feat: support tencent v3 api (#1542)

Co-authored-by: lihangfu <hfli8@iflytek.com>

common/config/config.go

@@ -117,10 +117,10 @@ var ValidThemes = map[string]bool{
 // All duration's unit is seconds
 // Shouldn't larger then RateLimitKeyExpirationDuration
 var (
-	GlobalApiRateLimitNum            = env.Int("GLOBAL_API_RATE_LIMIT", 180)
+	GlobalApiRateLimitNum            = env.Int("GLOBAL_API_RATE_LIMIT", 240)
 	GlobalApiRateLimitDuration int64 = 3 * 60
 
-	GlobalWebRateLimitNum            = env.Int("GLOBAL_WEB_RATE_LIMIT", 60)
+	GlobalWebRateLimitNum            = env.Int("GLOBAL_WEB_RATE_LIMIT", 120)
 	GlobalWebRateLimitDuration int64 = 3 * 60
 
 	UploadRateLimitNum            = 10

common/logger/logger.go

@@ -43,11 +43,19 @@ func SysLog(s string) {
 	_, _ = fmt.Fprintf(gin.DefaultWriter, "[SYS] %v | %s \n", t.Format("2006/01/02 - 15:04:05"), s)
 }
 
+func SysLogf(format string, a ...any) {
+	SysLog(fmt.Sprintf(format, a...))
+}
+
 func SysError(s string) {
 	t := time.Now()
 	_, _ = fmt.Fprintf(gin.DefaultErrorWriter, "[SYS] %v | %s \n", t.Format("2006/01/02 - 15:04:05"), s)
 }
 
+func SysErrorf(format string, a ...any) {
+	SysError(fmt.Sprintf(format, a...))
+}
+
 func Debug(ctx context.Context, msg string) {
 	if config.DebugEnabled {
 		logHelper(ctx, loggerDEBUG, msg)

controller/user.go

@@ -6,8 +6,6 @@ import (
 	"github.com/songquanpeng/one-api/common"
 	"github.com/songquanpeng/one-api/common/config"
 	"github.com/songquanpeng/one-api/common/ctxkey"
-	"github.com/songquanpeng/one-api/common/helper"
-	"github.com/songquanpeng/one-api/common/logger"
 	"github.com/songquanpeng/one-api/common/random"
 	"github.com/songquanpeng/one-api/model"
 	"net/http"
@@ -111,7 +109,6 @@ func Logout(c *gin.Context) {
 }
 
 func Register(c *gin.Context) {
-	ctx := c.Request.Context()
 	if !config.RegisterEnabled {
 		c.JSON(http.StatusOK, gin.H{
 			"message": "管理员关闭了新用户注册",
@@ -176,28 +173,7 @@ func Register(c *gin.Context) {
 		})
 		return
 	}
-	go func() {
+
-		err := user.ValidateAndFill()
-		if err != nil {
-			logger.Errorf(ctx, "user.ValidateAndFill failed: %w", err)
-			return
-		}
-		cleanToken := model.Token{
-			UserId:         user.Id,
-			Name:           "default",
-			Key:            random.GenerateKey(),
-			CreatedTime:    helper.GetTimestamp(),
-			AccessedTime:   helper.GetTimestamp(),
-			ExpiredTime:    -1,
-			RemainQuota:    -1,
-			UnlimitedQuota: true,
-		}
-		err = cleanToken.Insert()
-		if err != nil {
-			logger.Errorf(ctx, "cleanToken.Insert failed: %w", err)
-			return
-		}
-	}()
 	c.JSON(http.StatusOK, gin.H{
 		"success": true,
 		"message": "",

main.go

@@ -24,7 +24,7 @@ var buildFS embed.FS
 
 func main() {
 	logger.SetupLogger()
-	logger.SysLog(fmt.Sprintf("One API %s started", common.Version))
+	logger.SysLogf("One API %s started", common.Version)
 	if os.Getenv("GIN_MODE") != "debug" {
 		gin.SetMode(gin.ReleaseMode)
 	}

middleware/distributor.go

@@ -67,26 +67,28 @@ func SetupContextForSelectedChannel(c *gin.Context, channel *model.Channel, mode
 	c.Set(ctxkey.BaseURL, channel.GetBaseURL())
 	cfg, _ := channel.LoadConfig()
 	// this is for backward compatibility
-	switch channel.Type {
+	if channel.Other != nil {
-	case channeltype.Azure:
+		switch channel.Type {
-		if cfg.APIVersion == "" {
+		case channeltype.Azure:
-			cfg.APIVersion = channel.Other
+			if cfg.APIVersion == "" {
-		}
+				cfg.APIVersion = *channel.Other
-	case channeltype.Xunfei:
+			}
-		if cfg.APIVersion == "" {
+		case channeltype.Xunfei:
-			cfg.APIVersion = channel.Other
+			if cfg.APIVersion == "" {
-		}
+				cfg.APIVersion = *channel.Other
-	case channeltype.Gemini:
+			}
-		if cfg.APIVersion == "" {
+		case channeltype.Gemini:
-			cfg.APIVersion = channel.Other
+			if cfg.APIVersion == "" {
-		}
+				cfg.APIVersion = *channel.Other
-	case channeltype.AIProxyLibrary:
+			}
-		if cfg.LibraryID == "" {
+		case channeltype.AIProxyLibrary:
-			cfg.LibraryID = channel.Other
+			if cfg.LibraryID == "" {
-		}
+				cfg.LibraryID = *channel.Other
-	case channeltype.Ali:
+			}
-		if cfg.Plugin == "" {
+		case channeltype.Ali:
-			cfg.Plugin = channel.Other
+			if cfg.Plugin == "" {
+				cfg.Plugin = *channel.Other
+			}
 		}
 	}
 	c.Set(ctxkey.Config, cfg)

model/channel.go

@@ -27,7 +27,7 @@ type Channel struct {
 	TestTime           int64   `json:"test_time" gorm:"bigint"`
 	ResponseTime       int     `json:"response_time"` // in milliseconds
 	BaseURL            *string `json:"base_url" gorm:"column:base_url;default:''"`
-	Other              string  `json:"other"`   // DEPRECATED: please save config to field Config
+	Other              *string `json:"other"`   // DEPRECATED: please save config to field Config
 	Balance            float64 `json:"balance"` // in USD
 	BalanceUpdatedTime int64   `json:"balance_updated_time" gorm:"bigint"`
 	Models             string  `json:"models"`

model/user.go

@@ -6,6 +6,7 @@ import (
 	"github.com/songquanpeng/one-api/common"
 	"github.com/songquanpeng/one-api/common/blacklist"
 	"github.com/songquanpeng/one-api/common/config"
+	"github.com/songquanpeng/one-api/common/helper"
 	"github.com/songquanpeng/one-api/common/logger"
 	"github.com/songquanpeng/one-api/common/random"
 	"gorm.io/gorm"
@@ -140,6 +141,22 @@ func (user *User) Insert(inviterId int) error {
 			RecordLog(inviterId, LogTypeSystem, fmt.Sprintf("邀请用户赠送 %s", common.LogQuota(config.QuotaForInviter)))
 		}
 	}
+	// create default token
+	cleanToken := Token{
+		UserId:         user.Id,
+		Name:           "default",
+		Key:            random.GenerateKey(),
+		CreatedTime:    helper.GetTimestamp(),
+		AccessedTime:   helper.GetTimestamp(),
+		ExpiredTime:    -1,
+		RemainQuota:    -1,
+		UnlimitedQuota: true,
+	}
+	result.Error = cleanToken.Insert()
+	if result.Error != nil {
+		// do not block
+		logger.SysError(fmt.Sprintf("create default token for user %d failed: %s", user.Id, result.Error.Error()))
+	}
 	return nil
 }
 

relay/adaptor/cloudflare/main.go

@@ -17,15 +17,21 @@ import (
 )
 
 func ConvertRequest(textRequest model.GeneralOpenAIRequest) *Request {
-	lastMessage := textRequest.Messages[len(textRequest.Messages)-1]
+    var promptBuilder strings.Builder
-	return &Request{
+    for _, message := range textRequest.Messages {
-		MaxTokens:   textRequest.MaxTokens,
+        promptBuilder.WriteString(message.StringContent())
-		Prompt:      lastMessage.StringContent(),
+        promptBuilder.WriteString("\n")  // 添加换行符来分隔每个消息
-		Stream:      textRequest.Stream,
+    }
-		Temperature: textRequest.Temperature,
+
-	}
+    return &Request{
+        MaxTokens:   textRequest.MaxTokens,
+        Prompt:      promptBuilder.String(),
+        Stream:      textRequest.Stream,
+        Temperature: textRequest.Temperature,
+    }
 }
 
+
 func ResponseCloudflare2OpenAI(cloudflareResponse *Response) *openai.TextResponse {
 	choice := openai.TextResponseChoice{
 		Index: 0,

relay/adaptor/gemini/main.go

@@ -245,8 +245,10 @@ func responseGeminiChat2OpenAI(response *ChatResponse) *openai.TextResponse {
 func streamResponseGeminiChat2OpenAI(geminiResponse *ChatResponse) *openai.ChatCompletionsStreamResponse {
 	var choice openai.ChatCompletionsStreamResponseChoice
 	choice.Delta.Content = geminiResponse.GetResponseText()
-	choice.FinishReason = &constant.StopFinishReason
+	//choice.FinishReason = &constant.StopFinishReason
 	var response openai.ChatCompletionsStreamResponse
+	response.Id = fmt.Sprintf("chatcmpl-%s", random.GetUUID())
+	response.Created = helper.GetTimestamp()
 	response.Object = "chat.completion.chunk"
 	response.Model = "gemini"
 	response.Choices = []openai.ChatCompletionsStreamResponseChoice{choice}

relay/adaptor/tencent/adaptor.go

@@ -2,35 +2,43 @@ package tencent
 
 import (
 	"errors"
-	"fmt"
 	"github.com/gin-gonic/gin"
+	"github.com/songquanpeng/one-api/common/helper"
 	"github.com/songquanpeng/one-api/relay/adaptor"
 	"github.com/songquanpeng/one-api/relay/adaptor/openai"
 	"github.com/songquanpeng/one-api/relay/meta"
 	"github.com/songquanpeng/one-api/relay/model"
 	"io"
 	"net/http"
+	"strconv"
 	"strings"
 )
 
 // https://cloud.tencent.com/document/api/1729/101837
 
 type Adaptor struct {
-	Sign string
+	Sign      string
+	Action    string
+	Version   string
+	Timestamp int64
 }
 
 func (a *Adaptor) Init(meta *meta.Meta) {
-
+	a.Action = "ChatCompletions"
+	a.Version = "2023-09-01"
+	a.Timestamp = helper.GetTimestamp()
 }
 
 func (a *Adaptor) GetRequestURL(meta *meta.Meta) (string, error) {
-	return fmt.Sprintf("%s/hyllm/v1/chat/completions", meta.BaseURL), nil
+	return meta.BaseURL + "/", nil
 }
 
 func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, meta *meta.Meta) error {
 	adaptor.SetupCommonRequestHeader(c, req, meta)
 	req.Header.Set("Authorization", a.Sign)
-	req.Header.Set("X-TC-Action", meta.ActualModelName)
+	req.Header.Set("X-TC-Action", a.Action)
+	req.Header.Set("X-TC-Version", a.Version)
+	req.Header.Set("X-TC-Timestamp", strconv.FormatInt(a.Timestamp, 10))
 	return nil
 }
 
@@ -40,15 +48,13 @@ func (a *Adaptor) ConvertRequest(c *gin.Context, relayMode int, request *model.G
 	}
 	apiKey := c.Request.Header.Get("Authorization")
 	apiKey = strings.TrimPrefix(apiKey, "Bearer ")
-	appId, secretId, secretKey, err := ParseConfig(apiKey)
+	_, secretId, secretKey, err := ParseConfig(apiKey)
 	if err != nil {
 		return nil, err
 	}
 	tencentRequest := ConvertRequest(*request)
-	tencentRequest.AppId = appId
-	tencentRequest.SecretId = secretId
 	// we have to calculate the sign here
-	a.Sign = GetSign(*tencentRequest, secretKey)
+	a.Sign = GetSign(*tencentRequest, a, secretId, secretKey)
 	return tencentRequest, nil
 }
 

relay/adaptor/tencent/constants.go

@@ -1,7 +1,8 @@
 package tencent
 
 var ModelList = []string{
-	"ChatPro",
+	"hunyuan-lite",
-	"ChatStd",
+	"hunyuan-standard",
-	"hunyuan",
+	"hunyuan-standard-256K",
+	"hunyuan-pro",
 }

relay/adaptor/tencent/main.go

@@ -3,8 +3,8 @@ package tencent
 import (
 	"bufio"
 	"crypto/hmac"
-	"crypto/sha1"
+	"crypto/sha256"
-	"encoding/base64"
+	"encoding/hex"
 	"encoding/json"
 	"errors"
 	"fmt"
@@ -19,34 +19,26 @@ import (
 	"github.com/songquanpeng/one-api/relay/model"
 	"io"
 	"net/http"
-	"sort"
 	"strconv"
 	"strings"
+	"time"
 )
 
-// https://cloud.tencent.com/document/product/1729/97732
-
 func ConvertRequest(request model.GeneralOpenAIRequest) *ChatRequest {
-	messages := make([]Message, 0, len(request.Messages))
+	messages := make([]*Message, 0, len(request.Messages))
 	for i := 0; i < len(request.Messages); i++ {
 		message := request.Messages[i]
-		messages = append(messages, Message{
+		messages = append(messages, &Message{
 			Content: message.StringContent(),
 			Role:    message.Role,
 		})
 	}
-	stream := 0
-	if request.Stream {
-		stream = 1
-	}
 	return &ChatRequest{
-		Timestamp:   helper.GetTimestamp(),
+		Model:       &request.Model,
-		Expired:     helper.GetTimestamp() + 24*60*60,
+		Stream:      &request.Stream,
-		QueryID:     random.GetUUID(),
-		Temperature: request.Temperature,
-		TopP:        request.TopP,
-		Stream:      stream,
 		Messages:    messages,
+		TopP:        &request.TopP,
+		Temperature: &request.Temperature,
 	}
 }
 
@@ -54,7 +46,11 @@ func responseTencent2OpenAI(response *ChatResponse) *openai.TextResponse {
 	fullTextResponse := openai.TextResponse{
 		Object:  "chat.completion",
 		Created: helper.GetTimestamp(),
-		Usage:   response.Usage,
+		Usage: model.Usage{
+			PromptTokens:     response.Usage.PromptTokens,
+			CompletionTokens: response.Usage.CompletionTokens,
+			TotalTokens:      response.Usage.TotalTokens,
+		},
 	}
 	if len(response.Choices) > 0 {
 		choice := openai.TextResponseChoice{
@@ -154,6 +150,7 @@ func StreamHandler(c *gin.Context, resp *http.Response) (*model.ErrorWithStatusC
 
 func Handler(c *gin.Context, resp *http.Response) (*model.ErrorWithStatusCode, *model.Usage) {
 	var TencentResponse ChatResponse
+	var responseP ChatResponseP
 	responseBody, err := io.ReadAll(resp.Body)
 	if err != nil {
 		return openai.ErrorWrapper(err, "read_response_body_failed", http.StatusInternalServerError), nil
@@ -162,10 +159,11 @@ func Handler(c *gin.Context, resp *http.Response) (*model.ErrorWithStatusCode, *
 	if err != nil {
 		return openai.ErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), nil
 	}
-	err = json.Unmarshal(responseBody, &TencentResponse)
+	err = json.Unmarshal(responseBody, &responseP)
 	if err != nil {
 		return openai.ErrorWrapper(err, "unmarshal_response_body_failed", http.StatusInternalServerError), nil
 	}
+	TencentResponse = responseP.Response
 	if TencentResponse.Error.Code != 0 {
 		return &model.ErrorWithStatusCode{
 			Error: model.Error{
@@ -202,29 +200,62 @@ func ParseConfig(config string) (appId int64, secretId string, secretKey string,
 	return
 }
 
