Merge branch 'songquanpeng:main' into main

README.md

@@ -87,7 +87,7 @@ _✨ 通过标准的 OpenAI API 格式访问所有的大模型，开箱即用 
 5. 支持**多机部署**，[详见此处](#多机部署)。
 6. 支持**令牌管理**，设置令牌的过期时间和额度。
 7. 支持**兑换码管理**，支持批量生成和导出兑换码，可使用兑换码为账户进行充值。
-8. 支持**通道管理**，批量创建通道。
+8. 支持**渠道管理**，批量创建渠道。
 9. 支持**用户分组**以及**渠道分组**，支持为不同分组设置不同的倍率。
 10. 支持渠道**设置模型列表**。
 11. 支持**查看额度明细**。
@@ -421,16 +421,16 @@ https://openai.justsong.cn
    + 检查你的接口地址和 API Key 有没有填对。
    + 检查是否启用了 HTTPS，浏览器会拦截 HTTPS 域名下的 HTTP 请求。
 6. 报错：`当前分组负载已饱和，请稍后再试`
-   + 上游通道 429 了。
+   + 上游渠道 429 了。
 7. 升级之后我的数据会丢失吗？
    + 如果使用 MySQL，不会。
    + 如果使用 SQLite，需要按照我所给的部署命令挂载 volume 持久化 one-api.db 数据库文件，否则容器重启后数据会丢失。
 8. 升级之前数据库需要做变更吗？
    + 一般情况下不需要，系统将在初始化的时候自动调整。
    + 如果需要的话，我会在更新日志中说明，并给出脚本。
 9. 手动修改数据库后报错：`数据库一致性已被破坏，请联系管理员`？
-   + 这是检测到 ability 表里有些记录的通道 id 是不存在的，这大概率是因为你删了 channel 表里的记录但是没有同步在 ability 表里清理无效的通道。
-   + 对于每一个通道，其所支持的模型都需要有一个专门的 ability 表的记录，表示该通道支持该模型。
+   + 这是检测到 ability 表里有些记录的渠道 id 是不存在的，这大概率是因为你删了 channel 表里的记录但是没有同步在 ability 表里清理无效的渠道。
+   + 对于每一个渠道，其所支持的模型都需要有一个专门的 ability 表的记录，表示该渠道支持该模型。
 
 ## 相关项目
 * [FastGPT](https://github.com/labring/FastGPT): 基于 LLM 大语言模型的知识库问答系统

common/model-ratio.go

@@ -81,9 +81,12 @@ var ModelRatio = map[string]float64{
 	"bge-large-en":    0.002 * RMB,
 	"bge-large-8k":    0.002 * RMB,
 	// https://ai.google.dev/pricing
-	"PaLM-2":            1,
-	"gemini-pro":        1, // $0.00025 / 1k characters -> $0.001 / 1k tokens
-	"gemini-pro-vision": 1, // $0.00025 / 1k characters -> $0.001 / 1k tokens
+	"PaLM-2":                    1,
+	"gemini-pro":                1, // $0.00025 / 1k characters -> $0.001 / 1k tokens
+	"gemini-pro-vision":         1, // $0.00025 / 1k characters -> $0.001 / 1k tokens
+	"gemini-1.0-pro-vision-001": 1,
+	"gemini-1.0-pro-001":        1,
+	"gemini-1.5-pro":            1,
 	// https://open.bigmodel.cn/pricing
 	"glm-4":                     0.1 * RMB,
 	"glm-4v":                    0.1 * RMB,
@@ -249,6 +252,9 @@ func GetCompletionRatio(name string) float64 {
 	if strings.HasPrefix(name, "mistral-") {
 		return 3
 	}
+	if strings.HasPrefix(name, "gemini-") {
+		return 3
+	}
 	switch name {
 	case "llama2-70b-4096":
 		return 0.8 / 0.7

controller/channel-test.go

@@ -197,7 +197,7 @@ func testChannels(notify bool, scope string) error {
 		testAllChannelsRunning = false
 		testAllChannelsLock.Unlock()
 		if notify {
-			err := message.Notify(message.ByAll, "通道测试完成", "", "通道测试完成，如果没有收到禁用通知，说明所有通道都正常")
+			err := message.Notify(message.ByAll, "渠道测试完成", "", "渠道测试完成，如果没有收到禁用通知，说明所有渠道都正常")
 			if err != nil {
 				logger.SysError(fmt.Sprintf("failed to send email: %s", err.Error()))
 			}

