {
+			return 4
+		}
+
 		if strings.HasSuffix(name, "preview") || strings.HasPrefix(name, "gpt-4-turbo") || strings.HasPrefix(name, "gpt-4o") {
 			return 3
 		}
 		return 2
 	}
+	if name == "chatgpt-4o-latest" {
+		return 3
+	}
 	if strings.HasPrefix(name, "claude-instant-1") {
 		return 3
 	} else if strings.HasPrefix(name, "claude-2") {
@@ -337,6 +375,10 @@ func GetCompletionRatio(name string) float64 {
 			return 3
 		case "command-r-plus":
 			return 5
+		case "command-r-08-2024":
+			return 4
+		case "command-r-plus-08-2024":
+			return 4
 		default:
 			return 2
 		}

common/str.go

@@ -0,0 +1,70 @@
+package common
+
+import (
+	"encoding/json"
+	"math/rand"
+	"strconv"
+	"unsafe"
+)
+
+func GetStringIfEmpty(str string, defaultValue string) string {
+	if str == "" {
+		return defaultValue
+	}
+	return str
+}
+
+func GetRandomString(length int) string {
+	//rand.Seed(time.Now().UnixNano())
+	key := make([]byte, length)
+	for i := 0; i < length; i++ {
+		key[i] = keyChars[rand.Intn(len(keyChars))]
+	}
+	return string(key)
+}
+
+func MapToJsonStr(m map[string]interface{}) string {
+	bytes, err := json.Marshal(m)
+	if err != nil {
+		return ""
+	}
+	return string(bytes)
+}
+
+func StrToMap(str string) map[string]interface{} {
+	m := make(map[string]interface{})
+	err := json.Unmarshal([]byte(str), &m)
+	if err != nil {
+		return nil
+	}
+	return m
+}
+
+func IsJsonStr(str string) bool {
+	var js map[string]interface{}
+	return json.Unmarshal([]byte(str), &js) == nil
+}
+
+func String2Int(str string) int {
+	num, err := strconv.Atoi(str)
+	if err != nil {
+		return 0
+	}
+	return num
+}
+
+func StringsContains(strs []string, str string) bool {
+	for _, s := range strs {
+		if s == str {
+			return true
+		}
+	}
+	return false
+}
+
+// StringToByteSlice []byte only read, panic on append
+func StringToByteSlice(s string) []byte {
+	tmp1 := (*[2]uintptr)(unsafe.Pointer(&s))
+	tmp2 := [3]uintptr{tmp1[0], tmp1[1], tmp1[1]}
+	return *(*[]byte)(unsafe.Pointer(&tmp2))
+}

common/utils.go

@@ -2,7 +2,6 @@ package common
 
 import (
 	"context"
-	"encoding/json"
 	"errors"
 	"fmt"
 	"github.com/google/uuid"
@@ -18,7 +17,6 @@ import (
 	"strconv"
 	"strings"
 	"time"
-	"unsafe"
 )
 
 func OpenBrowser(url string) {
@@ -164,15 +162,6 @@ func GenerateKey() string {
 	return string(key)
 }
 
-func GetRandomString(length int) string {
-	//rand.Seed(time.Now().UnixNano())
-	key := make([]byte, length)
-	for i := 0; i < length; i++ {
-		key[i] = keyChars[rand.Intn(len(keyChars))]
-	}
-	return string(key)
-}
-
 func GetRandomInt(max int) int {
 	//rand.Seed(time.Now().UnixNano())
 	return rand.Intn(max)
@@ -199,60 +188,11 @@ func MessageWithRequestId(message string, id string) string {
 	return fmt.Sprintf("%s (request id: %s)", message, id)
 }
 
-func String2Int(str string) int {
-	num, err := strconv.Atoi(str)
-	if err != nil {
-		return 0
-	}
-	return num
-}
-
-func StringsContains(strs []string, str string) bool {
-	for _, s := range strs {
-		if s == str {
-			return true
-		}
-	}
-	return false
-}
-
-// StringToByteSlice []byte only read, panic on append
-func StringToByteSlice(s string) []byte {
-	tmp1 := (*[2]uintptr)(unsafe.Pointer(&s))
-	tmp2 := [3]uintptr{tmp1[0], tmp1[1], tmp1[1]}
-	return *(*[]byte)(unsafe.Pointer(&tmp2))
-}
-
 func RandomSleep() {
 	// Sleep for 0-3000 ms
 	time.Sleep(time.Duration(rand.Intn(3000)) * time.Millisecond)
 }
 
-func MapToJsonStr(m map[string]interface{}) string {
-	bytes, err := json.Marshal(m)
-	if err != nil {
-		return ""
-	}
-	return string(bytes)
-}
-
-func MapToJsonStrFloat(m map[string]float64) string {
-	bytes, err := json.Marshal(m)
-	if err != nil {
-		return ""
-	}
-	return string(bytes)
-}
-
-func StrToMap(str string) map[string]interface{} {
-	m := make(map[string]interface{})
-	err := json.Unmarshal([]byte(str), &m)
-	if err != nil {
-		return nil
-	}
-	return m
-}
-
 func GetProxiedHttpClient(proxyUrl string) (*http.Client, error) {
 	if "" == proxyUrl {
 		return &http.Client{}, nil

constant/env.go

@@ -1,7 +1,10 @@
 package constant
 
 import (
+	"fmt"
 	"one-api/common"
+	"os"
+	"strings"
 )
 
 var StreamingTimeout = common.GetEnvOrDefault("STREAMING_TIMEOUT", 30)
@@ -9,3 +12,35 @@ var DifyDebug = common.GetEnvOrDefaultBool("DIFY_DEBUG", true)
 
 // ForceStreamOption 覆盖请求参数，强制返回usage信息
 var ForceStreamOption = common.GetEnvOrDefaultBool("FORCE_STREAM_OPTION", true)
+
+var GetMediaToken = common.GetEnvOrDefaultBool("GET_MEDIA_TOKEN", true)
+
+var GetMediaTokenNotStream = common.GetEnvOrDefaultBool("GET_MEDIA_TOKEN_NOT_STREAM", true)
+
+var UpdateTask = common.GetEnvOrDefaultBool("UPDATE_TASK", true)
+
+var GeminiModelMap = map[string]string{
+	"gemini-1.5-pro-latest":   "v1beta",
+	"gemini-1.5-pro-001":      "v1beta",
+	"gemini-1.5-pro":          "v1beta",
+	"gemini-1.5-pro-exp-0801": "v1beta",
+	"gemini-1.5-flash-latest": "v1beta",
+	"gemini-1.5-flash-001":    "v1beta",
+	"gemini-1.5-flash":        "v1beta",
+	"gemini-ultra":            "v1beta",
+}
+
+func InitEnv() {
+	modelVersionMapStr := strings.TrimSpace(os.Getenv("GEMINI_MODEL_MAP"))
+	if modelVersionMapStr == "" {
+		return
+	}
+	for _, pair := range strings.Split(modelVersionMapStr, ",") {
+		parts := strings.Split(pair, ":")
+		if len(parts) == 2 {
+			GeminiModelMap[parts[0]] = parts[1]
+		} else {
+			common.SysError(fmt.Sprintf("invalid model version map: %s", pair))
+		}
+	}
+}

constant/midjourney.go

@@ -27,6 +27,7 @@ const (
 	MjActionLowVariation  = "LOW_VARIATION"
 	MjActionPan           = "PAN"
 	MjActionSwapFace      = "SWAP_FACE"
+	MjActionUpload        = "UPLOAD"
 )
 
 var MidjourneyModel2Action = map[string]string{
@@ -45,4 +46,5 @@ var MidjourneyModel2Action = map[string]string{
 	"mj_low_variation":  MjActionLowVariation,
 	"mj_pan":            MjActionPan,
 	"swap_face":         MjActionSwapFace,
+	"mj_upload":         MjActionUpload,
 }

controller/channel-test.go

@@ -5,6 +5,7 @@ import (
 	"encoding/json"
 	"errors"
 	"fmt"
+	"github.com/bytedance/gopkg/util/gopool"
 	"io"
 	"math"
 	"net/http"
@@ -12,6 +13,7 @@ import (
 	"net/url"
 	"one-api/common"
 	"one-api/dto"
+	"one-api/middleware"
 	"one-api/model"
 	"one-api/relay"
 	relaycommon "one-api/relay/common"
@@ -24,7 +26,7 @@ import (
 	"github.com/gin-gonic/gin"
 )
 
-func testChannel(channel *model.Channel, testModel string) (err error, openaiErr *dto.OpenAIError) {
+func testChannel(channel *model.Channel, testModel string) (err error, openAIErrorWithStatusCode *dto.OpenAIErrorWithStatusCode) {
 	tik := time.Now()
 	if channel.Type == common.ChannelTypeMidjourney {
 		return errors.New("midjourney channel test is not supported"), nil
@@ -40,29 +42,7 @@ func testChannel(channel *model.Channel, testModel string) (err error, openaiErr
 		Body:   nil,
 		Header: make(http.Header),
 	}
-	c.Request.Header.Set("Authorization", "Bearer "+channel.Key)
-	c.Request.Header.Set("Content-Type", "application/json")
-	c.Set("channel", channel.Type)
-	c.Set("base_url", channel.GetBaseURL())
-	switch channel.Type {
-	case common.ChannelTypeAzure:
-		c.Set("api_version", channel.Other)
-	case common.ChannelTypeXunfei:
-		c.Set("api_version", channel.Other)
-	//case common.ChannelTypeAIProxyLibrary:
-	//	c.Set("library_id", channel.Other)
-	case common.ChannelTypeGemini:
-		c.Set("api_version", channel.Other)
-	case common.ChannelTypeAli:
-		c.Set("plugin", channel.Other)
-	}
 
-	meta := relaycommon.GenRelayInfo(c)
-	apiType, _ := constant.ChannelType2APIType(channel.Type)
-	adaptor := relay.GetAdaptor(apiType)
-	if adaptor == nil {
-		return fmt.Errorf("invalid api type: %d, adaptor is nil", apiType), nil
-	}
 	if testModel == "" {
 		if channel.TestModel != nil && *channel.TestModel != "" {
 			testModel = *channel.TestModel
@@ -79,8 +59,7 @@ func testChannel(channel *model.Channel, testModel string) (err error, openaiErr
 			modelMap := make(map[string]string)
 			err := json.Unmarshal([]byte(modelMapping), &modelMap)
 			if err != nil {
-				openaiErr := service.OpenAIErrorWrapperLocal(err, "unmarshal_model_mapping_failed", http.StatusInternalServerError).Error
+				return err, service.OpenAIErrorWrapperLocal(err, "unmarshal_model_mapping_failed", http.StatusInternalServerError)
-				return err, &openaiErr
 			}
 			if modelMap[testModel] != "" {
 				testModel = modelMap[testModel]
@@ -88,14 +67,28 @@ func testChannel(channel *model.Channel, testModel string) (err error, openaiErr
 		}
 	}
 
+	c.Request.Header.Set("Authorization", "Bearer "+channel.Key)
+	c.Request.Header.Set("Content-Type", "application/json")
+	c.Set("channel", channel.Type)
+	c.Set("base_url", channel.GetBaseURL())
+
+	middleware.SetupContextForSelectedChannel(c, channel, testModel)
+
+	meta := relaycommon.GenRelayInfo(c)
+	apiType, _ := constant.ChannelType2APIType(channel.Type)
+	adaptor := relay.GetAdaptor(apiType)
+	if adaptor == nil {
+		return fmt.Errorf("invalid api type: %d, adaptor is nil", apiType), nil
+	}
+
 	request := buildTestRequest()
 	request.Model = testModel
 	meta.UpstreamModelName = testModel
 	common.SysLog(fmt.Sprintf("testing channel %d with model %s", channel.Id, testModel))
 
-	adaptor.Init(meta, *request)
+	adaptor.Init(meta)
 
-	convertedRequest, err := adaptor.ConvertRequest(c, constant.RelayModeChatCompletions, request)
+	convertedRequest, err := adaptor.ConvertRequest(c, meta, request)
 	if err != nil {
 		return err, nil
 	}
@@ -110,12 +103,12 @@ func testChannel(channel *model.Channel, testModel string) (err error, openaiErr
 		return err, nil
 	}
 	if resp != nil && resp.StatusCode != http.StatusOK {
-		err := relaycommon.RelayErrorHandler(resp)
+		err := service.RelayErrorHandler(resp)
-		return fmt.Errorf("status code %d: %s", resp.StatusCode, err.Error.Message), &err.Error
+		return fmt.Errorf("status code %d: %s", resp.StatusCode, err.Error.Message), err
 	}
 	usage, respErr := adaptor.DoResponse(c, resp, meta)
 	if respErr != nil {
-		return fmt.Errorf("%s", respErr.Error.Message), &respErr.Error
+		return fmt.Errorf("%s", respErr.Error.Message), respErr
 	}
 	if usage == nil {
 		return errors.New("usage is nil"), nil
@@ -225,11 +218,11 @@ func testAllChannels(notify bool) error {
 	if disableThreshold == 0 {
 		disableThreshold = 10000000 // a impossible value
 	}
-	go func() {
+	gopool.Go(func() {
 		for _, channel := range channels {
 			isChannelEnabled := channel.Status == common.ChannelStatusEnabled
 			tik := time.Now()
-			err, openaiErr := testChannel(channel, "")
+			err, openaiWithStatusErr := testChannel(channel, "")
 			tok := time.Now()
 			milliseconds := tok.Sub(tik).Milliseconds()
 
@@ -238,27 +231,29 @@ func testAllChannels(notify bool) error {
 				err = errors.New(fmt.Sprintf("响应时间 %.2fs 超过阈值 %.2fs", float64(milliseconds)/1000.0, float64(disableThreshold)/1000.0))
 				ban = true
 			}
-			if openaiErr != nil {
+
-				err = errors.New(fmt.Sprintf("type %s, code %v, message %s", openaiErr.Type, openaiErr.Code, openaiErr.Message))
+			// request error disables the channel
-				ban = true
+			if openaiWithStatusErr != nil {
+				oaiErr := openaiWithStatusErr.Error
+				err = errors.New(fmt.Sprintf("type %s, httpCode %d, code %v, message %s", oaiErr.Type, openaiWithStatusErr.StatusCode, oaiErr.Code, oaiErr.Message))
+				ban = service.ShouldDisableChannel(channel.Type, openaiWithStatusErr)
 			}
+
 			// parse *int to bool
-			if channel.AutoBan != nil && *channel.AutoBan == 0 {
+			if !channel.GetAutoBan() {
 				ban = false
 			}
-			if openaiErr != nil {
+
-				openAiErrWithStatus := dto.OpenAIErrorWithStatusCode{
+			// disable channel
-					StatusCode: -1,
+			if ban && isChannelEnabled {
-					Error:      *openaiErr,
+				service.DisableChannel(channel.Id, channel.Name, err.Error())
-					LocalError: false,
-				}
-				if isChannelEnabled && service.ShouldDisableChannel(channel.Type, &openAiErrWithStatus) && ban {
-					service.DisableChannel(channel.Id, channel.Name, err.Error())
-				}
-				if !isChannelEnabled && service.ShouldEnableChannel(err, openaiErr, channel.Status) {
-					service.EnableChannel(channel.Id, channel.Name)
-				}
 			}
+
+			// enable channel
+			if !isChannelEnabled && service.ShouldEnableChannel(err, openaiWithStatusErr, channel.Status) {
+				service.EnableChannel(channel.Id, channel.Name)
+			}
+
 			channel.UpdateResponseTime(milliseconds)
 			time.Sleep(common.RequestInterval)
 		}
@@ -271,7 +266,7 @@ func testAllChannels(notify bool) error {
 				common.SysError(fmt.Sprintf("failed to send email: %s", err.Error()))
 			}
 		}
-	}()
+	})
 	return nil
 }
 

controller/channel.go

@@ -198,6 +198,28 @@ func AddChannel(c *gin.Context) {
 	}
 	channel.CreatedTime = common.GetTimestamp()
 	keys := strings.Split(channel.Key, "\n")
+	if channel.Type == common.ChannelTypeVertexAi {
+		if channel.Other == "" {
+			c.JSON(http.StatusOK, gin.H{
+				"success": false,
+				"message": "部署地区不能为空",
+			})
+			return
+		} else {
+			if common.IsJsonStr(channel.Other) {
+				// must have default
+				regionMap := common.StrToMap(channel.Other)
+				if regionMap["default"] == nil {
+					c.JSON(http.StatusOK, gin.H{
+						"success": false,
+						"message": "部署地区必须包含default字段",
+					})
+					return
+				}
+			}
+		}
+		keys = []string{channel.Key}
+	}
 	channels := make([]model.Channel, 0, len(keys))
 	for _, key := range keys {
 		if key == "" {
@@ -297,6 +319,27 @@ func UpdateChannel(c *gin.Context) {
 		})
 		return
 	}
+	if channel.Type == common.ChannelTypeVertexAi {
+		if channel.Other == "" {
+			c.JSON(http.StatusOK, gin.H{
+				"success": false,
+				"message": "部署地区不能为空",
+			})
+			return
+		} else {
+			if common.IsJsonStr(channel.Other) {
+				// must have default
+				regionMap := common.StrToMap(channel.Other)
+				if regionMap["default"] == nil {
+					c.JSON(http.StatusOK, gin.H{
+						"success": false,
+						"message": "部署地区必须包含default字段",
+					})
+					return
+				}
+			}
+		}
+	}
 	err = channel.Update()
 	if err != nil {
 		c.JSON(http.StatusOK, gin.H{

controller/log.go

@@ -1,18 +1,19 @@
 package controller
 
 import (
-	"github.com/gin-gonic/gin"
 	"net/http"
 	"one-api/common"
 	"one-api/model"
 	"strconv"
+
+	"github.com/gin-gonic/gin"
 )
 
 func GetAllLogs(c *gin.Context) {
 	p, _ := strconv.Atoi(c.Query("p"))
 	pageSize, _ := strconv.Atoi(c.Query("page_size"))
-	if p < 0 {
+	if p < 1 {
-		p = 0
+		p = 1
 	}
 	if pageSize < 0 {
 		pageSize = common.ItemsPerPage
@@ -24,7 +25,7 @@ func GetAllLogs(c *gin.Context) {
 	tokenName := c.Query("token_name")
 	modelName := c.Query("model_name")
 	channel, _ := strconv.Atoi(c.Query("channel"))
-	logs, total, err := model.GetAllLogs(logType, startTimestamp, endTimestamp, modelName, username, tokenName, p*pageSize, pageSize, channel)
+	logs, total, err := model.GetAllLogs(logType, startTimestamp, endTimestamp, modelName, username, tokenName, (p-1)*pageSize, pageSize, channel)
 	if err != nil {
 		c.JSON(http.StatusOK, gin.H{
 			"success": false,
@@ -35,17 +36,20 @@ func GetAllLogs(c *gin.Context) {
 	c.JSON(http.StatusOK, gin.H{
 		"success": true,
 		"message": "",
-		"total":   total,
+		"data": map[string]any{
-		"data":    logs,
+			"items":     logs,
+			"total":     total,
+			"page":      p,
+			"page_size": pageSize,
+		},
 	})
-	return
 }
 
 func GetUserLogs(c *gin.Context) {
 	p, _ := strconv.Atoi(c.Query("p"))
 	pageSize, _ := strconv.Atoi(c.Query("page_size"))
-	if p < 0 {
+	if p < 1 {
-		p = 0
+		p = 1
 	}
 	if pageSize < 0 {
 		pageSize = common.ItemsPerPage
@@ -59,7 +63,7 @@ func GetUserLogs(c *gin.Context) {
 	endTimestamp, _ := strconv.ParseInt(c.Query("end_timestamp"), 10, 64)
 	tokenName := c.Query("token_name")
 	modelName := c.Query("model_name")
-	logs, total, err := model.GetUserLogs(userId, logType, startTimestamp, endTimestamp, modelName, tokenName, p*pageSize, pageSize)
+	logs, total, err := model.GetUserLogs(userId, logType, startTimestamp, endTimestamp, modelName, tokenName, (p-1)*pageSize, pageSize)
 	if err != nil {
 		c.JSON(http.StatusOK, gin.H{
 			"success": false,
@@ -70,8 +74,12 @@ func GetUserLogs(c *gin.Context) {
 	c.JSON(http.StatusOK, gin.H{
 		"success": true,
 		"message": "",
-		"total":   total,
+		"data": map[string]any{
-		"data":    logs,
+			"items":     logs,
+			"total":     total,
+			"page":      p,
+			"page_size": pageSize,
+		},
 	})
 	return
 }

controller/midjourney.go

@@ -146,28 +146,26 @@ func UpdateMidjourneyTaskBulk() {
 					buttonStr, _ := json.Marshal(responseItem.Buttons)
 					task.Buttons = string(buttonStr)
 				}
-
+				shouldReturnQuota := false
 				if (task.Progress != "100%" && responseItem.FailReason != "") || (task.Progress == "100%" && task.Status == "FAILURE") {
 					common.LogInfo(ctx, task.MjId+" 构建失败，"+task.FailReason)
 					task.Progress = "100%"
-					err = model.CacheUpdateUserQuota(task.UserId)
+					if task.Quota != 0 {
-					if err != nil {
+						shouldReturnQuota = true
-						common.LogError(ctx, "error update user quota cache: "+err.Error())
-					} else {
-						quota := task.Quota
-						if quota != 0 {
-							err = model.IncreaseUserQuota(task.UserId, quota)
-							if err != nil {
-								common.LogError(ctx, "fail to increase user quota: "+err.Error())
-							}
-							logContent := fmt.Sprintf("构图失败 %s，补偿 %s", task.MjId, common.LogQuota(quota))
-							model.RecordLog(task.UserId, model.LogTypeSystem, logContent)
-						}
 					}
 				}
 				err = task.Update()
 				if err != nil {
 					common.LogError(ctx, "UpdateMidjourneyTask task error: "+err.Error())
+				} else {
+					if shouldReturnQuota {
+						err = model.IncreaseUserQuota(task.UserId, task.Quota)
+						if err != nil {
+							common.LogError(ctx, "fail to increase user quota: "+err.Error())
+						}
+						logContent := fmt.Sprintf("构图失败 %s，补偿 %s", task.MjId, common.LogQuota(task.Quota))
+						model.RecordLog(task.UserId, model.LogTypeSystem, logContent)
+					}
 				}
 			}
 		}

controller/model.go

@@ -131,7 +131,7 @@ func init() {
 		}
 		meta := &relaycommon.RelayInfo{ChannelType: i}
 		adaptor := relay.GetAdaptor(apiType)
-		adaptor.Init(meta, dto.GeneralOpenAIRequest{})
+		adaptor.Init(meta)
 		channelId2Models[i] = adaptor.GetModelList()
 	}
 }

controller/relay.go

@@ -2,6 +2,7 @@ package controller
 
 import (
 	"bytes"
+	"errors"
 	"fmt"
 	"github.com/gin-gonic/gin"
 	"io"
@@ -22,13 +23,13 @@ func relayHandler(c *gin.Context, relayMode int) *dto.OpenAIErrorWithStatusCode
 	var err *dto.OpenAIErrorWithStatusCode
 	switch relayMode {
 	case relayconstant.RelayModeImagesGenerations:
-		err = relay.RelayImageHelper(c, relayMode)
+		err = relay.ImageHelper(c, relayMode)
 	case relayconstant.RelayModeAudioSpeech:
 		fallthrough
 	case relayconstant.RelayModeAudioTranslation:
 		fallthrough
 	case relayconstant.RelayModeAudioTranscription:
-		err = relay.AudioHelper(c, relayMode)
+		err = relay.AudioHelper(c)
 	case relayconstant.RelayModeRerank:
 		err = relay.RerankHelper(c, relayMode)
 	default:
@@ -39,43 +40,35 @@ func relayHandler(c *gin.Context, relayMode int) *dto.OpenAIErrorWithStatusCode
 
 func Relay(c *gin.Context) {
 	relayMode := constant.Path2RelayMode(c.Request.URL.Path)
-	retryTimes := common.RetryTimes
 	requestId := c.GetString(common.RequestIdKey)
-	channelId := c.GetInt("channel_id")
-	channelType := c.GetInt("channel_type")
 	group := c.GetString("group")
 	originalModel := c.GetString("original_model")
-	openaiErr := relayHandler(c, relayMode)
+	var openaiErr *dto.OpenAIErrorWithStatusCode
-	c.Set("use_channel", []string{fmt.Sprintf("%d", channelId)})
+
-	if openaiErr != nil {
+	for i := 0; i <= common.RetryTimes; i++ {
-		go processChannelError(c, channelId, channelType, openaiErr)
+		channel, err := getChannel(c, group, originalModel, i)
-	} else {
-		retryTimes = 0
-	}
-	for i := 0; shouldRetry(c, channelId, openaiErr, retryTimes) && i < retryTimes; i++ {
-		channel, err := model.CacheGetRandomSatisfiedChannel(group, originalModel, i)
 		if err != nil {
-			common.LogError(c.Request.Context(), fmt.Sprintf("CacheGetRandomSatisfiedChannel failed: %s", err.Error()))
+			common.LogError(c, err.Error())
+			openaiErr = service.OpenAIErrorWrapperLocal(err, "get_channel_failed", http.StatusInternalServerError)
 			break
 		}
-		channelId = channel.Id
-		useChannel := c.GetStringSlice("use_channel")
-		useChannel = append(useChannel, fmt.Sprintf("%d", channelId))
-		c.Set("use_channel", useChannel)
-		common.LogInfo(c.Request.Context(), fmt.Sprintf("using channel #%d to retry (remain times %d)", channel.Id, i))
-		middleware.SetupContextForSelectedChannel(c, channel, originalModel)
 
-		requestBody, err := common.GetRequestBody(c)
+		openaiErr = relayRequest(c, relayMode, channel)
-		c.Request.Body = io.NopCloser(bytes.NewBuffer(requestBody))
+
-		openaiErr = relayHandler(c, relayMode)
+		if openaiErr == nil {
-		if openaiErr != nil {
+			return // 成功处理请求，直接返回
-			go processChannelError(c, channelId, channel.Type, openaiErr)
+		}
+
+		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.Request.Context(), retryLogStr)
+		common.LogInfo(c, retryLogStr)
 	}
 
 	if openaiErr != nil {
@@ -89,10 +82,48 @@ func Relay(c *gin.Context) {
 	}
 }
 
-func shouldRetry(c *gin.Context, channelId int, openaiErr *dto.OpenAIErrorWithStatusCode, retryTimes int) bool {
+func relayRequest(c *gin.Context, relayMode int, channel *model.Channel) *dto.OpenAIErrorWithStatusCode {
+	addUsedChannel(c, channel.Id)
+	requestBody, _ := common.GetRequestBody(c)
+	c.Request.Body = io.NopCloser(bytes.NewBuffer(requestBody))
+	return relayHandler(c, relayMode)
+}
+
+func addUsedChannel(c *gin.Context, channelId int) {
+	useChannel := c.GetStringSlice("use_channel")
+	useChannel = append(useChannel, fmt.Sprintf("%d", channelId))
+	c.Set("use_channel", useChannel)
+}
+
+func getChannel(c *gin.Context, group, originalModel string, retryCount int) (*model.Channel, error) {
+	if retryCount == 0 {
+		autoBan := c.GetBool("auto_ban")
+		autoBanInt := 1
+		if !autoBan {
+			autoBanInt = 0
+		}
+		return &model.Channel{
+			Id:      c.GetInt("channel_id"),
+			Type:    c.GetInt("channel_type"),
+			Name:    c.GetString("channel_name"),
+			AutoBan: &autoBanInt,
+		}, nil
+	}
+	channel, err := model.CacheGetRandomSatisfiedChannel(group, originalModel, retryCount)
+	if err != nil {
+		return nil, errors.New(fmt.Sprintf("获取重试渠道失败: %s", err.Error()))
+	}
+	middleware.SetupContextForSelectedChannel(c, channel, originalModel)
+	return channel, nil
+}
+
+func shouldRetry(c *gin.Context, openaiErr *dto.OpenAIErrorWithStatusCode, retryTimes int) bool {
 	if openaiErr == nil {
 		return false
 	}
+	if openaiErr.LocalError {
+		return false
+	}
 	if retryTimes <= 0 {
 		return false
 	}
@@ -113,26 +144,27 @@ func shouldRetry(c *gin.Context, channelId int, openaiErr *dto.OpenAIErrorWithSt
 		return true
 	}
 	if openaiErr.StatusCode == http.StatusBadRequest {
+		channelType := c.GetInt("channel_type")
+		if channelType == common.ChannelTypeAnthropic {
+			return true
+		}
 		return false
 	}
 	if openaiErr.StatusCode == 408 {
 		// azure处理超时不重试
 		return false
 	}
-	if openaiErr.LocalError {
-		return false
-	}
 	if openaiErr.StatusCode/100 == 2 {
 		return false
 	}
 	return true
 }
 
-func processChannelError(c *gin.Context, channelId int, channelType int, err *dto.OpenAIErrorWithStatusCode) {
+func processChannelError(c *gin.Context, channelId int, channelType int, channelName string, autoBan bool, err *dto.OpenAIErrorWithStatusCode) {
-	autoBan := c.GetBool("auto_ban")
+	// 不要使用context获取渠道信息，异步处理时可能会出现渠道信息不一致的情况
-	common.LogError(c.Request.Context(), fmt.Sprintf("relay error (channel #%d, status code: %d): %s", channelId, err.StatusCode, err.Error.Message))
+	// do not use context to get channel info, there may be inconsistent channel info when processing asynchronously
+	common.LogError(c, fmt.Sprintf("relay error (channel #%d, status code: %d): %s", channelId, err.StatusCode, err.Error.Message))
 	if service.ShouldDisableChannel(channelType, err) && autoBan {
-		channelName := c.GetString("channel_name")
 		service.DisableChannel(channelId, channelName, err.Error.Message)
 	}
 }
@@ -208,14 +240,14 @@ func RelayTask(c *gin.Context) {
 	for i := 0; shouldRetryTaskRelay(c, channelId, taskErr, retryTimes) && i < retryTimes; i++ {
 		channel, err := model.CacheGetRandomSatisfiedChannel(group, originalModel, i)
 		if err != nil {
-			common.LogError(c.Request.Context(), fmt.Sprintf("CacheGetRandomSatisfiedChannel failed: %s", err.Error()))
+			common.LogError(c, fmt.Sprintf("CacheGetRandomSatisfiedChannel failed: %s", err.Error()))
 			break
 		}
 		channelId = channel.Id
 		useChannel := c.GetStringSlice("use_channel")
 		useChannel = append(useChannel, fmt.Sprintf("%d", channelId))
 		c.Set("use_channel", useChannel)
-		common.LogInfo(c.Request.Context(), fmt.Sprintf("using channel #%d to retry (remain times %d)", channel.Id, i))
+		common.LogInfo(c, fmt.Sprintf("using channel #%d to retry (remain times %d)", channel.Id, i))
 		middleware.SetupContextForSelectedChannel(c, channel, originalModel)
 
 		requestBody, err := common.GetRequestBody(c)
@@ -225,7 +257,7 @@ func RelayTask(c *gin.Context) {
 	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.Request.Context(), retryLogStr)
+		common.LogInfo(c, retryLogStr)
 	}
 	if taskErr != nil {
 		if taskErr.StatusCode == http.StatusTooManyRequests {

controller/user.go

@@ -806,11 +806,11 @@ type topUpRequest struct {
 	Key string `json:"key"`
 }
 
-var lock = sync.Mutex{}
+var topUpLock = sync.Mutex{}
 
 func TopUp(c *gin.Context) {
-	lock.Lock()
+	topUpLock.Lock()
-	defer lock.Unlock()
+	defer topUpLock.Unlock()
 	req := topUpRequest{}
 	err := c.ShouldBindJSON(&req)
 	if err != nil {

dto/audio.go

@@ -1,13 +1,34 @@
 package dto
 
-type TextToSpeechRequest struct {
+type AudioRequest struct {
-	Model          string  `json:"model" binding:"required"`
+	Model          string  `json:"model"`
-	Input          string  `json:"input" binding:"required"`
+	Input          string  `json:"input"`
-	Voice          string  `json:"voice" binding:"required"`
+	Voice          string  `json:"voice"`
-	Speed          float64 `json:"speed"`
+	Speed          float64 `json:"speed,omitempty"`
-	ResponseFormat string  `json:"response_format"`
+	ResponseFormat string  `json:"response_format,omitempty"`
 }
 
 type AudioResponse struct {
 	Text string `json:"text"`
 }
+
+type WhisperVerboseJSONResponse struct {
+	Task     string    `json:"task,omitempty"`
+	Language string    `json:"language,omitempty"`
+	Duration float64   `json:"duration,omitempty"`
+	Text     string    `json:"text,omitempty"`
+	Segments []Segment `json:"segments,omitempty"`
+}
+
+type Segment struct {
+	Id               int     `json:"id"`
+	Seek             int     `json:"seek"`
+	Start            float64 `json:"start"`
+	End              float64 `json:"end"`
+	Text             string  `json:"text"`
+	Tokens           []int   `json:"tokens"`
+	Temperature      float64 `json:"temperature"`
+	AvgLogprob       float64 `json:"avg_logprob"`
+	CompressionRatio float64 `json:"compression_ratio"`
+	NoSpeechProb     float64 `json:"no_speech_prob"`
+}

dto/dalle.go

@@ -12,9 +12,11 @@ type ImageRequest struct {
 }
 
 type ImageResponse struct {
-	Created int `json:"created"`
+	Data    []ImageData `json:"data"`
-	Data    []struct {
+	Created int64       `json:"created"`
-		Url     string `json:"url"`
+}
-		B64Json string `json:"b64_json"`
+type ImageData struct {
-	}
+	Url           string `json:"url"`
+	B64Json       string `json:"b64_json"`
+	RevisedPrompt string `json:"revised_prompt"`
 }

dto/midjourney.go

@@ -33,6 +33,12 @@ type MidjourneyResponse struct {
 	Result      string      `json:"result"`
 }
 
+type MidjourneyUploadResponse struct {
+	Code        int      `json:"code"`
+	Description string   `json:"description"`
+	Result      []string `json:"result"`
+}
+
 type MidjourneyResponseWithStatusCode struct {
 	StatusCode int `json:"statusCode"`
 	Response   MidjourneyResponse

dto/rerank.go

@@ -1,14 +1,17 @@
 package dto
 
 type RerankRequest struct {
-	Documents []any  `json:"documents"`
+	Documents       []any  `json:"documents"`
-	Query     string `json:"query"`
+	Query           string `json:"query"`
-	Model     string `json:"model"`
+	Model           string `json:"model"`
-	TopN      int    `json:"top_n"`
+	TopN            int    `json:"top_n"`
+	ReturnDocuments bool   `json:"return_documents,omitempty"`
+	MaxChunkPerDoc  int    `json:"max_chunk_per_doc,omitempty"`
+	OverLapTokens   int    `json:"overlap_tokens,omitempty"`
 }
 
 type RerankResponseDocument struct {
-	Document       any     `json:"document"`
+	Document       any     `json:"document,omitempty"`
 	Index          int     `json:"index"`
 	RelevanceScore float64 `json:"relevance_score"`
 }

dto/text_request.go

@@ -2,39 +2,38 @@ package dto
 
 import "encoding/json"
 