-func GetSign(req ChatRequest, secretKey string) string {
+func sha256hex(s string) string {
-	params := make([]string, 0)
+	b := sha256.Sum256([]byte(s))
-	params = append(params, "app_id="+strconv.FormatInt(req.AppId, 10))
+	return hex.EncodeToString(b[:])
-	params = append(params, "secret_id="+req.SecretId)
+}
-	params = append(params, "timestamp="+strconv.FormatInt(req.Timestamp, 10))
+
-	params = append(params, "query_id="+req.QueryID)
+func hmacSha256(s, key string) string {
-	params = append(params, "temperature="+strconv.FormatFloat(req.Temperature, 'f', -1, 64))
+	hashed := hmac.New(sha256.New, []byte(key))
-	params = append(params, "top_p="+strconv.FormatFloat(req.TopP, 'f', -1, 64))
+	hashed.Write([]byte(s))
-	params = append(params, "stream="+strconv.Itoa(req.Stream))
+	return string(hashed.Sum(nil))
-	params = append(params, "expired="+strconv.FormatInt(req.Expired, 10))
+}
-
+
-	var messageStr string
+func GetSign(req ChatRequest, adaptor *Adaptor, secId, secKey string) string {
-	for _, msg := range req.Messages {
+	// build canonical request string
-		messageStr += fmt.Sprintf(`{"role":"%s","content":"%s"},`, msg.Role, msg.Content)
+	host := "hunyuan.tencentcloudapi.com"
-	}
+	httpRequestMethod := "POST"
-	messageStr = strings.TrimSuffix(messageStr, ",")
+	canonicalURI := "/"
-	params = append(params, "messages=["+messageStr+"]")
+	canonicalQueryString := ""
-
+	canonicalHeaders := fmt.Sprintf("content-type:%s\nhost:%s\nx-tc-action:%s\n",
-	sort.Strings(params)
+		"application/json", host, strings.ToLower(adaptor.Action))
-	url := "hunyuan.cloud.tencent.com/hyllm/v1/chat/completions?" + strings.Join(params, "&")
+	signedHeaders := "content-type;host;x-tc-action"
-	mac := hmac.New(sha1.New, []byte(secretKey))
+	payload, _ := json.Marshal(req)
-	signURL := url
+	hashedRequestPayload := sha256hex(string(payload))
-	mac.Write([]byte(signURL))
+	canonicalRequest := fmt.Sprintf("%s\n%s\n%s\n%s\n%s\n%s",
-	sign := mac.Sum([]byte(nil))
+		httpRequestMethod,
-	return base64.StdEncoding.EncodeToString(sign)
+		canonicalURI,
+		canonicalQueryString,
+		canonicalHeaders,
+		signedHeaders,
+		hashedRequestPayload)
+	// build string to sign
+	algorithm := "TC3-HMAC-SHA256"
+	requestTimestamp := strconv.FormatInt(adaptor.Timestamp, 10)
+	timestamp, _ := strconv.ParseInt(requestTimestamp, 10, 64)
+	t := time.Unix(timestamp, 0).UTC()
+	// must be the format 2006-01-02, ref to package time for more info
+	date := t.Format("2006-01-02")
+	credentialScope := fmt.Sprintf("%s/%s/tc3_request", date, "hunyuan")
+	hashedCanonicalRequest := sha256hex(canonicalRequest)
+	string2sign := fmt.Sprintf("%s\n%s\n%s\n%s",
+		algorithm,
+		requestTimestamp,
+		credentialScope,
+		hashedCanonicalRequest)
+
+	// sign string
+	secretDate := hmacSha256(date, "TC3"+secKey)
+	secretService := hmacSha256("hunyuan", secretDate)
+	secretKey := hmacSha256("tc3_request", secretService)
+	signature := hex.EncodeToString([]byte(hmacSha256(string2sign, secretKey)))
+
+	// build authorization
+	authorization := fmt.Sprintf("%s Credential=%s/%s, SignedHeaders=%s, Signature=%s",
+		algorithm,
+		secId,
+		credentialScope,
+		signedHeaders,
+		signature)
+	return authorization
 }

relay/adaptor/tencent/model.go

@@ -1,63 +1,75 @@
 package tencent
 
-import (
-	"github.com/songquanpeng/one-api/relay/model"
-)
-
 type Message struct {
-	Role    string `json:"role"`
+	Role    string `json:"Role"`
-	Content string `json:"content"`
+	Content string `json:"Content"`
 }
 
 type ChatRequest struct {
-	AppId    int64  `json:"app_id"`    // 腾讯云账号的 APPID
+	// 模型名称，可选值包括 hunyuan-lite、hunyuan-standard、hunyuan-standard-256K、hunyuan-pro。
-	SecretId string `json:"secret_id"` // 官网 SecretId
+	// 各模型介绍请阅读 [产品概述](https://cloud.tencent.com/document/product/1729/104753) 中的说明。
-	// Timestamp当前 UNIX 时间戳，单位为秒，可记录发起 API 请求的时间。
+	//
-	// 例如1529223702，如果与当前时间相差过大，会引起签名过期错误
+	// 注意：
-	Timestamp int64 `json:"timestamp"`
+	// 不同的模型计费不同，请根据 [购买指南](https://cloud.tencent.com/document/product/1729/97731) 按需调用。
-	// Expired 签名的有效期，是一个符合 UNIX Epoch 时间戳规范的数值，
+	Model *string `json:"Model"`
-	// 单位为秒；Expired 必须大于 Timestamp 且 Expired-Timestamp 小于90天
+	// 聊天上下文信息。
-	Expired int64  `json:"expired"`
+	// 说明：
-	QueryID string `json:"query_id"` //请求 Id，用于问题排查
+	// 1. 长度最多为 40，按对话时间从旧到新在数组中排列。
-	// Temperature 较高的数值会使输出更加随机，而较低的数值会使其更加集中和确定
+	// 2. Message.Role 可选值：system、user、assistant。
-	// 默认 1.0，取值区间为[0.0,2.0]，非必要不建议使用,不合理的取值会影响效果
+	// 其中，system 角色可选，如存在则必须位于列表的最开始。user 和 assistant 需交替出现（一问一答），以 user 提问开始和结束，且 Content 不能为空。Role 的顺序示例：[system（可选） user assistant user assistant user ...]。
-	// 建议该参数和 top_p 只设置1个，不要同时更改 top_p
+	// 3. Messages 中 Content 总长度不能超过模型输入长度上限（可参考 [产品概述](https://cloud.tencent.com/document/product/1729/104753) 文档），超过则会截断最前面的内容，只保留尾部内容。
-	Temperature float64 `json:"temperature"`
+	Messages []*Message `json:"Messages"`
-	// TopP 影响输出文本的多样性，取值越大，生成文本的多样性越强
+	// 流式调用开关。
-	// 默认1.0，取值区间为[0.0, 1.0]，非必要不建议使用, 不合理的取值会影响效果
+	// 说明：
-	// 建议该参数和 temperature 只设置1个，不要同时更改
+	// 1. 未传值时默认为非流式调用（false）。
-	TopP float64 `json:"top_p"`
+	// 2. 流式调用时以 SSE 协议增量返回结果（返回值取 Choices[n].Delta 中的值，需要拼接增量数据才能获得完整结果）。
-	// Stream 0：同步，1：流式 （默认，协议：SSE)
+	// 3. 非流式调用时：
-	// 同步请求超时：60s，如果内容较长建议使用流式
+	// 调用方式与普通 HTTP 请求无异。
-	Stream int `json:"stream"`
+	// 接口响应耗时较长，**如需更低时延建议设置为 true**。
-	// Messages 会话内容, 长度最多为40, 按对话时间从旧到新在数组中排列
+	// 只返回一次最终结果（返回值取 Choices[n].Message 中的值）。
-	// 输入 content 总数最大支持 3000 token。
+	//
-	Messages []Message `json:"messages"`
+	// 注意：
+	// 通过 SDK 调用时，流式和非流式调用需用**不同的方式**获取返回值，具体参考 SDK 中的注释或示例（在各语言 SDK 代码仓库的 examples/hunyuan/v20230901/ 目录中）。
+	Stream *bool `json:"Stream"`
+	// 说明：
+	// 1. 影响输出文本的多样性，取值越大，生成文本的多样性越强。
+	// 2. 取值区间为 [0.0, 1.0]，未传值时使用各模型推荐值。
+	// 3. 非必要不建议使用，不合理的取值会影响效果。
+	TopP *float64 `json:"TopP"`
+	// 说明：
+	// 1. 较高的数值会使输出更加随机，而较低的数值会使其更加集中和确定。
+	// 2. 取值区间为 [0.0, 2.0]，未传值时使用各模型推荐值。
+	// 3. 非必要不建议使用，不合理的取值会影响效果。
+	Temperature *float64 `json:"Temperature"`
 }
 
 type Error struct {
-	Code    int    `json:"code"`
+	Code    int    `json:"Code"`
-	Message string `json:"message"`
+	Message string `json:"Message"`
 }
 
 type Usage struct {
-	InputTokens  int `json:"input_tokens"`
+	PromptTokens     int `json:"PromptTokens"`
-	OutputTokens int `json:"output_tokens"`
+	CompletionTokens int `json:"CompletionTokens"`
-	TotalTokens  int `json:"total_tokens"`
+	TotalTokens      int `json:"TotalTokens"`
 }
 
 type ResponseChoices struct {
-	FinishReason string  `json:"finish_reason,omitempty"` // 流式结束标志位，为 stop 则表示尾包
+	FinishReason string  `json:"FinishReason,omitempty"` // 流式结束标志位，为 stop 则表示尾包
-	Messages     Message `json:"messages,omitempty"`      // 内容，同步模式返回内容，流模式为 null 输出 content 内容总数最多支持 1024token。
+	Messages     Message `json:"Message,omitempty"`      // 内容，同步模式返回内容，流模式为 null 输出 content 内容总数最多支持 1024token。
-	Delta        Message `json:"delta,omitempty"`         // 内容，流模式返回内容，同步模式为 null 输出 content 内容总数最多支持 1024token。
+	Delta        Message `json:"Delta,omitempty"`        // 内容，流模式返回内容，同步模式为 null 输出 content 内容总数最多支持 1024token。
 }
 
 type ChatResponse struct {
-	Choices []ResponseChoices `json:"choices,omitempty"` // 结果
+	Choices []ResponseChoices `json:"Choices,omitempty"` // 结果
-	Created string            `json:"created,omitempty"` // unix 时间戳的字符串
+	Created int64             `json:"Created,omitempty"` // unix 时间戳的字符串
-	Id      string            `json:"id,omitempty"`      // 会话 id
+	Id      string            `json:"Id,omitempty"`      // 会话 id
-	Usage   model.Usage       `json:"usage,omitempty"`   // token 数量
+	Usage   Usage             `json:"Usage,omitempty"`   // token 数量
-	Error   Error             `json:"error,omitempty"`   // 错误信息 注意：此字段可能返回 null，表示取不到有效值
+	Error   Error             `json:"Error,omitempty"`   // 错误信息 注意：此字段可能返回 null，表示取不到有效值
-	Note    string            `json:"note,omitempty"`    // 注释
+	Note    string            `json:"Note,omitempty"`    // 注释
-	ReqID   string            `json:"req_id,omitempty"`  // 唯一请求 Id，每次请求都会返回。用于反馈接口入参
+	ReqID   string            `json:"Req_id,omitempty"`  // 唯一请求 Id，每次请求都会返回。用于反馈接口入参
+}
+
+type ChatResponseP struct {
+	Response ChatResponse `json:"Response,omitempty"`
 }

relay/adaptor/xunfei/adaptor.go

@@ -27,14 +27,6 @@ func (a *Adaptor) GetRequestURL(meta *meta.Meta) (string, error) {
 
 func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, meta *meta.Meta) error {
 	adaptor.SetupCommonRequestHeader(c, req, meta)
-	version := parseAPIVersionByModelName(meta.ActualModelName)
-	if version == "" {
-		version = a.meta.Config.APIVersion
-	}
-	if version == "" {
-		version = "v1.1"
-	}
-	a.meta.Config.APIVersion = version
 	// check DoResponse for auth part
 	return nil
 }
@@ -69,6 +61,14 @@ func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, meta *meta.Met
 	if a.request == nil {
 		return nil, openai.ErrorWrapper(errors.New("request is nil"), "request_is_nil", http.StatusBadRequest)
 	}
+	version := parseAPIVersionByModelName(meta.ActualModelName)
+	if version == "" {
+		version = a.meta.Config.APIVersion
+	}
+	if version == "" {
+		version = "v1.1"
+	}
+	a.meta.Config.APIVersion = version
 	if meta.IsStream {
 		err, usage = StreamHandler(c, meta, *a.request, splits[0], splits[1], splits[2])
 	} else {

relay/adaptor/xunfei/main.go

@@ -5,7 +5,14 @@ import (
 	"crypto/sha256"
 	"encoding/base64"
 	"encoding/json"
+	"errors"
 	"fmt"
+	"io"
+	"net/http"
+	"net/url"
+	"strings"
+	"time"
+
 	"github.com/gin-gonic/gin"
 	"github.com/gorilla/websocket"
 	"github.com/songquanpeng/one-api/common"
@@ -16,11 +23,6 @@ import (
 	"github.com/songquanpeng/one-api/relay/constant"
 	"github.com/songquanpeng/one-api/relay/meta"
 	"github.com/songquanpeng/one-api/relay/model"
-	"io"
-	"net/http"
-	"net/url"
-	"strings"
-	"time"
 )
 
 // https://console.xfyun.cn/services/cbm
@@ -28,11 +30,7 @@ import (
 
 func requestOpenAI2Xunfei(request model.GeneralOpenAIRequest, xunfeiAppId string, domain string) *ChatRequest {
 	messages := make([]Message, 0, len(request.Messages))
-	var lastToolCalls []model.Tool
 	for _, message := range request.Messages {
-		if message.ToolCalls != nil {
-			lastToolCalls = message.ToolCalls
-		}
 		messages = append(messages, Message{
 			Role:    message.Role,
 			Content: message.StringContent(),
@@ -45,9 +43,10 @@ func requestOpenAI2Xunfei(request model.GeneralOpenAIRequest, xunfeiAppId string
 	xunfeiRequest.Parameter.Chat.TopK = request.N
 	xunfeiRequest.Parameter.Chat.MaxTokens = request.MaxTokens
 	xunfeiRequest.Payload.Message.Text = messages
-	if len(lastToolCalls) != 0 {
+
-		for _, toolCall := range lastToolCalls {
+	if strings.HasPrefix(domain, "generalv3") {
-			xunfeiRequest.Payload.Functions.Text = append(xunfeiRequest.Payload.Functions.Text, toolCall.Function)
+		xunfeiRequest.Payload.Functions = &Functions{
+			Text: request.Tools,
 		}
 	}
 
@@ -203,7 +202,7 @@ func Handler(c *gin.Context, meta *meta.Meta, textRequest model.GeneralOpenAIReq
 		}
 	}
 	if len(xunfeiResponse.Payload.Choices.Text) == 0 {
-		return openai.ErrorWrapper(err, "xunfei_empty_response_detected", http.StatusInternalServerError), nil
+		return openai.ErrorWrapper(errors.New("xunfei empty response detected"), "xunfei_empty_response_detected", http.StatusInternalServerError), nil
 	}
 	xunfeiResponse.Payload.Choices.Text[0].Content = content
 

relay/adaptor/xunfei/model.go

@@ -9,6 +9,10 @@ type Message struct {
 	Content string `json:"content"`
 }
 
+type Functions struct {
+	Text []model.Tool `json:"text,omitempty"`
+}
+
 type ChatRequest struct {
 	Header struct {
 		AppId string `json:"app_id"`
@@ -26,9 +30,7 @@ type ChatRequest struct {
 		Message struct {
 			Text []Message `json:"text"`
 		} `json:"message"`
-		Functions struct {
+		Functions *Functions `json:"functions,omitempty"`
-			Text []model.Function `json:"text,omitempty"`
-		} `json:"functions,omitempty"`
 	} `json:"payload"`
 }
 

relay/channeltype/url.go

@@ -24,7 +24,7 @@ var ChannelBaseURLs = []string{
 	"https://openrouter.ai/api",                 // 20
 	"https://api.aiproxy.io",                    // 21
 	"https://fastgpt.run/api/openapi",           // 22
-	"https://hunyuan.cloud.tencent.com",         // 23
+	"https://hunyuan.tencentcloudapi.com",       // 23
 	"https://generativelanguage.googleapis.com", // 24
 	"https://api.moonshot.cn",                   // 25
 	"https://api.baichuan-ai.com",               // 26

web/berry/src/views/Channel/component/EditModal.js

@@ -163,7 +163,7 @@ const EditModal = ({ open, channelId, onCancel, onOk }) => {
       values.other = 'v2.1';
     }
     if (values.key === '') {
-      if (values.config.ak !== '' && values.config.sk !== '' && values.config.region !== '') {
+      if (values.config.ak && values.config.sk && values.config.region) {
         values.key = `${values.config.ak}|${values.config.sk}|${values.config.region}`;
       }
     }

web/default/src/pages/Channel/EditChannel.js

@@ -181,9 +181,6 @@ const EditChannel = () => {
     if (localInputs.type === 3 && localInputs.other === '') {
       localInputs.other = '2024-03-01-preview';
     }
-    if (localInputs.type === 18 && localInputs.other === '') {
-      localInputs.other = 'v2.1';
-    }
     let res;
     localInputs.models = localInputs.models.join(',');
     localInputs.group = localInputs.groups.join(',');