controller/token.go

@@ -142,6 +142,7 @@ func AddToken(c *gin.Context) {
 	c.JSON(http.StatusOK, gin.H{
 		"success": true,
 		"message": "",
+		"data":    cleanToken,
 	})
 	return
 }

i18n/en.json

@@ -8,12 +8,12 @@
   "确认删除": "Confirm Delete",
   "确认绑定": "Confirm Binding",
   "您正在删除自己的帐户，将清空所有数据且不可恢复": "You are deleting your account, all data will be cleared and unrecoverable.",
-  "\"通道「%s」（#%d）已被禁用\"": "\"Channel %s (#%d) has been disabled\"",
-  "通道「%s」（#%d）已被禁用，原因：%s": "Channel %s (#%d) has been disabled, reason: %s",
+  "\"渠道「%s」（#%d）已被禁用\"": "\"Channel %s (#%d) has been disabled\"",
+  "渠道「%s」（#%d）已被禁用，原因：%s": "Channel %s (#%d) has been disabled, reason: %s",
   "测试已在运行中": "Test is already running",
   "响应时间 %.2fs 超过阈值 %.2fs": "Response time %.2fs exceeds threshold %.2fs",
-  "通道测试完成": "Channel test completed",
-  "通道测试完成，如果没有收到禁用通知，说明所有通道都正常": "Channel test completed, if you have not received the disable notification, it means that all channels are normal",
+  "渠道测试完成": "Channel test completed",
+  "渠道测试完成，如果没有收到禁用通知，说明所有渠道都正常": "Channel test completed, if you have not received the disable notification, it means that all channels are normal",
   "无法连接至 GitHub 服务器，请稍后重试！": "Unable to connect to GitHub server, please try again later!",
   "返回值非法，用户字段为空，请稍后重试！": "The return value is illegal, the user field is empty, please try again later!",
   "管理员未开启通过 GitHub 登录以及注册": "The administrator did not turn on login and registration via GitHub",
@@ -119,11 +119,11 @@
   " 个月 ": " M ",
   " 年 ": " y ",
   "未测试": "Not tested",
-  "通道 ${name} 测试成功，耗时 ${time.toFixed(2)} 秒。": "Channel ${name} test succeeded, time consumed ${time.toFixed(2)} s.",
-  "已成功开始测试所有通道，请刷新页面查看结果。": "All channels have been successfully tested, please refresh the page to view the results.",
-  "已成功开始测试所有已启用通道，请刷新页面查看结果。": "All enabled channels have been successfully tested, please refresh the page to view the results.",
-  "通道 ${name} 余额更新成功！": "Channel ${name} balance updated successfully!",
-  "已更新完毕所有已启用通道余额！": "The balance of all enabled channels has been updated!",
+  "渠道 ${name} 测试成功，耗时 ${time.toFixed(2)} 秒。": "Channel ${name} test succeeded, time consumed ${time.toFixed(2)} s.",
+  "已成功开始测试所有渠道，请刷新页面查看结果。": "All channels have been successfully tested, please refresh the page to view the results.",
+  "已成功开始测试所有已启用渠道，请刷新页面查看结果。": "All enabled channels have been successfully tested, please refresh the page to view the results.",
+  "渠道 ${name} 余额更新成功！": "Channel ${name} balance updated successfully!",
+  "已更新完毕所有已启用渠道余额！": "The balance of all enabled channels has been updated!",
   "搜索渠道的 ID，名称和密钥 ...": "Search for channel ID, name and key ...",
   "名称": "Name",
   "分组": "Group",
@@ -141,9 +141,9 @@
   "启用": "Enable",
   "编辑": "Edit",
   "添加新的渠道": "Add a new channel",
-  "测试所有通道": "Test all channels",
-  "测试所有已启用通道": "Test all enabled channels",
-  "更新所有已启用通道余额": "Update the balance of all enabled channels",
+  "测试所有渠道": "Test all channels",
+  "测试所有已启用渠道": "Test all enabled channels",
+  "更新所有已启用渠道余额": "Update the balance of all enabled channels",
   "刷新": "Refresh",
   "处理中...": "Processing...",
   "绑定成功！": "Binding succeeded!",
@@ -207,11 +207,11 @@
   "监控设置": "Monitoring Settings",
   "最长响应时间": "Longest Response Time",
   "单位秒": "Unit in seconds",
-  "当运行通道全部测试时": "When all operating channels are tested",
-  "超过此时间将自动禁用通道": "Channels will be automatically disabled if this time is exceeded",
+  "当运行渠道全部测试时": "When all operating channels are tested",
+  "超过此时间将自动禁用渠道": "Channels will be automatically disabled if this time is exceeded",
   "额度提醒阈值": "Quota reminder threshold",
   "低于此额度时将发送邮件提醒用户": "Email will be sent to remind users when the quota is below this",
-  "失败时自动禁用通道": "Automatically disable the channel when it fails",
+  "失败时自动禁用渠道": "Automatically disable the channel when it fails",
   "保存监控设置": "Save Monitoring Settings",
   "额度设置": "Quota Settings",
   "新用户初始额度": "Initial quota for new users",
@@ -405,7 +405,7 @@
   "镜像": "Mirror",
   "请输入镜像站地址，格式为：https://domain.com，可不填，不填则使用渠道默认值": "Please enter the mirror site address, the format is: https://domain.com, it can be left blank, if left blank, the default value of the channel will be used",
   "模型": "Model",
-  "请选择该通道所支持的模型": "Please select the model supported by the channel",
+  "请选择该渠道所支持的模型": "Please select the model supported by the channel",
   "填入基础模型": "Fill in the basic model",
   "填入所有模型": "Fill in all models",
   "清除所有模型": "Clear all models",
@@ -515,7 +515,7 @@
   "请输入自定义渠道的 Base URL": "Please enter the Base URL of the custom channel",
   "Homepage URL 填": "Fill in the Homepage URL",
   "Authorization callback URL 填": "Fill in the Authorization callback URL",
-  "请为通道命名": "Please name the channel",
+  "请为渠道命名": "Please name the channel",
   "此项可选，用于修改请求体中的模型名称，为一个 JSON 字符串，键为请求中模型名称，值为要替换的模型名称，例如：": "This is optional, used to modify the model name in the request body, it's a JSON string, the key is the model name in the request, and the value is the model name to be replaced, for example:",
   "模型重定向": "Model redirection",
   "请输入渠道对应的鉴权密钥": "Please enter the authentication key corresponding to the channel",