-type ResponseFormat struct {
-	Type string `json:"type,omitempty"`
-}
-
 type GeneralOpenAIRequest struct {
-	Model            string          `json:"model,omitempty"`
+	Model             string         `json:"model,omitempty"`
-	Messages         []Message       `json:"messages,omitempty"`
+	Messages          []Message      `json:"messages,omitempty"`
-	Prompt           any             `json:"prompt,omitempty"`
+	Prompt            any            `json:"prompt,omitempty"`
-	BestOf           int             `json:"best_of,omitempty"`
+	BestOf            int            `json:"best_of,omitempty"`
-	Echo             bool            `json:"echo,omitempty"`
+	Echo              bool           `json:"echo,omitempty"`
-	Stream           bool            `json:"stream,omitempty"`
+	Stream            bool           `json:"stream,omitempty"`
-	StreamOptions    *StreamOptions  `json:"stream_options,omitempty"`
+	StreamOptions     *StreamOptions `json:"stream_options,omitempty"`
-	Suffix           string          `json:"suffix,omitempty"`
+	Suffix            string         `json:"suffix,omitempty"`
-	MaxTokens        uint            `json:"max_tokens,omitempty"`
+	MaxTokens         uint           `json:"max_tokens,omitempty"`
-	Temperature      float64         `json:"temperature,omitempty"`
+	Temperature       float64        `json:"temperature,omitempty"`
-	TopP             float64         `json:"top_p,omitempty"`
+	TopP              float64        `json:"top_p,omitempty"`
-	TopK             int             `json:"top_k,omitempty"`
+	TopK              int            `json:"top_k,omitempty"`
-	Stop             any             `json:"stop,omitempty"`
+	Stop              any            `json:"stop,omitempty"`
-	N                int             `json:"n,omitempty"`
+	N                 int            `json:"n,omitempty"`
-	Input            any             `json:"input,omitempty"`
+	Input             any            `json:"input,omitempty"`
-	Instruction      string          `json:"instruction,omitempty"`
+	Instruction       string         `json:"instruction,omitempty"`
-	Size             string          `json:"size,omitempty"`
+	Size              string         `json:"size,omitempty"`
-	Functions        any             `json:"functions,omitempty"`
+	Functions         any            `json:"functions,omitempty"`
-	FrequencyPenalty float64         `json:"frequency_penalty,omitempty"`
+	FrequencyPenalty  float64        `json:"frequency_penalty,omitempty"`
-	PresencePenalty  float64         `json:"presence_penalty,omitempty"`
+	PresencePenalty   float64        `json:"presence_penalty,omitempty"`
-	ResponseFormat   *ResponseFormat `json:"response_format,omitempty"`
+	ResponseFormat    any            `json:"response_format,omitempty"`
-	Seed             float64         `json:"seed,omitempty"`
+	Seed              float64        `json:"seed,omitempty"`
-	Tools            any             `json:"tools,omitempty"`
+	Tools             []ToolCall     `json:"tools,omitempty"`
-	ToolChoice       any             `json:"tool_choice,omitempty"`
+	ToolChoice        any            `json:"tool_choice,omitempty"`
-	User             string          `json:"user,omitempty"`
+	User              string         `json:"user,omitempty"`
-	LogitBias        any             `json:"logit_bias,omitempty"`
+	LogitBias         any            `json:"logit_bias,omitempty"`
-	LogProbs         any             `json:"logprobs,omitempty"`
+	LogProbs          any            `json:"logprobs,omitempty"`
-	TopLogProbs      int             `json:"top_logprobs,omitempty"`
+	TopLogProbs       int            `json:"top_logprobs,omitempty"`
+	Dimensions        int            `json:"dimensions,omitempty"`
+	ParallelToolCalls bool           `json:"parallel_Tool_Calls,omitempty"`
+	EncodingFormat    string         `json:"encoding_format,omitempty"`
 }
 
 type OpenAITools struct {
@@ -52,8 +51,8 @@ type StreamOptions struct {
 	IncludeUsage bool `json:"include_usage,omitempty"`
 }
 
-func (r GeneralOpenAIRequest) GetMaxTokens() int64 {
+func (r GeneralOpenAIRequest) GetMaxTokens() int {
-	return int64(r.MaxTokens)
+	return int(r.MaxTokens)
 }
 
 func (r GeneralOpenAIRequest) ParseInput() []string {
@@ -107,6 +106,11 @@ func (m Message) StringContent() string {
 	return string(m.Content)
 }
 
+func (m *Message) SetStringContent(content string) {
+	jsonContent, _ := json.Marshal(content)
+	m.Content = jsonContent
+}
+
 func (m Message) IsStringContent() bool {
 	var stringContent string
 	if err := json.Unmarshal(m.Content, &stringContent); err == nil {
@@ -146,7 +150,7 @@ func (m Message) ParseContent() []MediaMessage {
 					if ok {
 						subObj["detail"] = detail.(string)
 					} else {
-						subObj["detail"] = "auto"
+						subObj["detail"] = "high"
 					}
 					contentList = append(contentList, MediaMessage{
 						Type: ContentTypeImageURL,
@@ -155,7 +159,16 @@ func (m Message) ParseContent() []MediaMessage {
 							Detail: subObj["detail"].(string),
 						},
 					})
+				} else if url, ok := contentMap["image_url"].(string); ok {
+					contentList = append(contentList, MediaMessage{
+						Type: ContentTypeImageURL,
+						ImageUrl: MessageImageUrl{
+							Url:    url,
+							Detail: "high",
+						},
+					})
 				}
+
 			}
 		}
 		return contentList

dto/text_response.go

@@ -41,9 +41,9 @@ type OpenAITextResponse struct {
 }
 
 type OpenAIEmbeddingResponseItem struct {
-	Object    string    `json:"object"`
+	Object    string `json:"object"`
-	Index     int       `json:"index"`
+	Index     int    `json:"index"`
-	Embedding []float64 `json:"embedding"`
+	Embedding any    `json:"embedding"`
 }
 
 type OpenAIEmbeddingResponse struct {
@@ -66,10 +66,6 @@ type ChatCompletionsStreamResponseChoiceDelta struct {
 	ToolCalls []ToolCall `json:"tool_calls,omitempty"`
 }
 
-func (c *ChatCompletionsStreamResponseChoiceDelta) IsEmpty() bool {
-	return c.Content == nil && len(c.ToolCalls) == 0
-}
-
 func (c *ChatCompletionsStreamResponseChoiceDelta) SetContentString(s string) {
 	c.Content = &s
 }
@@ -90,9 +86,11 @@ type ToolCall struct {
 }
 
 type FunctionCall struct {
-	Name string `json:"name,omitempty"`
+	Description string `json:"description,omitempty"`
+	Name        string `json:"name,omitempty"`
 	// call function with arguments in JSON format
-	Arguments string `json:"arguments,omitempty"`
+	Parameters any    `json:"parameters,omitempty"` // request
+	Arguments  string `json:"arguments,omitempty"`
 }
 
 type ChatCompletionsStreamResponse struct {
@@ -105,6 +103,17 @@ type ChatCompletionsStreamResponse struct {
 	Usage             *Usage                                `json:"usage"`
 }
 
+func (c *ChatCompletionsStreamResponse) GetSystemFingerprint() string {
+	if c.SystemFingerprint == nil {
+		return ""
+	}
+	return *c.SystemFingerprint
+}
+
+func (c *ChatCompletionsStreamResponse) SetSystemFingerprint(s string) {
+	c.SystemFingerprint = &s
+}
+
 type ChatCompletionsStreamResponseSimple struct {
 	Choices []ChatCompletionsStreamResponseChoice `json:"choices"`
 	Usage   *Usage                                `json:"usage"`

go.mod

@@ -1,13 +1,16 @@
 module one-api
 
 // +heroku goVersion go1.18
-go 1.18
+go 1.21
+
+toolchain go1.22.4
 
 require (
 	github.com/anknown/ahocorasick v0.0.0-20190904063843-d75dbd5169c0
 	github.com/aws/aws-sdk-go-v2 v1.26.1
 	github.com/aws/aws-sdk-go-v2/credentials v1.17.11
 	github.com/aws/aws-sdk-go-v2/service/bedrockruntime v1.7.4
+	github.com/bytedance/gopkg v0.0.0-20220118071334-3db87571198b
 	github.com/gin-contrib/cors v1.4.0
 	github.com/gin-contrib/gzip v0.0.6
 	github.com/gin-contrib/sessions v0.0.5
@@ -24,8 +27,9 @@ require (
 	github.com/samber/lo v1.39.0
 	github.com/shirou/gopsutil v3.21.11+incompatible
 	github.com/stripe/stripe-go/v76 v76.21.0
-	golang.org/x/crypto v0.21.0
+	golang.org/x/crypto v0.26.0
 	golang.org/x/image v0.15.0
+	golang.org/x/net v0.28.0
 	gorm.io/driver/mysql v1.4.3
 	gorm.io/driver/postgres v1.5.2
 	gorm.io/driver/sqlite v1.4.3
@@ -39,7 +43,7 @@ require (
 	github.com/aws/aws-sdk-go-v2/internal/endpoints/v2 v2.6.5 // indirect
 	github.com/aws/smithy-go v1.20.2 // indirect
 	github.com/bytedance/sonic v1.9.1 // indirect
-	github.com/cespare/xxhash/v2 v2.1.2 // indirect
+	github.com/cespare/xxhash/v2 v2.3.0 // indirect
 	github.com/chenzhuoyu/base64x v0.0.0-20221115062448-fe3a3abad311 // indirect
 	github.com/dgryski/go-rendezvous v0.0.0-20200823014737-9f7001d12a5f // indirect
 	github.com/dlclark/regexp2 v1.11.0 // indirect
@@ -50,6 +54,7 @@ require (
 	github.com/go-playground/universal-translator v0.18.1 // indirect
 	github.com/go-sql-driver/mysql v1.6.0 // indirect
 	github.com/goccy/go-json v0.10.2 // indirect
+	github.com/google/go-cmp v0.6.0 // indirect
 	github.com/gorilla/context v1.1.1 // indirect
 	github.com/gorilla/securecookie v1.1.1 // indirect
 	github.com/gorilla/sessions v1.2.1 // indirect
@@ -67,6 +72,7 @@ require (
 	github.com/modern-go/concurrent v0.0.0-20180306012644-bacd9c7ef1dd // indirect
 	github.com/modern-go/reflect2 v1.0.2 // indirect
 	github.com/pelletier/go-toml/v2 v2.0.8 // indirect
+	github.com/stretchr/testify v1.9.0 // indirect
 	github.com/tklauser/go-sysconf v0.3.12 // indirect
 	github.com/tklauser/numcpus v0.6.1 // indirect
 	github.com/twitchyliquid64/golang-asm v0.15.1 // indirect
@@ -74,10 +80,9 @@ require (
 	github.com/yusufpapurcu/wmi v1.2.3 // indirect
 	golang.org/x/arch v0.3.0 // indirect
 	golang.org/x/exp v0.0.0-20240404231335-c0f41cb1a7a0 // indirect
-	golang.org/x/net v0.21.0 // indirect
+	golang.org/x/sync v0.8.0 // indirect
-	golang.org/x/sync v0.7.0 // indirect
+	golang.org/x/sys v0.24.0 // indirect
-	golang.org/x/sys v0.18.0 // indirect
+	golang.org/x/text v0.17.0 // indirect
-	golang.org/x/text v0.14.0 // indirect
+	google.golang.org/protobuf v1.34.2 // indirect
-	google.golang.org/protobuf v1.30.0 // indirect
 	gopkg.in/yaml.v3 v3.0.1 // indirect
 )

go.sum

@@ -16,11 +16,13 @@ github.com/aws/aws-sdk-go-v2/service/bedrockruntime v1.7.4 h1:JgHnonzbnA3pbqj76w
 github.com/aws/aws-sdk-go-v2/service/bedrockruntime v1.7.4/go.mod h1:nZspkhg+9p8iApLFoyAqfyuMP0F38acy2Hm3r5r95Cg=
 github.com/aws/smithy-go v1.20.2 h1:tbp628ireGtzcHDDmLT/6ADHidqnwgF57XOXZe6tp4Q=
 github.com/aws/smithy-go v1.20.2/go.mod h1:krry+ya/rV9RDcV/Q16kpu6ypI4K2czasz0NC3qS14E=
+github.com/bytedance/gopkg v0.0.0-20220118071334-3db87571198b h1:LTGVFpNmNHhj0vhOlfgWueFJ32eK9blaIlHR2ciXOT0=
+github.com/bytedance/gopkg v0.0.0-20220118071334-3db87571198b/go.mod h1:2ZlV9BaUH4+NXIBF0aMdKKAnHTzqH+iMU4KUjAbL23Q=
 github.com/bytedance/sonic v1.5.0/go.mod h1:ED5hyg4y6t3/9Ku1R6dU/4KyJ48DZ4jPhfY1O2AihPM=
 github.com/bytedance/sonic v1.9.1 h1:6iJ6NqdoxCDr6mbY8h18oSO+cShGSMRGCEo7F2h0x8s=
 github.com/bytedance/sonic v1.9.1/go.mod h1:i736AoUSYt75HyZLoJW9ERYxcy6eaN6h4BZXU064P/U=
-github.com/cespare/xxhash/v2 v2.1.2 h1:YRXhKfTDauu4ajMg1TPgFO5jnlC2HCbmLXMcTG5cbYE=
+github.com/cespare/xxhash/v2 v2.3.0 h1:UL815xU9SqsFlibzuggzjXhog7bL6oX9BbNZnL2UFvs=
-github.com/cespare/xxhash/v2 v2.1.2/go.mod h1:VGX0DQ3Q6kWi7AoAeZDth3/j3BFtOZR5XLFGgcrjCOs=
+github.com/cespare/xxhash/v2 v2.3.0/go.mod h1:VGX0DQ3Q6kWi7AoAeZDth3/j3BFtOZR5XLFGgcrjCOs=
 github.com/chenzhuoyu/base64x v0.0.0-20211019084208-fb5309c8db06/go.mod h1:DH46F32mSOjUmXrMHnKwZdA8wcEefY7UVqBKYGjpdQY=
 github.com/chenzhuoyu/base64x v0.0.0-20221115062448-fe3a3abad311 h1:qSGYFH7+jGhDF8vLC+iwCD4WpbV1EBDSzWkJODFLams=
 github.com/chenzhuoyu/base64x v0.0.0-20221115062448-fe3a3abad311/go.mod h1:b583jCggY9gE99b6G5LEC39OIiVsWj+R97kbl5odCEk=
@@ -30,11 +32,10 @@ github.com/davecgh/go-spew v1.1.1 h1:vj9j/u1bqnvCEfJOwUhtlOARqs3+rkHYY13jYWTU97c
 github.com/davecgh/go-spew v1.1.1/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
 github.com/dgryski/go-rendezvous v0.0.0-20200823014737-9f7001d12a5f h1:lO4WD4F/rVNCu3HqELle0jiPLLBs70cWOduZpkS1E78=
 github.com/dgryski/go-rendezvous v0.0.0-20200823014737-9f7001d12a5f/go.mod h1:cuUVRXasLTGF7a8hSLbxyZXjz+1KgoB3wDUb6vlszIc=
-github.com/dlclark/regexp2 v1.10.0 h1:+/GIL799phkJqYW+3YbOd8LCcbHzT0Pbo8zl70MHsq0=
-github.com/dlclark/regexp2 v1.10.0/go.mod h1:DHkYz0B9wPfa6wondMfaivmHpzrQ3v9q8cnmRbL6yW8=
 github.com/dlclark/regexp2 v1.11.0 h1:G/nrcoOa7ZXlpoa/91N3X7mM3r8eIlMBBJZvsz/mxKI=
 github.com/dlclark/regexp2 v1.11.0/go.mod h1:DHkYz0B9wPfa6wondMfaivmHpzrQ3v9q8cnmRbL6yW8=
 github.com/fsnotify/fsnotify v1.4.9 h1:hsms1Qyu0jgnwNXIxa+/V/PDsU6CfLf6CNO8H7IWoS4=
+github.com/fsnotify/fsnotify v1.4.9/go.mod h1:znqG4EE+3YCdAaPaxE2ZRY/06pZUdp0tY4IgpuI1SZQ=
 github.com/gabriel-vasile/mimetype v1.4.3 h1:in2uUcidCuFcDKtdcBxlR0rJ1+fsokWf+uqxgUFjbI0=
 github.com/gabriel-vasile/mimetype v1.4.3/go.mod h1:d8uq/6HKRL6CGdk+aubisF/M5GcPfT7nKyLpA0lbSSk=
 github.com/gin-contrib/cors v1.4.0 h1:oJ6gwtUl3lqV0WEIwM/LxPF1QZ5qe2lGWdY2+bz7y0g=
@@ -55,6 +56,7 @@ github.com/go-ole/go-ole v1.2.6 h1:/Fpf6oFPoeFik9ty7siob0G6Ke8QvQEuVcuChpwXzpY=
 github.com/go-ole/go-ole v1.2.6/go.mod h1:pprOEPIfldk/42T2oK7lQ4v4JSDwmV0As9GaiUsvbm0=
 github.com/go-playground/assert/v2 v2.0.1/go.mod h1:VDjEfimB/XKnb+ZQfWdccd7VUvScMdVu0Titje2rxJ4=
 github.com/go-playground/assert/v2 v2.2.0 h1:JvknZsQTYeFEAhQwI4qEt9cyV5ONwRHC+lYKSsYSR8s=
+github.com/go-playground/assert/v2 v2.2.0/go.mod h1:VDjEfimB/XKnb+ZQfWdccd7VUvScMdVu0Titje2rxJ4=
 github.com/go-playground/locales v0.13.0/go.mod h1:taPMhCMXrRLJO55olJkUXHZBHCxTMfnGwq/HNwmWNS8=
 github.com/go-playground/locales v0.14.0/go.mod h1:sawfccIbzZTqEDETgFXqTho0QybSa7l++s0DH+LDiLs=
 github.com/go-playground/locales v0.14.1 h1:EWaQ/wswjilfKLTECiXz7Rh+3BjFhfDFKv/oXslEjJA=
@@ -79,10 +81,9 @@ github.com/golang-jwt/jwt v3.2.2+incompatible/go.mod h1:8pz2t5EyA70fFQQSrl6XZXzq
 github.com/golang/protobuf v1.3.3/go.mod h1:vzj43D7+SQXF/4pzW/hwtAqwc6iTitCiVSaWz5lYuqw=
 github.com/golang/protobuf v1.5.0/go.mod h1:FsONVRAS9T7sI+LIUmWTfcYkHO4aIWwzhcaSAoJOfIk=
 github.com/google/go-cmp v0.5.5/go.mod h1:v8dTdLbMG2kIc/vJvl+f65V22dbkXbowE6jgT/gNBxE=
-github.com/google/go-cmp v0.5.8 h1:e6P7q2lk1O+qJJb4BtCQXlK8vWEO8V1ZeuEdJNOqZyg=
+github.com/google/go-cmp v0.6.0 h1:ofyhxvXcZhMsU5ulbFiLKl/XBFqE1GSq7atu8tAmTRI=
+github.com/google/go-cmp v0.6.0/go.mod h1:17dUlkBOakJ0+DkrSSNjCkIjxS6bF9zb3elmeNGIjoY=
 github.com/google/gofuzz v1.0.0/go.mod h1:dBl0BpW6vV/+mYPU4Po3pmUjxk6FQPldtuIdl/M65Eg=
-github.com/google/uuid v1.3.0 h1:t6JiXgmwXMjEs8VusXIJk2BXHsn+wx8BZdTaoZ5fu7I=
-github.com/google/uuid v1.3.0/go.mod h1:TIyPZe4MgqvfeYDBFedMoGGpEw/LqOeaOT+nhxU+yHo=
 github.com/google/uuid v1.6.0 h1:NIvaJDMOsjHA8n1jAhLSgzrAzy1Hgr+hNrb57e+94F0=
 github.com/google/uuid v1.6.0/go.mod h1:TIyPZe4MgqvfeYDBFedMoGGpEw/LqOeaOT+nhxU+yHo=
 github.com/gorilla/context v1.1.1 h1:AWwleXJkX/nhcU9bZSnZoi3h/qGYqQAGhq6zZe/aQW8=
@@ -126,8 +127,6 @@ github.com/leodido/go-urn v1.2.0/go.mod h1:+8+nEpDfqqsY+g338gtMEUOtuK+4dEMhiQEgx
 github.com/leodido/go-urn v1.2.1/go.mod h1:zt4jvISO2HfUBqxjfIshjdMTYS56ZS/qv49ictyFfxY=
 github.com/leodido/go-urn v1.4.0 h1:WT9HwE9SGECu3lg4d/dIA+jxlljEa1/ffXKmRjqdmIQ=
 github.com/leodido/go-urn v1.4.0/go.mod h1:bvxc+MVxLKB4z00jd1z+Dvzr47oO32F/QSNjSBOlFxI=
-github.com/linux-do/tiktoken-go v0.7.0 h1:Kcm/miJ5gp77srtF8GQWnfq7W9kTaXEuHZg/g9IVEu8=
-github.com/linux-do/tiktoken-go v0.7.0/go.mod h1:9Vkdtp0ngi4USmrdSx984iuIQ5IMr0hnUdz4jZZTJb8=
 github.com/mattn/go-isatty v0.0.12/go.mod h1:cbi8OIDigv2wuxKPP5vlRcQ1OAZbq2CE4Kysco4FUpU=
 github.com/mattn/go-isatty v0.0.14/go.mod h1:7GGIvUiUoEMVVmxf/4nioHXj79iQHKdU27kJ6hsGG94=
 github.com/mattn/go-isatty v0.0.20 h1:xfD0iDuEKnDkl03q4limB+vH+GxLEtL/jb4xVJSWWEY=
@@ -142,16 +141,17 @@ github.com/modern-go/reflect2 v0.0.0-20180701023420-4b7aa43c6742/go.mod h1:bx2lN
 github.com/modern-go/reflect2 v1.0.2 h1:xBagoLtFs94CBntxluKeaWgTMpvLxC4ur3nMaC9Gz0M=
 github.com/modern-go/reflect2 v1.0.2/go.mod h1:yWuevngMOJpCy52FWWMvUC8ws7m/LJsjYzDa0/r8luk=
 github.com/nxadm/tail v1.4.8 h1:nPr65rt6Y5JFSKQO7qToXr7pePgD6Gwiw05lkbyAQTE=
+github.com/nxadm/tail v1.4.8/go.mod h1:+ncqLTQzXmGhMZNUePPaPqPvBxHAIsmXswZKocGu+AU=
 github.com/onsi/ginkgo v1.16.5 h1:8xi0RTUf59SOSfEtZMvwTvXYMzG4gV23XVHOZiXNtnE=
+github.com/onsi/ginkgo v1.16.5/go.mod h1:+E8gABHa3K6zRBolWtd+ROzc/U5bkGt0FwiG042wbpU=
 github.com/onsi/gomega v1.18.1 h1:M1GfJqGRrBrrGGsbxzV5dqM2U2ApXefZCQpkukxYRLE=
+github.com/onsi/gomega v1.18.1/go.mod h1:0q+aL8jAiMXy9hbwj2mr5GziHiwhAIQpFmmtT5hitRs=
 github.com/pelletier/go-toml/v2 v2.0.1/go.mod h1:r9LEWfGN8R5k0VXJ+0BkIe7MYkRdwZOjgMj2KwnJFUo=
 github.com/pelletier/go-toml/v2 v2.0.8 h1:0ctb6s9mE31h0/lhu+J6OPmVeDxJn+kYnJc2jZR9tGQ=
 github.com/pelletier/go-toml/v2 v2.0.8/go.mod h1:vuYfssBdrU2XDZ9bYydBu6t+6a6PYNcZljzZR9VXg+4=
 github.com/pkg/diff v0.0.0-20210226163009-20ebb0f2a09e/go.mod h1:pJLUxLENpZxwdsKMEsNbx1VGcRFpLqf3715MtcvvzbA=
 github.com/pkg/errors v0.9.1 h1:FEBLx1zS214owpjy7qsBeixbURkuhQAwrK5UwLGTwt4=
 github.com/pkg/errors v0.9.1/go.mod h1:bwawxfHBFNV+L2hUp1rHADufV3IMtnDRdf1r5NINEl0=
-github.com/pkoukk/tiktoken-go v0.1.6 h1:JF0TlJzhTbrI30wCvFuiw6FzP2+/bR+FIxUdgEAcUsw=
-github.com/pkoukk/tiktoken-go v0.1.6/go.mod h1:9NiV+i9mJKGj1rYOT+njbv+ZwA/zJxYdewGl6qVatpg=
 github.com/pkoukk/tiktoken-go v0.1.7 h1:qOBHXX4PHtvIvmOtyg1EeKlwFRiMKAcoMp4Q+bLQDmw=
 github.com/pkoukk/tiktoken-go v0.1.7/go.mod h1:9NiV+i9mJKGj1rYOT+njbv+ZwA/zJxYdewGl6qVatpg=
 github.com/pmezard/go-difflib v1.0.0 h1:4DBwDE0NGyQoBHbLQYPwSUPoCMWR5BEzIk/f1lZbAQM=
@@ -174,7 +174,8 @@ github.com/stretchr/testify v1.7.1/go.mod h1:6Fq8oRcR53rry900zMqJjRRixrwX3KX962/
 github.com/stretchr/testify v1.8.0/go.mod h1:yNjHg4UonilssWZ8iaSj1OCr/vHnekPRkoO+kdMU+MU=
 github.com/stretchr/testify v1.8.1/go.mod h1:w2LPCIKwWwSfY2zedu0+kehJoqGctiVI29o6fzry7u4=
 github.com/stretchr/testify v1.8.3/go.mod h1:sz/lmYIOXD/1dqDmKjjqLyZ2RngseejIcXlSw2iwfAo=
-github.com/stretchr/testify v1.8.4 h1:CcVxjf3Q8PM0mHUKJCdn+eZZtm5yQwehR5yeSVQQcUk=
+github.com/stretchr/testify v1.9.0 h1:HtqpIVDClZ4nwg75+f6Lvsy/wHu+3BoSGCbBAcpTsTg=
+github.com/stretchr/testify v1.9.0/go.mod h1:r2ic/lqez/lEtzL7wO/rwa5dbSLXVDPFyf8C91i36aY=
 github.com/stripe/stripe-go/v76 v76.21.0 h1:O3GHImHS4oUI3qWMOClHN3zAQF5/oswS/NB7leV1fsU=
 github.com/stripe/stripe-go/v76 v76.21.0/go.mod h1:rw1MxjlAKKcZ+3FOXgTHgwiOa2ya6CPq6ykpJ0Q6Po4=
 github.com/tklauser/go-sysconf v0.3.12 h1:0QaGUFOdQaIVdPgfITYzaTegZvdCjmYO52cSFAEVmqU=
@@ -195,18 +196,19 @@ golang.org/x/arch v0.0.0-20210923205945-b76863e36670/go.mod h1:5om86z9Hs0C8fWVUu
 golang.org/x/arch v0.3.0 h1:02VY4/ZcO/gBOH6PUaoiptASxtXU10jazRCP865E97k=
 golang.org/x/arch v0.3.0/go.mod h1:5om86z9Hs0C8fWVUuoMHwpExlXzs5Tkyp9hOrfG7pp8=
 golang.org/x/crypto v0.0.0-20210711020723-a769d52b0f97/go.mod h1:GvvjBRRGRdwPK5ydBHafDWAxML/pGHZbMvKqRZ5+Abc=
-golang.org/x/crypto v0.21.0 h1:X31++rzVUdKhX5sWmSOFZxx8UW/ldWx55cbf08iNAMA=
+golang.org/x/crypto v0.26.0 h1:RrRspgV4mU+YwB4FYnuBoKsUapNIL5cohGAmSH3azsw=
-golang.org/x/crypto v0.21.0/go.mod h1:0BP7YvVV9gBbVKyeTG0Gyn+gZm94bibOW5BjDEYAOMs=
+golang.org/x/crypto v0.26.0/go.mod h1:GY7jblb9wI+FOo5y8/S2oY4zWP07AkOJ4+jxCqdqn54=
 golang.org/x/exp v0.0.0-20240404231335-c0f41cb1a7a0 h1:985EYyeCOxTpcgOTJpflJUwOeEz0CQOdPt73OzpE9F8=
 golang.org/x/exp v0.0.0-20240404231335-c0f41cb1a7a0/go.mod h1:/lliqkxwWAhPjf5oSOIJup2XcqJaw8RGS6k3TGEc7GI=
 golang.org/x/image v0.15.0 h1:kOELfmgrmJlw4Cdb7g/QGuB3CvDrXbqEIww/pNtNBm8=
 golang.org/x/image v0.15.0/go.mod h1:HUYqC05R2ZcZ3ejNQsIHQDQiwWM4JBqmm6MKANTp4LE=
 golang.org/x/net v0.0.0-20210226172049-e18ecbb05110/go.mod h1:m0MpNAwzfU5UDzcl9v0D8zg8gWTRqZa9RBIspLL5mdg=
 golang.org/x/net v0.0.0-20210520170846-37e1c6afe023/go.mod h1:9nx3DQGgdP8bBQD5qxJ1jj9UTztislL4KSBs9R2vV5Y=
-golang.org/x/net v0.21.0 h1:AQyQV4dYCvJ7vGmJyKki9+PBdyvhkSd8EIx/qb0AYv4=
+golang.org/x/net v0.28.0 h1:a9JDOJc5GMUJ0+UDqmLT86WiEy7iWyIhz8gz8E4e5hE=
-golang.org/x/net v0.21.0/go.mod h1:bIjVDfnllIU7BJ2DNgfnXvpSvtn8VRwhlsaeUTyUS44=
+golang.org/x/net v0.28.0/go.mod h1:yqtgsTWOOnlGLG9GFRrK3++bGOUEkNBoHZc8MEDWPNg=
-golang.org/x/sync v0.7.0 h1:YsImfSBoP9QPYL0xyKJPq0gcaJdG3rInoqxTWbfQu9M=
+golang.org/x/sync v0.0.0-20210220032951-036812b2e83c/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
-golang.org/x/sync v0.7.0/go.mod h1:Czt+wKu1gCyEFDUtn0jG5QVvpJ6rzVqr5aXyt9drQfk=
+golang.org/x/sync v0.8.0 h1:3NFvSEYkUoMifnESzZl15y791HH1qU2xm6eCJU5ZPXQ=
+golang.org/x/sync v0.8.0/go.mod h1:Czt+wKu1gCyEFDUtn0jG5QVvpJ6rzVqr5aXyt9drQfk=
 golang.org/x/sys v0.0.0-20190916202348-b4ddaad3f8a3/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.0.0-20200116001909-b77594299b42/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.0.0-20201119102817-f84b799fce68/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
@@ -214,30 +216,32 @@ golang.org/x/sys v0.0.0-20210423082822-04245dca01da/go.mod h1:h1NjWce9XRLGQEsW7w
 golang.org/x/sys v0.0.0-20210615035016-665e8c7367d1/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.0.0-20210630005230-0f9fa26af87c/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.0.0-20210806184541-e5e7981a1069/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
+golang.org/x/sys v0.0.0-20220110181412-a018aaa089fe/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.0.0-20220704084225-05e143d24a9e/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.6.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.8.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.11.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
-golang.org/x/sys v0.18.0 h1:DBdB3niSjOA/O0blCZBqDefyWNYveAYMNF1Wum0DYQ4=
+golang.org/x/sys v0.24.0 h1:Twjiwq9dn6R1fQcyiK+wQyHWfaz/BJB+YIpzU/Cv3Xg=
-golang.org/x/sys v0.18.0/go.mod h1:/VUhepiaJMQUp4+oa/7Zr1D23ma6VTLIYjOOTFZPUcA=
+golang.org/x/sys v0.24.0/go.mod h1:/VUhepiaJMQUp4+oa/7Zr1D23ma6VTLIYjOOTFZPUcA=
 golang.org/x/term v0.0.0-20201126162022-7de9c90e9dd1/go.mod h1:bj7SfCRtBDWHUb9snDiAeCFNEtKQo2Wmx5Cou7ajbmo=
 golang.org/x/text v0.3.2/go.mod h1:bEr9sfX3Q8Zfm5fL9x+3itogRgK3+ptLWKqgva+5dAk=
 golang.org/x/text v0.3.3/go.mod h1:5Zoc/QRtKVWzQhOtBMvqHzDpF6irO9z98xDceosuGiQ=
 golang.org/x/text v0.3.6/go.mod h1:5Zoc/QRtKVWzQhOtBMvqHzDpF6irO9z98xDceosuGiQ=
-golang.org/x/text v0.14.0 h1:ScX5w1eTa3QqT8oi6+ziP7dTV1S2+ALU0bI+0zXKWiQ=
+golang.org/x/text v0.17.0 h1:XtiM5bkSOt+ewxlOE/aE/AKEHibwj/6gvWMl9Rsh0Qc=
-golang.org/x/text v0.14.0/go.mod h1:18ZOQIKpY8NJVqYksKHtTdi31H5itFRjB5/qKTNYzSU=
+golang.org/x/text v0.17.0/go.mod h1:BuEKDfySbSR4drPmRPG/7iBdf8hvFMuRexcpahXilzY=
 golang.org/x/tools v0.0.0-20180917221912-90fa682c2a6e/go.mod h1:n7NCudcB/nEzxVGmLbDWY5pfWTLqBcC2KZ6jyYvM4mQ=
 golang.org/x/xerrors v0.0.0-20191204190536-9bdfabe68543/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=
 google.golang.org/protobuf v1.26.0-rc.1/go.mod h1:jlhhOSvTdKEhbULTjvd4ARK9grFBp09yW+WbY/TyQbw=
 google.golang.org/protobuf v1.28.0/go.mod h1:HV8QOd/L58Z+nl8r43ehVNZIU/HEI6OcFqwMG9pJV4I=
-google.golang.org/protobuf v1.30.0 h1:kPPoIgf3TsEvrm0PFe15JQ+570QVxYzEvvHqChK+cng=
+google.golang.org/protobuf v1.34.2 h1:6xV6lTsCfpGD21XK49h7MhtcApnLqkfYgPcdHftf6hg=
-google.golang.org/protobuf v1.30.0/go.mod h1:HV8QOd/L58Z+nl8r43ehVNZIU/HEI6OcFqwMG9pJV4I=
+google.golang.org/protobuf v1.34.2/go.mod h1:qYOHts0dSfpeUzUFpOMr/WGzszTmLH+DiWniOlNbLDw=
 gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0=
 gopkg.in/check.v1 v1.0.0-20180628173108-788fd7840127/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0=
 gopkg.in/check.v1 v1.0.0-20201130134442-10cb98267c6c h1:Hei/4ADfdWqJk1ZMxUNpqntNwaWcugrBjAiHlqqRiVk=
 gopkg.in/check.v1 v1.0.0-20201130134442-10cb98267c6c/go.mod h1:JHkPIbrfpd72SG/EVd6muEfDQjcINNoR0C8j2r3qZ4Q=
 gopkg.in/errgo.v2 v2.1.0/go.mod h1:hNsd1EY+bozCKY1Ytp96fpM3vjJbqLJn88ws8XvfDNI=
 gopkg.in/tomb.v1 v1.0.0-20141024135613-dd632973f1e7 h1:uRGJdciOHaEIrze2W8Q3AKkepLTh2hOroT7a+7czfdQ=
+gopkg.in/tomb.v1 v1.0.0-20141024135613-dd632973f1e7/go.mod h1:dt/ZhP58zS4L8KSrWDmTeBkI65Dw0HsyUHuEVlX15mw=
 gopkg.in/yaml.v2 v2.2.2/go.mod h1:hI93XBmqTisBFMUTm0b8Fm+jr3Dg1NNxqwp+5A1VGuI=
 gopkg.in/yaml.v2 v2.2.8/go.mod h1:hI93XBmqTisBFMUTm0b8Fm+jr3Dg1NNxqwp+5A1VGuI=
 gopkg.in/yaml.v2 v2.4.0 h1:D8xgwECY7CYvx+Y2n4sBz93Jn9JRvxdiyyo8CTfuKaY=

main.go

@@ -3,12 +3,14 @@ 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"
+	"one-api/constant"
 	"one-api/controller"
 	"one-api/middleware"
 	"one-api/model"
@@ -40,6 +42,11 @@ func main() {
 	if err != nil {
 		common.FatalLog("failed to initialize database: " + err.Error())
 	}
+	// Initialize SQL Database
+	err = model.InitLogDB()
+	if err != nil {
+		common.FatalLog("failed to initialize database: " + err.Error())
+	}
 	defer func() {
 		err := model.CloseDB()
 		if err != nil {
@@ -53,6 +60,8 @@ func main() {
 		common.FatalLog("failed to initialize Redis: " + err.Error())
 	}
 