monitor/channel.go

@@ -31,25 +31,25 @@ func notifyRootUser(subject string, content string) {
 func DisableChannel(channelId int, channelName string, reason string) {
 	model.UpdateChannelStatusById(channelId, common.ChannelStatusAutoDisabled)
 	logger.SysLog(fmt.Sprintf("channel #%d has been disabled: %s", channelId, reason))
-	subject := fmt.Sprintf("通道「%s」（#%d）已被禁用", channelName, channelId)
-	content := fmt.Sprintf("通道「%s」（#%d）已被禁用，原因：%s", channelName, channelId, reason)
+	subject := fmt.Sprintf("渠道「%s」（#%d）已被禁用", channelName, channelId)
+	content := fmt.Sprintf("渠道「%s」（#%d）已被禁用，原因：%s", channelName, channelId, reason)
 	notifyRootUser(subject, content)
 }
 
 func MetricDisableChannel(channelId int, successRate float64) {
 	model.UpdateChannelStatusById(channelId, common.ChannelStatusAutoDisabled)
 	logger.SysLog(fmt.Sprintf("channel #%d has been disabled due to low success rate: %.2f", channelId, successRate*100))
-	subject := fmt.Sprintf("通道 #%d 已被禁用", channelId)
-	content := fmt.Sprintf("该渠道在最近 %d 次调用中成功率为 %.2f%%，低于阈值 %.2f%%，因此被系统自动禁用。",
-		config.MetricQueueSize, successRate*100, config.MetricSuccessRateThreshold*100)
+	subject := fmt.Sprintf("渠道 #%d 已被禁用", channelId)
+	content := fmt.Sprintf("该渠道（#%d）在最近 %d 次调用中成功率为 %.2f%%，低于阈值 %.2f%%，因此被系统自动禁用。",
+		channelId, config.MetricQueueSize, successRate*100, config.MetricSuccessRateThreshold*100)
 	notifyRootUser(subject, content)
 }
 
 // EnableChannel enable & notify
 func EnableChannel(channelId int, channelName string) {
 	model.UpdateChannelStatusById(channelId, common.ChannelStatusEnabled)
 	logger.SysLog(fmt.Sprintf("channel #%d has been enabled", channelId))
-	subject := fmt.Sprintf("通道「%s」（#%d）已被启用", channelName, channelId)
-	content := fmt.Sprintf("通道「%s」（#%d）已被启用", channelName, channelId)
+	subject := fmt.Sprintf("渠道「%s」（#%d）已被启用", channelName, channelId)
+	content := fmt.Sprintf("渠道「%s」（#%d）已被启用", channelName, channelId)
 	notifyRootUser(subject, content)
 }

relay/channel/gemini/constants.go

@@ -3,6 +3,6 @@ package gemini
 // https://ai.google.dev/models/gemini
 
 var ModelList = []string{
-	"gemini-pro", "gemini-1.0-pro-001",
+	"gemini-pro", "gemini-1.0-pro-001", "gemini-1.5-pro",
 	"gemini-pro-vision", "gemini-1.0-pro-vision-001",
 }

relay/channel/ollama/adaptor.go

@@ -3,13 +3,14 @@ package ollama
 import (
 	"errors"
 	"fmt"
+	"io"
+	"net/http"
+
 	"github.com/gin-gonic/gin"
 	"github.com/songquanpeng/one-api/relay/channel"
 	"github.com/songquanpeng/one-api/relay/constant"
 	"github.com/songquanpeng/one-api/relay/model"
 	"github.com/songquanpeng/one-api/relay/util"
-	"io"
-	"net/http"
 )
 
 type Adaptor struct {
@@ -22,6 +23,9 @@ func (a *Adaptor) Init(meta *util.RelayMeta) {
 func (a *Adaptor) GetRequestURL(meta *util.RelayMeta) (string, error) {
 	// https://github.com/ollama/ollama/blob/main/docs/api.md
 	fullRequestURL := fmt.Sprintf("%s/api/chat", meta.BaseURL)
+	if meta.Mode == constant.RelayModeEmbeddings {
+		fullRequestURL = fmt.Sprintf("%s/api/embeddings", meta.BaseURL)
+	}
 	return fullRequestURL, nil
 }
 
@@ -37,7 +41,8 @@ func (a *Adaptor) ConvertRequest(c *gin.Context, relayMode int, request *model.G
 	}
 	switch relayMode {
 	case constant.RelayModeEmbeddings:
-		return nil, errors.New("not supported")
+		ollamaEmbeddingRequest := ConvertEmbeddingRequest(*request)
+		return ollamaEmbeddingRequest, nil
 	default:
 		return ConvertRequest(*request), nil
 	}
@@ -51,7 +56,12 @@ func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, meta *util.Rel
 	if meta.IsStream {
 		err, usage = StreamHandler(c, resp)
 	} else {
-		err, usage = Handler(c, resp)
+		switch meta.Mode {
+		case constant.RelayModeEmbeddings:
+			err, usage = EmbeddingHandler(c, resp)
+		default:
+			err, usage = Handler(c, resp)
+		}
 	}
 	return
 }

relay/channel/ollama/main.go

@@ -5,16 +5,17 @@ import (
 	"context"
 	"encoding/json"
 	"fmt"
+	"io"
+	"net/http"
+	"strings"
+
 	"github.com/gin-gonic/gin"
 	"github.com/songquanpeng/one-api/common"
 	"github.com/songquanpeng/one-api/common/helper"
 	"github.com/songquanpeng/one-api/common/logger"
 	"github.com/songquanpeng/one-api/relay/channel/openai"
 	"github.com/songquanpeng/one-api/relay/constant"
 	"github.com/songquanpeng/one-api/relay/model"
-	"io"
-	"net/http"
-	"strings"
 )
 
 func ConvertRequest(request model.GeneralOpenAIRequest) *ChatRequest {
@@ -139,6 +140,64 @@ func StreamHandler(c *gin.Context, resp *http.Response) (*model.ErrorWithStatusC
 	return nil, &usage
 }
 