+	// Initialize constants
+	constant.InitEnv()
 	// Initialize options
 	model.InitOptionMap()
 	if common.RedisEnabled {
@@ -89,11 +98,11 @@ func main() {
 		}
 		go controller.AutomaticallyTestChannels(frequency)
 	}
-	if common.IsMasterNode {
+	if common.IsMasterNode && constant.UpdateTask {
-		common.SafeGoroutine(func() {
+		gopool.Go(func() {
 			controller.UpdateMidjourneyTaskBulk()
 		})
-		common.SafeGoroutine(func() {
+		gopool.Go(func() {
 			controller.UpdateTaskBulk()
 		})
 	}

middleware/auth.go

@@ -6,6 +6,7 @@ import (
 	"net/http"
 	"one-api/common"
 	"one-api/model"
+	"strconv"
 	"strings"
 )
 
@@ -16,6 +17,7 @@ func authHelper(c *gin.Context, minRole int) {
 	id := session.Get("id")
 	status := session.Get("status")
 	linuxDoEnable := session.Get("linuxdo_enable")
+	useAccessToken := false
 	if username == nil {
 		// Check access token
 		accessToken := c.Request.Header.Get("Authorization")
@@ -35,6 +37,7 @@ func authHelper(c *gin.Context, minRole int) {
 			id = user.Id
 			status = user.Status
 			linuxDoEnable = user.LinuxDoId == "" || user.LinuxDoLevel >= common.LinuxDoMinLevel
+			useAccessToken = true
 		} else {
 			c.JSON(http.StatusOK, gin.H{
 				"success": false,
@@ -44,6 +47,36 @@ func authHelper(c *gin.Context, minRole int) {
 			return
 		}
 	}
+	if !useAccessToken {
+		// get header New-Api-User
+		apiUserIdStr := c.Request.Header.Get("New-Api-User")
+		if apiUserIdStr == "" {
+			c.JSON(http.StatusUnauthorized, gin.H{
+				"success": false,
+				"message": "无权进行此操作，请刷新页面或清空缓存后重试",
+			})
+			c.Abort()
+			return
+		}
+		apiUserId, err := strconv.Atoi(apiUserIdStr)
+		if err != nil {
+			c.JSON(http.StatusUnauthorized, gin.H{
+				"success": false,
+				"message": "无权进行此操作，登录信息无效，请重新登录",
+			})
+			c.Abort()
+			return
+
+		}
+		if id != apiUserId {
+			c.JSON(http.StatusUnauthorized, gin.H{
+				"success": false,
+				"message": "无权进行此操作，与登录用户不匹配，请重新登录",
+			})
+			c.Abort()
+			return
+		}
+	}
 	if status.(int) == common.UserStatusDisabled {
 		c.JSON(http.StatusOK, gin.H{
 			"success": false,
@@ -120,6 +153,12 @@ func TokenAuth() func(c *gin.Context) {
 			key = parts[0]
 		}
 		token, err := model.ValidateUserToken(key)
+		if token != nil {
+			id := c.GetInt("id")
+			if id == 0 {
+				c.Set("id", token.UserId)
+			}
+		}
 		if err != nil {
 			abortWithOpenAiMessage(c, http.StatusUnauthorized, err.Error())
 			return

middleware/distributor.go

@@ -1,6 +1,7 @@
 package middleware
 
 import (
+	"errors"
 	"fmt"
 	"net/http"
 	"one-api/common"
@@ -25,6 +26,10 @@ func Distribute() func(c *gin.Context) {
 		var channel *model.Channel
 		channelId, ok := c.Get("specific_channel_id")
 		modelRequest, shouldSelectChannel, err := getModelRequest(c)
+		if err != nil {
+			abortWithOpenAiMessage(c, http.StatusBadRequest, "Invalid request, "+err.Error())
+			return
+		}
 		userGroup, _ := model.CacheGetUserGroup(userId)
 		c.Set("group", userGroup)
 		if ok {
@@ -141,7 +146,7 @@ func getModelRequest(c *gin.Context) (*ModelRequest, bool, error) {
 	}
 	if err != nil {
 		abortWithOpenAiMessage(c, http.StatusBadRequest, "无效的请求, "+err.Error())
-		return nil, false, err
+		return nil, false, errors.New("无效的请求, " + err.Error())
 	}
 	if strings.HasPrefix(c.Request.URL.Path, "/v1/moderations") {
 		if modelRequest.Model == "" {
@@ -154,18 +159,22 @@ func getModelRequest(c *gin.Context) (*ModelRequest, bool, error) {
 		}
 	}
 	if strings.HasPrefix(c.Request.URL.Path, "/v1/images/generations") {
-		if modelRequest.Model == "" {
+		modelRequest.Model = common.GetStringIfEmpty(modelRequest.Model, "dall-e")
-			modelRequest.Model = "dall-e"
-		}
 	}
 	if strings.HasPrefix(c.Request.URL.Path, "/v1/audio") {
-		if modelRequest.Model == "" {
+		relayMode := relayconstant.RelayModeAudioSpeech
-			if strings.HasPrefix(c.Request.URL.Path, "/v1/audio/speech") {
+		if strings.HasPrefix(c.Request.URL.Path, "/v1/audio/speech") {
-				modelRequest.Model = "tts-1"
+			modelRequest.Model = common.GetStringIfEmpty(modelRequest.Model, "tts-1")
-			} else {
+		} else if strings.HasPrefix(c.Request.URL.Path, "/v1/audio/translations") {
-				modelRequest.Model = "whisper-1"
+			modelRequest.Model = common.GetStringIfEmpty(modelRequest.Model, c.PostForm("model"))
-			}
+			modelRequest.Model = common.GetStringIfEmpty(modelRequest.Model, "whisper-1")
+			relayMode = relayconstant.RelayModeAudioTranslation
+		} else if strings.HasPrefix(c.Request.URL.Path, "/v1/audio/transcriptions") {
+			modelRequest.Model = common.GetStringIfEmpty(modelRequest.Model, c.PostForm("model"))
+			modelRequest.Model = common.GetStringIfEmpty(modelRequest.Model, "whisper-1")
+			relayMode = relayconstant.RelayModeAudioTranscription
 		}
+		c.Set("relay_mode", relayMode)
 	}
 	return &modelRequest, shouldSelectChannel, nil
 }
@@ -175,19 +184,13 @@ func SetupContextForSelectedChannel(c *gin.Context, channel *model.Channel, mode
 	if channel == nil {
 		return
 	}
-	c.Set("channel", channel.Type)
 	c.Set("channel_id", channel.Id)
 	c.Set("channel_name", channel.Name)
 	c.Set("channel_type", channel.Type)
-	ban := true
-	// parse *int to bool
-	if channel.AutoBan != nil && *channel.AutoBan == 0 {
-		ban = false
-	}
 	if nil != channel.OpenAIOrganization && "" != *channel.OpenAIOrganization {
 		c.Set("channel_organization", *channel.OpenAIOrganization)
 	}
-	c.Set("auto_ban", ban)
+	c.Set("auto_ban", channel.GetAutoBan())
 	c.Set("model_mapping", channel.GetModelMapping())
 	c.Set("status_code_mapping", channel.GetStatusCodeMapping())
 	c.Request.Header.Set("Authorization", fmt.Sprintf("Bearer %s", channel.Key))
@@ -196,13 +199,15 @@ func SetupContextForSelectedChannel(c *gin.Context, channel *model.Channel, mode
 	switch channel.Type {
 	case common.ChannelTypeAzure:
 		c.Set("api_version", channel.Other)
+	case common.ChannelTypeVertexAi:
+		c.Set("region", channel.Other)
 	case common.ChannelTypeXunfei:
 		c.Set("api_version", channel.Other)
-	//case common.ChannelTypeAIProxyLibrary:
-	//	c.Set("library_id", channel.Other)
 	case common.ChannelTypeGemini:
 		c.Set("api_version", channel.Other)
 	case common.ChannelTypeAli:
 		c.Set("plugin", channel.Other)
+	case common.ChannelCloudflare:
+		c.Set("api_version", channel.Other)
 	}
 }

middleware/utils.go

@@ -1,11 +1,13 @@
 package middleware
 
 import (
+	"fmt"
 	"github.com/gin-gonic/gin"
 	"one-api/common"
 )
 
 func abortWithOpenAiMessage(c *gin.Context, statusCode int, message string) {
+	userId := c.GetInt("id")
 	c.JSON(statusCode, gin.H{
 		"error": gin.H{
 			"message": common.MessageWithRequestId(message, c.GetString(common.RequestIdKey)),
@@ -13,7 +15,7 @@ func abortWithOpenAiMessage(c *gin.Context, statusCode int, message string) {
 		},
 	})
 	c.Abort()
-	common.LogError(c.Request.Context(), message)
+	common.LogError(c.Request.Context(), fmt.Sprintf("user %d | %s", userId, message))
 }
 
 func abortWithMidjourneyMessage(c *gin.Context, statusCode int, code int, description string) {

model/channel.go

@@ -61,6 +61,13 @@ func (channel *Channel) SetOtherInfo(otherInfo map[string]interface{}) {
 	channel.OtherInfo = string(otherInfoBytes)
 }
 
+func (channel *Channel) GetAutoBan() bool {
+	if channel.AutoBan == nil {
+		return false
+	}
+	return *channel.AutoBan == 1
+}
+
 func (channel *Channel) Save() error {
 	return DB.Save(channel).Error
 }
@@ -99,16 +106,23 @@ func SearchChannels(keyword string, group string, model string) ([]*Channel, err
 	// 构造WHERE子句
 	var whereClause string
 	var args []interface{}
-	if group != "" {
+	if group != "" && group != "null" {
-		whereClause = "(id = ? OR name LIKE ? OR " + keyCol + " = ?) AND " + groupCol + " LIKE ? AND " + modelsCol + " LIKE ?"
+		var groupCondition string
-		args = append(args, common.String2Int(keyword), "%"+keyword+"%", keyword, "%"+group+"%", "%"+model+"%")
+		if common.UsingMySQL {
+			groupCondition = `CONCAT(',', ` + groupCol + `, ',') LIKE ?`
+		} else {
+			// sqlite, PostgreSQL
+			groupCondition = `(',' || ` + groupCol + ` || ',') LIKE ?`
+		}
+		whereClause = "(id = ? OR name LIKE ? OR " + keyCol + " = ?) AND " + modelsCol + ` LIKE ? AND ` + groupCondition
+		args = append(args, common.String2Int(keyword), "%"+keyword+"%", keyword, "%"+model+"%", "%,"+group+",%")
 	} else {
 		whereClause = "(id = ? OR name LIKE ? OR " + keyCol + " = ?) AND " + modelsCol + " LIKE ?"
 		args = append(args, common.String2Int(keyword), "%"+keyword+"%", keyword, "%"+model+"%")
 	}
 
 	// 执行查询
-	err := baseQuery.Where(whereClause, args...).Find(&channels).Error
+	err := baseQuery.Where(whereClause, args...).Order("priority desc").Find(&channels).Error
 	if err != nil {
 		return nil, err
 	}

model/log.go

@@ -3,9 +3,12 @@ package model
 import (
 	"context"
 	"fmt"
-	"gorm.io/gorm"
 	"one-api/common"
 	"strings"
+	"time"
+
+	"github.com/bytedance/gopkg/util/gopool"
+	"gorm.io/gorm"
 )
 
 type Log struct {
@@ -36,7 +39,7 @@ const (
 )
 
 func GetLogByKey(key string) (logs []*Log, err error) {
-	err = DB.Joins("left join tokens on tokens.id = logs.token_id").Where("tokens.key = ?", strings.TrimPrefix(key, "sk-")).Find(&logs).Error
+	err = LOG_DB.Joins("left join tokens on tokens.id = logs.token_id").Where("tokens.key = ?", strings.TrimPrefix(key, "sk-")).Find(&logs).Error
 	return logs, err
 }
 
@@ -52,7 +55,7 @@ func RecordLog(userId int, logType int, content string) {
 		Type:      logType,
 		Content:   content,
 	}
-	err := DB.Create(log).Error
+	err := LOG_DB.Create(log).Error
 	if err != nil {
 		common.SysError("failed to record log: " + err.Error())
 	}
@@ -82,12 +85,12 @@ func RecordConsumeLog(ctx context.Context, userId int, channelId int, promptToke
 		IsStream:         isStream,
 		Other:            otherStr,
 	}
-	err := DB.Create(log).Error
+	err := LOG_DB.Create(log).Error
 	if err != nil {
 		common.LogError(ctx, "failed to record log: "+err.Error())
 	}
 	if common.DataExportEnabled {
-		common.SafeGoroutine(func() {
+		gopool.Go(func() {
 			LogQuotaData(userId, username, modelName, quota, common.GetTimestamp(), promptTokens+completionTokens)
 		})
 	}
@@ -96,12 +99,12 @@ func RecordConsumeLog(ctx context.Context, userId int, channelId int, promptToke
 func GetAllLogs(logType int, startTimestamp int64, endTimestamp int64, modelName string, username string, tokenName string, startIdx int, num int, channel int) (logs []*Log, total int64, err error) {
 	var tx *gorm.DB
 	if logType == LogTypeUnknown {
-		tx = DB
+		tx = LOG_DB
 	} else {
-		tx = DB.Where("type = ?", logType)
+		tx = LOG_DB.Where("type = ?", logType)
 	}
 	if modelName != "" {
-		tx = tx.Where("model_name = ?", modelName)
+		tx = tx.Where("model_name like ?", modelName)
 	}
 	if username != "" {
 		tx = tx.Where("username = ?", username)
@@ -118,25 +121,26 @@ func GetAllLogs(logType int, startTimestamp int64, endTimestamp int64, modelName
 	if channel != 0 {
 		tx = tx.Where("channel_id = ?", channel)
 	}
-
 	err = tx.Model(&Log{}).Count(&total).Error
 	if err != nil {
 		return nil, 0, err
 	}
-
 	err = tx.Order("id desc").Limit(num).Offset(startIdx).Find(&logs).Error
+	if err != nil {
+		return nil, 0, err
+	}
 	return logs, total, err
 }
 
 func GetUserLogs(userId int, logType int, startTimestamp int64, endTimestamp int64, modelName string, tokenName string, startIdx int, num int) (logs []*Log, total int64, err error) {
 	var tx *gorm.DB
 	if logType == LogTypeUnknown {
-		tx = DB.Where("user_id = ?", userId)
+		tx = LOG_DB.Where("user_id = ?", userId)
 	} else {
-		tx = DB.Where("user_id = ? and type = ?", userId, logType)
+		tx = LOG_DB.Where("user_id = ? and type = ?", userId, logType)
 	}
 	if modelName != "" {
-		tx = tx.Where("model_name = ?", modelName)
+		tx = tx.Where("model_name like ?", modelName)
 	}
 	if tokenName != "" {
 		tx = tx.Where("token_name = ?", tokenName)
@@ -147,14 +151,11 @@ func GetUserLogs(userId int, logType int, startTimestamp int64, endTimestamp int
 	if endTimestamp != 0 {
 		tx = tx.Where("created_at <= ?", endTimestamp)
 	}
-
 	err = tx.Model(&Log{}).Count(&total).Error
 	if err != nil {
 		return nil, 0, err
 	}
-
 	err = tx.Order("id desc").Limit(num).Offset(startIdx).Omit("id").Find(&logs).Error
-	return logs, total, err
 	for i := range logs {
 		var otherMap map[string]interface{}
 		otherMap = common.StrToMap(logs[i].Other)
@@ -168,12 +169,12 @@ func GetUserLogs(userId int, logType int, startTimestamp int64, endTimestamp int
 }
 
 func SearchAllLogs(keyword string) (logs []*Log, err error) {
-	err = DB.Where("type = ? or content LIKE ?", keyword, keyword+"%").Order("id desc").Limit(common.MaxRecentItems).Find(&logs).Error
+	err = LOG_DB.Where("type = ? or content LIKE ?", keyword, keyword+"%").Order("id desc").Limit(common.MaxRecentItems).Find(&logs).Error
 	return logs, err
 }
 
 func SearchUserLogs(userId int, keyword string) (logs []*Log, err error) {
-	err = DB.Where("user_id = ? and type = ?", userId, keyword).Order("id desc").Limit(common.MaxRecentItems).Omit("id").Find(&logs).Error
+	err = LOG_DB.Where("user_id = ? and type = ?", userId, keyword).Order("id desc").Limit(common.MaxRecentItems).Omit("id").Find(&logs).Error
 	return logs, err
 }
 
@@ -184,12 +185,18 @@ type Stat struct {
 }
 
 func SumUsedQuota(logType int, startTimestamp int64, endTimestamp int64, modelName string, username string, tokenName string, channel int) (stat Stat) {
-	tx := DB.Table("logs").Select("sum(quota) quota, count(*) rpm, sum(prompt_tokens) + sum(completion_tokens) tpm")
+	tx := LOG_DB.Table("logs").Select("sum(quota) quota")
+
+	// 为rpm和tpm创建单独的查询
+	rpmTpmQuery := LOG_DB.Table("logs").Select("count(*) rpm, sum(prompt_tokens) + sum(completion_tokens) tpm")
+
 	if username != "" {
 		tx = tx.Where("username = ?", username)
+		rpmTpmQuery = rpmTpmQuery.Where("username = ?", username)
 	}
 	if tokenName != "" {
 		tx = tx.Where("token_name = ?", tokenName)
+		rpmTpmQuery = rpmTpmQuery.Where("token_name = ?", tokenName)
 	}
 	if startTimestamp != 0 {
 		tx = tx.Where("created_at >= ?", startTimestamp)
@@ -198,17 +205,29 @@ func SumUsedQuota(logType int, startTimestamp int64, endTimestamp int64, modelNa
 		tx = tx.Where("created_at <= ?", endTimestamp)
 	}
 	if modelName != "" {
-		tx = tx.Where("model_name = ?", modelName)
+		tx = tx.Where("model_name like ?", modelName)
+		rpmTpmQuery = rpmTpmQuery.Where("model_name like ?", modelName)
 	}
 	if channel != 0 {
 		tx = tx.Where("channel_id = ?", channel)
+		rpmTpmQuery = rpmTpmQuery.Where("channel_id = ?", channel)
 	}
-	tx.Where("type = ?", LogTypeConsume).Scan(&stat)
+
+	tx = tx.Where("type = ?", LogTypeConsume)
+	rpmTpmQuery = rpmTpmQuery.Where("type = ?", LogTypeConsume)
+
+	// 只统计最近60秒的rpm和tpm
+	rpmTpmQuery = rpmTpmQuery.Where("created_at >= ?", time.Now().Add(-60*time.Second).Unix())
+
+	// 执行查询
+	tx.Scan(&stat)
+	rpmTpmQuery.Scan(&stat)
+
 	return stat
 }
 
 func SumUsedToken(logType int, startTimestamp int64, endTimestamp int64, modelName string, username string, tokenName string) (token int) {
-	tx := DB.Table("logs").Select("ifnull(sum(prompt_tokens),0) + ifnull(sum(completion_tokens),0)")
+	tx := LOG_DB.Table("logs").Select("ifnull(sum(prompt_tokens),0) + ifnull(sum(completion_tokens),0)")
 	if username != "" {
 		tx = tx.Where("username = ?", username)
 	}
@@ -229,6 +248,6 @@ func SumUsedToken(logType int, startTimestamp int64, endTimestamp int64, modelNa
 }
 
 func DeleteOldLog(targetTimestamp int64) (int64, error) {
-	result := DB.Where("created_at < ?", targetTimestamp).Delete(&Log{})
+	result := LOG_DB.Where("created_at < ?", targetTimestamp).Delete(&Log{})
 	return result.RowsAffected, result.Error
 }

model/main.go

@@ -15,6 +15,8 @@ import (
 
 var DB *gorm.DB
 
+var LOG_DB *gorm.DB
+
 func createRootAccountIfNeed() error {
 	var user User
 	//if user.Status != common.UserStatusEnabled {
@@ -38,9 +40,9 @@ func createRootAccountIfNeed() error {
 	return nil
 }
 
-func chooseDB() (*gorm.DB, error) {
+func chooseDB(envName string) (*gorm.DB, error) {
-	if os.Getenv("SQL_DSN") != "" {
+	dsn := os.Getenv(envName)
-		dsn := os.Getenv("SQL_DSN")
+	if dsn != "" {
 		if strings.HasPrefix(dsn, "postgres://") {
 			// Use PostgreSQL
 			common.SysLog("using PostgreSQL as database")
@@ -52,6 +54,13 @@ func chooseDB() (*gorm.DB, error) {
 				PrepareStmt: true, // precompile SQL
 			})
 		}
+		if strings.HasPrefix(dsn, "local") {
+			common.SysLog("SQL_DSN not set, using SQLite as database")
+			common.UsingSQLite = true
+			return gorm.Open(sqlite.Open(common.SQLitePath), &gorm.Config{
+				PrepareStmt: true, // precompile SQL
+			})
+		}
 		// Use MySQL
 		common.SysLog("using MySQL as database")
 		// check parseTime
@@ -76,7 +85,7 @@ func chooseDB() (*gorm.DB, error) {
 }
 
 func InitDB() (err error) {
-	db, err := chooseDB()
+	db, err := chooseDB("SQL_DSN")
 	if err == nil {
 		if common.DebugEnabled {
 			db = db.Debug()
@@ -100,52 +109,7 @@ func InitDB() (err error) {
 		//	_, _ = sqlDB.Exec("ALTER TABLE midjourneys MODIFY status VARCHAR(20);")   // TODO: delete this line when most users have upgraded
 		//}
 		common.SysLog("database migration started")
-		err = db.AutoMigrate(&Channel{})
+		err = migrateDB()
-		if err != nil {
-			return err
-		}
-		err = db.AutoMigrate(&Token{})
-		if err != nil {
-			return err
-		}
-		err = db.AutoMigrate(&User{})
-		if err != nil {
-			return err
-		}
-		err = db.AutoMigrate(&Option{})
-		if err != nil {
-			return err
-		}
-		err = db.AutoMigrate(&Redemption{})
-		if err != nil {
-			return err
-		}
-		err = db.AutoMigrate(&Ability{})
-		if err != nil {
-			return err
-		}
-		err = db.AutoMigrate(&Log{})
-		if err != nil {
-			return err
-		}
-		err = db.AutoMigrate(&Midjourney{})
-		if err != nil {
-			return err
-		}
-		err = db.AutoMigrate(&TopUp{})
-		if err != nil {
-			return err
-		}
-		err = db.AutoMigrate(&QuotaData{})
-		if err != nil {
-			return err
-		}
-		err = db.AutoMigrate(&Task{})
-		if err != nil {
-			return err
-		}
-		common.SysLog("database migrated")
-		err = createRootAccountIfNeed()
 		return err
 	} else {
 		common.FatalLog(err)
@@ -153,8 +117,103 @@ func InitDB() (err error) {
 	return err
 }
 
-func CloseDB() error {
+func InitLogDB() (err error) {
-	sqlDB, err := DB.DB()
+	if os.Getenv("LOG_SQL_DSN") == "" {
+		LOG_DB = DB
+		return
+	}
+	db, err := chooseDB("LOG_SQL_DSN")
+	if err == nil {
+		if common.DebugEnabled {
+			db = db.Debug()
+		}
+		LOG_DB = db
+		sqlDB, err := LOG_DB.DB()
+		if err != nil {
+			return err
+		}
+		sqlDB.SetMaxIdleConns(common.GetEnvOrDefault("SQL_MAX_IDLE_CONNS", 100))
+		sqlDB.SetMaxOpenConns(common.GetEnvOrDefault("SQL_MAX_OPEN_CONNS", 1000))
+		sqlDB.SetConnMaxLifetime(time.Second * time.Duration(common.GetEnvOrDefault("SQL_MAX_LIFETIME", 60)))
+
+		if !common.IsMasterNode {
+			return nil
+		}
+		//if common.UsingMySQL {
+		//	_, _ = sqlDB.Exec("DROP INDEX idx_channels_key ON channels;")             // TODO: delete this line when most users have upgraded
+		//	_, _ = sqlDB.Exec("ALTER TABLE midjourneys MODIFY action VARCHAR(40);")   // TODO: delete this line when most users have upgraded
+		//	_, _ = sqlDB.Exec("ALTER TABLE midjourneys MODIFY progress VARCHAR(30);") // TODO: delete this line when most users have upgraded
+		//	_, _ = sqlDB.Exec("ALTER TABLE midjourneys MODIFY status VARCHAR(20);")   // TODO: delete this line when most users have upgraded
+		//}
+		common.SysLog("database migration started")
+		err = migrateLOGDB()
+		return err
+	} else {
+		common.FatalLog(err)
+	}
+	return err
+}
+
+func migrateDB() error {
+	err := DB.AutoMigrate(&Channel{})
+	if err != nil {
+		return err
+	}
+	err = DB.AutoMigrate(&Token{})
+	if err != nil {
+		return err
+	}
+	err = DB.AutoMigrate(&User{})
+	if err != nil {
+		return err
+	}
+	err = DB.AutoMigrate(&Option{})
+	if err != nil {
+		return err
+	}
+	err = DB.AutoMigrate(&Redemption{})
+	if err != nil {
+		return err
+	}
+	err = DB.AutoMigrate(&Ability{})
+	if err != nil {
+		return err
+	}
+	err = DB.AutoMigrate(&Log{})
+	if err != nil {
+		return err
+	}
+	err = DB.AutoMigrate(&Midjourney{})
+	if err != nil {
+		return err
+	}
+	err = DB.AutoMigrate(&TopUp{})
+	if err != nil {
+		return err
+	}
+	err = DB.AutoMigrate(&QuotaData{})
+	if err != nil {
+		return err
+	}
+	err = DB.AutoMigrate(&Task{})
+	if err != nil {
+		return err
+	}
+	common.SysLog("database migrated")
+	err = createRootAccountIfNeed()
+	return err
+}
+
+func migrateLOGDB() error {
+	var err error
+	if err = LOG_DB.AutoMigrate(&Log{}); err != nil {
+		return err
+	}
+	return nil
+}
+
+func closeDB(db *gorm.DB) error {
+	sqlDB, err := db.DB()
 	if err != nil {
 		return err
 	}
@@ -162,6 +221,16 @@ func CloseDB() error {
 	return err
 }
 
+func CloseDB() error {
+	if LOG_DB != DB {
+		err := closeDB(LOG_DB)
+		if err != nil {
+			return err
+		}
+	}
+	return closeDB(DB)
+}
+
 var (
 	lastPingTime time.Time
 	pingMutex    sync.Mutex

model/token.go

@@ -50,12 +50,12 @@ func ValidateUserToken(key string) (token *Token, err error) {
 		if token.Status == common.TokenStatusExhausted {
 			keyPrefix := key[:3]
 			keySuffix := key[len(key)-3:]
-			return nil, errors.New("该令牌额度已用尽 TokenStatusExhausted[sk-" + keyPrefix + "***" + keySuffix + "]")
+			return token, errors.New("该令牌额度已用尽 TokenStatusExhausted[sk-" + keyPrefix + "***" + keySuffix + "]")
 		} else if token.Status == common.TokenStatusExpired {
-			return nil, errors.New("该令牌已过期")
+			return token, errors.New("该令牌已过期")
 		}
 		if token.Status != common.TokenStatusEnabled {
-			return nil, errors.New("该令牌状态不可用")
+			return token, errors.New("该令牌状态不可用")
 		}
 		if token.ExpiredTime != -1 && token.ExpiredTime < common.GetTimestamp() {
 			if !common.RedisEnabled {
@@ -65,7 +65,7 @@ func ValidateUserToken(key string) (token *Token, err error) {
 					common.SysError("failed to update token status" + err.Error())
 				}
 			}
-			return nil, errors.New("该令牌已过期")
+			return token, errors.New("该令牌已过期")
 		}
 		if !token.UnlimitedQuota && token.RemainQuota <= 0 {
 			if !common.RedisEnabled {
@@ -78,7 +78,7 @@ func ValidateUserToken(key string) (token *Token, err error) {
 			}
 			keyPrefix := key[:3]
 			keySuffix := key[len(key)-3:]
-			return nil, errors.New(fmt.Sprintf("[sk-%s***%s] 该令牌额度已用尽 !token.UnlimitedQuota && token.RemainQuota = %d", keyPrefix, keySuffix, token.RemainQuota))
+			return token, errors.New(fmt.Sprintf("[sk-%s***%s] 该令牌额度已用尽 !token.UnlimitedQuota && token.RemainQuota = %d", keyPrefix, keySuffix, token.RemainQuota))
 		}
 		return token, nil
 	}

model/utils.go

@@ -2,6 +2,7 @@ package model
 
 import (
 	"errors"
+	"github.com/bytedance/gopkg/util/gopool"
 	"gorm.io/gorm"
 	"one-api/common"
 	"sync"
@@ -28,12 +29,12 @@ func init() {
 }
 
 func InitBatchUpdater() {
-	go func() {
+	gopool.Go(func() {
 		for {
 			time.Sleep(time.Duration(common.BatchUpdateInterval) * time.Second)
 			batchUpdate()
 		}
-	}()
+	})
 }
 
 func addNewRecord(type_ int, id int, value int) {

relay/channel/adapter.go

@@ -10,12 +10,13 @@ import (
 
 type Adaptor interface {
 	// Init IsStream bool
-	Init(info *relaycommon.RelayInfo, request dto.GeneralOpenAIRequest)
+	Init(info *relaycommon.RelayInfo)
-	InitRerank(info *relaycommon.RelayInfo, request dto.RerankRequest)
 	GetRequestURL(info *relaycommon.RelayInfo) (string, error)
 	SetupRequestHeader(c *gin.Context, req *http.Request, info *relaycommon.RelayInfo) error
-	ConvertRequest(c *gin.Context, relayMode int, request *dto.GeneralOpenAIRequest) (any, 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)
 	GetModelList() []string

relay/channel/ali/adaptor.go

@@ -8,6 +8,7 @@ import (
 	"net/http"
 	"one-api/dto"
 	"one-api/relay/channel"
+	"one-api/relay/channel/openai"
 	relaycommon "one-api/relay/common"
 	"one-api/relay/constant"
 )
@@ -15,17 +16,18 @@ import (
 type Adaptor struct {
 }
 
-func (a *Adaptor) InitRerank(info *relaycommon.RelayInfo, request dto.RerankRequest) {
+func (a *Adaptor) Init(info *relaycommon.RelayInfo) {
-}
-
-func (a *Adaptor) Init(info *relaycommon.RelayInfo, request dto.GeneralOpenAIRequest) {
-
 }
 
 func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {
-	fullRequestURL := fmt.Sprintf("%s/api/v1/services/aigc/text-generation/generation", info.BaseUrl)
+	var fullRequestURL string
-	if info.RelayMode == constant.RelayModeEmbeddings {
+	switch info.RelayMode {
+	case constant.RelayModeEmbeddings:
 		fullRequestURL = fmt.Sprintf("%s/api/v1/services/embeddings/text-embedding/text-embedding", info.BaseUrl)
+	case constant.RelayModeImagesGenerations:
+		fullRequestURL = fmt.Sprintf("%s/api/v1/services/aigc/text2image/image-synthesis", info.BaseUrl)
+	default:
+		fullRequestURL = fmt.Sprintf("%s/compatible-mode/v1/chat/completions", info.BaseUrl)
 	}
 	return fullRequestURL, nil
 }
@@ -42,22 +44,32 @@ func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, info *re
 	return nil
 }
 
-func (a *Adaptor) ConvertRequest(c *gin.Context, relayMode int, request *dto.GeneralOpenAIRequest) (any, error) {
+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")
 	}
-	switch relayMode {
+	switch info.RelayMode {
 	case constant.RelayModeEmbeddings:
 		baiduEmbeddingRequest := embeddingRequestOpenAI2Ali(*request)
 		return baiduEmbeddingRequest, nil
 	default:
-		baiduRequest := requestOpenAI2Ali(*request)
+		aliReq := requestOpenAI2Ali(*request)
-		return baiduRequest, nil
+		return aliReq, nil
 	}
 }
 
+func (a *Adaptor) ConvertImageRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.ImageRequest) (any, error) {
+	aliRequest := oaiImage2Ali(request)
+	return aliRequest, nil
+}
+
 func (a *Adaptor) ConvertRerankRequest(c *gin.Context, relayMode int, request dto.RerankRequest) (any, error) {
-	return nil, nil
+	return nil, errors.New("not implemented")
+}
+
+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) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (*http.Response, error) {
@@ -65,14 +77,16 @@ func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, request
 }
 
 func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage *dto.Usage, err *dto.OpenAIErrorWithStatusCode) {
-	if info.IsStream {
+	switch info.RelayMode {
-		err, usage = aliStreamHandler(c, resp)
+	case constant.RelayModeImagesGenerations:
-	} else {
+		err, usage = aliImageHandler(c, resp, info)
-		switch info.RelayMode {
+	case constant.RelayModeEmbeddings:
-		case constant.RelayModeEmbeddings:
+		err, usage = aliEmbeddingHandler(c, resp)
-			err, usage = aliEmbeddingHandler(c, resp)
+	default:
-		default:
+		if info.IsStream {
-			err, usage = aliHandler(c, resp)
+			err, usage = openai.OaiStreamHandler(c, resp, info)
+		} else {
+			err, usage = openai.OpenaiHandler(c, resp, info.PromptTokens, info.UpstreamModelName)
 		}
 	}
 	return

relay/channel/ali/dto.go

@@ -36,8 +36,8 @@ type AliEmbeddingRequest struct {
 }
 
 type AliEmbedding struct {
-	Embedding []float64 `json:"embedding"`
+	Embedding any `json:"embedding"`
-	TextIndex int       `json:"text_index"`
+	TextIndex int `json:"text_index"`
 }
 
 type AliEmbeddingResponse struct {
@@ -60,13 +60,40 @@ type AliUsage struct {
 	TotalTokens  int `json:"total_tokens"`
 }
 
-type AliOutput struct {
+type TaskResult struct {
-	Text         string `json:"text"`
+	B64Image string `json:"b64_image,omitempty"`
-	FinishReason string `json:"finish_reason"`
+	Url      string `json:"url,omitempty"`
+	Code     string `json:"code,omitempty"`
+	Message  string `json:"message,omitempty"`
 }
 
-type AliChatResponse struct {
+type AliOutput struct {
+	TaskId       string       `json:"task_id,omitempty"`
+	TaskStatus   string       `json:"task_status,omitempty"`
+	Text         string       `json:"text"`
+	FinishReason string       `json:"finish_reason"`
+	Message      string       `json:"message,omitempty"`
+	Code         string       `json:"code,omitempty"`
+	Results      []TaskResult `json:"results,omitempty"`
+}
+
+type AliResponse struct {
 	Output AliOutput `json:"output"`
 	Usage  AliUsage  `json:"usage"`
 	AliError
 }
+
+type AliImageRequest struct {
+	Model string `json:"model"`
+	Input struct {
+		Prompt         string `json:"prompt"`
+		NegativePrompt string `json:"negative_prompt,omitempty"`
+	} `json:"input"`
+	Parameters struct {
+		Size  string `json:"size,omitempty"`
+		N     int    `json:"n,omitempty"`
+		Steps string `json:"steps,omitempty"`
+		Scale string `json:"scale,omitempty"`
+	} `json:"parameters,omitempty"`
+	ResponseFormat string `json:"response_format,omitempty"`
+}

relay/channel/ali/image.go

@@ -0,0 +1,177 @@
+package ali
+
+import (
+	"encoding/json"
+	"errors"
+	"fmt"
+	"github.com/gin-gonic/gin"
+	"io"
+	"net/http"
+	"one-api/common"
+	"one-api/dto"
+	relaycommon "one-api/relay/common"
+	"one-api/service"
+	"strings"
+	"time"
+)
+
+func oaiImage2Ali(request dto.ImageRequest) *AliImageRequest {
+	var imageRequest AliImageRequest
+	imageRequest.Input.Prompt = request.Prompt
+	imageRequest.Model = request.Model
+	imageRequest.Parameters.Size = strings.Replace(request.Size, "x", "*", -1)
+	imageRequest.Parameters.N = request.N
+	imageRequest.ResponseFormat = request.ResponseFormat
+
+	return &imageRequest
+}
+
+func updateTask(info *relaycommon.RelayInfo, taskID string, key string) (*AliResponse, error, []byte) {
+	url := fmt.Sprintf("/api/v1/tasks/%s", taskID)
+
+	var aliResponse AliResponse
+
+	req, err := http.NewRequest("GET", url, nil)
+	if err != nil {
+		return &aliResponse, err, nil
+	}
+
+	req.Header.Set("Authorization", "Bearer "+key)
+
+	client := &http.Client{}
+	resp, err := client.Do(req)
+	if err != nil {
+		common.SysError("updateTask client.Do err: " + err.Error())
+		return &aliResponse, err, nil
+	}
+	defer resp.Body.Close()
+
+	responseBody, err := io.ReadAll(resp.Body)
+
+	var response AliResponse
+	err = json.Unmarshal(responseBody, &response)
+	if err != nil {
+		common.SysError("updateTask NewDecoder err: " + err.Error())
+		return &aliResponse, err, nil
+	}
+
+	return &response, nil, responseBody
+}
+
+func asyncTaskWait(info *relaycommon.RelayInfo, taskID string, key string) (*AliResponse, []byte, error) {
+	waitSeconds := 3
+	step := 0
+	maxStep := 20
+
+	var taskResponse AliResponse
+	var responseBody []byte
+
+	for {
+		step++
+		rsp, err, body := updateTask(info, taskID, key)
+		responseBody = body
+		if err != nil {
+			return &taskResponse, responseBody, err
+		}
+
+		if rsp.Output.TaskStatus == "" {
+			return &taskResponse, responseBody, nil
+		}
+
+		switch rsp.Output.TaskStatus {
+		case "FAILED":
+			fallthrough
+		case "CANCELED":
+			fallthrough
+		case "SUCCEEDED":
+			fallthrough
+		case "UNKNOWN":
+			return rsp, responseBody, nil
+		}
+		if step >= maxStep {
+			break
+		}
+		time.Sleep(time.Duration(waitSeconds) * time.Second)
+	}
+
+	return nil, nil, fmt.Errorf("aliAsyncTaskWait timeout")
+}
+
+func responseAli2OpenAIImage(c *gin.Context, response *AliResponse, info *relaycommon.RelayInfo, responseFormat string) *dto.ImageResponse {
+	imageResponse := dto.ImageResponse{
+		Created: info.StartTime.Unix(),
+	}
+
+	for _, data := range response.Output.Results {
+		var b64Json string
+		if responseFormat == "b64_json" {
+			_, b64, err := common.GetImageFromUrl(data.Url)
+			if err != nil {
+				common.LogError(c, "get_image_data_failed: "+err.Error())
+				continue
+			}
+			b64Json = b64
+		} else {
+			b64Json = data.B64Image
+		}
+
+		imageResponse.Data = append(imageResponse.Data, dto.ImageData{
+			Url:           data.Url,
+			B64Json:       b64Json,
+			RevisedPrompt: "",
+		})
+	}
+	return &imageResponse
+}
+
+func aliImageHandler(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (*dto.OpenAIErrorWithStatusCode, *dto.Usage) {
+	apiKey := c.Request.Header.Get("Authorization")
+	apiKey = strings.TrimPrefix(apiKey, "Bearer ")
+	responseFormat := c.GetString("response_format")
+
+	var aliTaskResponse AliResponse
+	responseBody, err := io.ReadAll(resp.Body)
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "read_response_body_failed", http.StatusInternalServerError), nil
+	}
+	err = resp.Body.Close()
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), nil
+	}
+	err = json.Unmarshal(responseBody, &aliTaskResponse)
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "unmarshal_response_body_failed", http.StatusInternalServerError), nil
+	}
+
+	if aliTaskResponse.Message != "" {
+		common.LogError(c, "ali_async_task_failed: "+aliTaskResponse.Message)
+		return service.OpenAIErrorWrapper(errors.New(aliTaskResponse.Message), "ali_async_task_failed", http.StatusInternalServerError), nil
+	}
+
+	aliResponse, _, err := asyncTaskWait(info, aliTaskResponse.Output.TaskId, apiKey)
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "ali_async_task_wait_failed", http.StatusInternalServerError), nil
+	}
+
+	if aliResponse.Output.TaskStatus != "SUCCEEDED" {
+		return &dto.OpenAIErrorWithStatusCode{
+			Error: dto.OpenAIError{
+				Message: aliResponse.Output.Message,
+				Type:    "ali_error",
+				Param:   "",
+				Code:    aliResponse.Output.Code,
+			},
+			StatusCode: resp.StatusCode,
+		}, nil
+	}
+
+	fullTextResponse := responseAli2OpenAIImage(c, aliResponse, info, responseFormat)
+	jsonResponse, err := json.Marshal(fullTextResponse)
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "marshal_response_body_failed", http.StatusInternalServerError), nil
+	}
+	c.Writer.Header().Set("Content-Type", "application/json")
+	c.Writer.WriteHeader(resp.StatusCode)
+	_, err = c.Writer.Write(jsonResponse)
+	return nil, nil
+}

relay/channel/ali/text.go

@@ -16,34 +16,13 @@ import (
 
 const EnableSearchModelSuffix = "-internet"
 
-func requestOpenAI2Ali(request dto.GeneralOpenAIRequest) *AliChatRequest {
+func requestOpenAI2Ali(request dto.GeneralOpenAIRequest) *dto.GeneralOpenAIRequest {
-	messages := make([]AliMessage, 0, len(request.Messages))
+	if request.TopP >= 1 {
-	//prompt := ""
+		request.TopP = 0.999
-	for i := 0; i < len(request.Messages); i++ {
+	} else if request.TopP <= 0 {
-		message := request.Messages[i]
+		request.TopP = 0.001
-		messages = append(messages, AliMessage{
-			Content: message.StringContent(),
-			Role:    strings.ToLower(message.Role),
-		})
-	}
-	enableSearch := false
-	aliModel := request.Model
-	if strings.HasSuffix(aliModel, EnableSearchModelSuffix) {
-		enableSearch = true
-		aliModel = strings.TrimSuffix(aliModel, EnableSearchModelSuffix)
-	}
-	return &AliChatRequest{
-		Model: request.Model,
-		Input: AliInput{
-			//Prompt:  prompt,
-			Messages: messages,
-		},
-		Parameters: AliParameters{
-			IncrementalOutput: request.Stream,
-			Seed:              uint64(request.Seed),
-			EnableSearch:      enableSearch,
-		},
 	}
+	return &request
 }
 
 func embeddingRequestOpenAI2Ali(request dto.GeneralOpenAIRequest) *AliEmbeddingRequest {
@@ -110,7 +89,7 @@ func embeddingResponseAli2OpenAI(response *AliEmbeddingResponse) *dto.OpenAIEmbe
 	return &openAIEmbeddingResponse
 }
 
-func responseAli2OpenAI(response *AliChatResponse) *dto.OpenAITextResponse {
+func responseAli2OpenAI(response *AliResponse) *dto.OpenAITextResponse {
 	content, _ := json.Marshal(response.Output.Text)
 	choice := dto.OpenAITextResponseChoice{
 		Index: 0,
@@ -134,7 +113,7 @@ func responseAli2OpenAI(response *AliChatResponse) *dto.OpenAITextResponse {
 	return &fullTextResponse
 }
 
-func streamResponseAli2OpenAI(aliResponse *AliChatResponse) *dto.ChatCompletionsStreamResponse {
+func streamResponseAli2OpenAI(aliResponse *AliResponse) *dto.ChatCompletionsStreamResponse {
 	var choice dto.ChatCompletionsStreamResponseChoice
 	choice.Delta.SetContentString(aliResponse.Output.Text)
 	if aliResponse.Output.FinishReason != "null" {
@@ -154,18 +133,7 @@ func streamResponseAli2OpenAI(aliResponse *AliChatResponse) *dto.ChatCompletions
 func aliStreamHandler(c *gin.Context, resp *http.Response) (*dto.OpenAIErrorWithStatusCode, *dto.Usage) {
 	var usage dto.Usage
 	scanner := bufio.NewScanner(resp.Body)
-	scanner.Split(func(data []byte, atEOF bool) (advance int, token []byte, err error) {
+	scanner.Split(bufio.ScanLines)
-		if atEOF && len(data) == 0 {
-			return 0, nil, nil
-		}
-		if i := strings.Index(string(data), "\n"); i >= 0 {
-			return i + 1, data[0:i], nil
-		}
-		if atEOF {
-			return len(data), data, nil
-		}
-		return 0, nil, nil
-	})
 	dataChan := make(chan string)
 	stopChan := make(chan bool)
 	go func() {
@@ -187,7 +155,7 @@ func aliStreamHandler(c *gin.Context, resp *http.Response) (*dto.OpenAIErrorWith
 	c.Stream(func(w io.Writer) bool {
 		select {
 		case data := <-dataChan:
-			var aliResponse AliChatResponse
+			var aliResponse AliResponse
 			err := json.Unmarshal([]byte(data), &aliResponse)
 			if err != nil {
 				common.SysError("error unmarshalling stream response: " + err.Error())
@@ -221,7 +189,7 @@ func aliStreamHandler(c *gin.Context, resp *http.Response) (*dto.OpenAIErrorWith
 }
 
 func aliHandler(c *gin.Context, resp *http.Response) (*dto.OpenAIErrorWithStatusCode, *dto.Usage) {
-	var aliResponse AliChatResponse
+	var aliResponse AliResponse
 	responseBody, err := io.ReadAll(resp.Body)
 	if err != nil {
 		return service.OpenAIErrorWrapper(err, "read_response_body_failed", http.StatusInternalServerError), nil

relay/channel/api_request.go

@@ -7,14 +7,19 @@ import (
 	"io"
 	"net/http"
 	"one-api/relay/common"
+	"one-api/relay/constant"
 	"one-api/service"
 )
 
 func SetupApiRequestHeader(info *common.RelayInfo, c *gin.Context, req *http.Request) {
-	req.Header.Set("Content-Type", c.Request.Header.Get("Content-Type"))
+	if info.RelayMode == constant.RelayModeAudioTranscription || info.RelayMode == constant.RelayModeAudioTranslation {
-	req.Header.Set("Accept", c.Request.Header.Get("Accept"))
+		// multipart/form-data
-	if info.IsStream && c.Request.Header.Get("Accept") == "" {
+	} else {
-		req.Header.Set("Accept", "text/event-stream")
+		req.Header.Set("Content-Type", c.Request.Header.Get("Content-Type"))
+		req.Header.Set("Accept", c.Request.Header.Get("Accept"))
+		if info.IsStream && c.Request.Header.Get("Accept") == "" {
+			req.Header.Set("Accept", "text/event-stream")
+		}
 	}
 }
 
@@ -38,6 +43,29 @@ func DoApiRequest(a Adaptor, c *gin.Context, info *common.RelayInfo, requestBody
 	return resp, nil
 }
 
+func DoFormRequest(a Adaptor, c *gin.Context, info *common.RelayInfo, requestBody io.Reader) (*http.Response, error) {
+	fullRequestURL, err := a.GetRequestURL(info)
+	if err != nil {
+		return nil, fmt.Errorf("get request url failed: %w", err)
+	}
+	req, err := http.NewRequest(c.Request.Method, fullRequestURL, requestBody)
+	if err != nil {
+		return nil, fmt.Errorf("new request failed: %w", err)
+	}
+	// set form data
+	req.Header.Set("Content-Type", c.Request.Header.Get("Content-Type"))
+
+	err = a.SetupRequestHeader(c, req, info)
+	if err != nil {
+		return nil, fmt.Errorf("setup request header failed: %w", err)
+	}
+	resp, err := doRequest(c, req)
+	if err != nil {
+		return nil, fmt.Errorf("do request failed: %w", err)
+	}
+	return resp, 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

@@ -20,12 +20,17 @@ type Adaptor struct {
 	RequestMode int
 }
 
-func (a *Adaptor) InitRerank(info *relaycommon.RelayInfo, request dto.RerankRequest) {
+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) Init(info *relaycommon.RelayInfo, request dto.GeneralOpenAIRequest) {
+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) {
 	if strings.HasPrefix(info.UpstreamModelName, "claude-3") {
 		a.RequestMode = RequestModeMessage
 	} else {
@@ -41,7 +46,7 @@ func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, info *re
 	return nil
 }
 
-func (a *Adaptor) ConvertRequest(c *gin.Context, relayMode int, request *dto.GeneralOpenAIRequest) (any, error) {
+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")
 	}

relay/channel/aws/relay-aws.go

@@ -222,9 +222,11 @@ func awsStreamHandler(c *gin.Context, resp *http.Response, info *relaycommon.Rel
 		}
 	}
 	service.Done(c)
-	err = resp.Body.Close()
+	if resp != nil {
-	if err != nil {
+		err = resp.Body.Close()
-		return service.OpenAIErrorWrapperLocal(err, "close_response_body_failed", http.StatusInternalServerError), nil
+		if err != nil {
+			return service.OpenAIErrorWrapperLocal(err, "close_response_body_failed", http.StatusInternalServerError), nil
+		}
 	}
 	return nil, &usage
 }

relay/channel/baidu/adaptor.go

@@ -16,12 +16,17 @@ import (
 type Adaptor struct {
 }
 
-func (a *Adaptor) InitRerank(info *relaycommon.RelayInfo, request dto.RerankRequest) {
+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) Init(info *relaycommon.RelayInfo, request dto.GeneralOpenAIRequest) {
+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) {
 
 }
 
@@ -99,11 +104,11 @@ func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, info *re
 	return nil
 }
 
-func (a *Adaptor) ConvertRequest(c *gin.Context, relayMode int, request *dto.GeneralOpenAIRequest) (any, error) {
+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")
 	}
-	switch relayMode {
+	switch info.RelayMode {
 	case constant.RelayModeEmbeddings:
 		baiduEmbeddingRequest := embeddingRequestOpenAI2Baidu(*request)
 		return baiduEmbeddingRequest, nil

relay/channel/baidu/dto.go

@@ -50,9 +50,9 @@ type BaiduEmbeddingRequest struct {
 }
 
 type BaiduEmbeddingData struct {
-	Object    string    `json:"object"`
+	Object    string `json:"object"`
-	Embedding []float64 `json:"embedding"`
+	Embedding any    `json:"embedding"`
-	Index     int       `json:"index"`
+	Index     int    `json:"index"`
 }
 
 type BaiduEmbeddingResponse struct {

relay/channel/claude/adaptor.go

@@ -21,12 +21,17 @@ type Adaptor struct {
 	RequestMode int
 }
 
-func (a *Adaptor) InitRerank(info *relaycommon.RelayInfo, request dto.RerankRequest) {
+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) Init(info *relaycommon.RelayInfo, request dto.GeneralOpenAIRequest) {
+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) {
 	if strings.HasPrefix(info.UpstreamModelName, "claude-3") {
 		a.RequestMode = RequestModeMessage
 	} else {
@@ -53,7 +58,7 @@ func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, info *re
 	return nil
 }
 
-func (a *Adaptor) ConvertRequest(c *gin.Context, relayMode int, request *dto.GeneralOpenAIRequest) (any, error) {
+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")
 	}
@@ -74,9 +79,9 @@ func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, request
 
 func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage *dto.Usage, err *dto.OpenAIErrorWithStatusCode) {
 	if info.IsStream {
-		err, usage = claudeStreamHandler(c, resp, info, a.RequestMode)
+		err, usage = ClaudeStreamHandler(c, resp, info, a.RequestMode)
 	} else {
-		err, usage = claudeHandler(a.RequestMode, c, resp, info.PromptTokens, info.UpstreamModelName)
+		err, usage = ClaudeHandler(c, resp, a.RequestMode, info)
 	}
 	return
 }

relay/channel/claude/dto.go

@@ -5,11 +5,18 @@ type ClaudeMetadata struct {
 }
 
 type ClaudeMediaMessage struct {
-	Type       string               `json:"type"`
+	Type        string               `json:"type"`
-	Text       string               `json:"text,omitempty"`
+	Text        string               `json:"text,omitempty"`
-	Source     *ClaudeMessageSource `json:"source,omitempty"`
+	Source      *ClaudeMessageSource `json:"source,omitempty"`
-	Usage      *ClaudeUsage         `json:"usage,omitempty"`
+	Usage       *ClaudeUsage         `json:"usage,omitempty"`
-	StopReason *string              `json:"stop_reason,omitempty"`
+	StopReason  *string              `json:"stop_reason,omitempty"`
+	PartialJson string               `json:"partial_json,omitempty"`
+	// tool_calls
+	Id        string `json:"id,omitempty"`
+	Name      string `json:"name,omitempty"`
+	Input     any    `json:"input,omitempty"`
+	Content   string `json:"content,omitempty"`
+	ToolUseId string `json:"tool_use_id,omitempty"`
 }
 
 type ClaudeMessageSource struct {
@@ -23,6 +30,18 @@ type ClaudeMessage struct {
 	Content any    `json:"content"`
 }
 
+type Tool struct {
+	Name        string                 `json:"name"`
+	Description string                 `json:"description,omitempty"`
+	InputSchema map[string]interface{} `json:"input_schema"`
+}
+
+type InputSchema struct {
+	Type       string `json:"type"`
+	Properties any    `json:"properties,omitempty"`
+	Required   any    `json:"required,omitempty"`
+}
+
 type ClaudeRequest struct {
 	Model             string          `json:"model"`
 	Prompt            string          `json:"prompt,omitempty"`
@@ -35,7 +54,9 @@ type ClaudeRequest struct {
 	TopP              float64         `json:"top_p,omitempty"`
 	TopK              int             `json:"top_k,omitempty"`
 	//ClaudeMetadata    `json:"metadata,omitempty"`
-	Stream bool `json:"stream,omitempty"`
+	Stream     bool   `json:"stream,omitempty"`
+	Tools      []Tool `json:"tools,omitempty"`
+	ToolChoice any    `json:"tool_choice,omitempty"`
 }
 
 type ClaudeError struct {
@@ -44,24 +65,20 @@ type ClaudeError struct {
 }
 
 type ClaudeResponse struct {
-	Id         string               `json:"id"`
+	Id           string               `json:"id"`
-	Type       string               `json:"type"`
+	Type         string               `json:"type"`
-	Content    []ClaudeMediaMessage `json:"content"`
+	Content      []ClaudeMediaMessage `json:"content"`
-	Completion string               `json:"completion"`
+	Completion   string               `json:"completion"`
-	StopReason string               `json:"stop_reason"`
+	StopReason   string               `json:"stop_reason"`
-	Model      string               `json:"model"`
+	Model        string               `json:"model"`
-	Error      ClaudeError          `json:"error"`
+	Error        ClaudeError          `json:"error"`
-	Usage      ClaudeUsage          `json:"usage"`
+	Usage        ClaudeUsage          `json:"usage"`
-	Index      int                  `json:"index"`   // stream only
+	Index        int                  `json:"index"` // stream only
-	Delta      *ClaudeMediaMessage  `json:"delta"`   // stream only
+	ContentBlock *ClaudeMediaMessage  `json:"content_block"`
-	Message    *ClaudeResponse      `json:"message"` // stream only: message_start
+	Delta        *ClaudeMediaMessage  `json:"delta"`   // stream only
+	Message      *ClaudeResponse      `json:"message"` // stream only: message_start
 }
 
-//type ClaudeResponseChoice struct {
-//	Index   int                `json:"index"`
-//	Type    string             `json:"type"`
-//}
-
 type ClaudeUsage struct {
 	InputTokens  int `json:"input_tokens"`
 	OutputTokens int `json:"output_tokens"`

relay/channel/claude/relay-claude.go

@@ -8,12 +8,10 @@ import (
 	"io"
 	"net/http"
 	"one-api/common"
-	"one-api/constant"
 	"one-api/dto"
 	relaycommon "one-api/relay/common"
 	"one-api/service"
 	"strings"
-	"time"
 )
 
 func stopReasonClaude2OpenAI(reason string) string {
@@ -30,6 +28,7 @@ func stopReasonClaude2OpenAI(reason string) string {
 }
 
 func RequestOpenAI2ClaudeComplete(textRequest dto.GeneralOpenAIRequest) *ClaudeRequest {
+
 	claudeRequest := ClaudeRequest{
 		Model:         textRequest.Model,
 		Prompt:        "",
@@ -60,6 +59,28 @@ func RequestOpenAI2ClaudeComplete(textRequest dto.GeneralOpenAIRequest) *ClaudeR
 }
 
 func RequestOpenAI2ClaudeMessage(textRequest dto.GeneralOpenAIRequest) (*ClaudeRequest, error) {
+	claudeTools := make([]Tool, 0, len(textRequest.Tools))
+
+	for _, tool := range textRequest.Tools {
+		if params, ok := tool.Function.Parameters.(map[string]any); ok {
+			claudeTool := Tool{
+				Name:        tool.Function.Name,
+				Description: tool.Function.Description,
+			}
+			claudeTool.InputSchema = make(map[string]interface{})
+			claudeTool.InputSchema["type"] = params["type"].(string)
+			claudeTool.InputSchema["properties"] = params["properties"]
+			claudeTool.InputSchema["required"] = params["required"]
+			for s, a := range params {
+				if s == "type" || s == "properties" || s == "required" {
+					continue
+				}
+				claudeTool.InputSchema[s] = a
+			}
+			claudeTools = append(claudeTools, claudeTool)
+		}
+	}
+
 	claudeRequest := ClaudeRequest{
 		Model:         textRequest.Model,
 		MaxTokens:     textRequest.MaxTokens,
@@ -68,10 +89,24 @@ func RequestOpenAI2ClaudeMessage(textRequest dto.GeneralOpenAIRequest) (*ClaudeR
 		TopP:          textRequest.TopP,
 		TopK:          textRequest.TopK,
 		Stream:        textRequest.Stream,
+		Tools:         claudeTools,
 	}
 	if claudeRequest.MaxTokens == 0 {
 		claudeRequest.MaxTokens = 4096
 	}
+	if textRequest.Stop != nil {
+		// stop maybe string/array string, convert to array string
+		switch textRequest.Stop.(type) {
+		case string:
+			claudeRequest.StopSequences = []string{textRequest.Stop.(string)}
+		case []interface{}:
+			stopSequences := make([]string, 0)
+			for _, stop := range textRequest.Stop.([]interface{}) {
+				stopSequences = append(stopSequences, stop.(string))
+			}
+			claudeRequest.StopSequences = stopSequences
+		}
+	}
 	formatMessages := make([]dto.Message, 0)
 	var lastMessage *dto.Message
 	for i, message := range textRequest.Messages {
@@ -104,6 +139,7 @@ func RequestOpenAI2ClaudeMessage(textRequest dto.GeneralOpenAIRequest) (*ClaudeR
 	}
 
 	claudeMessages := make([]ClaudeMessage, 0)
+	isFirstMessage := true
 	for _, message := range formatMessages {
 		if message.Role == "system" {
 			if message.IsStringContent() {
@@ -119,6 +155,22 @@ func RequestOpenAI2ClaudeMessage(textRequest dto.GeneralOpenAIRequest) (*ClaudeR
 				claudeRequest.System = content
 			}
 		} else {
+			if isFirstMessage {
+				isFirstMessage = false
+				if message.Role != "user" {
+					// fix: first message is assistant, add user message
+					claudeMessage := ClaudeMessage{
+						Role: "user",
+						Content: []ClaudeMediaMessage{
+							{
+								Type: "text",
+								Text: "...",
+							},
+						},
+					}
+					claudeMessages = append(claudeMessages, claudeMessage)
+				}
+			}
 			claudeMessage := ClaudeMessage{
 				Role: message.Role,
 			}
@@ -171,6 +223,7 @@ func StreamResponseClaude2OpenAI(reqMode int, claudeResponse *ClaudeResponse) (*
 	response.Object = "chat.completion.chunk"
 	response.Model = claudeResponse.Model
 	response.Choices = make([]dto.ChatCompletionsStreamResponseChoice, 0)
+	tools := make([]dto.ToolCall, 0)
 	var choice dto.ChatCompletionsStreamResponseChoice
 	if reqMode == RequestModeCompletion {
 		choice.Delta.SetContentString(claudeResponse.Completion)
@@ -186,10 +239,33 @@ func StreamResponseClaude2OpenAI(reqMode int, claudeResponse *ClaudeResponse) (*
 			choice.Delta.SetContentString("")
 			choice.Delta.Role = "assistant"
 		} else if claudeResponse.Type == "content_block_start" {
-			return nil, nil
+			if claudeResponse.ContentBlock != nil {
+				//choice.Delta.SetContentString(claudeResponse.ContentBlock.Text)
+				if claudeResponse.ContentBlock.Type == "tool_use" {
+					tools = append(tools, dto.ToolCall{
+						ID:   claudeResponse.ContentBlock.Id,
+						Type: "function",
+						Function: dto.FunctionCall{
+							Name:      claudeResponse.ContentBlock.Name,
+							Arguments: "",
+						},
+					})
+				}
+			} else {
+				return nil, nil
+			}
 		} else if claudeResponse.Type == "content_block_delta" {
-			choice.Index = claudeResponse.Index
+			if claudeResponse.Delta != nil {
-			choice.Delta.SetContentString(claudeResponse.Delta.Text)
+				choice.Index = claudeResponse.Index
+				choice.Delta.SetContentString(claudeResponse.Delta.Text)
+				if claudeResponse.Delta.Type == "input_json_delta" {
+					tools = append(tools, dto.ToolCall{
+						Function: dto.FunctionCall{
+							Arguments: claudeResponse.Delta.PartialJson,
+						},
+					})
+				}
+			}
 		} else if claudeResponse.Type == "message_delta" {
 			finishReason := stopReasonClaude2OpenAI(*claudeResponse.Delta.StopReason)
 			if finishReason != "null" {
@@ -205,6 +281,10 @@ func StreamResponseClaude2OpenAI(reqMode int, claudeResponse *ClaudeResponse) (*
 	if claudeUsage == nil {
 		claudeUsage = &ClaudeUsage{}
 	}
+	if len(tools) > 0 {
+		choice.Delta.Content = nil // compatible with other OpenAI derivative applications, like LobeOpenAICompatibleFactory ...
+		choice.Delta.ToolCalls = tools
+	}
 	response.Choices = append(response.Choices, choice)
 
 	return &response, claudeUsage
@@ -217,6 +297,11 @@ func ResponseClaude2OpenAI(reqMode int, claudeResponse *ClaudeResponse) *dto.Ope
 		Object:  "chat.completion",
 		Created: common.GetTimestamp(),
 	}
+	var responseText string
+	if len(claudeResponse.Content) > 0 {
+		responseText = claudeResponse.Content[0].Text
+	}
+	tools := make([]dto.ToolCall, 0)
 	if reqMode == RequestModeCompletion {
 		content, _ := json.Marshal(strings.TrimPrefix(claudeResponse.Completion, " "))
 		choice := dto.OpenAITextResponseChoice{
@@ -231,114 +316,96 @@ func ResponseClaude2OpenAI(reqMode int, claudeResponse *ClaudeResponse) *dto.Ope
 		choices = append(choices, choice)
 	} else {
 		fullTextResponse.Id = claudeResponse.Id
-		for i, message := range claudeResponse.Content {
+		for _, message := range claudeResponse.Content {
-			content, _ := json.Marshal(message.Text)
+			if message.Type == "tool_use" {
-			choice := dto.OpenAITextResponseChoice{
+				args, _ := json.Marshal(message.Input)
-				Index: i,
+				tools = append(tools, dto.ToolCall{
-				Message: dto.Message{
+					ID:   message.Id,
-					Role:    "assistant",
+					Type: "function", // compatible with other OpenAI derivative applications
-					Content: content,
+					Function: dto.FunctionCall{
-				},
+						Name:      message.Name,
-				FinishReason: stopReasonClaude2OpenAI(claudeResponse.StopReason),
+						Arguments: string(args),
+					},
+				})
 			}
-			choices = append(choices, choice)
 		}
 	}
-
+	choice := dto.OpenAITextResponseChoice{
+		Index: 0,
+		Message: dto.Message{
+			Role: "assistant",
+		},
+		FinishReason: stopReasonClaude2OpenAI(claudeResponse.StopReason),
+	}
+	choice.SetStringContent(responseText)
+	if len(tools) > 0 {
+		choice.Message.ToolCalls = tools
+	}
+	choices = append(choices, choice)
 	fullTextResponse.Choices = choices
 	return &fullTextResponse
 }
 
-func claudeStreamHandler(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo, requestMode int) (*dto.OpenAIErrorWithStatusCode, *dto.Usage) {
+func ClaudeStreamHandler(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo, requestMode int) (*dto.OpenAIErrorWithStatusCode, *dto.Usage) {
 	responseId := fmt.Sprintf("chatcmpl-%s", common.GetUUID())
 	var usage *dto.Usage
 	usage = &dto.Usage{}
 	responseText := ""
 	createdTime := common.GetTimestamp()
 	scanner := bufio.NewScanner(resp.Body)
-	scanner.Split(func(data []byte, atEOF bool) (advance int, token []byte, err error) {
+	scanner.Split(bufio.ScanLines)
-		if atEOF && len(data) == 0 {
+	service.SetEventStreamHeaders(c)
-			return 0, nil, nil
+
+	for scanner.Scan() {
+		data := scanner.Text()
+		info.SetFirstResponseTime()
+		if len(data) < 6 || !strings.HasPrefix(data, "data:") {
+			continue
 		}
-		if i := strings.Index(string(data), "\n"); i >= 0 {
+		data = strings.TrimPrefix(data, "data:")
-			return i + 1, data[0:i], nil
+		data = strings.TrimSpace(data)
+		var claudeResponse ClaudeResponse
+		err := json.Unmarshal([]byte(data), &claudeResponse)
+		if err != nil {
+			common.SysError("error unmarshalling stream response: " + err.Error())
+			continue
 		}
-		if atEOF {
+
-			return len(data), data, nil
+		response, claudeUsage := StreamResponseClaude2OpenAI(requestMode, &claudeResponse)
+		if response == nil {
+			continue
 		}
-		return 0, nil, nil
+		if requestMode == RequestModeCompletion {
-	})
+			responseText += claudeResponse.Completion
-	dataChan := make(chan string, 5)
+			responseId = response.Id
-	stopChan := make(chan bool, 2)
+		} else {
-	go func() {
+			if claudeResponse.Type == "message_start" {
-		for scanner.Scan() {
+				// message_start, 获取usage
-			data := scanner.Text()
+				responseId = claudeResponse.Message.Id
-			if !strings.HasPrefix(data, "data: ") {
+				info.UpstreamModelName = claudeResponse.Message.Model
+				usage.PromptTokens = claudeUsage.InputTokens
+			} else if claudeResponse.Type == "content_block_delta" {
+				responseText += claudeResponse.Delta.Text
+			} else if claudeResponse.Type == "message_delta" {
+				usage.CompletionTokens = claudeUsage.OutputTokens
+				usage.TotalTokens = claudeUsage.InputTokens + claudeUsage.OutputTokens
+			} else if claudeResponse.Type == "content_block_start" {
+
+			} else {
 				continue
 			}
-			data = strings.TrimPrefix(data, "data: ")
-			if !common.SafeSendStringTimeout(dataChan, data, constant.StreamingTimeout) {
-				// send data timeout, stop the stream
-				common.LogError(c, "send data timeout, stop the stream")
-				break
-			}
 		}
-		stopChan <- true
+		//response.Id = responseId
-	}()
+		response.Id = responseId
-	isFirst := true
+		response.Created = createdTime
-	service.SetEventStreamHeaders(c)
+		response.Model = info.UpstreamModelName
-	c.Stream(func(w io.Writer) bool {
-		select {
-		case data := <-dataChan:
-			if isFirst {
-				isFirst = false
-				info.FirstResponseTime = time.Now()
-			}
-			// some implementations may add \r at the end of data
-			data = strings.TrimSuffix(data, "\r")
-			var claudeResponse ClaudeResponse
-			err := json.Unmarshal([]byte(data), &claudeResponse)
-			if err != nil {
-				common.SysError("error unmarshalling stream response: " + err.Error())
-				return true
-			}
 
-			response, claudeUsage := StreamResponseClaude2OpenAI(requestMode, &claudeResponse)
+		err = service.ObjectData(c, response)
-			if response == nil {
+		if err != nil {
-				return true
+			common.LogError(c, "send_stream_response_failed: "+err.Error())
-			}
-			if requestMode == RequestModeCompletion {
-				responseText += claudeResponse.Completion
-				responseId = response.Id
-			} else {
-				if claudeResponse.Type == "message_start" {
-					// message_start, 获取usage
-					responseId = claudeResponse.Message.Id
-					info.UpstreamModelName = claudeResponse.Message.Model
-					usage.PromptTokens = claudeUsage.InputTokens
-				} else if claudeResponse.Type == "content_block_delta" {
-					responseText += claudeResponse.Delta.Text
-				} else if claudeResponse.Type == "message_delta" {
-					usage.CompletionTokens = claudeUsage.OutputTokens
-					usage.TotalTokens = claudeUsage.InputTokens + claudeUsage.OutputTokens
-				} else {
-					return true
-				}
-			}
-			//response.Id = responseId
-			response.Id = responseId
-			response.Created = createdTime
-			response.Model = info.UpstreamModelName
-
-			err = service.ObjectData(c, response)
-			if err != nil {
-				common.SysError(err.Error())
-			}
-			return true
-		case <-stopChan:
-			return false
 		}
-	})
+	}
+
 	if requestMode == RequestModeCompletion {
 		usage, _ = service.ResponseText2Usage(responseText, info.UpstreamModelName, info.PromptTokens)
 	} else {
@@ -357,14 +424,11 @@ func claudeStreamHandler(c *gin.Context, resp *http.Response, info *relaycommon.
 		}
 	}
 	service.Done(c)
-	err := resp.Body.Close()
+	resp.Body.Close()
-	if err != nil {
-		return service.OpenAIErrorWrapperLocal(err, "close_response_body_failed", http.StatusInternalServerError), nil
-	}
 	return nil, usage
 }
 
-func claudeHandler(requestMode int, c *gin.Context, resp *http.Response, promptTokens int, model string) (*dto.OpenAIErrorWithStatusCode, *dto.Usage) {
+func ClaudeHandler(c *gin.Context, resp *http.Response, requestMode int, info *relaycommon.RelayInfo) (*dto.OpenAIErrorWithStatusCode, *dto.Usage) {
 	responseBody, err := io.ReadAll(resp.Body)
 	if err != nil {
 		return service.OpenAIErrorWrapper(err, "read_response_body_failed", http.StatusInternalServerError), nil
@@ -390,15 +454,15 @@ func claudeHandler(requestMode int, c *gin.Context, resp *http.Response, promptT
 		}, nil
 	}
 	fullTextResponse := ResponseClaude2OpenAI(requestMode, &claudeResponse)
-	completionTokens, err := service.CountTokenText(claudeResponse.Completion, model)
+	completionTokens, err := service.CountTokenText(claudeResponse.Completion, info.OriginModelName)
 	if err != nil {
 		return service.OpenAIErrorWrapper(err, "count_token_text_failed", http.StatusInternalServerError), nil
 	}
 	usage := dto.Usage{}
 	if requestMode == RequestModeCompletion {
-		usage.PromptTokens = promptTokens
+		usage.PromptTokens = info.PromptTokens
 		usage.CompletionTokens = completionTokens
-		usage.TotalTokens = promptTokens + completionTokens
+		usage.TotalTokens = info.PromptTokens + completionTokens
 	} else {
 		usage.PromptTokens = claudeResponse.Usage.InputTokens
 		usage.CompletionTokens = claudeResponse.Usage.OutputTokens

relay/channel/cloudflare/adaptor.go

@@ -0,0 +1,105 @@
+package cloudflare
+
+import (
+	"bytes"
+	"errors"
+	"fmt"
+	"github.com/gin-gonic/gin"
+	"io"
+	"net/http"
+	"one-api/dto"
+	"one-api/relay/channel"
+	relaycommon "one-api/relay/common"
+	"one-api/relay/constant"
+)
+
+type Adaptor struct {
+}
+
+func (a *Adaptor) Init(info *relaycommon.RelayInfo) {
+}
+
+func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {
+	switch info.RelayMode {
+	case constant.RelayModeChatCompletions:
+		return fmt.Sprintf("%s/client/v4/accounts/%s/ai/v1/chat/completions", info.BaseUrl, info.ApiVersion), nil
+	case constant.RelayModeEmbeddings:
+		return fmt.Sprintf("%s/client/v4/accounts/%s/ai/v1/embeddings", info.BaseUrl, info.ApiVersion), nil
+	default:
+		return fmt.Sprintf("%s/client/v4/accounts/%s/ai/run/%s", info.BaseUrl, info.ApiVersion, info.UpstreamModelName), nil
+	}
+}
+
+func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, info *relaycommon.RelayInfo) error {
+	channel.SetupApiRequestHeader(info, c, req)
+	req.Header.Set("Authorization", fmt.Sprintf("Bearer %s", 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")
+	}
+	switch info.RelayMode {
+	case constant.RelayModeCompletions:
+		return convertCf2CompletionsRequest(*request), nil
+	default:
+		return request, nil
+	}
+}
+
+func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (*http.Response, error) {
+	return channel.DoApiRequest(a, c, info, requestBody)
+}
+
+func (a *Adaptor) ConvertRerankRequest(c *gin.Context, relayMode int, request dto.RerankRequest) (any, error) {
+	return request, nil
+}
+
+func (a *Adaptor) ConvertAudioRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.AudioRequest) (io.Reader, error) {
+	// 添加文件字段
+	file, _, err := c.Request.FormFile("file")
+	if err != nil {
+		return nil, errors.New("file is required")
+	}
+	defer file.Close()
+	// 打开临时文件用于保存上传的文件内容
+	requestBody := &bytes.Buffer{}
+
+	// 将上传的文件内容复制到临时文件
+	if _, err := io.Copy(requestBody, file); err != nil {
+		return nil, err
+	}
+	return requestBody, nil
+}
+
+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) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage *dto.Usage, err *dto.OpenAIErrorWithStatusCode) {
+	switch info.RelayMode {
+	case constant.RelayModeEmbeddings:
+		fallthrough
+	case constant.RelayModeChatCompletions:
+		if info.IsStream {
+			err, usage = cfStreamHandler(c, resp, info)
+		} else {
+			err, usage = cfHandler(c, resp, info)
+		}
+	case constant.RelayModeAudioTranslation:
+		fallthrough
+	case constant.RelayModeAudioTranscription:
+		err, usage = cfSTTHandler(c, resp, info)
+	}
+	return
+}
+
+func (a *Adaptor) GetModelList() []string {
+	return ModelList
+}
+
+func (a *Adaptor) GetChannelName() string {
+	return ChannelName
+}

relay/channel/cloudflare/constant.go

@@ -0,0 +1,39 @@
+package cloudflare
+
+var ModelList = []string{
+	"@cf/meta/llama-3.1-8b-instruct",
+	"@cf/meta/llama-2-7b-chat-fp16",
+	"@cf/meta/llama-2-7b-chat-int8",
+	"@cf/mistral/mistral-7b-instruct-v0.1",
+	"@hf/thebloke/deepseek-coder-6.7b-base-awq",
+	"@hf/thebloke/deepseek-coder-6.7b-instruct-awq",
+	"@cf/deepseek-ai/deepseek-math-7b-base",
+	"@cf/deepseek-ai/deepseek-math-7b-instruct",
+	"@cf/thebloke/discolm-german-7b-v1-awq",
+	"@cf/tiiuae/falcon-7b-instruct",
+	"@cf/google/gemma-2b-it-lora",
+	"@hf/google/gemma-7b-it",
+	"@cf/google/gemma-7b-it-lora",
+	"@hf/nousresearch/hermes-2-pro-mistral-7b",
+	"@hf/thebloke/llama-2-13b-chat-awq",
+	"@cf/meta-llama/llama-2-7b-chat-hf-lora",
+	"@cf/meta/llama-3-8b-instruct",
+	"@hf/thebloke/llamaguard-7b-awq",
+	"@hf/thebloke/mistral-7b-instruct-v0.1-awq",
+	"@hf/mistralai/mistral-7b-instruct-v0.2",
+	"@cf/mistral/mistral-7b-instruct-v0.2-lora",
+	"@hf/thebloke/neural-chat-7b-v3-1-awq",
+	"@cf/openchat/openchat-3.5-0106",
+	"@hf/thebloke/openhermes-2.5-mistral-7b-awq",
+	"@cf/microsoft/phi-2",
+	"@cf/qwen/qwen1.5-0.5b-chat",
+	"@cf/qwen/qwen1.5-1.8b-chat",
+	"@cf/qwen/qwen1.5-14b-chat-awq",
+	"@cf/qwen/qwen1.5-7b-chat-awq",
+	"@cf/defog/sqlcoder-7b-2",
+	"@hf/nexusflow/starling-lm-7b-beta",
+	"@cf/tinyllama/tinyllama-1.1b-chat-v1.0",
+	"@hf/thebloke/zephyr-7b-beta-awq",
+}
+
+var ChannelName = "cloudflare"

relay/channel/cloudflare/dto.go

@@ -0,0 +1,21 @@
+package cloudflare
+
+import "one-api/dto"
+
+type CfRequest struct {
+	Messages    []dto.Message `json:"messages,omitempty"`
+	Lora        string        `json:"lora,omitempty"`
+	MaxTokens   int           `json:"max_tokens,omitempty"`
+	Prompt      string        `json:"prompt,omitempty"`
+	Raw         bool          `json:"raw,omitempty"`
+	Stream      bool          `json:"stream,omitempty"`
+	Temperature float64       `json:"temperature,omitempty"`
+}
+
+type CfAudioResponse struct {
+	Result CfSTTResult `json:"result"`
+}
+
+type CfSTTResult struct {
+	Text string `json:"text"`
+}

relay/channel/cloudflare/relay_cloudflare.go

@@ -0,0 +1,156 @@
+package cloudflare
+
+import (
+	"bufio"
+	"encoding/json"
+	"github.com/gin-gonic/gin"
+	"io"
+	"net/http"
+	"one-api/common"
+	"one-api/dto"
+	relaycommon "one-api/relay/common"
+	"one-api/service"
+	"strings"
+	"time"
+)
+
+func convertCf2CompletionsRequest(textRequest dto.GeneralOpenAIRequest) *CfRequest {
+	p, _ := textRequest.Prompt.(string)
+	return &CfRequest{
+		Prompt:      p,
+		MaxTokens:   textRequest.GetMaxTokens(),
+		Stream:      textRequest.Stream,
+		Temperature: textRequest.Temperature,
+	}
+}
+
+func cfStreamHandler(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (*dto.OpenAIErrorWithStatusCode, *dto.Usage) {
+	scanner := bufio.NewScanner(resp.Body)
+	scanner.Split(bufio.ScanLines)
+
+	service.SetEventStreamHeaders(c)
+	id := service.GetResponseID(c)
+	var responseText string
+	isFirst := true
+
+	for scanner.Scan() {
+		data := scanner.Text()
+		if len(data) < len("data: ") {
+			continue
+		}
+		data = strings.TrimPrefix(data, "data: ")
+		data = strings.TrimSuffix(data, "\r")
+
+		if data == "[DONE]" {
+			break
+		}
+
+		var response dto.ChatCompletionsStreamResponse
+		err := json.Unmarshal([]byte(data), &response)
+		if err != nil {
+			common.LogError(c, "error_unmarshalling_stream_response: "+err.Error())
+			continue
+		}
+		for _, choice := range response.Choices {
+			choice.Delta.Role = "assistant"
+			responseText += choice.Delta.GetContentString()
+		}
+		response.Id = id
+		response.Model = info.UpstreamModelName
+		err = service.ObjectData(c, response)
+		if isFirst {
+			isFirst = false
+			info.FirstResponseTime = time.Now()
+		}
+		if err != nil {
+			common.LogError(c, "error_rendering_stream_response: "+err.Error())
+		}
+	}
+
+	if err := scanner.Err(); err != nil {
+		common.LogError(c, "error_scanning_stream_response: "+err.Error())
+	}
+	usage, _ := service.ResponseText2Usage(responseText, info.UpstreamModelName, info.PromptTokens)
+	if info.ShouldIncludeUsage {
+		response := service.GenerateFinalUsageResponse(id, info.StartTime.Unix(), info.UpstreamModelName, *usage)
+		err := service.ObjectData(c, response)
+		if err != nil {
+			common.LogError(c, "error_rendering_final_usage_response: "+err.Error())
+		}
+	}
+	service.Done(c)
+
+	err := resp.Body.Close()
+	if err != nil {
+		common.LogError(c, "close_response_body_failed: "+err.Error())
+	}
+
+	return nil, usage
+}
+
+func cfHandler(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (*dto.OpenAIErrorWithStatusCode, *dto.Usage) {
+	responseBody, err := io.ReadAll(resp.Body)
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "read_response_body_failed", http.StatusInternalServerError), nil
+	}
+	err = resp.Body.Close()
+	if err != nil {
+		return service.OpenAIErrorWrapperLocal(err, "close_response_body_failed", http.StatusInternalServerError), nil
+	}
+	var response dto.TextResponse
+	err = json.Unmarshal(responseBody, &response)
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "unmarshal_response_body_failed", http.StatusInternalServerError), nil
+	}
+	response.Model = info.UpstreamModelName
+	var responseText string
+	for _, choice := range response.Choices {
+		responseText += choice.Message.StringContent()
+	}
+	usage, _ := service.ResponseText2Usage(responseText, info.UpstreamModelName, info.PromptTokens)
+	response.Usage = *usage
+	response.Id = service.GetResponseID(c)
+	jsonResponse, err := json.Marshal(response)
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "marshal_response_body_failed", http.StatusInternalServerError), nil
+	}
+	c.Writer.Header().Set("Content-Type", "application/json")
+	c.Writer.WriteHeader(resp.StatusCode)
+	_, _ = c.Writer.Write(jsonResponse)
+	return nil, usage
+}
+
+func cfSTTHandler(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (*dto.OpenAIErrorWithStatusCode, *dto.Usage) {
+	var cfResp CfAudioResponse
+	responseBody, err := io.ReadAll(resp.Body)
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "read_response_body_failed", http.StatusInternalServerError), nil
+	}
+	err = resp.Body.Close()
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), nil
+	}
+	err = json.Unmarshal(responseBody, &cfResp)
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "unmarshal_response_body_failed", http.StatusInternalServerError), nil
+	}
+
+	audioResp := &dto.AudioResponse{
+		Text: cfResp.Result.Text,
+	}
+
+	jsonResponse, err := json.Marshal(audioResp)
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "marshal_response_body_failed", http.StatusInternalServerError), nil
+	}
+	c.Writer.Header().Set("Content-Type", "application/json")
+	c.Writer.WriteHeader(resp.StatusCode)
+	_, _ = c.Writer.Write(jsonResponse)
+
+	usage := &dto.Usage{}
+	usage.PromptTokens = info.PromptTokens
+	usage.CompletionTokens, _ = service.CountTokenText(cfResp.Result.Text, info.UpstreamModelName)
+	usage.TotalTokens = usage.PromptTokens + usage.CompletionTokens
+
+	return nil, usage
+}

relay/channel/cohere/adaptor.go

@@ -1,6 +1,7 @@
 package cohere
 
 import (
+	"errors"
 	"fmt"
 	"github.com/gin-gonic/gin"
 	"io"
@@ -14,10 +15,17 @@ import (
 type Adaptor struct {
 }
 
-func (a *Adaptor) InitRerank(info *relaycommon.RelayInfo, request dto.RerankRequest) {
+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) Init(info *relaycommon.RelayInfo, request dto.GeneralOpenAIRequest) {
+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) {
@@ -34,7 +42,7 @@ func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, info *re
 	return nil
 }
 
-func (a *Adaptor) ConvertRequest(c *gin.Context, relayMode int, request *dto.GeneralOpenAIRequest) (any, error) {
+func (a *Adaptor) ConvertRequest(c *gin.Context, info *relaycommon.RelayInfo, request *dto.GeneralOpenAIRequest) (any, error) {
 	return requestOpenAI2Cohere(*request), nil
 }
 

relay/channel/cohere/constant.go

@@ -1,7 +1,10 @@
 package cohere
 
 var ModelList = []string{
-	"command-r", "command-r-plus", "command-light", "command-light-nightly", "command", "command-nightly",
+	"command-r", "command-r-plus",
+	"command-r-08-2024", "command-r-plus-08-2024",
+	"c4ai-aya-23-35b", "c4ai-aya-23-8b",
+	"command-light", "command-light-nightly", "command", "command-nightly",
 	"rerank-english-v3.0", "rerank-multilingual-v3.0", "rerank-english-v2.0", "rerank-multilingual-v2.0",
 }
 

relay/channel/cohere/dto.go

@@ -7,7 +7,7 @@ type CohereRequest struct {
 	ChatHistory []ChatHistory `json:"chat_history"`
 	Message     string        `json:"message"`
 	Stream      bool          `json:"stream"`
-	MaxTokens   int64         `json:"max_tokens"`
+	MaxTokens   int           `json:"max_tokens"`
 }
 
 type ChatHistory struct {

relay/channel/dify/adaptor.go

@@ -14,12 +14,17 @@ import (
 type Adaptor struct {
 }
 
-func (a *Adaptor) InitRerank(info *relaycommon.RelayInfo, request dto.RerankRequest) {
+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) Init(info *relaycommon.RelayInfo, request dto.GeneralOpenAIRequest) {
+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) {
@@ -32,7 +37,7 @@ func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, info *re
 	return nil
 }
 
-func (a *Adaptor) ConvertRequest(c *gin.Context, relayMode int, request *dto.GeneralOpenAIRequest) (any, error) {
+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")
 	}

relay/channel/dify/relay-dify.go

@@ -53,7 +53,7 @@ func streamResponseDify2OpenAI(difyResponse DifyChunkChatCompletionResponse) *dt
 		choice.Delta.SetContentString("Workflow: " + difyResponse.Data.WorkflowId + "\n")
 	} else if constant.DifyDebug && difyResponse.Event == "node_started" {
 		choice.Delta.SetContentString("Node: " + difyResponse.Data.NodeId + "\n")
-	} else if difyResponse.Event == "message" {
+	} else if difyResponse.Event == "message" || difyResponse.Event == "agent_message" {
 		choice.Delta.SetContentString(difyResponse.Answer)
 	}
 	response.Choices = append(response.Choices, choice)

relay/channel/gemini/adaptor.go

@@ -6,6 +6,7 @@ import (
 	"github.com/gin-gonic/gin"
 	"io"
 	"net/http"
+	"one-api/constant"
 	"one-api/dto"
 	"one-api/relay/channel"
 	relaycommon "one-api/relay/common"
@@ -14,22 +15,23 @@ import (
 type Adaptor struct {
 }
 
-func (a *Adaptor) InitRerank(info *relaycommon.RelayInfo, request dto.RerankRequest) {
+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) Init(info *relaycommon.RelayInfo, request dto.GeneralOpenAIRequest) {
+func (a *Adaptor) ConvertImageRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.ImageRequest) (any, error) {
+	//TODO implement me
+	return nil, errors.New("not implemented")
 }
 
-// 定义一个映射，存储模型名称和对应的版本
+func (a *Adaptor) Init(info *relaycommon.RelayInfo) {
-var modelVersionMap = map[string]string{
+
-	"gemini-1.5-pro-latest":   "v1beta",
-	"gemini-1.5-flash-latest": "v1beta",
-	"gemini-ultra":            "v1beta",
 }
 
 func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {
 	// 从映射中获取模型名称对应的版本，如果找不到就使用 info.ApiVersion 或默认的版本 "v1"
-	version, beta := modelVersionMap[info.UpstreamModelName]
+	version, beta := constant.GeminiModelMap[info.UpstreamModelName]
 	if !beta {
 		if info.ApiVersion != "" {
 			version = info.ApiVersion
@@ -40,7 +42,7 @@ func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {
 
 	action := "generateContent"
 	if info.IsStream {
-		action = "streamGenerateContent"
+		action = "streamGenerateContent?alt=sse"
 	}
 	return fmt.Sprintf("%s/%s/models/%s:%s", info.BaseUrl, version, info.UpstreamModelName, action), nil
 }
@@ -51,7 +53,7 @@ func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, info *re
 	return nil
 }
 
-func (a *Adaptor) ConvertRequest(c *gin.Context, relayMode int, request *dto.GeneralOpenAIRequest) (any, error) {
+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")
 	}
@@ -68,9 +70,9 @@ func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, request
 
 func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage *dto.Usage, err *dto.OpenAIErrorWithStatusCode) {
 	if info.IsStream {
-		err, usage = geminiChatStreamHandler(c, resp, info)
+		err, usage = GeminiChatStreamHandler(c, resp, info)
 	} else {
-		err, usage = geminiChatHandler(c, resp, info.PromptTokens, info.UpstreamModelName)
+		err, usage = GeminiChatHandler(c, resp)
 	}
 	return
 }

relay/channel/gemini/dto.go

@@ -12,9 +12,15 @@ type GeminiInlineData struct {
 	Data     string `json:"data"`
 }
 
+type FunctionCall struct {
+	FunctionName string `json:"name"`
+	Arguments    any    `json:"args"`
+}
+
 type GeminiPart struct {
-	Text       string            `json:"text,omitempty"`
+	Text         string            `json:"text,omitempty"`
-	InlineData *GeminiInlineData `json:"inlineData,omitempty"`
+	InlineData   *GeminiInlineData `json:"inlineData,omitempty"`
+	FunctionCall *FunctionCall     `json:"functionCall,omitempty"`
 }
 
 type GeminiChatContent struct {

relay/channel/gemini/relay-gemini.go

@@ -4,18 +4,14 @@ import (
 	"bufio"
 	"encoding/json"
 	"fmt"
+	"github.com/gin-gonic/gin"
 	"io"
-	"log"
 	"net/http"
 	"one-api/common"
-	"one-api/constant"
 	"one-api/dto"
 	relaycommon "one-api/relay/common"
 	"one-api/service"
 	"strings"
-	"time"
-
-	"github.com/gin-gonic/gin"
 )
 
 // Setting safety to the lowest possible values since Gemini is already powerless enough
@@ -46,7 +42,17 @@ func CovertGemini2OpenAI(textRequest dto.GeneralOpenAIRequest) *GeminiChatReques
 			MaxOutputTokens: textRequest.MaxTokens,
 		},
 	}
-	if textRequest.Functions != nil {
+	if textRequest.Tools != nil {
+		functions := make([]dto.FunctionCall, 0, len(textRequest.Tools))
+		for _, tool := range textRequest.Tools {
+			functions = append(functions, tool.Function)
+		}
+		geminiRequest.Tools = []GeminiChatTools{
+			{
+				FunctionDeclarations: functions,
+			},
+		}
+	} else if textRequest.Functions != nil {
 		geminiRequest.Tools = []GeminiChatTools{
 			{
 				FunctionDeclarations: textRequest.Functions,
@@ -77,13 +83,28 @@ func CovertGemini2OpenAI(textRequest dto.GeneralOpenAIRequest) *GeminiChatReques
 				if imageNum > GeminiVisionMaxImageNum {
 					continue
 				}
-				mimeType, data, _ := common.GetImageFromUrl(part.ImageUrl.(dto.MessageImageUrl).Url)
+				// 判断是否是url
-				parts = append(parts, GeminiPart{
+				if strings.HasPrefix(part.ImageUrl.(dto.MessageImageUrl).Url, "http") {
-					InlineData: &GeminiInlineData{
+					// 是url，获取图片的类型和base64编码的数据
-						MimeType: mimeType,
+					mimeType, data, _ := common.GetImageFromUrl(part.ImageUrl.(dto.MessageImageUrl).Url)
-						Data:     data,
+					parts = append(parts, GeminiPart{
-					},
+						InlineData: &GeminiInlineData{
-				})
+							MimeType: mimeType,
+							Data:     data,
+						},
+					})
+				} else {
+					_, format, base64String, err := common.DecodeBase64ImageData(part.ImageUrl.(dto.MessageImageUrl).Url)
+					if err != nil {
+						continue
+					}
+					parts = append(parts, GeminiPart{
+						InlineData: &GeminiInlineData{
+							MimeType: "image/" + format,
+							Data:     base64String,
+						},
+					})
+				}
 			}
 		}
 		content.Parts = parts
@@ -126,6 +147,30 @@ func (g *GeminiChatResponse) GetResponseText() string {
 	return ""
 }
 
+func getToolCalls(candidate *GeminiChatCandidate) []dto.ToolCall {
+	var toolCalls []dto.ToolCall
+
+	item := candidate.Content.Parts[0]
+	if item.FunctionCall == nil {
+		return toolCalls
+	}
+	argsBytes, err := json.Marshal(item.FunctionCall.Arguments)
+	if err != nil {
+		//common.SysError("getToolCalls failed: " + err.Error())
+		return toolCalls
+	}
+	toolCall := dto.ToolCall{
+		ID:   fmt.Sprintf("call_%s", common.GetUUID()),
+		Type: "function",
+		Function: dto.FunctionCall{
+			Arguments: string(argsBytes),
+			Name:      item.FunctionCall.FunctionName,
+		},
+	}
+	toolCalls = append(toolCalls, toolCall)
+	return toolCalls
+}
+
 func responseGeminiChat2OpenAI(response *GeminiChatResponse) *dto.OpenAITextResponse {
 	fullTextResponse := dto.OpenAITextResponse{
 		Id:      fmt.Sprintf("chatcmpl-%s", common.GetUUID()),
@@ -144,8 +189,11 @@ func responseGeminiChat2OpenAI(response *GeminiChatResponse) *dto.OpenAITextResp
 			FinishReason: relaycommon.StopFinishReason,
 		}
 		if len(candidate.Content.Parts) > 0 {
-			content, _ = json.Marshal(candidate.Content.Parts[0].Text)
+			if candidate.Content.Parts[0].FunctionCall != nil {
-			choice.Message.Content = content
+				choice.Message.ToolCalls = getToolCalls(&candidate)
+			} else {
+				choice.Message.SetStringContent(candidate.Content.Parts[0].Text)
+			}
 		}
 		fullTextResponse.Choices = append(fullTextResponse.Choices, choice)
 	}
@@ -154,8 +202,17 @@ func responseGeminiChat2OpenAI(response *GeminiChatResponse) *dto.OpenAITextResp
 
 func streamResponseGeminiChat2OpenAI(geminiResponse *GeminiChatResponse) *dto.ChatCompletionsStreamResponse {
 	var choice dto.ChatCompletionsStreamResponseChoice
-	choice.Delta.SetContentString(geminiResponse.GetResponseText())
+	//choice.Delta.SetContentString(geminiResponse.GetResponseText())
-	choice.FinishReason = &relaycommon.StopFinishReason
+	if len(geminiResponse.Candidates) > 0 && len(geminiResponse.Candidates[0].Content.Parts) > 0 {
+		respFirst := geminiResponse.Candidates[0].Content.Parts[0]
+		if respFirst.FunctionCall != nil {
+			// function response
+			choice.Delta.ToolCalls = getToolCalls(&geminiResponse.Candidates[0])
+		} else {
+			// text response
+			choice.Delta.SetContentString(respFirst.Text)
+		}
+	}
 	var response dto.ChatCompletionsStreamResponse
 	response.Object = "chat.completion.chunk"
 	response.Model = "gemini"
@@ -163,110 +220,66 @@ func streamResponseGeminiChat2OpenAI(geminiResponse *GeminiChatResponse) *dto.Ch
 	return &response
 }
 
-func geminiChatStreamHandler(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (*dto.OpenAIErrorWithStatusCode, *dto.Usage) {
+func GeminiChatStreamHandler(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (*dto.OpenAIErrorWithStatusCode, *dto.Usage) {
 	responseText := ""
-	responseJson := ""
 	id := fmt.Sprintf("chatcmpl-%s", common.GetUUID())
 	createAt := common.GetTimestamp()
 	var usage = &dto.Usage{}
-	dataChan := make(chan string, 5)
-	stopChan := make(chan bool, 2)
 	scanner := bufio.NewScanner(resp.Body)
-	scanner.Split(func(data []byte, atEOF bool) (advance int, token []byte, err error) {
+	scanner.Split(bufio.ScanLines)
-		if atEOF && len(data) == 0 {
+
-			return 0, nil, nil
-		}
-		if i := strings.Index(string(data), "\n"); i >= 0 {
-			return i + 1, data[0:i], nil
-		}
-		if atEOF {
-			return len(data), data, nil
-		}
-		return 0, nil, nil
-	})
-	go func() {
-		for scanner.Scan() {
-			data := scanner.Text()
-			responseJson += data
-			data = strings.TrimSpace(data)
-			if !strings.HasPrefix(data, "\"text\": \"") {
-				continue
-			}
-			data = strings.TrimPrefix(data, "\"text\": \"")
-			data = strings.TrimSuffix(data, "\"")
-			if !common.SafeSendStringTimeout(dataChan, data, constant.StreamingTimeout) {
-				// send data timeout, stop the stream
-				common.LogError(c, "send data timeout, stop the stream")
-				break
-			}
-		}
-		stopChan <- true
-	}()
-	isFirst := true
 	service.SetEventStreamHeaders(c)
-	c.Stream(func(w io.Writer) bool {
+	for scanner.Scan() {
-		select {
+		data := scanner.Text()
-		case data := <-dataChan:
+		info.SetFirstResponseTime()
-			if isFirst {
+		data = strings.TrimSpace(data)
-				isFirst = false
+		if !strings.HasPrefix(data, "data: ") {
-				info.FirstResponseTime = time.Now()
+			continue
-			}
-			// this is used to prevent annoying \ related format bug
-			data = fmt.Sprintf("{\"content\": \"%s\"}", data)
-			type dummyStruct struct {
-				Content string `json:"content"`
-			}
-			var dummy dummyStruct
-			err := json.Unmarshal([]byte(data), &dummy)
-			responseText += dummy.Content
-			var choice dto.ChatCompletionsStreamResponseChoice
-			choice.Delta.SetContentString(dummy.Content)
-			response := dto.ChatCompletionsStreamResponse{
-				Id:      id,
-				Object:  "chat.completion.chunk",
-				Created: createAt,
-				Model:   info.UpstreamModelName,
-				Choices: []dto.ChatCompletionsStreamResponseChoice{choice},
-			}
-			jsonResponse, err := json.Marshal(response)
-			if err != nil {
-				common.SysError("error marshalling stream response: " + err.Error())
-				return true
-			}
-			c.Render(-1, common.CustomEvent{Data: "data: " + string(jsonResponse)})
-			return true
-		case <-stopChan:
-			return false
 		}
-	})
+		data = strings.TrimPrefix(data, "data: ")
-	var geminiChatResponses []GeminiChatResponse
+		data = strings.TrimSuffix(data, "\"")
-	err := json.Unmarshal([]byte(responseJson), &geminiChatResponses)
+		var geminiResponse GeminiChatResponse
-	if err != nil {
+		err := json.Unmarshal([]byte(data), &geminiResponse)
-		log.Printf("cannot get gemini usage: %s", err.Error())
+		if err != nil {
-		usage, _ = service.ResponseText2Usage(responseText, info.UpstreamModelName, info.PromptTokens)
+			common.LogError(c, "error unmarshalling stream response: "+err.Error())
-	} else {
+			continue
-		for _, response := range geminiChatResponses {
+		}
-			usage.PromptTokens = response.UsageMetadata.PromptTokenCount
+
-			usage.CompletionTokens = response.UsageMetadata.CandidatesTokenCount
+		response := streamResponseGeminiChat2OpenAI(&geminiResponse)
+		if response == nil {
+			continue
+		}
+		response.Id = id
+		response.Created = createAt
+		responseText += response.Choices[0].Delta.GetContentString()
+		if geminiResponse.UsageMetadata.TotalTokenCount != 0 {
+			usage.PromptTokens = geminiResponse.UsageMetadata.PromptTokenCount
+			usage.CompletionTokens = geminiResponse.UsageMetadata.CandidatesTokenCount
+		}
+		err = service.ObjectData(c, response)
+		if err != nil {
+			common.LogError(c, err.Error())
 		}
-		usage.TotalTokens = usage.PromptTokens + usage.CompletionTokens
 	}
+
+	response := service.GenerateStopResponse(id, createAt, info.UpstreamModelName, relaycommon.StopFinishReason)
+	service.ObjectData(c, response)
+
+	usage.TotalTokens = usage.PromptTokens + usage.CompletionTokens
+
 	if info.ShouldIncludeUsage {
-		response := service.GenerateFinalUsageResponse(id, createAt, info.UpstreamModelName, *usage)
+		response = service.GenerateFinalUsageResponse(id, createAt, info.UpstreamModelName, *usage)
 		err := service.ObjectData(c, response)
 		if err != nil {
 			common.SysError("send final response failed: " + err.Error())
 		}
 	}
 	service.Done(c)
-	err = resp.Body.Close()
+	resp.Body.Close()
-	if err != nil {
-		return service.OpenAIErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), usage
-	}
 	return nil, usage
 }
 
-func geminiChatHandler(c *gin.Context, resp *http.Response, promptTokens int, model string) (*dto.OpenAIErrorWithStatusCode, *dto.Usage) {
+func GeminiChatHandler(c *gin.Context, resp *http.Response) (*dto.OpenAIErrorWithStatusCode, *dto.Usage) {
 	responseBody, err := io.ReadAll(resp.Body)
 	if err != nil {
 		return service.OpenAIErrorWrapper(err, "read_response_body_failed", http.StatusInternalServerError), nil

relay/channel/jina/adaptor.go

@@ -15,17 +15,24 @@ import (
 type Adaptor struct {
 }
 
-func (a *Adaptor) InitRerank(info *relaycommon.RelayInfo, request dto.RerankRequest) {
+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) Init(info *relaycommon.RelayInfo, request dto.GeneralOpenAIRequest) {
+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) {
 	if info.RelayMode == constant.RelayModeRerank {
 		return fmt.Sprintf("%s/v1/rerank", info.BaseUrl), nil
 	} else if info.RelayMode == constant.RelayModeEmbeddings {
-		return fmt.Sprintf("%s/v1/embeddings ", info.BaseUrl), nil
+		return fmt.Sprintf("%s/v1/embeddings", info.BaseUrl), nil
 	}
 	return "", errors.New("invalid relay mode")
 }
@@ -36,7 +43,7 @@ func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, info *re
 	return nil
 }
 
-func (a *Adaptor) ConvertRequest(c *gin.Context, relayMode int, request *dto.GeneralOpenAIRequest) (any, error) {
+func (a *Adaptor) ConvertRequest(c *gin.Context, info *relaycommon.RelayInfo, request *dto.GeneralOpenAIRequest) (any, error) {
 	return request, nil
 }
 
@@ -51,6 +58,8 @@ func (a *Adaptor) ConvertRerankRequest(c *gin.Context, relayMode int, request dt
 func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage *dto.Usage, err *dto.OpenAIErrorWithStatusCode) {
 	if info.RelayMode == constant.RelayModeRerank {
 		err, usage = jinaRerankHandler(c, resp)
+	} else if info.RelayMode == constant.RelayModeEmbeddings {
+		err, usage = jinaEmbeddingHandler(c, resp)
 	}
 	return
 }

relay/channel/jina/relay-jina.go

@@ -33,3 +33,28 @@ func jinaRerankHandler(c *gin.Context, resp *http.Response) (*dto.OpenAIErrorWit
 	_, err = c.Writer.Write(jsonResponse)
 	return nil, &jinaResp.Usage
 }
+
+func jinaEmbeddingHandler(c *gin.Context, resp *http.Response) (*dto.OpenAIErrorWithStatusCode, *dto.Usage) {
+	responseBody, err := io.ReadAll(resp.Body)
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "read_response_body_failed", http.StatusInternalServerError), nil
+	}
+	err = resp.Body.Close()
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), nil
+	}
+	var jinaResp dto.OpenAIEmbeddingResponse
+	err = json.Unmarshal(responseBody, &jinaResp)
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "unmarshal_response_body_failed", http.StatusInternalServerError), nil
+	}
+
+	jsonResponse, err := json.Marshal(jinaResp)
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "marshal_response_body_failed", http.StatusInternalServerError), nil
+	}
+	c.Writer.Header().Set("Content-Type", "application/json")
+	c.Writer.WriteHeader(resp.StatusCode)
+	_, err = c.Writer.Write(jsonResponse)
+	return nil, &jinaResp.Usage
+}

relay/channel/ollama/adaptor.go

@@ -10,16 +10,22 @@ import (
 	"one-api/relay/channel/openai"
 	relaycommon "one-api/relay/common"
 	relayconstant "one-api/relay/constant"
-	"one-api/service"
 )
 
 type Adaptor struct {
 }
 
-func (a *Adaptor) InitRerank(info *relaycommon.RelayInfo, request dto.RerankRequest) {
+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) Init(info *relaycommon.RelayInfo, request dto.GeneralOpenAIRequest) {
+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) {
@@ -36,11 +42,11 @@ func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, info *re
 	return nil
 }
 
-func (a *Adaptor) ConvertRequest(c *gin.Context, relayMode int, request *dto.GeneralOpenAIRequest) (any, error) {
+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")
 	}
-	switch relayMode {
+	switch info.RelayMode {
 	case relayconstant.RelayModeEmbeddings:
 		return requestOpenAI2Embeddings(*request), nil
 	default:
@@ -58,11 +64,7 @@ func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, request
 
 func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage *dto.Usage, err *dto.OpenAIErrorWithStatusCode) {
 	if info.IsStream {
-		var responseText string
+		err, usage = openai.OaiStreamHandler(c, resp, info)
-		err, usage, responseText, _ = openai.OpenaiStreamHandler(c, resp, info)
-		if usage == nil || usage.TotalTokens == 0 || (usage.PromptTokens+usage.CompletionTokens) == 0 {
-			usage, _ = service.ResponseText2Usage(responseText, info.UpstreamModelName, info.PromptTokens)
-		}
 	} else {
 		if info.RelayMode == relayconstant.RelayModeEmbeddings {
 			err, usage = ollamaEmbeddingHandler(c, resp, info.PromptTokens, info.UpstreamModelName, info.RelayMode)

relay/channel/ollama/dto.go

@@ -3,14 +3,18 @@ package ollama
 import "one-api/dto"
 
 type OllamaRequest struct {
-	Model       string        `json:"model,omitempty"`
+	Model            string         `json:"model,omitempty"`
-	Messages    []dto.Message `json:"messages,omitempty"`
+	Messages         []dto.Message  `json:"messages,omitempty"`
-	Stream      bool          `json:"stream,omitempty"`
+	Stream           bool           `json:"stream,omitempty"`
-	Temperature float64       `json:"temperature,omitempty"`
+	Temperature      float64        `json:"temperature,omitempty"`
-	Seed        float64       `json:"seed,omitempty"`
+	Seed             float64        `json:"seed,omitempty"`
-	Topp        float64       `json:"top_p,omitempty"`
+	Topp             float64        `json:"top_p,omitempty"`
-	TopK        int           `json:"top_k,omitempty"`
+	TopK             int            `json:"top_k,omitempty"`
-	Stop        any           `json:"stop,omitempty"`
+	Stop             any            `json:"stop,omitempty"`
+	Tools            []dto.ToolCall `json:"tools,omitempty"`
+	ResponseFormat   any            `json:"response_format,omitempty"`
+	FrequencyPenalty float64        `json:"frequency_penalty,omitempty"`
+	PresencePenalty  float64        `json:"presence_penalty,omitempty"`
 }
 
 type OllamaEmbeddingRequest struct {
@@ -19,8 +23,5 @@ type OllamaEmbeddingRequest struct {
 }
 
 type OllamaEmbeddingResponse struct {
-	Embedding []float64 `json:"embedding,omitempty"`
+	Embedding any `json:"embedding,omitempty"`
 }
-
-//type OllamaOptions struct {
-//}

relay/channel/ollama/relay-ollama.go

@@ -28,14 +28,18 @@ func requestOpenAI2Ollama(request dto.GeneralOpenAIRequest) *OllamaRequest {
 		Stop, _ = request.Stop.([]string)
 	}
 	return &OllamaRequest{
-		Model:       request.Model,
+		Model:            request.Model,
-		Messages:    messages,
+		Messages:         messages,
-		Stream:      request.Stream,
+		Stream:           request.Stream,
-		Temperature: request.Temperature,
+		Temperature:      request.Temperature,
-		Seed:        request.Seed,
+		Seed:             request.Seed,
-		Topp:        request.TopP,
+		Topp:             request.TopP,
-		TopK:        request.TopK,
+		TopK:             request.TopK,
-		Stop:        Stop,
+		Stop:             Stop,
+		Tools:            request.Tools,
+		ResponseFormat:   request.ResponseFormat,
+		FrequencyPenalty: request.FrequencyPenalty,
+		PresencePenalty:  request.PresencePenalty,
 	}
 }
 

relay/channel/openai/adaptor.go

@@ -1,10 +1,13 @@
 package openai
 
 import (
+	"bytes"
+	"encoding/json"
 	"errors"
 	"fmt"
 	"github.com/gin-gonic/gin"
 	"io"
+	"mime/multipart"
 	"net/http"
 	"one-api/common"
 	"one-api/dto"
@@ -14,22 +17,16 @@ import (
 	"one-api/relay/channel/minimax"
 	"one-api/relay/channel/moonshot"
 	relaycommon "one-api/relay/common"
-	"one-api/service"
+	"one-api/relay/constant"
 	"strings"
 )
 
 type Adaptor struct {
-	ChannelType int
+	ChannelType    int
+	ResponseFormat string
 }
 
-func (a *Adaptor) ConvertRerankRequest(c *gin.Context, relayMode int, request dto.RerankRequest) (any, error) {
+func (a *Adaptor) Init(info *relaycommon.RelayInfo) {
-	return nil, nil
-}
-
-func (a *Adaptor) InitRerank(info *relaycommon.RelayInfo, request dto.RerankRequest) {
-}
-
-func (a *Adaptor) Init(info *relaycommon.RelayInfo, request dto.GeneralOpenAIRequest) {
 	a.ChannelType = info.ChannelType
 }
 
@@ -74,28 +71,84 @@ func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, info *re
 	return nil
 }
 
-func (a *Adaptor) ConvertRequest(c *gin.Context, relayMode int, request *dto.GeneralOpenAIRequest) (any, error) {
+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")
 	}
+	if info.ChannelType != common.ChannelTypeOpenAI {
+		request.StreamOptions = nil
+	}
+	return request, nil
+}
+
+func (a *Adaptor) ConvertRerankRequest(c *gin.Context, relayMode int, request dto.RerankRequest) (any, error) {
+	return nil, errors.New("not implemented")
+}
+
+func (a *Adaptor) ConvertAudioRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.AudioRequest) (io.Reader, error) {
+	a.ResponseFormat = request.ResponseFormat
+	if info.RelayMode == constant.RelayModeAudioSpeech {
+		jsonData, err := json.Marshal(request)
+		if err != nil {
+			return nil, fmt.Errorf("error marshalling object: %w", err)
+		}
+		return bytes.NewReader(jsonData), nil
+	} else {
+		var requestBody bytes.Buffer
+		writer := multipart.NewWriter(&requestBody)
+
+		writer.WriteField("model", request.Model)
+
+		// 添加文件字段
+		file, header, err := c.Request.FormFile("file")
+		if err != nil {
+			return nil, errors.New("file is required")
+		}
+		defer file.Close()
+
+		part, err := writer.CreateFormFile("file", header.Filename)
+		if err != nil {
+			return nil, errors.New("create form file failed")
+		}
+		if _, err := io.Copy(part, file); err != nil {
+			return nil, errors.New("copy file failed")
+		}
+
+		// 关闭 multipart 编写器以设置分界线
+		writer.Close()
+		c.Request.Header.Set("Content-Type", writer.FormDataContentType())
+		return &requestBody, nil
+	}
+}
+
+func (a *Adaptor) ConvertImageRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.ImageRequest) (any, error) {
 	return request, nil
 }
 
 func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (*http.Response, error) {
-	return channel.DoApiRequest(a, c, info, requestBody)
+	if info.RelayMode == constant.RelayModeAudioTranscription || info.RelayMode == constant.RelayModeAudioTranslation {
+		return channel.DoFormRequest(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) {
-	if info.IsStream {
+	switch info.RelayMode {
-		var responseText string
+	case constant.RelayModeAudioSpeech:
-		var toolCount int
+		err, usage = OpenaiTTSHandler(c, resp, info)
-		err, usage, responseText, toolCount = OpenaiStreamHandler(c, resp, info)
+	case constant.RelayModeAudioTranslation:
-		if usage == nil || usage.TotalTokens == 0 || (usage.PromptTokens+usage.CompletionTokens) == 0 {
+		fallthrough
-			usage, _ = service.ResponseText2Usage(responseText, info.UpstreamModelName, info.PromptTokens)
+	case constant.RelayModeAudioTranscription:
-			usage.CompletionTokens += toolCount * 7
+		err, usage = OpenaiSTTHandler(c, resp, info, a.ResponseFormat)
+	case constant.RelayModeImagesGenerations:
+		err, usage = OpenaiTTSHandler(c, resp, info)
+	default:
+		if info.IsStream {
+			err, usage = OaiStreamHandler(c, resp, info)
+		} else {
+			err, usage = OpenaiHandler(c, resp, info.PromptTokens, info.UpstreamModelName)
 		}
-	} else {
-		err, usage = OpenaiHandler(c, resp, info.PromptTokens, info.UpstreamModelName)
 	}
 	return
 }

relay/channel/openai/constant.go

@@ -8,7 +8,9 @@ var ModelList = []string{
 	"gpt-4-32k", "gpt-4-32k-0314", "gpt-4-32k-0613",
 	"gpt-4-turbo-preview", "gpt-4-turbo", "gpt-4-turbo-2024-04-09",
 	"gpt-4-vision-preview",
-	"gpt-4o", "gpt-4o-2024-05-13",
+	"chatgpt-4o-latest",
+	"gpt-4o", "gpt-4o-2024-05-13", "gpt-4o-2024-08-06",
+	"gpt-4o-mini", "gpt-4o-mini-2024-07-18",
 	"text-embedding-ada-002", "text-embedding-3-small", "text-embedding-3-large",
 	"text-curie-001", "text-babbage-001", "text-ada-001", "text-davinci-002", "text-davinci-003",
 	"text-moderation-latest", "text-moderation-stable",

relay/channel/openai/relay-openai.go

@@ -4,6 +4,8 @@ import (
 	"bufio"
 	"bytes"
 	"encoding/json"
+	"fmt"
+	"github.com/bytedance/gopkg/util/gopool"
 	"github.com/gin-gonic/gin"
 	"io"
 	"net/http"
@@ -18,34 +20,36 @@ import (
 	"time"
 )
 
-func OpenaiStreamHandler(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (*dto.OpenAIErrorWithStatusCode, *dto.Usage, string, int) {
+func OaiStreamHandler(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (*dto.OpenAIErrorWithStatusCode, *dto.Usage) {
-	//checkSensitive := constant.ShouldCheckCompletionSensitive()
+	containStreamUsage := false
+	var responseId string
+	var createAt int64 = 0
+	var systemFingerprint string
+	model := info.UpstreamModelName
+
 	var responseTextBuilder strings.Builder
-	var usage dto.Usage
+	var usage = &dto.Usage{}
+	var streamItems []string // store stream items
+
 	toolCount := 0
 	scanner := bufio.NewScanner(resp.Body)
-	scanner.Split(func(data []byte, atEOF bool) (advance int, token []byte, err error) {
+	scanner.Split(bufio.ScanLines)
-		if atEOF && len(data) == 0 {
+
-			return 0, nil, nil
+	service.SetEventStreamHeaders(c)
-		}
+
-		if i := strings.Index(string(data), "\n"); i >= 0 {
+	ticker := time.NewTicker(time.Duration(constant.StreamingTimeout) * time.Second)
-			return i + 1, data[0:i], nil
+	defer ticker.Stop()
-		}
+
-		if atEOF {
+	stopChan := make(chan bool)
-			return len(data), data, nil
-		}
-		return 0, nil, nil
-	})
-	dataChan := make(chan string, 5)
-	stopChan := make(chan bool, 2)
 	defer close(stopChan)
-	defer close(dataChan)
+	var (
-	var wg sync.WaitGroup
+		lastStreamData string
-	go func() {
+		mu             sync.Mutex
-		wg.Add(1)
+	)
-		defer wg.Done()
+	gopool.Go(func() {
-		var streamItems []string // store stream items
 		for scanner.Scan() {
+			info.SetFirstResponseTime()
+			ticker.Reset(time.Duration(constant.StreamingTimeout) * time.Second)
 			data := scanner.Text()
 			if len(data) < 6 { // ignore blank line or wrong format
 				continue
@@ -53,55 +57,67 @@ func OpenaiStreamHandler(c *gin.Context, resp *http.Response, info *relaycommon.
 			if data[:6] != "data: " && data[:6] != "[DONE]" {
 				continue
 			}
-			if !common.SafeSendStringTimeout(dataChan, data, constant.StreamingTimeout) {
+			mu.Lock()
-				// send data timeout, stop the stream
-				common.LogError(c, "send data timeout, stop the stream")
-				break
-			}
 			data = data[6:]
 			if !strings.HasPrefix(data, "[DONE]") {
-				streamItems = append(streamItems, data)
+				if lastStreamData != "" {
-			}
+					err := service.StringData(c, lastStreamData)
-		}
+					if err != nil {
-		// 计算token
+						common.LogError(c, "streaming error: "+err.Error())
-		streamResp := "[" + strings.Join(streamItems, ",") + "]"
-		switch info.RelayMode {
-		case relayconstant.RelayModeChatCompletions:
-			var streamResponses []dto.ChatCompletionsStreamResponseSimple
-			err := json.Unmarshal(common.StringToByteSlice(streamResp), &streamResponses)
-			if err != nil {
-				// 一次性解析失败，逐个解析
-				common.SysError("error unmarshalling stream response: " + err.Error())
-				for _, item := range streamItems {
-					var streamResponse dto.ChatCompletionsStreamResponseSimple
-					err := json.Unmarshal(common.StringToByteSlice(item), &streamResponse)
-					if err == nil {
-						if streamResponse.Usage != nil {
-							if streamResponse.Usage.TotalTokens != 0 {
-								usage = *streamResponse.Usage
-							}
-						}
-						for _, choice := range streamResponse.Choices {
-							responseTextBuilder.WriteString(choice.Delta.GetContentString())
-							if choice.Delta.ToolCalls != nil {
-								if len(choice.Delta.ToolCalls) > toolCount {
-									toolCount = len(choice.Delta.ToolCalls)
-								}
-								for _, tool := range choice.Delta.ToolCalls {
-									responseTextBuilder.WriteString(tool.Function.Name)
-									responseTextBuilder.WriteString(tool.Function.Arguments)
-								}
-							}
-						}
 					}
 				}
-			} else {
+				lastStreamData = data
-				for _, streamResponse := range streamResponses {
+				streamItems = append(streamItems, data)
-					if streamResponse.Usage != nil {
+			}
-						if streamResponse.Usage.TotalTokens != 0 {
+			mu.Unlock()
-							usage = *streamResponse.Usage
+		}
-						}
+		common.SafeSendBool(stopChan, true)
-					}
+	})
+
+	select {
+	case <-ticker.C:
+		// 超时处理逻辑
+		common.LogError(c, "streaming timeout")
+	case <-stopChan:
+		// 正常结束
+	}
+
+	shouldSendLastResp := true
+	var lastStreamResponse dto.ChatCompletionsStreamResponse
+	err := json.Unmarshal(common.StringToByteSlice(lastStreamData), &lastStreamResponse)
+	if err == nil {
+		responseId = lastStreamResponse.Id
+		createAt = lastStreamResponse.Created
+		systemFingerprint = lastStreamResponse.GetSystemFingerprint()
+		model = lastStreamResponse.Model
+		if service.ValidUsage(lastStreamResponse.Usage) {
+			containStreamUsage = true
+			usage = lastStreamResponse.Usage
+			if !info.ShouldIncludeUsage {
+				shouldSendLastResp = false
+			}
+		}
+	}
+	if shouldSendLastResp {
+		service.StringData(c, lastStreamData)
+	}
+
+	// 计算token
+	streamResp := "[" + strings.Join(streamItems, ",") + "]"
+	switch info.RelayMode {
+	case relayconstant.RelayModeChatCompletions:
+		var streamResponses []dto.ChatCompletionsStreamResponse
+		err := json.Unmarshal(common.StringToByteSlice(streamResp), &streamResponses)
+		if err != nil {
+			// 一次性解析失败，逐个解析
+			common.SysError("error unmarshalling stream response: " + err.Error())
+			for _, item := range streamItems {
+				var streamResponse dto.ChatCompletionsStreamResponse
+				err := json.Unmarshal(common.StringToByteSlice(item), &streamResponse)
+				if err == nil {
+					//if service.ValidUsage(streamResponse.Usage) {
+					//	usage = streamResponse.Usage
+					//}
 					for _, choice := range streamResponse.Choices {
 						responseTextBuilder.WriteString(choice.Delta.GetContentString())
 						if choice.Delta.ToolCalls != nil {
@@ -116,67 +132,65 @@ func OpenaiStreamHandler(c *gin.Context, resp *http.Response, info *relaycommon.
 					}
 				}
 			}
-		case relayconstant.RelayModeCompletions:
+		} else {
-			var streamResponses []dto.CompletionsStreamResponse
+			for _, streamResponse := range streamResponses {
-			err := json.Unmarshal(common.StringToByteSlice(streamResp), &streamResponses)
+				//if service.ValidUsage(streamResponse.Usage) {
-			if err != nil {
+				//	usage = streamResponse.Usage
-				// 一次性解析失败，逐个解析
+				//	containStreamUsage = true
-				common.SysError("error unmarshalling stream response: " + err.Error())
+				//}
-				for _, item := range streamItems {
+				for _, choice := range streamResponse.Choices {
-					var streamResponse dto.CompletionsStreamResponse
+					responseTextBuilder.WriteString(choice.Delta.GetContentString())
-					err := json.Unmarshal(common.StringToByteSlice(item), &streamResponse)
+					if choice.Delta.ToolCalls != nil {
-					if err == nil {
+						if len(choice.Delta.ToolCalls) > toolCount {
-						for _, choice := range streamResponse.Choices {
+							toolCount = len(choice.Delta.ToolCalls)
-							responseTextBuilder.WriteString(choice.Text)
+						}
+						for _, tool := range choice.Delta.ToolCalls {
+							responseTextBuilder.WriteString(tool.Function.Name)
+							responseTextBuilder.WriteString(tool.Function.Arguments)
 						}
 					}
 				}
-			} else {
+			}
-				for _, streamResponse := range streamResponses {
+		}
+	case relayconstant.RelayModeCompletions:
+		var streamResponses []dto.CompletionsStreamResponse
+		err := json.Unmarshal(common.StringToByteSlice(streamResp), &streamResponses)
+		if err != nil {
+			// 一次性解析失败，逐个解析
+			common.SysError("error unmarshalling stream response: " + err.Error())
+			for _, item := range streamItems {
+				var streamResponse dto.CompletionsStreamResponse
+				err := json.Unmarshal(common.StringToByteSlice(item), &streamResponse)
+				if err == nil {
 					for _, choice := range streamResponse.Choices {
 						responseTextBuilder.WriteString(choice.Text)
 					}
 				}
 			}
-		}
+		} else {
-		if len(dataChan) > 0 {
+			for _, streamResponse := range streamResponses {
-			// wait data out
+				for _, choice := range streamResponse.Choices {
-			time.Sleep(2 * time.Second)
+					responseTextBuilder.WriteString(choice.Text)
-		}
+				}
-		common.SafeSendBool(stopChan, true)
-	}()
-	service.SetEventStreamHeaders(c)
-	isFirst := true
-	ticker := time.NewTicker(time.Duration(constant.StreamingTimeout) * time.Second)
-	defer ticker.Stop()
-	c.Stream(func(w io.Writer) bool {
-		select {
-		case <-ticker.C:
-			common.LogError(c, "reading data from upstream timeout")
-			return false
-		case data := <-dataChan:
-			if isFirst {
-				isFirst = false
-				info.FirstResponseTime = time.Now()
 			}
-			ticker.Reset(time.Duration(constant.StreamingTimeout) * time.Second)
-			if strings.HasPrefix(data, "data: [DONE]") {
-				data = data[:12]
-			}
-			// some implementations may add \r at the end of data
-			data = strings.TrimSuffix(data, "\r")
-			c.Render(-1, common.CustomEvent{Data: data})
-			return true
-		case <-stopChan:
-			return false
 		}
-	})
-	err := resp.Body.Close()
-	if err != nil {
-		return service.OpenAIErrorWrapperLocal(err, "close_response_body_failed", http.StatusInternalServerError), nil, "", toolCount
 	}
-	wg.Wait()
+
-	return nil, &usage, responseTextBuilder.String(), toolCount
+	if !containStreamUsage {
+		usage, _ = service.ResponseText2Usage(responseTextBuilder.String(), info.UpstreamModelName, info.PromptTokens)
+		usage.CompletionTokens += toolCount * 7
+	}
+
+	if info.ShouldIncludeUsage && !containStreamUsage {
+		response := service.GenerateFinalUsageResponse(responseId, createAt, model, *usage)
+		response.SetSystemFingerprint(systemFingerprint)
+		service.ObjectData(c, response)
+	}
+
+	service.Done(c)
+
+	resp.Body.Close()
+	return nil, usage
 }
 
 func OpenaiHandler(c *gin.Context, resp *http.Response, promptTokens int, model string) (*dto.OpenAIErrorWithStatusCode, *dto.Usage) {
@@ -213,11 +227,7 @@ func OpenaiHandler(c *gin.Context, resp *http.Response, promptTokens int, model
 	if err != nil {
 		return service.OpenAIErrorWrapper(err, "copy_response_body_failed", http.StatusInternalServerError), nil
 	}
-	err = resp.Body.Close()
+	resp.Body.Close()
-	if err != nil {
-		return service.OpenAIErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), nil
-	}
-
 	if simpleResponse.Usage.TotalTokens == 0 || (simpleResponse.Usage.PromptTokens == 0 && simpleResponse.Usage.CompletionTokens == 0) {
 		completionTokens := 0
 		for _, choice := range simpleResponse.Choices {
@@ -232,3 +242,134 @@ func OpenaiHandler(c *gin.Context, resp *http.Response, promptTokens int, model
 	}
 	return nil, &simpleResponse.Usage
 }
+
+func OpenaiTTSHandler(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (*dto.OpenAIErrorWithStatusCode, *dto.Usage) {
+	responseBody, err := io.ReadAll(resp.Body)
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "read_response_body_failed", http.StatusInternalServerError), nil
+	}
+	err = resp.Body.Close()
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), nil
+	}
+	// Reset response body
+	resp.Body = io.NopCloser(bytes.NewBuffer(responseBody))
+	// We shouldn't set the header before we parse the response body, because the parse part may fail.
+	// And then we will have to send an error response, but in this case, the header has already been set.
+	// So the httpClient will be confused by the response.
+	// For example, Postman will report error, and we cannot check the response at all.
+	for k, v := range resp.Header {
+		c.Writer.Header().Set(k, v[0])
+	}
+	c.Writer.WriteHeader(resp.StatusCode)
+	_, err = io.Copy(c.Writer, resp.Body)
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "copy_response_body_failed", http.StatusInternalServerError), nil
+	}
+	err = resp.Body.Close()
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), nil
+	}
+
+	usage := &dto.Usage{}
+	usage.PromptTokens = info.PromptTokens
+	usage.TotalTokens = info.PromptTokens
+	return nil, usage
+}
+
+func OpenaiSTTHandler(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo, responseFormat string) (*dto.OpenAIErrorWithStatusCode, *dto.Usage) {
+	var audioResp dto.AudioResponse
+	responseBody, err := io.ReadAll(resp.Body)
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "read_response_body_failed", http.StatusInternalServerError), nil
+	}
+	err = resp.Body.Close()
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), nil
+	}
+	err = json.Unmarshal(responseBody, &audioResp)
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "unmarshal_response_body_failed", http.StatusInternalServerError), nil
+	}
+
+	// Reset response body
+	resp.Body = io.NopCloser(bytes.NewBuffer(responseBody))
+	// We shouldn't set the header before we parse the response body, because the parse part may fail.
+	// And then we will have to send an error response, but in this case, the header has already been set.
+	// So the httpClient will be confused by the response.
+	// For example, Postman will report error, and we cannot check the response at all.
+	for k, v := range resp.Header {
+		c.Writer.Header().Set(k, v[0])
+	}
+	c.Writer.WriteHeader(resp.StatusCode)
+	_, err = io.Copy(c.Writer, resp.Body)
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "copy_response_body_failed", http.StatusInternalServerError), nil
+	}
+	resp.Body.Close()
+
+	var text string
+	switch responseFormat {
+	case "json":
+		text, err = getTextFromJSON(responseBody)
+	case "text":
+		text, err = getTextFromText(responseBody)
+	case "srt":
+		text, err = getTextFromSRT(responseBody)
+	case "verbose_json":
+		text, err = getTextFromVerboseJSON(responseBody)
+	case "vtt":
+		text, err = getTextFromVTT(responseBody)
+	}
+
+	usage := &dto.Usage{}
+	usage.PromptTokens = info.PromptTokens
+	usage.CompletionTokens, _ = service.CountTokenText(text, info.UpstreamModelName)
+	usage.TotalTokens = usage.PromptTokens + usage.CompletionTokens
+	return nil, usage
+}
+
+func getTextFromVTT(body []byte) (string, error) {
+	return getTextFromSRT(body)
+}
+
+func getTextFromVerboseJSON(body []byte) (string, error) {
+	var whisperResponse dto.WhisperVerboseJSONResponse
+	if err := json.Unmarshal(body, &whisperResponse); err != nil {
+		return "", fmt.Errorf("unmarshal_response_body_failed err :%w", err)
+	}
+	return whisperResponse.Text, nil
+}
+
+func getTextFromSRT(body []byte) (string, error) {
+	scanner := bufio.NewScanner(strings.NewReader(string(body)))
+	var builder strings.Builder
+	var textLine bool
+	for scanner.Scan() {
+		line := scanner.Text()
+		if textLine {
+			builder.WriteString(line)
+			textLine = false
+			continue
+		} else if strings.Contains(line, "-->") {
+			textLine = true
+			continue
+		}
+	}
+	if err := scanner.Err(); err != nil {
+		return "", err
+	}
+	return builder.String(), nil
+}
+
+func getTextFromText(body []byte) (string, error) {
+	return strings.TrimSuffix(string(body), "\n"), nil
+}
+
+func getTextFromJSON(body []byte) (string, error) {
+	var whisperResponse dto.AudioResponse
+	if err := json.Unmarshal(body, &whisperResponse); err != nil {
+		return "", fmt.Errorf("unmarshal_response_body_failed err :%w", err)
+	}
+	return whisperResponse.Text, nil
+}

relay/channel/palm/adaptor.go

@@ -15,12 +15,17 @@ import (
 type Adaptor struct {
 }
 
-func (a *Adaptor) InitRerank(info *relaycommon.RelayInfo, request dto.RerankRequest) {
+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) Init(info *relaycommon.RelayInfo, request dto.GeneralOpenAIRequest) {
+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) {
@@ -33,7 +38,7 @@ func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, info *re
 	return nil
 }
 
-func (a *Adaptor) ConvertRequest(c *gin.Context, relayMode int, request *dto.GeneralOpenAIRequest) (any, error) {
+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")
 	}

relay/channel/perplexity/adaptor.go

@@ -10,18 +10,22 @@ import (
 	"one-api/relay/channel"
 	"one-api/relay/channel/openai"
 	relaycommon "one-api/relay/common"
-	"one-api/service"
 )
 
 type Adaptor struct {
 }
 
-func (a *Adaptor) InitRerank(info *relaycommon.RelayInfo, request dto.RerankRequest) {
+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) Init(info *relaycommon.RelayInfo, request dto.GeneralOpenAIRequest) {
+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) {
@@ -34,7 +38,7 @@ func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, info *re
 	return nil
 }
 
-func (a *Adaptor) ConvertRequest(c *gin.Context, relayMode int, request *dto.GeneralOpenAIRequest) (any, error) {
+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")
 	}
@@ -54,11 +58,7 @@ func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, request
 
 func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage *dto.Usage, err *dto.OpenAIErrorWithStatusCode) {
 	if info.IsStream {
-		var responseText string
+		err, usage = openai.OaiStreamHandler(c, resp, info)
-		err, usage, responseText, _ = openai.OpenaiStreamHandler(c, resp, info)
-		if usage == nil || usage.TotalTokens == 0 || (usage.PromptTokens+usage.CompletionTokens) == 0 {
-			usage, _ = service.ResponseText2Usage(responseText, info.UpstreamModelName, info.PromptTokens)
-		}
 	} else {
 		err, usage = openai.OpenaiHandler(c, resp, info.PromptTokens, info.UpstreamModelName)
 	}

relay/channel/siliconflow/adaptor.go

@@ -0,0 +1,83 @@
+package siliconflow
+
+import (
+	"errors"
+	"fmt"
+	"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"
+	"one-api/relay/constant"
+)
+
+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) {
+	if info.RelayMode == constant.RelayModeRerank {
+		return fmt.Sprintf("%s/v1/rerank", info.BaseUrl), nil
+	} else if info.RelayMode == constant.RelayModeEmbeddings {
+		return fmt.Sprintf("%s/v1/embeddings", info.BaseUrl), nil
+	} else if info.RelayMode == constant.RelayModeChatCompletions {
+		return fmt.Sprintf("%s/v1/chat/completions", info.BaseUrl), nil
+	}
+	return "", errors.New("invalid relay mode")
+}
+
+func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, info *relaycommon.RelayInfo) error {
+	channel.SetupApiRequestHeader(info, c, req)
+	req.Header.Set("Authorization", fmt.Sprintf("Bearer %s", info.ApiKey))
+	return nil
+}
+
+func (a *Adaptor) ConvertRequest(c *gin.Context, info *relaycommon.RelayInfo, request *dto.GeneralOpenAIRequest) (any, error) {
+	return request, nil
+}
+
+func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (*http.Response, error) {
+	return channel.DoApiRequest(a, c, info, requestBody)
+}
+
+func (a *Adaptor) ConvertRerankRequest(c *gin.Context, relayMode int, request dto.RerankRequest) (any, error) {
+	return request, nil
+}
+
+func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage *dto.Usage, err *dto.OpenAIErrorWithStatusCode) {
+	switch info.RelayMode {
+	case constant.RelayModeRerank:
+		err, usage = siliconflowRerankHandler(c, resp)
+	case constant.RelayModeChatCompletions:
+		if info.IsStream {
+			err, usage = openai.OaiStreamHandler(c, resp, info)
+		} else {
+			err, usage = openai.OpenaiHandler(c, resp, info.PromptTokens, info.UpstreamModelName)
+		}
+	case constant.RelayModeEmbeddings:
+		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/siliconflow/constant.go

@@ -0,0 +1,51 @@
+package siliconflow
+
+var ModelList = []string{
+	"THUDM/glm-4-9b-chat",
+	//"stabilityai/stable-diffusion-xl-base-1.0",
+	//"TencentARC/PhotoMaker",
+	"InstantX/InstantID",
+	//"stabilityai/stable-diffusion-2-1",
+	//"stabilityai/sd-turbo",
+	//"stabilityai/sdxl-turbo",
+	"ByteDance/SDXL-Lightning",
+	"deepseek-ai/deepseek-llm-67b-chat",
+	"Qwen/Qwen1.5-14B-Chat",
+	"Qwen/Qwen1.5-7B-Chat",
+	"Qwen/Qwen1.5-110B-Chat",
+	"Qwen/Qwen1.5-32B-Chat",
+	"01-ai/Yi-1.5-6B-Chat",
+	"01-ai/Yi-1.5-9B-Chat-16K",
+	"01-ai/Yi-1.5-34B-Chat-16K",
+	"THUDM/chatglm3-6b",
+	"deepseek-ai/DeepSeek-V2-Chat",
+	"Qwen/Qwen2-72B-Instruct",
+	"Qwen/Qwen2-7B-Instruct",
+	"Qwen/Qwen2-57B-A14B-Instruct",
+	//"stabilityai/stable-diffusion-3-medium",
+	"deepseek-ai/DeepSeek-Coder-V2-Instruct",
+	"Qwen/Qwen2-1.5B-Instruct",
+	"internlm/internlm2_5-7b-chat",
+	"BAAI/bge-large-en-v1.5",
+	"BAAI/bge-large-zh-v1.5",
+	"Pro/Qwen/Qwen2-7B-Instruct",
+	"Pro/Qwen/Qwen2-1.5B-Instruct",
+	"Pro/Qwen/Qwen1.5-7B-Chat",
+	"Pro/THUDM/glm-4-9b-chat",
+	"Pro/THUDM/chatglm3-6b",
+	"Pro/01-ai/Yi-1.5-9B-Chat-16K",
+	"Pro/01-ai/Yi-1.5-6B-Chat",
+	"Pro/google/gemma-2-9b-it",
+	"Pro/internlm/internlm2_5-7b-chat",
+	"Pro/meta-llama/Meta-Llama-3-8B-Instruct",
+	"Pro/mistralai/Mistral-7B-Instruct-v0.2",
+	"black-forest-labs/FLUX.1-schnell",
+	"iic/SenseVoiceSmall",
+	"netease-youdao/bce-embedding-base_v1",
+	"BAAI/bge-m3",
+	"internlm/internlm2_5-20b-chat",
+	"Qwen/Qwen2-Math-72B-Instruct",
+	"netease-youdao/bce-reranker-base_v1",
+	"BAAI/bge-reranker-v2-m3",
+}
+var ChannelName = "siliconflow"

relay/channel/siliconflow/dto.go

@@ -0,0 +1,17 @@
+package siliconflow
+
+import "one-api/dto"
+
+type SFTokens struct {
+	InputTokens  int `json:"input_tokens"`
+	OutputTokens int `json:"output_tokens"`
+}
+
+type SFMeta struct {
+	Tokens SFTokens `json:"tokens"`
+}
+
+type SFRerankResponse struct {
+	Results []dto.RerankResponseDocument `json:"results"`
+	Meta    SFMeta                       `json:"meta"`
+}

relay/channel/siliconflow/relay-siliconflow.go

@@ -0,0 +1,44 @@
+package siliconflow
+
+import (
+	"encoding/json"
+	"github.com/gin-gonic/gin"
+	"io"
+	"net/http"
+	"one-api/dto"
+	"one-api/service"
+)
+
+func siliconflowRerankHandler(c *gin.Context, resp *http.Response) (*dto.OpenAIErrorWithStatusCode, *dto.Usage) {
+	responseBody, err := io.ReadAll(resp.Body)
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "read_response_body_failed", http.StatusInternalServerError), nil
+	}
+	err = resp.Body.Close()
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), nil
+	}
+	var siliconflowResp SFRerankResponse
+	err = json.Unmarshal(responseBody, &siliconflowResp)
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "unmarshal_response_body_failed", http.StatusInternalServerError), nil
+	}
+	usage := &dto.Usage{
+		PromptTokens:     siliconflowResp.Meta.Tokens.InputTokens,
+		CompletionTokens: siliconflowResp.Meta.Tokens.OutputTokens,
+		TotalTokens:      siliconflowResp.Meta.Tokens.InputTokens + siliconflowResp.Meta.Tokens.OutputTokens,
+	}
+	rerankResp := &dto.RerankResponse{
+		Results: siliconflowResp.Results,
+		Usage:   *usage,
+	}
+
+	jsonResponse, err := json.Marshal(rerankResp)
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "marshal_response_body_failed", http.StatusInternalServerError), nil
+	}
+	c.Writer.Header().Set("Content-Type", "application/json")
+	c.Writer.WriteHeader(resp.StatusCode)
+	_, err = c.Writer.Write(jsonResponse)
+	return nil, usage
+}

relay/channel/tencent/adaptor.go

@@ -23,12 +23,17 @@ type Adaptor struct {
 	Timestamp int64
 }
 
-func (a *Adaptor) InitRerank(info *relaycommon.RelayInfo, request dto.RerankRequest) {
+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) Init(info *relaycommon.RelayInfo, request dto.GeneralOpenAIRequest) {
+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) {
 	a.Action = "ChatCompletions"
 	a.Version = "2023-09-01"
 	a.Timestamp = common.GetTimestamp()
@@ -47,7 +52,7 @@ func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, info *re
 	return nil
 }
 
-func (a *Adaptor) ConvertRequest(c *gin.Context, relayMode int, request *dto.GeneralOpenAIRequest) (any, error) {
+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")
 	}

relay/channel/vertex/adaptor.go

@@ -0,0 +1,184 @@
+package vertex
+
+import (
+	"encoding/json"
+	"errors"
+	"fmt"
+	"github.com/gin-gonic/gin"
+	"github.com/jinzhu/copier"
+	"io"
+	"net/http"
+	"one-api/dto"
+	"one-api/relay/channel"
+	"one-api/relay/channel/claude"
+	"one-api/relay/channel/gemini"
+	"one-api/relay/channel/openai"
+	relaycommon "one-api/relay/common"
+	"strings"
+)
+
+const (
+	RequestModeClaude = 1
+	RequestModeGemini = 2
+	RequestModeLlama  = 3
+)
+
+var claudeModelMap = map[string]string{
+	"claude-3-sonnet-20240229":   "claude-3-sonnet@20240229",
+	"claude-3-opus-20240229":     "claude-3-opus@20240229",
+	"claude-3-haiku-20240307":    "claude-3-haiku@20240307",
+	"claude-3-5-sonnet-20240620": "claude-3-5-sonnet@20240620",
+}
+
+const anthropicVersion = "vertex-2023-10-16"
+
+type Adaptor struct {
+	RequestMode        int
+	AccountCredentials Credentials
+}
+
+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) {
+	if strings.HasPrefix(info.UpstreamModelName, "claude") {
+		a.RequestMode = RequestModeClaude
+	} else if strings.HasPrefix(info.UpstreamModelName, "gemini") {
+		a.RequestMode = RequestModeGemini
+	} else if strings.Contains(info.UpstreamModelName, "llama") {
+		a.RequestMode = RequestModeLlama
+	}
+}
+
+func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {
+	adc := &Credentials{}
+	if err := json.Unmarshal([]byte(info.ApiKey), adc); err != nil {
+		return "", fmt.Errorf("failed to decode credentials file: %w", err)
+	}
+	region := GetModelRegion(info.ApiVersion, info.OriginModelName)
+	a.AccountCredentials = *adc
+	suffix := ""
+	if a.RequestMode == RequestModeGemini {
+		if info.IsStream {
+			suffix = "streamGenerateContent?alt=sse"
+		} else {
+			suffix = "generateContent"
+		}
+		return fmt.Sprintf(
+			"https://%s-aiplatform.googleapis.com/v1/projects/%s/locations/%s/publishers/google/models/%s:%s",
+			region,
+			adc.ProjectID,
+			region,
+			info.UpstreamModelName,
+			suffix,
+		), nil
+	} else if a.RequestMode == RequestModeClaude {
+		if info.IsStream {
+			suffix = "streamRawPredict?alt=sse"
+		} else {
+			suffix = "rawPredict"
+		}
+		if v, ok := claudeModelMap[info.UpstreamModelName]; ok {
+			info.UpstreamModelName = v
+		}
+		return fmt.Sprintf(
+			"https://%s-aiplatform.googleapis.com/v1/projects/%s/locations/%s/publishers/anthropic/models/%s:%s",
+			region,
+			adc.ProjectID,
+			region,
+			info.UpstreamModelName,
+			suffix,
+		), nil
+	} else if a.RequestMode == RequestModeLlama {
+		return fmt.Sprintf(
+			"https://%s-aiplatform.googleapis.com/v1beta1/projects/%s/locations/%s/endpoints/openapi/chat/completions",
+			region,
+			adc.ProjectID,
+			region,
+		), nil
+	}
+	return "", errors.New("unsupported request mode")
+}
+
+func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, 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)
+	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")
+	}
+	if a.RequestMode == RequestModeClaude {
+		claudeReq, err := claude.RequestOpenAI2ClaudeMessage(*request)
+		if err != nil {
+			return nil, err
+		}
+		vertexClaudeReq := &VertexAIClaudeRequest{
+			AnthropicVersion: anthropicVersion,
+		}
+		if err = copier.Copy(vertexClaudeReq, claudeReq); err != nil {
+			return nil, errors.New("failed to copy claude request")
+		}
+		c.Set("request_model", request.Model)
+		return vertexClaudeReq, nil
+	} else if a.RequestMode == RequestModeGemini {
+		geminiRequest := gemini.CovertGemini2OpenAI(*request)
+		c.Set("request_model", request.Model)
+		return geminiRequest, nil
+	} else if a.RequestMode == RequestModeLlama {
+		return request, nil
+	}
+	return nil, errors.New("unsupported request mode")
+}
+
+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) (*http.Response, 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) {
+	if info.IsStream {
+		switch a.RequestMode {
+		case RequestModeClaude:
+			err, usage = claude.ClaudeStreamHandler(c, resp, info, claude.RequestModeMessage)
+		case RequestModeGemini:
+			err, usage = gemini.GeminiChatStreamHandler(c, resp, info)
+		case RequestModeLlama:
+			err, usage = openai.OaiStreamHandler(c, resp, info)
+		}
+	} else {
+		switch a.RequestMode {
+		case RequestModeClaude:
+			err, usage = claude.ClaudeHandler(c, resp, claude.RequestModeMessage, info)
+		case RequestModeGemini:
+			err, usage = gemini.GeminiChatHandler(c, resp)
+		case RequestModeLlama:
+			err, usage = openai.OpenaiHandler(c, resp, info.PromptTokens, info.OriginModelName)
+		}
+	}
+	return
+}
+
+func (a *Adaptor) GetModelList() []string {
+	return ModelList
+}
+
+func (a *Adaptor) GetChannelName() string {
+	return ChannelName
+}

relay/channel/vertex/constants.go

@@ -0,0 +1,15 @@
+package vertex
+
+var ModelList = []string{
+	"claude-3-sonnet-20240229",
+	"claude-3-opus-20240229",
+	"claude-3-haiku-20240307",
+	"claude-3-5-sonnet-20240620",
+
+	//"gemini-1.5-pro-latest", "gemini-1.5-flash-latest",
+	"gemini-1.5-pro-001", "gemini-1.5-flash-001", "gemini-pro", "gemini-pro-vision",
+
+	"meta/llama3-405b-instruct-maas",
+}
+
+var ChannelName = "vertex-ai"

relay/channel/vertex/dto.go

@@ -0,0 +1,17 @@
+package vertex
+
+import "one-api/relay/channel/claude"
+
+type VertexAIClaudeRequest struct {
+	AnthropicVersion string                 `json:"anthropic_version"`
+	Messages         []claude.ClaudeMessage `json:"messages"`
+	System           string                 `json:"system,omitempty"`
+	MaxTokens        int                    `json:"max_tokens,omitempty"`
+	StopSequences    []string               `json:"stop_sequences,omitempty"`
+	Stream           bool                   `json:"stream,omitempty"`
+	Temperature      float64                `json:"temperature,omitempty"`
+	TopP             float64                `json:"top_p,omitempty"`
+	TopK             int                    `json:"top_k,omitempty"`
+	Tools            []claude.Tool          `json:"tools,omitempty"`
+	ToolChoice       any                    `json:"tool_choice,omitempty"`
+}

relay/channel/vertex/relay-vertex.go

@@ -0,0 +1,16 @@
+package vertex
+
+import "one-api/common"
+
+func GetModelRegion(other string, localModelName string) string {
+	// if other is json string
+	if common.IsJsonStr(other) {
+		m := common.StrToMap(other)
+		if m[localModelName] != nil {
+			return m[localModelName].(string)
+		} else {
+			return m["default"].(string)
+		}
+	}
+	return other
+}

relay/channel/vertex/service_account.go

@@ -0,0 +1,122 @@
+package vertex
+
+import (
+	"crypto/rsa"
+	"crypto/x509"
+	"encoding/json"
+	"encoding/pem"
+	"errors"
+	"github.com/bytedance/gopkg/cache/asynccache"
+	"github.com/golang-jwt/jwt"
+	"net/http"
+	"net/url"
+	relaycommon "one-api/relay/common"
+	"strings"
+
+	"fmt"
+	"time"
+)
+
+type Credentials struct {
+	ProjectID    string `json:"project_id"`
+	PrivateKeyID string `json:"private_key_id"`
+	PrivateKey   string `json:"private_key"`
+	ClientEmail  string `json:"client_email"`
+	ClientID     string `json:"client_id"`
+}
+
+var Cache = asynccache.NewAsyncCache(asynccache.Options{
+	RefreshDuration: time.Minute * 35,
+	EnableExpire:    true,
+	ExpireDuration:  time.Minute * 30,
+	Fetcher: func(key string) (interface{}, error) {
+		return nil, errors.New("not found")
+	},
+})
+
+func getAccessToken(a *Adaptor, info *relaycommon.RelayInfo) (string, error) {
+	cacheKey := fmt.Sprintf("access-token-%d", info.ChannelId)
+	val, err := Cache.Get(cacheKey)
+	if err == nil {
+		return val.(string), nil
+	}
+
+	signedJWT, err := createSignedJWT(a.AccountCredentials.ClientEmail, a.AccountCredentials.PrivateKey)
+	if err != nil {
+		return "", fmt.Errorf("failed to create signed JWT: %w", err)
+	}
+	newToken, err := exchangeJwtForAccessToken(signedJWT)
+	if err != nil {
+		return "", fmt.Errorf("failed to exchange JWT for access token: %w", err)
+	}
+	if err := Cache.SetDefault(cacheKey, newToken); err {
+		return newToken, nil
+	}
+	return newToken, nil
+}
+
+func createSignedJWT(email, privateKeyPEM string) (string, error) {
+
+	privateKeyPEM = strings.ReplaceAll(privateKeyPEM, "-----BEGIN PRIVATE KEY-----", "")
+	privateKeyPEM = strings.ReplaceAll(privateKeyPEM, "-----END PRIVATE KEY-----", "")
+	privateKeyPEM = strings.ReplaceAll(privateKeyPEM, "\r", "")
+	privateKeyPEM = strings.ReplaceAll(privateKeyPEM, "\n", "")
+	privateKeyPEM = strings.ReplaceAll(privateKeyPEM, "\\n", "")
+
+	block, _ := pem.Decode([]byte("-----BEGIN PRIVATE KEY-----\n" + privateKeyPEM + "\n-----END PRIVATE KEY-----"))
+	if block == nil {
+		return "", fmt.Errorf("failed to parse PEM block containing the private key")
+	}
+
+	privateKey, err := x509.ParsePKCS8PrivateKey(block.Bytes)
+	if err != nil {
+		return "", err
+	}
+
+	rsaPrivateKey, ok := privateKey.(*rsa.PrivateKey)
+	if !ok {
+		return "", fmt.Errorf("not an RSA private key")
+	}
+
+	now := time.Now()
+	claims := jwt.MapClaims{
+		"iss":   email,
+		"scope": "https://www.googleapis.com/auth/cloud-platform",
+		"aud":   "https://www.googleapis.com/oauth2/v4/token",
+		"exp":   now.Add(time.Minute * 35).Unix(),
+		"iat":   now.Unix(),
+	}
+
+	token := jwt.NewWithClaims(jwt.SigningMethodRS256, claims)
+	signedToken, err := token.SignedString(rsaPrivateKey)
+	if err != nil {
+		return "", err
+	}
+
+	return signedToken, nil
+}
+
+func exchangeJwtForAccessToken(signedJWT string) (string, error) {
+
+	authURL := "https://www.googleapis.com/oauth2/v4/token"
+	data := url.Values{}
+	data.Set("grant_type", "urn:ietf:params:oauth:grant-type:jwt-bearer")
+	data.Set("assertion", signedJWT)
+
+	resp, err := http.PostForm(authURL, data)
+	if err != nil {
+		return "", err
+	}
+	defer resp.Body.Close()
+
+	var result map[string]interface{}
+	if err := json.NewDecoder(resp.Body).Decode(&result); err != nil {
+		return "", err
+	}
+
+	if accessToken, ok := result["access_token"].(string); ok {
+		return accessToken, nil
+	}
+
+	return "", fmt.Errorf("failed to get access token: %v", result)
+}

relay/channel/xunfei/adaptor.go

@@ -16,12 +16,17 @@ type Adaptor struct {
 	request *dto.GeneralOpenAIRequest
 }
 
-func (a *Adaptor) InitRerank(info *relaycommon.RelayInfo, request dto.RerankRequest) {
+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) Init(info *relaycommon.RelayInfo, request dto.GeneralOpenAIRequest) {
+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) {
@@ -33,7 +38,7 @@ func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, info *re
 	return nil
 }
 
-func (a *Adaptor) ConvertRequest(c *gin.Context, relayMode int, request *dto.GeneralOpenAIRequest) (any, error) {
+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")
 	}

relay/channel/zhipu/adaptor.go

@@ -14,12 +14,17 @@ import (
 type Adaptor struct {
 }
 
-func (a *Adaptor) InitRerank(info *relaycommon.RelayInfo, request dto.RerankRequest) {
+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) Init(info *relaycommon.RelayInfo, request dto.GeneralOpenAIRequest) {
+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) {
@@ -37,7 +42,7 @@ func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, info *re
 	return nil
 }
 
-func (a *Adaptor) ConvertRequest(c *gin.Context, relayMode int, request *dto.GeneralOpenAIRequest) (any, error) {
+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")
 	}

relay/channel/zhipu/relay-zhipu.go

@@ -153,18 +153,7 @@ func streamMetaResponseZhipu2OpenAI(zhipuResponse *ZhipuStreamMetaResponse) (*dt
 func zhipuStreamHandler(c *gin.Context, resp *http.Response) (*dto.OpenAIErrorWithStatusCode, *dto.Usage) {
 	var usage *dto.Usage
 	scanner := bufio.NewScanner(resp.Body)
-	scanner.Split(func(data []byte, atEOF bool) (advance int, token []byte, err error) {
+	scanner.Split(bufio.ScanLines)
-		if atEOF && len(data) == 0 {
-			return 0, nil, nil
-		}
-		if i := strings.Index(string(data), "\n\n"); i >= 0 && strings.Index(string(data), ":") >= 0 {
-			return i + 2, data[0:i], nil
-		}
-		if atEOF {
-			return len(data), data, nil
-		}
-		return 0, nil, nil
-	})
 	dataChan := make(chan string)
 	metaChan := make(chan string)
 	stopChan := make(chan bool)

relay/channel/zhipu_4v/adaptor.go

@@ -10,18 +10,22 @@ import (
 	"one-api/relay/channel"
 	"one-api/relay/channel/openai"
 	relaycommon "one-api/relay/common"
-	"one-api/service"
 )
 
 type Adaptor struct {
 }
 
-func (a *Adaptor) InitRerank(info *relaycommon.RelayInfo, request dto.RerankRequest) {
+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) Init(info *relaycommon.RelayInfo, request dto.GeneralOpenAIRequest) {
+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) {
@@ -35,7 +39,7 @@ func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, info *re
 	return nil
 }
 
-func (a *Adaptor) ConvertRequest(c *gin.Context, relayMode int, request *dto.GeneralOpenAIRequest) (any, error) {
+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")
 	}
@@ -55,13 +59,7 @@ func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, request
 
 func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage *dto.Usage, err *dto.OpenAIErrorWithStatusCode) {
 	if info.IsStream {
-		var responseText string
+		err, usage = openai.OaiStreamHandler(c, resp, info)
-		var toolCount int
-		err, usage, responseText, toolCount = openai.OpenaiStreamHandler(c, resp, info)
-		if usage == nil || usage.TotalTokens == 0 || (usage.PromptTokens+usage.CompletionTokens) == 0 {
-			usage, _ = service.ResponseText2Usage(responseText, info.UpstreamModelName, info.PromptTokens)
-			usage.CompletionTokens += toolCount * 7
-		}
 	} else {
 		err, usage = openai.OpenaiHandler(c, resp, info.PromptTokens, info.UpstreamModelName)
 	}

relay/channel/zhipu_4v/constants.go

@@ -1,7 +1,7 @@
 package zhipu_4v
 
 var ModelList = []string{
-	"glm-4", "glm-4v", "glm-3-turbo",
+	"glm-4", "glm-4v", "glm-3-turbo", "glm-4-alltools", "glm-4-plus", "glm-4-0520", "glm-4-air", "glm-4-airx", "glm-4-long", "glm-4-flash", "glm-4v-plus",
 }
 
 var ChannelName = "zhipu_4v"

relay/common/relay_info.go

@@ -17,10 +17,12 @@ type RelayInfo struct {
 	TokenUnlimited       bool
 	StartTime            time.Time
 	FirstResponseTime    time.Time
+	setFirstResponse     bool
 	ApiType              int
 	IsStream             bool
 	RelayMode            int
 	UpstreamModelName    string
+	OriginModelName      string
 	RequestURLPath       string
 	ApiVersion           string
 	PromptTokens         int
@@ -32,7 +34,7 @@ type RelayInfo struct {
 }
 
 func GenRelayInfo(c *gin.Context) *RelayInfo {
-	channelType := c.GetInt("channel")
+	channelType := c.GetInt("channel_type")
 	channelId := c.GetInt("channel_id")
 
 	tokenId := c.GetInt("token_id")
@@ -56,6 +58,8 @@ func GenRelayInfo(c *gin.Context) *RelayInfo {
 		TokenUnlimited:    tokenUnlimited,
 		StartTime:         startTime,
 		FirstResponseTime: startTime.Add(-time.Second),
+		OriginModelName:   c.GetString("original_model"),
+		UpstreamModelName: c.GetString("original_model"),
 		ApiType:           apiType,
 		ApiVersion:        c.GetString("api_version"),
 		ApiKey:            strings.TrimPrefix(c.Request.Header.Get("Authorization"), "Bearer "),
@@ -67,8 +71,12 @@ func GenRelayInfo(c *gin.Context) *RelayInfo {
 	if info.ChannelType == common.ChannelTypeAzure {
 		info.ApiVersion = GetAPIVersion(c)
 	}
+	if info.ChannelType == common.ChannelTypeVertexAi {
+		info.ApiVersion = c.GetString("region")
+	}
 	if info.ChannelType == common.ChannelTypeOpenAI || info.ChannelType == common.ChannelTypeAnthropic ||
-		info.ChannelType == common.ChannelTypeAws || info.ChannelType == common.ChannelTypeGemini {
+		info.ChannelType == common.ChannelTypeAws || info.ChannelType == common.ChannelTypeGemini ||
+		info.ChannelType == common.ChannelCloudflare {
 		info.SupportStreamOptions = true
 	}
 	return info
@@ -82,6 +90,13 @@ func (info *RelayInfo) SetIsStream(isStream bool) {
 	info.IsStream = isStream
 }
 
+func (info *RelayInfo) SetFirstResponseTime() {
+	if !info.setFirstResponse {
+		info.FirstResponseTime = time.Now()
+		info.setFirstResponse = true
+	}
+}
+
 type TaskRelayInfo struct {
 	ChannelType       int
 	ChannelId         int
@@ -103,7 +118,7 @@ type TaskRelayInfo struct {
 }
 
 func GenTaskRelayInfo(c *gin.Context) *TaskRelayInfo {
-	channelType := c.GetInt("channel")
+	channelType := c.GetInt("channel_type")
 	channelId := c.GetInt("channel_id")
 
 	tokenId := c.GetInt("token_id")

relay/common/relay_utils.go

@@ -1,50 +1,17 @@
 package common
 
 import (
-	"encoding/json"
 	"fmt"
 	"github.com/gin-gonic/gin"
 	_ "image/gif"
 	_ "image/jpeg"
 	_ "image/png"
-	"io"
-	"net/http"
 	"one-api/common"
-	"one-api/dto"
-	"strconv"
 	"strings"
 )
 
 var StopFinishReason = "stop"
 
-func RelayErrorHandler(resp *http.Response) (OpenAIErrorWithStatusCode *dto.OpenAIErrorWithStatusCode) {
-	OpenAIErrorWithStatusCode = &dto.OpenAIErrorWithStatusCode{
-		StatusCode: resp.StatusCode,
-		Error: dto.OpenAIError{
-			Message: fmt.Sprintf("bad response status code %d", resp.StatusCode),
-			Type:    "upstream_error",
-			Code:    "bad_response_status_code",
-			Param:   strconv.Itoa(resp.StatusCode),
-		},
-	}
-	responseBody, err := io.ReadAll(resp.Body)
-	if err != nil {
-		return
-	}
-	err = resp.Body.Close()
-	if err != nil {
-		return
-	}
-	var textResponse dto.TextResponseWithError
-	err = json.Unmarshal(responseBody, &textResponse)
-	if err != nil {
-		OpenAIErrorWithStatusCode.Error.Message = fmt.Sprintf("error unmarshalling response body: %s", responseBody)
-		return
-	}
-	OpenAIErrorWithStatusCode.Error = textResponse.Error
-	return
-}
-
 func GetFullRequestURL(baseURL string, requestURL string, channelType int) string {
 	fullRequestURL := fmt.Sprintf("%s%s", baseURL, requestURL)
 

relay/constant/api_type.go

@@ -22,6 +22,9 @@ const (
 	APITypeCohere
 	APITypeDify
 	APITypeJina
+	APITypeCloudflare
+	APITypeSiliconFlow
+	APITypeVertexAi
 
 	APITypeDummy // this one is only for count, do not add any channel after this
 )
@@ -63,6 +66,12 @@ func ChannelType2APIType(channelType int) (int, bool) {
 		apiType = APITypeDify
 	case common.ChannelTypeJina:
 		apiType = APITypeJina
+	case common.ChannelCloudflare:
+		apiType = APITypeCloudflare
+	case common.ChannelTypeSiliconFlow:
+		apiType = APITypeSiliconFlow
+	case common.ChannelTypeVertexAi:
+		apiType = APITypeVertexAi
 	}
 	if apiType == -1 {
 		return APITypeOpenAI, false

relay/constant/relay_mode.go

@@ -13,6 +13,7 @@ const (
 	RelayModeModerations
 	RelayModeImagesGenerations
 	RelayModeEdits
+
 	RelayModeMidjourneyImagine
 	RelayModeMidjourneyDescribe
 	RelayModeMidjourneyBlend
@@ -22,16 +23,20 @@ const (
 	RelayModeMidjourneyTaskFetch
 	RelayModeMidjourneyTaskImageSeed
 	RelayModeMidjourneyTaskFetchByCondition
-	RelayModeAudioSpeech
-	RelayModeAudioTranscription
-	RelayModeAudioTranslation
 	RelayModeMidjourneyAction
 	RelayModeMidjourneyModal
 	RelayModeMidjourneyShorten
 	RelayModeSwapFace
+	RelayModeMidjourneyUpload
+
+	RelayModeAudioSpeech        // tts
+	RelayModeAudioTranscription // whisper
+	RelayModeAudioTranslation   // whisper
+
 	RelayModeSunoFetch
 	RelayModeSunoFetchByID
 	RelayModeSunoSubmit
+
 	RelayModeRerank
 )
 
@@ -77,6 +82,9 @@ func Path2RelayModeMidjourney(path string) int {
 	} else if strings.HasSuffix(path, "/mj/insight-face/swap") {
 		// midjourney plus
 		relayMode = RelayModeSwapFace
+	} else if strings.HasSuffix(path, "/submit/upload-discord-images") {
+		// midjourney plus
+		relayMode = RelayModeMidjourneyUpload
 	} else if strings.HasSuffix(path, "/mj/submit/imagine") {
 		relayMode = RelayModeMidjourneyImagine
 	} else if strings.HasSuffix(path, "/mj/submit/blend") {

relay/relay-audio.go

@@ -1,13 +1,10 @@
 package relay
 
 import (
-	"bytes"
-	"context"
 	"encoding/json"
 	"errors"
 	"fmt"
 	"github.com/gin-gonic/gin"
-	"io"
 	"net/http"
 	"one-api/common"
 	"one-api/constant"
@@ -16,69 +13,71 @@ import (
 	relaycommon "one-api/relay/common"
 	relayconstant "one-api/relay/constant"
 	"one-api/service"
-	"strings"
-	"time"
 )
 
-func AudioHelper(c *gin.Context, relayMode int) *dto.OpenAIErrorWithStatusCode {
+func getAndValidAudioRequest(c *gin.Context, info *relaycommon.RelayInfo) (*dto.AudioRequest, error) {
-	tokenId := c.GetInt("token_id")
+	audioRequest := &dto.AudioRequest{}
-	channelType := c.GetInt("channel")
+	err := common.UnmarshalBodyReusable(c, audioRequest)
-	channelId := c.GetInt("channel_id")
+	if err != nil {
-	userId := c.GetInt("id")
+		return nil, err
-	group := c.GetString("group")
-	startTime := time.Now()
-
-	var audioRequest dto.TextToSpeechRequest
-	if !strings.HasPrefix(c.Request.URL.Path, "/v1/audio/transcriptions") {
-		err := common.UnmarshalBodyReusable(c, &audioRequest)
-		if err != nil {
-			return service.OpenAIErrorWrapper(err, "bind_request_body_failed", http.StatusBadRequest)
-		}
-	} else {
-		audioRequest = dto.TextToSpeechRequest{
-			Model: "whisper-1",
-		}
 	}
-	//err := common.UnmarshalBodyReusable(c, &audioRequest)
+	switch info.RelayMode {
-
+	case relayconstant.RelayModeAudioSpeech:
-	// request validation
+		if audioRequest.Model == "" {
-	if audioRequest.Model == "" {
+			return nil, errors.New("model is required")
-		return service.OpenAIErrorWrapper(errors.New("model is required"), "required_field_missing", http.StatusBadRequest)
-	}
-
-	if strings.HasPrefix(audioRequest.Model, "tts-1") {
-		if audioRequest.Voice == "" {
-			return service.OpenAIErrorWrapper(errors.New("voice is required"), "required_field_missing", http.StatusBadRequest)
 		}
-	}
-	var err error
-	promptTokens := 0
-	preConsumedTokens := common.PreConsumedQuota
-	if strings.HasPrefix(audioRequest.Model, "tts-1") {
 		if constant.ShouldCheckPromptSensitive() {
-			err = service.CheckSensitiveInput(audioRequest.Input)
+			err := service.CheckSensitiveInput(audioRequest.Input)
 			if err != nil {
-				return service.OpenAIErrorWrapper(err, "sensitive_words_detected", http.StatusBadRequest)
+				return nil, err
 			}
 		}
+	default:
+		if audioRequest.Model == "" {
+			audioRequest.Model = c.PostForm("model")
+		}
+		if audioRequest.Model == "" {
+			return nil, errors.New("model is required")
+		}
+		if audioRequest.ResponseFormat == "" {
+			audioRequest.ResponseFormat = "json"
+		}
+	}
+	return audioRequest, nil
+}
+
+func AudioHelper(c *gin.Context) *dto.OpenAIErrorWithStatusCode {
+	relayInfo := relaycommon.GenRelayInfo(c)
+	audioRequest, err := getAndValidAudioRequest(c, relayInfo)
+
+	if err != nil {
+		common.LogError(c, fmt.Sprintf("getAndValidAudioRequest failed: %s", err.Error()))
+		return service.OpenAIErrorWrapper(err, "invalid_audio_request", http.StatusBadRequest)
+	}
+
+	promptTokens := 0
+	preConsumedTokens := common.PreConsumedQuota
+	if relayInfo.RelayMode == relayconstant.RelayModeAudioSpeech {
 		promptTokens, err = service.CountAudioToken(audioRequest.Input, audioRequest.Model)
 		if err != nil {
 			return service.OpenAIErrorWrapper(err, "count_audio_token_failed", http.StatusInternalServerError)
 		}
 		preConsumedTokens = promptTokens
+		relayInfo.PromptTokens = promptTokens
 	}
+
 	modelRatio := common.GetModelRatio(audioRequest.Model)
-	groupRatio := common.GetGroupRatio(group)
+	groupRatio := common.GetGroupRatio(relayInfo.Group)
 	ratio := modelRatio * groupRatio
 	preConsumedQuota := int(float64(preConsumedTokens) * ratio)
-	userQuota, err := model.CacheGetUserQuota(userId)
+	userQuota, err := model.CacheGetUserQuota(relayInfo.UserId)
 	if err != nil {
 		return service.OpenAIErrorWrapperLocal(err, "get_user_quota_failed", http.StatusInternalServerError)
 	}
 	if userQuota-preConsumedQuota < 0 {
-		return service.OpenAIErrorWrapperLocal(errors.New("user quota is not enough"), "insufficient_user_quota", http.StatusForbidden)
+		return service.OpenAIErrorWrapperLocal(errors.New(fmt.Sprintf("audio pre-consumed quota failed, user quota: %d, need quota: %d", userQuota, preConsumedQuota)), "insufficient_user_quota", http.StatusBadRequest)
 	}
-	err = model.CacheDecreaseUserQuota(userId, preConsumedQuota)
+	err = model.CacheDecreaseUserQuota(relayInfo.UserId, preConsumedQuota)
 	if err != nil {
 		return service.OpenAIErrorWrapperLocal(err, "decrease_user_quota_failed", http.StatusInternalServerError)
 	}
@@ -88,28 +87,12 @@ func AudioHelper(c *gin.Context, relayMode int) *dto.OpenAIErrorWithStatusCode {
 		preConsumedQuota = 0
 	}
 	if preConsumedQuota > 0 {
-		userQuota, err = model.PreConsumeTokenQuota(tokenId, preConsumedQuota)
+		userQuota, err = model.PreConsumeTokenQuota(relayInfo.TokenId, preConsumedQuota)
 		if err != nil {
 			return service.OpenAIErrorWrapperLocal(err, "pre_consume_token_quota_failed", http.StatusForbidden)
 		}
 	}
 
-	succeed := false
-	defer func() {
-		if succeed {
-			return
-		}
-		if preConsumedQuota > 0 {
-			// we need to roll back the pre-consumed quota
-			defer func() {
-				go func() {
-					// negative means add quota back for token & user
-					returnPreConsumedQuota(c, tokenId, userQuota, preConsumedQuota)
-				}()
-			}()
-		}
-	}()
-
 	// map model name
 	modelMapping := c.GetString("model_mapping")
 	if modelMapping != "" {
@@ -122,133 +105,44 @@ func AudioHelper(c *gin.Context, relayMode int) *dto.OpenAIErrorWithStatusCode {
 			audioRequest.Model = modelMap[audioRequest.Model]
 		}
 	}
+	relayInfo.UpstreamModelName = audioRequest.Model
 
-	baseURL := common.ChannelBaseURLs[channelType]
+	adaptor := GetAdaptor(relayInfo.ApiType)
-	requestURL := c.Request.URL.String()
+	if adaptor == nil {
-	if c.GetString("base_url") != "" {
+		return service.OpenAIErrorWrapperLocal(fmt.Errorf("invalid api type: %d", relayInfo.ApiType), "invalid_api_type", http.StatusBadRequest)
-		baseURL = c.GetString("base_url")
 	}
+	adaptor.Init(relayInfo)
 
-	fullRequestURL := relaycommon.GetFullRequestURL(baseURL, requestURL, channelType)
+	ioReader, err := adaptor.ConvertAudioRequest(c, relayInfo, *audioRequest)
-	if relayMode == relayconstant.RelayModeAudioTranscription && channelType == common.ChannelTypeAzure {
-		// https://learn.microsoft.com/en-us/azure/ai-services/openai/whisper-quickstart?tabs=command-line#rest-api
-		apiVersion := relaycommon.GetAPIVersion(c)
-		fullRequestURL = fmt.Sprintf("%s/openai/deployments/%s/audio/transcriptions?api-version=%s", baseURL, audioRequest.Model, apiVersion)
-	}
-
-	requestBody := c.Request.Body
-
-	req, err := http.NewRequest(c.Request.Method, fullRequestURL, requestBody)
 	if err != nil {
-		return service.OpenAIErrorWrapper(err, "new_request_failed", http.StatusInternalServerError)
+		return service.OpenAIErrorWrapperLocal(err, "convert_request_failed", http.StatusInternalServerError)
 	}
 
-	if relayMode == relayconstant.RelayModeAudioTranscription && channelType == common.ChannelTypeAzure {
+	resp, err := adaptor.DoRequest(c, relayInfo, ioReader)
-		// https://learn.microsoft.com/en-us/azure/ai-services/openai/whisper-quickstart?tabs=command-line#rest-api
-		apiKey := c.Request.Header.Get("Authorization")
-		apiKey = strings.TrimPrefix(apiKey, "Bearer ")
-		req.Header.Set("api-key", apiKey)
-		req.ContentLength = c.Request.ContentLength
-	} else {
-		req.Header.Set("Authorization", c.Request.Header.Get("Authorization"))
-	}
-
-	req.Header.Set("Content-Type", c.Request.Header.Get("Content-Type"))
-	req.Header.Set("Accept", c.Request.Header.Get("Accept"))
-
-	resp, err := service.GetHttpClient().Do(req)
 	if err != nil {
 		return service.OpenAIErrorWrapper(err, "do_request_failed", http.StatusInternalServerError)
 	}
 
-	err = req.Body.Close()
+	statusCodeMappingStr := c.GetString("status_code_mapping")
-	if err != nil {
+	if resp != nil {
-		return service.OpenAIErrorWrapper(err, "close_request_body_failed", http.StatusInternalServerError)
+		if resp.StatusCode != http.StatusOK {
-	}
+			returnPreConsumedQuota(c, relayInfo.TokenId, userQuota, preConsumedQuota)
-	err = c.Request.Body.Close()
+			openaiErr := service.RelayErrorHandler(resp)
-	if err != nil {
+			// reset status code 重置状态码
-		return service.OpenAIErrorWrapper(err, "close_request_body_failed", http.StatusInternalServerError)
+			service.ResetStatusCode(openaiErr, statusCodeMappingStr)
-	}
+			return openaiErr
-
-	if resp.StatusCode != http.StatusOK {
-		return relaycommon.RelayErrorHandler(resp)
-	}
-	succeed = true
-
-	var audioResponse dto.AudioResponse
-
-	defer func(ctx context.Context) {
-		go func() {
-			useTimeSeconds := time.Now().Unix() - startTime.Unix()
-			quota := 0
-			if strings.HasPrefix(audioRequest.Model, "tts-1") {
-				quota = promptTokens
-			} else {
-				quota, err = service.CountAudioToken(audioResponse.Text, audioRequest.Model)
-			}
-			quota = int(float64(quota) * ratio)
-			if ratio != 0 && quota <= 0 {
-				quota = 1
-			}
-			quotaDelta := quota - preConsumedQuota
-			err := model.PostConsumeTokenQuota(tokenId, userQuota, quotaDelta, preConsumedQuota, true)
-			if err != nil {
-				common.SysError("error consuming token remain quota: " + err.Error())
-			}
-			err = model.CacheUpdateUserQuota(userId)
-			if err != nil {
-				common.SysError("error update user quota cache: " + err.Error())
-			}
-			if quota != 0 {
-				tokenName := c.GetString("token_name")
-				logContent := fmt.Sprintf("模型倍率 %.2f，分组倍率 %.2f", modelRatio, groupRatio)
-				other := make(map[string]interface{})
-				other["model_ratio"] = modelRatio
-				other["group_ratio"] = groupRatio
-				model.RecordConsumeLog(ctx, userId, channelId, promptTokens, 0, audioRequest.Model, tokenName, quota, logContent, tokenId, userQuota, int(useTimeSeconds), false, other)
-				model.UpdateUserUsedQuotaAndRequestCount(userId, quota)
-				channelId := c.GetInt("channel_id")
-				model.UpdateChannelUsedQuota(channelId, quota)
-			}
-		}()
-	}(c.Request.Context())
-
-	responseBody, err := io.ReadAll(resp.Body)
-
-	if err != nil {
-		return service.OpenAIErrorWrapper(err, "read_response_body_failed", http.StatusInternalServerError)
-	}
-	err = resp.Body.Close()
-	if err != nil {
-		return service.OpenAIErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError)
-	}
-	if strings.HasPrefix(audioRequest.Model, "tts-1") {
-
-	} else {
-		err = json.Unmarshal(responseBody, &audioResponse)
-		if err != nil {
-			return service.OpenAIErrorWrapper(err, "unmarshal_response_body_failed", http.StatusInternalServerError)
-		}
-		contains, words := service.SensitiveWordContains(audioResponse.Text)
-		if contains {
-			return service.OpenAIErrorWrapper(errors.New("response contains sensitive words: "+strings.Join(words, ", ")), "response_contains_sensitive_words", http.StatusBadRequest)
 		}
 	}
 
-	resp.Body = io.NopCloser(bytes.NewBuffer(responseBody))
+	usage, openaiErr := adaptor.DoResponse(c, resp, relayInfo)
+	if openaiErr != nil {
+		returnPreConsumedQuota(c, relayInfo.TokenId, userQuota, preConsumedQuota)
+		// reset status code 重置状态码
+		service.ResetStatusCode(openaiErr, statusCodeMappingStr)
+		return openaiErr
+	}
 
-	for k, v := range resp.Header {
+	postConsumeQuota(c, relayInfo, audioRequest.Model, usage, ratio, preConsumedQuota, userQuota, modelRatio, groupRatio, 0, false, "")
-		c.Writer.Header().Set(k, v[0])
-	}
-	c.Writer.WriteHeader(resp.StatusCode)
 
-	_, err = io.Copy(c.Writer, resp.Body)
-	if err != nil {
-		return service.OpenAIErrorWrapper(err, "copy_response_body_failed", http.StatusInternalServerError)
-	}
-	err = resp.Body.Close()
-	if err != nil {
-		return service.OpenAIErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError)
-	}
 	return nil
 }

relay/relay-image.go

@@ -2,7 +2,6 @@ package relay
 
 import (
 	"bytes"
-	"context"
 	"encoding/json"
 	"errors"
 	"fmt"
@@ -14,72 +13,72 @@ import (
 	"one-api/dto"
 	"one-api/model"
 	relaycommon "one-api/relay/common"
-	relayconstant "one-api/relay/constant"
 	"one-api/service"
 	"strings"
-	"time"
 )
 
-func RelayImageHelper(c *gin.Context, relayMode int) *dto.OpenAIErrorWithStatusCode {
+func getAndValidImageRequest(c *gin.Context, info *relaycommon.RelayInfo) (*dto.ImageRequest, error) {
-	tokenId := c.GetInt("token_id")
+	imageRequest := &dto.ImageRequest{}
-	channelType := c.GetInt("channel")
+	err := common.UnmarshalBodyReusable(c, imageRequest)
-	channelId := c.GetInt("channel_id")
-	userId := c.GetInt("id")
-	group := c.GetString("group")
-	startTime := time.Now()
-
-	var imageRequest dto.ImageRequest
-	err := common.UnmarshalBodyReusable(c, &imageRequest)
 	if err != nil {
-		return service.OpenAIErrorWrapper(err, "bind_request_body_failed", http.StatusBadRequest)
+		return nil, err
 	}
-
+	if imageRequest.Prompt == "" {
-	if imageRequest.Model == "" {
+		return nil, errors.New("prompt is required")
-		imageRequest.Model = "dall-e-3"
 	}
-	if imageRequest.Size == "" {
+	if strings.Contains(imageRequest.Size, "×") {
-		imageRequest.Size = "1024x1024"
+		return nil, errors.New("size an unexpected error occurred in the parameter, please use 'x' instead of the multiplication sign '×'")
 	}
 	if imageRequest.N == 0 {
 		imageRequest.N = 1
 	}
-	// Prompt validation
+	if imageRequest.Size == "" {
-	if imageRequest.Prompt == "" {
+		imageRequest.Size = "1024x1024"
-		return service.OpenAIErrorWrapper(errors.New("prompt is required"), "required_field_missing", http.StatusBadRequest)
+	}
+	if imageRequest.Model == "" {
+		imageRequest.Model = "dall-e-2"
 	}
 
-	if constant.ShouldCheckPromptSensitive() {
-		err = service.CheckSensitiveInput(imageRequest.Prompt)
-		if err != nil {
-			return service.OpenAIErrorWrapper(err, "sensitive_words_detected", http.StatusBadRequest)
-		}
-	}
-
-	if strings.Contains(imageRequest.Size, "×") {
-		return service.OpenAIErrorWrapper(errors.New("size an unexpected error occurred in the parameter, please use 'x' instead of the multiplication sign '×'"), "invalid_field_value", http.StatusBadRequest)
-	}
 	// Not "256x256", "512x512", or "1024x1024"
 	if imageRequest.Model == "dall-e-2" || imageRequest.Model == "dall-e" {
 		if imageRequest.Size != "" && imageRequest.Size != "256x256" && imageRequest.Size != "512x512" && imageRequest.Size != "1024x1024" {
-			return service.OpenAIErrorWrapper(errors.New("size must be one of 256x256, 512x512, or 1024x1024, dall-e-3 1024x1792 or 1792x1024"), "invalid_field_value", http.StatusBadRequest)
+			return nil, errors.New("size must be one of 256x256, 512x512, or 1024x1024, dall-e-3 1024x1792 or 1792x1024")
 		}
 	} else if imageRequest.Model == "dall-e-3" {
 		if imageRequest.Size != "" && imageRequest.Size != "1024x1024" && imageRequest.Size != "1024x1792" && imageRequest.Size != "1792x1024" {
-			return service.OpenAIErrorWrapper(errors.New("size must be one of 256x256, 512x512, or 1024x1024, dall-e-3 1024x1792 or 1792x1024"), "invalid_field_value", http.StatusBadRequest)
+			return nil, errors.New("size must be one of 256x256, 512x512, or 1024x1024, dall-e-3 1024x1792 or 1792x1024")
 		}
-		if imageRequest.N != 1 {
+		if imageRequest.Quality == "" {
-			return service.OpenAIErrorWrapper(errors.New("n must be 1"), "invalid_field_value", http.StatusBadRequest)
+			imageRequest.Quality = "standard"
+		}
+		//if imageRequest.N != 1 {
+		//	return nil, errors.New("n must be 1")
+		//}
+	}
+	// N should between 1 and 10
+	//if imageRequest.N != 0 && (imageRequest.N < 1 || imageRequest.N > 10) {
+	//	return service.OpenAIErrorWrapper(errors.New("n must be between 1 and 10"), "invalid_field_value", http.StatusBadRequest)
+	//}
+	if constant.ShouldCheckPromptSensitive() {
+		err := service.CheckSensitiveInput(imageRequest.Prompt)
+		if err != nil {
+			return nil, err
 		}
 	}
+	return imageRequest, nil
+}
 
-	// N should between 1 and 10
+func ImageHelper(c *gin.Context, relayMode int) *dto.OpenAIErrorWithStatusCode {
-	if imageRequest.N != 0 && (imageRequest.N < 1 || imageRequest.N > 10) {
+	relayInfo := relaycommon.GenRelayInfo(c)
-		return service.OpenAIErrorWrapper(errors.New("n must be between 1 and 10"), "invalid_field_value", http.StatusBadRequest)
+
+	imageRequest, err := getAndValidImageRequest(c, relayInfo)
+	if err != nil {
+		common.LogError(c, fmt.Sprintf("getAndValidImageRequest failed: %s", err.Error()))
+		return service.OpenAIErrorWrapper(err, "invalid_image_request", http.StatusBadRequest)
 	}
 
 	// map model name
 	modelMapping := c.GetString("model_mapping")
-	isModelMapped := false
 	if modelMapping != "" {
 		modelMap := make(map[string]string)
 		err := json.Unmarshal([]byte(modelMapping), &modelMap)
@@ -88,31 +87,9 @@ func RelayImageHelper(c *gin.Context, relayMode int) *dto.OpenAIErrorWithStatusC
 		}
 		if modelMap[imageRequest.Model] != "" {
 			imageRequest.Model = modelMap[imageRequest.Model]
-			isModelMapped = true
 		}
 	}
-	baseURL := common.ChannelBaseURLs[channelType]
+	relayInfo.UpstreamModelName = imageRequest.Model
-	requestURL := c.Request.URL.String()
-	if c.GetString("base_url") != "" {
-		baseURL = c.GetString("base_url")
-	}
-	fullRequestURL := relaycommon.GetFullRequestURL(baseURL, requestURL, channelType)
-	if channelType == common.ChannelTypeAzure && relayMode == relayconstant.RelayModeImagesGenerations {
-		// https://learn.microsoft.com/en-us/azure/ai-services/openai/dall-e-quickstart?tabs=dalle3%2Ccommand-line&pivots=rest-api
-		apiVersion := relaycommon.GetAPIVersion(c)
-		// https://{resource_name}.openai.azure.com/openai/deployments/dall-e-3/images/generations?api-version=2023-06-01-preview
-		fullRequestURL = fmt.Sprintf("%s/openai/deployments/%s/images/generations?api-version=%s", baseURL, imageRequest.Model, apiVersion)
-	}
-	var requestBody io.Reader
-	if isModelMapped || channelType == common.ChannelTypeAzure { // make Azure channel request body
-		jsonStr, err := json.Marshal(imageRequest)
-		if err != nil {
-			return service.OpenAIErrorWrapper(err, "marshal_text_request_failed", http.StatusInternalServerError)
-		}
-		requestBody = bytes.NewBuffer(jsonStr)
-	} else {
-		requestBody = c.Request.Body
-	}
 
 	modelPrice, success := common.GetModelPrice(imageRequest.Model, true)
 	if !success {
@@ -121,8 +98,9 @@ func RelayImageHelper(c *gin.Context, relayMode int) *dto.OpenAIErrorWithStatusC
 		// per 1 modelRatio = $0.04 / 16
 		modelPrice = 0.0025 * modelRatio
 	}
-	groupRatio := common.GetGroupRatio(group)
+
-	userQuota, err := model.CacheGetUserQuota(userId)
+	groupRatio := common.GetGroupRatio(relayInfo.Group)
+	userQuota, err := model.CacheGetUserQuota(relayInfo.UserId)
 
 	sizeRatio := 1.0
 	// Size
@@ -144,104 +122,67 @@ func RelayImageHelper(c *gin.Context, relayMode int) *dto.OpenAIErrorWithStatusC
 		}
 	}
 
-	quota := int(modelPrice*groupRatio*common.QuotaPerUnit*sizeRatio*qualityRatio) * imageRequest.N
+	imageRatio := modelPrice * sizeRatio * qualityRatio * float64(imageRequest.N)
+	quota := int(imageRatio * groupRatio * common.QuotaPerUnit)
 
 	if userQuota-quota < 0 {
-		return service.OpenAIErrorWrapperLocal(errors.New("user quota is not enough"), "insufficient_user_quota", http.StatusForbidden)
+		return service.OpenAIErrorWrapperLocal(errors.New(fmt.Sprintf("image pre-consumed quota failed, user quota: %d, need quota: %d", userQuota, quota)), "insufficient_user_quota", http.StatusBadRequest)
 	}
 
-	req, err := http.NewRequest(c.Request.Method, fullRequestURL, requestBody)
+	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
+
+	convertedRequest, err := adaptor.ConvertImageRequest(c, relayInfo, *imageRequest)
 	if err != nil {
-		return service.OpenAIErrorWrapper(err, "new_request_failed", http.StatusInternalServerError)
+		return service.OpenAIErrorWrapperLocal(err, "convert_request_failed", http.StatusInternalServerError)
 	}
 
-	token := c.Request.Header.Get("Authorization")
+	jsonData, err := json.Marshal(convertedRequest)
-	if channelType == common.ChannelTypeAzure { // Azure authentication
+	if err != nil {
-		token = strings.TrimPrefix(token, "Bearer ")
+		return service.OpenAIErrorWrapperLocal(err, "json_marshal_failed", http.StatusInternalServerError)
-		req.Header.Set("api-key", token)
-	} else {
-		req.Header.Set("Authorization", token)
 	}
-	req.Header.Set("Content-Type", c.Request.Header.Get("Content-Type"))
+	requestBody = bytes.NewBuffer(jsonData)
-	req.Header.Set("Accept", c.Request.Header.Get("Accept"))
 
-	resp, err := service.GetHttpClient().Do(req)
+	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)
 	}
 
-	err = req.Body.Close()
+	if resp != nil {
-	if err != nil {
+		relayInfo.IsStream = relayInfo.IsStream || strings.HasPrefix(resp.Header.Get("Content-Type"), "text/event-stream")
-		return service.OpenAIErrorWrapper(err, "close_request_body_failed", http.StatusInternalServerError)
-	}
-	err = c.Request.Body.Close()
-	if err != nil {
-		return service.OpenAIErrorWrapper(err, "close_request_body_failed", http.StatusInternalServerError)
-	}
-
-	if resp.StatusCode != http.StatusOK {
-		return relaycommon.RelayErrorHandler(resp)
-	}
-
-	var textResponse dto.ImageResponse
-	defer func(ctx context.Context) {
-		useTimeSeconds := time.Now().Unix() - startTime.Unix()
 		if resp.StatusCode != http.StatusOK {
-			return
+			openaiErr := service.RelayErrorHandler(resp)
+			// reset status code 重置状态码
+			service.ResetStatusCode(openaiErr, statusCodeMappingStr)
+			return openaiErr
 		}
-		err := model.PostConsumeTokenQuota(tokenId, userQuota, quota, 0, true)
-		if err != nil {
-			common.SysError("error consuming token remain quota: " + err.Error())
-		}
-		err = model.CacheUpdateUserQuota(userId)
-		if err != nil {
-			common.SysError("error update user quota cache: " + err.Error())
-		}
-		if quota != 0 {
-			tokenName := c.GetString("token_name")
-			quality := "normal"
-			if imageRequest.Quality == "hd" {
-				quality = "hd"
-			}
-			logContent := fmt.Sprintf("模型价格 %.2f，分组倍率 %.2f, 大小 %s, 品质 %s", modelPrice, groupRatio, imageRequest.Size, quality)
-			other := make(map[string]interface{})
-			other["model_price"] = modelPrice
-			other["group_ratio"] = groupRatio
-			model.RecordConsumeLog(ctx, userId, channelId, 0, 0, imageRequest.Model, tokenName, quota, logContent, tokenId, userQuota, int(useTimeSeconds), false, other)
-			model.UpdateUserUsedQuotaAndRequestCount(userId, quota)
-			channelId := c.GetInt("channel_id")
-			model.UpdateChannelUsedQuota(channelId, quota)
-		}
-	}(c.Request.Context())
-
-	responseBody, err := io.ReadAll(resp.Body)
-
-	if err != nil {
-		return service.OpenAIErrorWrapper(err, "read_response_body_failed", http.StatusInternalServerError)
-	}
-	err = resp.Body.Close()
-	if err != nil {
-		return service.OpenAIErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError)
-	}
-	err = json.Unmarshal(responseBody, &textResponse)
-	if err != nil {
-		return service.OpenAIErrorWrapper(err, "unmarshal_response_body_failed", http.StatusInternalServerError)
 	}
 
-	resp.Body = io.NopCloser(bytes.NewBuffer(responseBody))
+	_, openaiErr := adaptor.DoResponse(c, resp, relayInfo)
+	if openaiErr != nil {
+		// reset status code 重置状态码
+		service.ResetStatusCode(openaiErr, statusCodeMappingStr)
+		return openaiErr
+	}
 
-	for k, v := range resp.Header {
+	usage := &dto.Usage{
-		c.Writer.Header().Set(k, v[0])
+		PromptTokens: imageRequest.N,
+		TotalTokens:  imageRequest.N,
 	}
-	c.Writer.WriteHeader(resp.StatusCode)
 
-	_, err = io.Copy(c.Writer, resp.Body)
+	quality := "standard"
-	if err != nil {
+	if imageRequest.Quality == "hd" {
-		return service.OpenAIErrorWrapper(err, "copy_response_body_failed", http.StatusInternalServerError)
+		quality = "hd"
-	}
-	err = resp.Body.Close()
-	if err != nil {
-		return service.OpenAIErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError)
 	}
+
+	logContent := fmt.Sprintf("大小 %s, 品质 %s", imageRequest.Size, quality)
+	postConsumeQuota(c, relayInfo, imageRequest.Model, usage, 0, 0, userQuota, 0, groupRatio, imageRatio, true, logContent)
+
 	return nil
 }

relay/relay-mj.go

@@ -382,6 +382,8 @@ func RelayMidjourneySubmit(c *gin.Context, relayMode int) *dto.MidjourneyRespons
 		midjRequest.Action = constant.MjActionShorten
 	} else if relayMode == relayconstant.RelayModeMidjourneyBlend { //绘画任务，此类任务可重复
 		midjRequest.Action = constant.MjActionBlend
+	} else if relayMode == relayconstant.RelayModeMidjourneyUpload { //绘画任务，此类任务可重复
+		midjRequest.Action = constant.MjActionUpload
 	} else if midjRequest.TaskId != "" { //放大、变换任务，此类任务，如果重复且已有结果，远端api会直接返回最终结果
 		mjId := ""
 		if relayMode == relayconstant.RelayModeMidjourneyChange {
@@ -547,7 +549,7 @@ func RelayMidjourneySubmit(c *gin.Context, relayMode int) *dto.MidjourneyRespons
 		if err != nil {
 			common.SysError("get_channel_null: " + err.Error())
 		}
-		if channel.AutoBan != nil && *channel.AutoBan == 1 && common.AutomaticDisableChannelEnabled {
+		if channel.GetAutoBan() && common.AutomaticDisableChannelEnabled {
 			model.UpdateChannelStatusById(midjourneyTask.ChannelId, 2, "No available account instance")
 		}
 	}
@@ -580,7 +582,10 @@ func RelayMidjourneySubmit(c *gin.Context, relayMode int) *dto.MidjourneyRespons
 			responseBody = []byte(newBody)
 		}
 	}
-
+	if midjResponse.Code == 1 && midjRequest.Action == "UPLOAD" {
+		midjourneyTask.Progress = "100%"
+		midjourneyTask.Status = "SUCCESS"
+	}
 	err = midjourneyTask.Insert()
 	if err != nil {
 		return &dto.MidjourneyResponse{
@@ -594,7 +599,6 @@ func RelayMidjourneySubmit(c *gin.Context, relayMode int) *dto.MidjourneyRespons
 		newBody := strings.Replace(string(responseBody), `"code":22`, `"code":1`, -1)
 		responseBody = []byte(newBody)
 	}
-
 	//resp.Body = io.NopCloser(bytes.NewBuffer(responseBody))
 	bodyReader := io.NopCloser(bytes.NewBuffer(responseBody))
 

relay/relay-text.go

@@ -52,7 +52,7 @@ func getAndValidateTextRequest(c *gin.Context, relayInfo *relaycommon.RelayInfo)
 		}
 	case relayconstant.RelayModeEmbeddings:
 	case relayconstant.RelayModeModerations:
-		if textRequest.Input == "" {
+		if textRequest.Input == "" || textRequest.Input == nil {
 			return nil, errors.New("field input is required")
 		}
 	case relayconstant.RelayModeEdits:
@@ -91,7 +91,7 @@ func TextHelper(c *gin.Context) *dto.OpenAIErrorWithStatusCode {
 		}
 	}
 	relayInfo.UpstreamModelName = textRequest.Model
-	modelPrice, success := common.GetModelPrice(textRequest.Model, false)
+	modelPrice, getModelPriceSuccess := common.GetModelPrice(textRequest.Model, false)
 	groupRatio := common.GetGroupRatio(relayInfo.Group)
 
 	var preConsumedQuota int
@@ -112,7 +112,7 @@ func TextHelper(c *gin.Context) *dto.OpenAIErrorWithStatusCode {
 		return service.OpenAIErrorWrapper(err, "count_token_messages_failed", http.StatusInternalServerError)
 	}
 
-	if !success {
+	if !getModelPriceSuccess {
 		preConsumedTokens := common.PreConsumedQuota
 		if textRequest.MaxTokens != 0 {
 			preConsumedTokens = promptTokens + int(textRequest.MaxTokens)
@@ -130,6 +130,12 @@ func TextHelper(c *gin.Context) *dto.OpenAIErrorWithStatusCode {
 		return openaiErr
 	}
 
+	includeUsage := false
+	// 判断用户是否需要返回使用情况
+	if textRequest.StreamOptions != nil && textRequest.StreamOptions.IncludeUsage {
+		includeUsage = true
+	}
+
 	// 如果不支持StreamOptions，将StreamOptions设置为nil
 	if !relayInfo.SupportStreamOptions || !textRequest.Stream {
 		textRequest.StreamOptions = nil
@@ -142,18 +148,18 @@ func TextHelper(c *gin.Context) *dto.OpenAIErrorWithStatusCode {
 		}
 	}
 
-	if textRequest.StreamOptions != nil && textRequest.StreamOptions.IncludeUsage {
+	if includeUsage {
-		relayInfo.ShouldIncludeUsage = textRequest.StreamOptions.IncludeUsage
+		relayInfo.ShouldIncludeUsage = true
 	}
 
 	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, *textRequest)
+	adaptor.Init(relayInfo)
 	var requestBody io.Reader
 
-	convertedRequest, err := adaptor.ConvertRequest(c, relayInfo.RelayMode, textRequest)
+	convertedRequest, err := adaptor.ConvertRequest(c, relayInfo, textRequest)
 	if err != nil {
 		return service.OpenAIErrorWrapperLocal(err, "convert_request_failed", http.StatusInternalServerError)
 	}
@@ -187,7 +193,7 @@ func TextHelper(c *gin.Context) *dto.OpenAIErrorWithStatusCode {
 		service.ResetStatusCode(openaiErr, statusCodeMappingStr)
 		return openaiErr
 	}
-	postConsumeQuota(c, relayInfo, textRequest.Model, usage, ratio, preConsumedQuota, userQuota, modelRatio, groupRatio, modelPrice, success)
+	postConsumeQuota(c, relayInfo, textRequest.Model, usage, ratio, preConsumedQuota, userQuota, modelRatio, groupRatio, modelPrice, getModelPriceSuccess, "")
 	return nil
 }
 
@@ -241,9 +247,12 @@ func preConsumeQuota(c *gin.Context, preConsumedQuota int, relayInfo *relaycommo
 	if err != nil {
 		return 0, 0, service.OpenAIErrorWrapperLocal(err, "get_user_quota_failed", http.StatusInternalServerError)
 	}
-	if userQuota <= 0 || userQuota-preConsumedQuota < 0 {
+	if userQuota <= 0 {
 		return 0, 0, service.OpenAIErrorWrapperLocal(errors.New("user quota is not enough"), "insufficient_user_quota", http.StatusForbidden)
 	}
+	if userQuota-preConsumedQuota < 0 {
+		return 0, 0, service.OpenAIErrorWrapperLocal(errors.New(fmt.Sprintf("chat pre-consumed quota failed, user quota: %d, need quota: %d", userQuota, preConsumedQuota)), "insufficient_user_quota", http.StatusBadRequest)
+	}
 	err = model.CacheDecreaseUserQuota(relayInfo.UserId, preConsumedQuota)
 	if err != nil {
 		return 0, 0, service.OpenAIErrorWrapperLocal(err, "decrease_user_quota_failed", http.StatusInternalServerError)
@@ -256,13 +265,13 @@ func preConsumeQuota(c *gin.Context, preConsumedQuota int, relayInfo *relaycommo
 			if tokenQuota > 100*preConsumedQuota {
 				// 令牌额度充足，信任令牌
 				preConsumedQuota = 0
-				common.LogInfo(c.Request.Context(), fmt.Sprintf("user %d quota %d and token %d quota %d are enough, trusted and no need to pre-consume", relayInfo.UserId, userQuota, relayInfo.TokenId, tokenQuota))
+				common.LogInfo(c, fmt.Sprintf("user %d quota %d and token %d quota %d are enough, trusted and no need to pre-consume", relayInfo.UserId, userQuota, relayInfo.TokenId, tokenQuota))
 			}
 		} else {
 			// in this case, we do not pre-consume quota
 			// because the user has enough quota
 			preConsumedQuota = 0
-			common.LogInfo(c.Request.Context(), fmt.Sprintf("user %d with unlimited token has enough quota %d, trusted and no need to pre-consume", relayInfo.UserId, userQuota))
+			common.LogInfo(c, fmt.Sprintf("user %d with unlimited token has enough quota %d, trusted and no need to pre-consume", relayInfo.UserId, userQuota))
 		}
 	}
 	if preConsumedQuota > 0 {
@@ -288,8 +297,15 @@ func returnPreConsumedQuota(c *gin.Context, tokenId int, userQuota int, preConsu
 
 func postConsumeQuota(ctx *gin.Context, relayInfo *relaycommon.RelayInfo, modelName string,
 	usage *dto.Usage, ratio float64, preConsumedQuota int, userQuota int, modelRatio float64, groupRatio float64,
-	modelPrice float64, usePrice bool) {
+	modelPrice float64, usePrice bool, extraContent string) {
-
+	if usage == nil {
+		usage = &dto.Usage{
+			PromptTokens:     relayInfo.PromptTokens,
+			CompletionTokens: 0,
+			TotalTokens:      relayInfo.PromptTokens,
+		}
+		extraContent += "  ，（可能是请求出错）"
+	}
 	useTimeSeconds := time.Now().Unix() - relayInfo.StartTime.Unix()
 	promptTokens := usage.PromptTokens
 	completionTokens := usage.CompletionTokens
@@ -309,8 +325,8 @@ func postConsumeQuota(ctx *gin.Context, relayInfo *relaycommon.RelayInfo, modelN
 	}
 	totalTokens := promptTokens + completionTokens
 	var logContent string
-	if modelPrice == -1 {
+	if !usePrice {
-		logContent = fmt.Sprintf("模型倍率 %.2f，分组倍率 %.2f，补全倍率 %.2f", modelRatio, groupRatio, completionRatio)
+		logContent = fmt.Sprintf("模型倍率 %.2f，补全倍率 %.2f，分组倍率 %.2f", modelRatio, completionRatio, groupRatio)
 	} else {
 		logContent = fmt.Sprintf("模型价格 %.2f，分组倍率 %.2f", modelPrice, groupRatio)
 	}
@@ -350,6 +366,9 @@ func postConsumeQuota(ctx *gin.Context, relayInfo *relaycommon.RelayInfo, modelN
 		logModel = "g-*"
 		logContent += fmt.Sprintf("，模型 %s", modelName)
 	}
+	if extraContent != "" {
+		logContent += ", " + extraContent
+	}
 	other := service.GenerateTextOtherInfo(ctx, relayInfo, modelRatio, groupRatio, completionRatio, modelPrice)
 	model.RecordConsumeLog(ctx, relayInfo.UserId, relayInfo.ChannelId, promptTokens, completionTokens, logModel,
 		tokenName, quota, logContent, relayInfo.TokenId, userQuota, int(useTimeSeconds), relayInfo.IsStream, other)

relay/relay_adaptor.go

@@ -7,6 +7,7 @@ import (
 	"one-api/relay/channel/aws"
 	"one-api/relay/channel/baidu"
 	"one-api/relay/channel/claude"
+	"one-api/relay/channel/cloudflare"
 	"one-api/relay/channel/cohere"
 	"one-api/relay/channel/dify"
 	"one-api/relay/channel/gemini"
@@ -15,8 +16,10 @@ import (
 	"one-api/relay/channel/openai"
 	"one-api/relay/channel/palm"
 	"one-api/relay/channel/perplexity"
+	"one-api/relay/channel/siliconflow"
 	"one-api/relay/channel/task/suno"
 	"one-api/relay/channel/tencent"
+	"one-api/relay/channel/vertex"
 	"one-api/relay/channel/xunfei"
 	"one-api/relay/channel/zhipu"
 	"one-api/relay/channel/zhipu_4v"
@@ -59,6 +62,12 @@ func GetAdaptor(apiType int) channel.Adaptor {
 		return &dify.Adaptor{}
 	case constant.APITypeJina:
 		return &jina.Adaptor{}
+	case constant.APITypeCloudflare:
+		return &cloudflare.Adaptor{}
+	case constant.APITypeSiliconFlow:
+		return &siliconflow.Adaptor{}
+	case constant.APITypeVertexAi:
+		return &vertex.Adaptor{}
 	}
 	return nil
 }

relay/relay_rerank.go

@@ -38,6 +38,23 @@ func RerankHelper(c *gin.Context, relayMode int) *dto.OpenAIErrorWithStatusCode
 	if len(rerankRequest.Documents) == 0 {
 		return service.OpenAIErrorWrapperLocal(fmt.Errorf("documents is empty"), "invalid_documents", 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[rerankRequest.Model] != "" {
+			rerankRequest.Model = modelMap[rerankRequest.Model]
+			// set upstream model name
+			//isModelMapped = true
+		}
+	}
+
 	relayInfo.UpstreamModelName = rerankRequest.Model
 	modelPrice, success := common.GetModelPrice(rerankRequest.Model, false)
 	groupRatio := common.GetGroupRatio(relayInfo.Group)
@@ -66,7 +83,7 @@ func RerankHelper(c *gin.Context, relayMode int) *dto.OpenAIErrorWithStatusCode
 	if adaptor == nil {
 		return service.OpenAIErrorWrapperLocal(fmt.Errorf("invalid api type: %d", relayInfo.ApiType), "invalid_api_type", http.StatusBadRequest)
 	}
-	adaptor.InitRerank(relayInfo, *rerankRequest)
+	adaptor.Init(relayInfo)
 
 	convertedRequest, err := adaptor.ConvertRerankRequest(c, relayInfo.RelayMode, *rerankRequest)
 	if err != nil {
@@ -99,6 +116,6 @@ func RerankHelper(c *gin.Context, relayMode int) *dto.OpenAIErrorWithStatusCode
 		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, ratio, preConsumedQuota, userQuota, modelRatio, groupRatio, modelPrice, success, "")
 	return nil
 }

router/relay-router.go

@@ -79,5 +79,6 @@ func registerMjRouterGroup(relayMjRouter *gin.RouterGroup) {
 		relayMjRouter.GET("/task/:id/image-seed", controller.RelayMidjourney)
 		relayMjRouter.POST("/task/list-by-condition", controller.RelayMidjourney)
 		relayMjRouter.POST("/insight-face/swap", controller.RelayMidjourney)
+		relayMjRouter.POST("/submit/upload-discord-images", controller.RelayMidjourney)
 	}
 }

service/channel.go

@@ -54,6 +54,8 @@ func ShouldDisableChannel(channelType int, err *relaymodel.OpenAIErrorWithStatus
 	switch err.Error.Type {
 	case "insufficient_quota":
 		return true
+	case "insufficient_user_quota":
+		return true
 	// https://docs.anthropic.com/claude/reference/errors
 	case "authentication_error":
 		return true
@@ -74,14 +76,14 @@ func ShouldDisableChannel(channelType int, err *relaymodel.OpenAIErrorWithStatus
 	return false
 }
 
-func ShouldEnableChannel(err error, openAIErr *relaymodel.OpenAIError, status int) bool {
+func ShouldEnableChannel(err error, openaiWithStatusErr *relaymodel.OpenAIErrorWithStatusCode, status int) bool {
 	if !common.AutomaticEnableChannelEnabled {
 		return false
 	}
 	if err != nil {
 		return false
 	}
-	if openAIErr != nil {
+	if openaiWithStatusErr != nil {
 		return false
 	}
 	if status != common.ChannelStatusAutoDisabled {

service/error.go

@@ -28,13 +28,11 @@ func MidjourneyErrorWithStatusCodeWrapper(code int, desc string, statusCode int)
 // OpenAIErrorWrapper wraps an error into an OpenAIErrorWithStatusCode
 func OpenAIErrorWrapper(err error, code string, statusCode int) *dto.OpenAIErrorWithStatusCode {
 	text := err.Error()
-	// 定义一个正则表达式匹配URL
+	lowerText := strings.ToLower(text)
-	if strings.Contains(text, "Post") || strings.Contains(text, "dial") {
+	if strings.Contains(lowerText, "post") || strings.Contains(lowerText, "dial") || strings.Contains(lowerText, "http") {
 		common.SysLog(fmt.Sprintf("error: %s", text))
 		text = "请求上游地址失败"
 	}
-	//避免暴露内部错误
-
 	openAIError := dto.OpenAIError{
 		Message: text,
 		Type:    "new_api_error",
@@ -56,10 +54,9 @@ func RelayErrorHandler(resp *http.Response) (errWithStatusCode *dto.OpenAIErrorW
 	errWithStatusCode = &dto.OpenAIErrorWithStatusCode{
 		StatusCode: resp.StatusCode,
 		Error: dto.OpenAIError{
-			Message: "",
+			Type:  "upstream_error",
-			Type:    "upstream_error",
+			Code:  "bad_response_status_code",
-			Code:    "bad_response_status_code",
+			Param: strconv.Itoa(resp.StatusCode),
-			Param:   strconv.Itoa(resp.StatusCode),
 		},
 	}
 	responseBody, err := io.ReadAll(resp.Body)
@@ -114,14 +111,12 @@ func TaskErrorWrapperLocal(err error, code string, statusCode int) *dto.TaskErro
 
 func TaskErrorWrapper(err error, code string, statusCode int) *dto.TaskError {
 	text := err.Error()
-
+	lowerText := strings.ToLower(text)
-	// 定义一个正则表达式匹配URL
+	if strings.Contains(lowerText, "post") || strings.Contains(lowerText, "dial") || strings.Contains(lowerText, "http") {
-	if strings.Contains(text, "Post") || strings.Contains(text, "dial") {
 		common.SysLog(fmt.Sprintf("error: %s", text))
 		text = "请求上游地址失败"
 	}
 	//避免暴露内部错误
-
 	taskError := &dto.TaskError{
 		Code:       code,
 		Message:    text,

service/midjourney.go

@@ -49,6 +49,8 @@ func GetMjRequestModel(relayMode int, midjRequest *dto.MidjourneyRequest) (strin
 			action = constant.MjActionModal
 		case relayconstant.RelayModeSwapFace:
 			action = constant.MjActionSwapFace
+		case relayconstant.RelayModeMidjourneyUpload:
+			action = constant.MjActionUpload
 		case relayconstant.RelayModeMidjourneySimpleChange:
 			params := ConvertSimpleChangeParams(midjRequest.Content)
 			if params == nil {
@@ -220,7 +222,7 @@ func DoMidjourneyHttpRequest(c *gin.Context, timeout time.Duration, fullRequestU
 		return MidjourneyErrorWithStatusCodeWrapper(constant.MjErrorUnknown, "close_request_body_failed", statusCode), nullBytes, err
 	}
 	var midjResponse dto.MidjourneyResponse
-
+	var midjourneyUploadsResponse dto.MidjourneyUploadResponse
 	responseBody, err := io.ReadAll(resp.Body)
 	if err != nil {
 		return MidjourneyErrorWithStatusCodeWrapper(constant.MjErrorUnknown, "read_response_body_failed", statusCode), nullBytes, err
@@ -230,13 +232,16 @@ func DoMidjourneyHttpRequest(c *gin.Context, timeout time.Duration, fullRequestU
 		return MidjourneyErrorWithStatusCodeWrapper(constant.MjErrorUnknown, "close_response_body_failed", statusCode), responseBody, err
 	}
 	respStr := string(responseBody)
-	log.Printf("responseBody: %s", respStr)
+	log.Printf("respStr: %s", respStr)
 	if respStr == "" {
 		return MidjourneyErrorWithStatusCodeWrapper(constant.MjErrorUnknown, "empty_response_body", statusCode), responseBody, nil
 	} else {
 		err = json.Unmarshal(responseBody, &midjResponse)
 		if err != nil {
-			return MidjourneyErrorWithStatusCodeWrapper(constant.MjErrorUnknown, "unmarshal_response_body_failed", statusCode), responseBody, err
+			err2 := json.Unmarshal(responseBody, &midjourneyUploadsResponse)
+			if err2 != nil {
+				return MidjourneyErrorWithStatusCodeWrapper(constant.MjErrorUnknown, "unmarshal_response_body_failed", statusCode), responseBody, err
+			}
 		}
 	}
 	//log.Printf("midjResponse: %v", midjResponse)

service/relay.go

@@ -0,0 +1,75 @@
+package service
+
+import (
+	"encoding/json"
+	"errors"
+	"fmt"
+	"github.com/gin-gonic/gin"
+	"net/http"
+	"one-api/common"
+	"one-api/dto"
+)
+
+func SetEventStreamHeaders(c *gin.Context) {
+	c.Writer.Header().Set("Content-Type", "text/event-stream")
+	c.Writer.Header().Set("Cache-Control", "no-cache")
+	c.Writer.Header().Set("Connection", "keep-alive")
+	c.Writer.Header().Set("Transfer-Encoding", "chunked")
+	c.Writer.Header().Set("X-Accel-Buffering", "no")
+}
+
+func StringData(c *gin.Context, str string) error {
+	//str = strings.TrimPrefix(str, "data: ")
+	//str = strings.TrimSuffix(str, "\r")
+	c.Render(-1, common.CustomEvent{Data: "data: " + str})
+	if flusher, ok := c.Writer.(http.Flusher); ok {
+		flusher.Flush()
+	} else {
+		return errors.New("streaming error: flusher not found")
+	}
+	return nil
+}
+
+func ObjectData(c *gin.Context, object interface{}) error {
+	jsonData, err := json.Marshal(object)
+	if err != nil {
+		return fmt.Errorf("error marshalling object: %w", err)
+	}
+	return StringData(c, string(jsonData))
+}
+
+func Done(c *gin.Context) {
+	_ = StringData(c, "[DONE]")
+}
+
+func GetResponseID(c *gin.Context) string {
+	logID := c.GetString("X-Oneapi-Request-Id")
+	return fmt.Sprintf("chatcmpl-%s", logID)
+}
+
+func GenerateStopResponse(id string, createAt int64, model string, finishReason string) *dto.ChatCompletionsStreamResponse {
+	return &dto.ChatCompletionsStreamResponse{
+		Id:                id,
+		Object:            "chat.completion.chunk",
+		Created:           createAt,
+		Model:             model,
+		SystemFingerprint: nil,
+		Choices: []dto.ChatCompletionsStreamResponseChoice{
+			{
+				FinishReason: &finishReason,
+			},
+		},
+	}
+}
+
+func GenerateFinalUsageResponse(id string, createAt int64, model string, usage dto.Usage) *dto.ChatCompletionsStreamResponse {
+	return &dto.ChatCompletionsStreamResponse{
+		Id:                id,
+		Object:            "chat.completion.chunk",
+		Created:           createAt,
+		Model:             model,
+		SystemFingerprint: nil,
+		Choices:           make([]dto.ChatCompletionsStreamResponseChoice, 0),
+		Usage:             &usage,
+	}
+}

service/sse.go

@@ -1,37 +0,0 @@
-package service
-
-import (
-	"encoding/json"
-	"fmt"
-	"github.com/gin-gonic/gin"
-	"one-api/common"
-	"strings"
-)
-
-func SetEventStreamHeaders(c *gin.Context) {
-	c.Writer.Header().Set("Content-Type", "text/event-stream")
-	c.Writer.Header().Set("Cache-Control", "no-cache")
-	c.Writer.Header().Set("Connection", "keep-alive")
-	c.Writer.Header().Set("Transfer-Encoding", "chunked")
-	c.Writer.Header().Set("X-Accel-Buffering", "no")
-}
-
-func StringData(c *gin.Context, str string) {
-	str = strings.TrimPrefix(str, "data: ")
-	str = strings.TrimSuffix(str, "\r")
-	c.Render(-1, common.CustomEvent{Data: "data: " + str})
-	c.Writer.Flush()
-}
-
-func ObjectData(c *gin.Context, object interface{}) error {
-	jsonData, err := json.Marshal(object)
-	if err != nil {
-		return fmt.Errorf("error marshalling object: %w", err)
-	}
-	StringData(c, string(jsonData))
-	return nil
-}
-
-func Done(c *gin.Context) {
-	StringData(c, "[DONE]")
-}