+func ConvertEmbeddingRequest(request model.GeneralOpenAIRequest) *EmbeddingRequest {
+	return &EmbeddingRequest{
+		Model:  request.Model,
+		Prompt: strings.Join(request.ParseInput(), " "),
+	}
+}
+
+func EmbeddingHandler(c *gin.Context, resp *http.Response) (*model.ErrorWithStatusCode, *model.Usage) {
+	var ollamaResponse EmbeddingResponse
+	err := json.NewDecoder(resp.Body).Decode(&ollamaResponse)
+	if err != nil {
+		return openai.ErrorWrapper(err, "unmarshal_response_body_failed", http.StatusInternalServerError), nil
+	}
+
+	err = resp.Body.Close()
+	if err != nil {
+		return openai.ErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), nil
+	}
+
+	if ollamaResponse.Error != "" {
+		return &model.ErrorWithStatusCode{
+			Error: model.Error{
+				Message: ollamaResponse.Error,
+				Type:    "ollama_error",
+				Param:   "",
+				Code:    "ollama_error",
+			},
+			StatusCode: resp.StatusCode,
+		}, nil
+	}
+
+	fullTextResponse := embeddingResponseOllama2OpenAI(&ollamaResponse)
+	jsonResponse, err := json.Marshal(fullTextResponse)
+	if err != nil {
+		return openai.ErrorWrapper(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, &fullTextResponse.Usage
+}
+
+func embeddingResponseOllama2OpenAI(response *EmbeddingResponse) *openai.EmbeddingResponse {
+	openAIEmbeddingResponse := openai.EmbeddingResponse{
+		Object: "list",
+		Data:   make([]openai.EmbeddingResponseItem, 0, 1),
+		Model:  "text-embedding-v1",
+		Usage:  model.Usage{TotalTokens: 0},
+	}
+
+	openAIEmbeddingResponse.Data = append(openAIEmbeddingResponse.Data, openai.EmbeddingResponseItem{
+		Object:    `embedding`,
+		Index:     0,
+		Embedding: response.Embedding,
+	})
+	return &openAIEmbeddingResponse
+}
+
 func Handler(c *gin.Context, resp *http.Response) (*model.ErrorWithStatusCode, *model.Usage) {
 	ctx := context.TODO()
 	var ollamaResponse ChatResponse

relay/channel/ollama/model.go

@@ -35,3 +35,13 @@ type ChatResponse struct {
 	EvalDuration    int     `json:"eval_duration,omitempty"`
 	Error           string  `json:"error,omitempty"`
 }
+
+type EmbeddingRequest struct {
+	Model  string `json:"model"`
+	Prompt string `json:"prompt"`
+}
+
+type EmbeddingResponse struct {
+	Error     string    `json:"error,omitempty"`
+	Embedding []float64 `json:"embedding,omitempty"`
+}

relay/channel/openai/adaptor.go

@@ -31,11 +31,8 @@ func (a *Adaptor) GetRequestURL(meta *util.RelayMeta) (string, error) {
 		task := strings.TrimPrefix(requestURL, "/v1/")
 		model_ := meta.ActualModelName
 		model_ = strings.Replace(model_, ".", "", -1)
-		// https://github.com/songquanpeng/one-api/issues/67
-		model_ = strings.TrimSuffix(model_, "-0301")
-		model_ = strings.TrimSuffix(model_, "-0314")
-		model_ = strings.TrimSuffix(model_, "-0613")
-
+		//https://github.com/songquanpeng/one-api/issues/1191
+		// {your endpoint}/openai/deployments/{your azure_model}/chat/completions?api-version={api_version}
 		requestURL = fmt.Sprintf("/openai/deployments/%s/%s", model_, task)
 		return util.GetFullRequestURL(meta.BaseURL, requestURL, meta.ChannelType), nil
 	case common.ChannelTypeMinimax: