Merge remote-tracking branch 'remotes/origin_songquanpeng/main'

# Conflicts:
#	relay/adaptor/openai/adaptor.go
#	relay/controller/text.go

.env.example

@@ -0,0 +1,3 @@
+PORT=3000
+DEBUG=false
+HTTPS_PROXY=http://localhost:7890

.github/workflows/ci.yml

@@ -0,0 +1,47 @@
+name: CI
+
+# This setup assumes that you run the unit tests with code coverage in the same
+# workflow that will also print the coverage report as comment to the pull request. 
+# Therefore, you need to trigger this workflow when a pull request is (re)opened or
+# when new code is pushed to the branch of the pull request. In addition, you also
+# need to trigger this workflow when new code is pushed to the main branch because 
+# we need to upload the code coverage results as artifact for the main branch as
+# well since it will be the baseline code coverage.
+# 
+# We do not want to trigger the workflow for pushes to *any* branch because this
+# would trigger our jobs twice on pull requests (once from "push" event and once
+# from "pull_request->synchronize")
+on:
+  pull_request:
+    types: [opened, reopened, synchronize]
+  push:
+    branches:
+      - 'main'
+
+jobs:
+  unit_tests:
+    name: "Unit tests"
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v4
+
+      - name: Setup Go
+        uses: actions/setup-go@v4
+        with:
+          go-version: ^1.22
+
+      # When you execute your unit tests, make sure to use the "-coverprofile" flag to write a 
+      # coverage profile to a file. You will need the name of the file (e.g. "coverage.txt")
+      # in the next step as well as the next job.
+      - name: Test
+        run: go test -cover -coverprofile=coverage.txt ./...
+      - uses: codecov/codecov-action@v4
+        with:
+          token: ${{ secrets.CODECOV_TOKEN }}
+
+  commit_lint:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v3
+      - uses: wagoid/commitlint-github-action@v6

.github/workflows/docker-image-amd64.yml

@@ -1,61 +0,0 @@
-name: Publish Docker image (amd64)
-
-on:
-  push:
-    tags:
-      - 'v*.*.*'
-  workflow_dispatch:
-    inputs:
-      name:
-        description: 'reason'
-        required: false
-jobs:
-  push_to_registries:
-    name: Push Docker image to multiple registries
-    runs-on: ubuntu-latest
-    permissions:
-      packages: write
-      contents: read
-    steps:
-      - name: Check out the repo
-        uses: actions/checkout@v3
-
-      - name: Check repository URL
-        run: |
-          REPO_URL=$(git config --get remote.origin.url)
-          if [[ $REPO_URL == *"pro" ]]; then
-            exit 1
-          fi        
-
-      - name: Save version info
-        run: |
-          git describe --tags > VERSION 
-
-      - name: Log in to Docker Hub
-        uses: docker/login-action@v2
-        with:
-          username: ${{ secrets.DOCKERHUB_USERNAME }}
-          password: ${{ secrets.DOCKERHUB_TOKEN }}
-
-      - name: Log in to the Container registry
-        uses: docker/login-action@v2
-        with:
-          registry: ghcr.io
-          username: ${{ github.actor }}
-          password: ${{ secrets.GITHUB_TOKEN }}
-
-      - name: Extract metadata (tags, labels) for Docker
-        id: meta
-        uses: docker/metadata-action@v4
-        with:
-          images: |
-            justsong/one-api
-            ghcr.io/${{ github.repository }}
-
-      - name: Build and push Docker images
-        uses: docker/build-push-action@v3
-        with:
-          context: .
-          push: true
-          tags: ${{ steps.meta.outputs.tags }}
-          labels: ${{ steps.meta.outputs.labels }}

.github/workflows/docker-image-en.yml

@@ -1,4 +1,4 @@
-name: Publish Docker image (amd64, English)
+name: Publish Docker image (English)
 
 on:
   push:
@@ -34,6 +34,13 @@ jobs:
       - name: Translate
         run: |
           python ./i18n/translate.py --repository_path . --json_file_path ./i18n/en.json
+
+      - name: Set up QEMU
+        uses: docker/setup-qemu-action@v2
+
+      - name: Set up Docker Buildx
+        uses: docker/setup-buildx-action@v2
+
       - name: Log in to Docker Hub
         uses: docker/login-action@v2
         with:
@@ -51,6 +58,7 @@ jobs:
         uses: docker/build-push-action@v3
         with:
           context: .
+          platforms: linux/amd64,linux/arm64
           push: true
           tags: ${{ steps.meta.outputs.tags }}
           labels: ${{ steps.meta.outputs.labels }}

.github/workflows/docker-image.yml

@@ -1,10 +1,9 @@
-name: Publish Docker image (arm64)
+name: Publish Docker image
 
 on:
   push:
     tags:
       - 'v*.*.*'
-      - '!*-alpha*'
   workflow_dispatch:
     inputs:
       name:

.gitignore

@@ -9,3 +9,4 @@ logs
 data
 /web/node_modules
 cmd.md
+.env

Dockerfile

@@ -1,4 +1,4 @@
-FROM node:16 as builder
+FROM --platform=$BUILDPLATFORM node:16 AS builder
 
 WORKDIR /web
 COPY ./VERSION .
@@ -17,7 +17,9 @@ WORKDIR /web/air
 RUN npm install --force
 RUN DISABLE_ESLINT_PLUGIN='true' REACT_APP_VERSION=$(cat VERSION) npm run build
 
-FROM golang AS builder2
+FROM golang:alpine AS builder2
+
+RUN apk add --no-cache g++
 
 ENV GO111MODULE=on \
     CGO_ENABLED=1 \
@@ -29,7 +31,7 @@ ADD go.mod go.sum ./
 RUN go mod download
 COPY . .
 COPY --from=builder /web/build ./web/build
-RUN go build -ldflags "-s -w -X 'github.com/songquanpeng/one-api/common.Version=$(cat VERSION)' -extldflags '-static'" -o one-api
+RUN go build -trimpath -ldflags "-s -w -X 'github.com/songquanpeng/one-api/common.Version=$(cat VERSION)' -extldflags '-static'" -o one-api
 
 FROM alpine
 

README.en.md

@@ -245,16 +245,41 @@ If the channel ID is not provided, load balancing will be used to distribute the
     + Example: `LOG_SQL_DSN=root:123456@tcp(localhost:3306)/oneapi-logs`
 5. `FRONTEND_BASE_URL`: When set, the specified frontend address will be used instead of the backend address.
     + Example: `FRONTEND_BASE_URL=https://openai.justsong.cn`
-6. `SYNC_FREQUENCY`: When set, the system will periodically sync configurations from the database, with the unit in seconds. If not set, no sync will happen.
+6. 'MEMORY_CACHE_ENABLED': Enabling memory caching can cause a certain delay in updating user quotas, with optional values of 'true' and 'false'. If not set, it defaults to 'false'.
+7. `SYNC_FREQUENCY`: When set, the system will periodically sync configurations from the database, with the unit in seconds. If not set, no sync will happen.
     + Example: `SYNC_FREQUENCY=60`
-7. `NODE_TYPE`: When set, specifies the node type. Valid values are `master` and `slave`. If not set, it defaults to `master`.
+8. `NODE_TYPE`: When set, specifies the node type. Valid values are `master` and `slave`. If not set, it defaults to `master`.
     + Example: `NODE_TYPE=slave`
-8. `CHANNEL_UPDATE_FREQUENCY`: When set, it periodically updates the channel balances, with the unit in minutes. If not set, no update will happen.
+9. `CHANNEL_UPDATE_FREQUENCY`: When set, it periodically updates the channel balances, with the unit in minutes. If not set, no update will happen.
     + Example: `CHANNEL_UPDATE_FREQUENCY=1440`
-9. `CHANNEL_TEST_FREQUENCY`: When set, it periodically tests the channels, with the unit in minutes. If not set, no test will happen.
+10. `CHANNEL_TEST_FREQUENCY`: When set, it periodically tests the channels, with the unit in minutes. If not set, no test will happen.
     + Example: `CHANNEL_TEST_FREQUENCY=1440`
-10. `POLLING_INTERVAL`: The time interval (in seconds) between requests when updating channel balances and testing channel availability. Default is no interval.
+11. `POLLING_INTERVAL`: The time interval (in seconds) between requests when updating channel balances and testing channel availability. Default is no interval.
     + Example: `POLLING_INTERVAL=5`
+12. `BATCH_UPDATE_ENABLED`: Enabling batch database update aggregation can cause a certain delay in updating user quotas. The optional values are 'true' and 'false', but if not set, it defaults to 'false'.
+    +Example: ` BATCH_UPDATE_ENABLED=true`
+    +If you encounter an issue with too many database connections, you can try enabling this option.
+13. `BATCH_UPDATE_INTERVAL=5`: The time interval for batch updating aggregates, measured in seconds, defaults to '5'.
+    +Example: ` BATCH_UPDATE_INTERVAL=5`
+14. Request frequency limit:
+    + `GLOBAL_API_RATE_LIMIT`: Global API rate limit (excluding relay requests), the maximum number of requests within three minutes per IP, default to 180.
+    + `GLOBAL_WEL_RATE_LIMIT`: Global web speed limit, the maximum number of requests within three minutes per IP, default to 60.
+15. Encoder cache settings:
+    +`TIKTOKEN_CACHE_DIR`: By default, when the program starts, it will download the encoding of some common word elements online, such as' gpt-3.5 turbo '. In some unstable network environments or offline situations, it may cause startup problems. This directory can be configured to cache data and can be migrated to an offline environment.
+    +`DATA_GYM_CACHE_DIR`: Currently, this configuration has the same function as' TIKTOKEN-CACHE-DIR ', but its priority is not as high as it.
+16. `RELAY_TIMEOUT`: Relay timeout setting, measured in seconds, with no default timeout time set.
+17. `RELAY_PROXY`: After setting up, use this proxy to request APIs.
+18. `USER_CONTENT_REQUEST_TIMEOUT`: The timeout period for users to upload and download content, measured in seconds.
+19. `USER_CONTENT_REQUEST_PROXY`: After setting up, use this agent to request content uploaded by users, such as images.
+20. `SQLITE_BUSY_TIMEOUT`: SQLite lock wait timeout setting, measured in milliseconds, default to '3000'.
+21. `GEMINI_SAFETY_SETTING`: Gemini's security settings are set to 'BLOCK-NONE' by default.
+22. `GEMINI_VERSION`: The Gemini version used by the One API, which defaults to 'v1'.
+23. `THE`: The system's theme setting, default to 'default', specific optional values refer to [here] (./web/README. md).
+24. `ENABLE_METRIC`: Whether to disable channels based on request success rate, default not enabled, optional values are 'true' and 'false'.
+25. `METRIC_QUEUE_SIZE`: Request success rate statistics queue size, default to '10'.
+26. `METRIC_SUCCESS_RATE_THRESHOLD`: Request success rate threshold, default to '0.8'.
+27. `INITIAL_ROOT_TOKEN`: If this value is set, a root user token with the value of the environment variable will be automatically created when the system starts for the first time.
+28. `INITIAL_ROOT_ACCESS_TOKEN`: If this value is set, a system management token will be automatically created for the root user with a value of the environment variable when the system starts for the first time.
 
 ### Command Line Parameters
 1. `--port <port_number>`: Specifies the port number on which the server listens. Defaults to `3000`.
@@ -287,7 +312,9 @@ If the channel ID is not provided, load balancing will be used to distribute the
     + Double-check that your interface address and API Key are correct.
 
 ## Related Projects
-[FastGPT](https://github.com/labring/FastGPT): Knowledge question answering system based on the LLM
+* [FastGPT](https://github.com/labring/FastGPT): Knowledge question answering system based on the LLM
+* [VChart](https://github.com/VisActor/VChart):  More than just a cross-platform charting library, but also an expressive data storyteller.
+* [VMind](https://github.com/VisActor/VMind):  Not just automatic, but also fantastic. Open-source solution for intelligent visualization.
 
 ## Note
 This project is an open-source project. Please use it in compliance with OpenAI's [Terms of Use](https://openai.com/policies/terms-of-use) and **applicable laws and regulations**. It must not be used for illegal purposes.

README.md

@@ -88,6 +88,8 @@ _✨ 通过标准的 OpenAI API 格式访问所有的大模型，开箱即用
    + [x] [Cloudflare Workers AI](https://developers.cloudflare.com/workers-ai/)
    + [x] [DeepL](https://www.deepl.com/)
    + [x] [together.ai](https://www.together.ai/)
+   + [x] [novita.ai](https://www.novita.ai/)
+   + [x] [硅基流动 SiliconCloud](https://siliconflow.cn/siliconcloud)
 2. 支持配置镜像以及众多[第三方代理服务](https://iamazing.cn/page/openai-api-third-party-services)。
 3. 支持通过**负载均衡**的方式访问多个渠道。
 4. 支持 **stream 模式**，可以通过流式传输实现打字机效果。
@@ -250,9 +252,9 @@ docker run --name chatgpt-web -d -p 3002:3002 -e OPENAI_API_BASE_URL=https://ope
 #### QChatGPT - QQ机器人
 项目主页：https://github.com/RockChinQ/QChatGPT
 
-根据文档完成部署后，在`config.py`设置配置项`openai_config`的`reverse_proxy`为 One API 后端地址，设置`api_key`为 One API 生成的key，并在配置项`completion_api_params`的`model`参数设置为 One API 支持的模型名称。
+根据[文档](https://qchatgpt.rockchin.top)完成部署后，在 `data/provider.json`设置`requester.openai-chat-completions.base-url`为 One API 实例地址，并填写 API Key 到 `keys.openai` 组中，设置 `model` 为要使用的模型名称。
 
-可安装 [Switcher 插件](https://github.com/RockChinQ/Switcher)在运行时切换所使用的模型。
+运行期间可以通过`!model`命令查看、切换可用模型。
 
 ### 部署到第三方平台
 <details>
@@ -340,6 +342,7 @@ graph LR
 不加的话将会使用负载均衡的方式使用多个渠道。
 
 ### 环境变量
+> One API 支持从 `.env` 文件中读取环境变量，请参照 `.env.example` 文件，使用时请将其重命名为 `.env`。
 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`
    + 如果数据库访问延迟很低，没有必要启用 Redis，启用后反而会出现数据滞后的问题。
@@ -369,32 +372,33 @@ graph LR
 9. `CHANNEL_UPDATE_FREQUENCY`：设置之后将定期更新渠道余额，单位为分钟，未设置则不进行更新。
    + 例子：`CHANNEL_UPDATE_FREQUENCY=1440`
 10. `CHANNEL_TEST_FREQUENCY`：设置之后将定期检查渠道，单位为分钟，未设置则不进行检查。 
-11. 例子：`CHANNEL_TEST_FREQUENCY=1440`
+   +例子：`CHANNEL_TEST_FREQUENCY=1440`
-12. `POLLING_INTERVAL`：批量更新渠道余额以及测试可用性时的请求间隔，单位为秒，默认无间隔。
+11. `POLLING_INTERVAL`：批量更新渠道余额以及测试可用性时的请求间隔，单位为秒，默认无间隔。
     + 例子：`POLLING_INTERVAL=5`
-13. `BATCH_UPDATE_ENABLED`：启用数据库批量更新聚合，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`。
+12. `BATCH_UPDATE_ENABLED`：启用数据库批量更新聚合，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：`BATCH_UPDATE_ENABLED=true`
     + 如果你遇到了数据库连接数过多的问题，可以尝试启用该选项。
-14. `BATCH_UPDATE_INTERVAL=5`：批量更新聚合的时间间隔，单位为秒，默认为 `5`。
+13. `BATCH_UPDATE_INTERVAL=5`：批量更新聚合的时间间隔，单位为秒，默认为 `5`。
     + 例子：`BATCH_UPDATE_INTERVAL=5`
-15. 请求频率限制：
+14. 请求频率限制：
     + `GLOBAL_API_RATE_LIMIT`：全局 API 速率限制（除中继请求外），单 ip 三分钟内的最大请求数，默认为 `180`。
     + `GLOBAL_WEB_RATE_LIMIT`：全局 Web 速率限制，单 ip 三分钟内的最大请求数，默认为 `60`。
-16. 编码器缓存设置：
+15. 编码器缓存设置：
     + `TIKTOKEN_CACHE_DIR`：默认程序启动时会联网下载一些通用的词元的编码，如：`gpt-3.5-turbo`，在一些网络环境不稳定，或者离线情况，可能会导致启动有问题，可以配置此目录缓存数据，可迁移到离线环境。
     + `DATA_GYM_CACHE_DIR`：目前该配置作用与 `TIKTOKEN_CACHE_DIR` 一致，但是优先级没有它高。
-17. `RELAY_TIMEOUT`：中继超时设置，单位为秒，默认不设置超时时间。
+16. `RELAY_TIMEOUT`：中继超时设置，单位为秒，默认不设置超时时间。
-18. `RELAY_PROXY`：设置后使用该代理来请求 API。
+17. `RELAY_PROXY`：设置后使用该代理来请求 API。
-19. `USER_CONTENT_REQUEST_TIMEOUT`：用户上传内容下载超时时间，单位为秒。
+18. `USER_CONTENT_REQUEST_TIMEOUT`：用户上传内容下载超时时间，单位为秒。
-20. `USER_CONTENT_REQUEST_PROXY`：设置后使用该代理来请求用户上传的内容，例如图片。
+19. `USER_CONTENT_REQUEST_PROXY`：设置后使用该代理来请求用户上传的内容，例如图片。
-21. `SQLITE_BUSY_TIMEOUT`：SQLite 锁等待超时设置，单位为毫秒，默认 `3000`。
+20. `SQLITE_BUSY_TIMEOUT`：SQLite 锁等待超时设置，单位为毫秒，默认 `3000`。
-22. `GEMINI_SAFETY_SETTING`：Gemini 的安全设置，默认 `BLOCK_NONE`。
+21. `GEMINI_SAFETY_SETTING`：Gemini 的安全设置，默认 `BLOCK_NONE`。
-23. `GEMINI_VERSION`：One API 所使用的 Gemini 版本，默认为 `v1`。
+22. `GEMINI_VERSION`：One API 所使用的 Gemini 版本，默认为 `v1`。
-24. `THEME`：系统的主题设置，默认为 `default`，具体可选值参考[此处](./web/README.md)。
+23. `THEME`：系统的主题设置，默认为 `default`，具体可选值参考[此处](./web/README.md)。
-25. `ENABLE_METRIC`：是否根据请求成功率禁用渠道，默认不开启，可选值为 `true` 和 `false`。
+24. `ENABLE_METRIC`：是否根据请求成功率禁用渠道，默认不开启，可选值为 `true` 和 `false`。
-26. `METRIC_QUEUE_SIZE`：请求成功率统计队列大小，默认为 `10`。
+25. `METRIC_QUEUE_SIZE`：请求成功率统计队列大小，默认为 `10`。
-27. `METRIC_SUCCESS_RATE_THRESHOLD`：请求成功率阈值，默认为 `0.8`。
+26. `METRIC_SUCCESS_RATE_THRESHOLD`：请求成功率阈值，默认为 `0.8`。
-28. `INITIAL_ROOT_TOKEN`：如果设置了该值，则在系统首次启动时会自动创建一个值为该环境变量值的 root 用户令牌。
+27. `INITIAL_ROOT_TOKEN`：如果设置了该值，则在系统首次启动时会自动创建一个值为该环境变量值的 root 用户令牌。
+28. `INITIAL_ROOT_ACCESS_TOKEN`：如果设置了该值，则在系统首次启动时会自动创建一个值为该环境变量的 root 用户创建系统管理令牌。
 
 ### 命令行参数
 1. `--port <port_number>`: 指定服务器监听的端口号，默认为 `3000`。
@@ -447,6 +451,8 @@ https://openai.justsong.cn
 ## 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
 * [ChatGPT Next Web](https://github.com/Yidadaa/ChatGPT-Next-Web):  一键拥有你自己的跨平台 ChatGPT 应用
+* [VChart](https://github.com/VisActor/VChart):  不只是开箱即用的多端图表库，更是生动灵活的数据故事讲述者。
+* [VMind](https://github.com/VisActor/VMind):  不仅自动，还很智能。开源智能可视化解决方案。
 
 ## 注意
 

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
@@ -143,8 +143,12 @@ var MetricFailChanSize = env.Int("METRIC_FAIL_CHAN_SIZE", 128)
 
 var InitialRootToken = os.Getenv("INITIAL_ROOT_TOKEN")
 
+var InitialRootAccessToken = os.Getenv("INITIAL_ROOT_ACCESS_TOKEN")
+
 var GeminiVersion = env.String("GEMINI_VERSION", "v1")
 
+var OnlyOneLogFile = env.Bool("ONLY_ONE_LOG_FILE", false)
+
 var RelayProxy = env.String("RELAY_PROXY", "")
 var UserContentRequestProxy = env.String("USER_CONTENT_REQUEST_PROXY", "")
 var UserContentRequestTimeout = env.Int("USER_CONTENT_REQUEST_TIMEOUT", 30)

common/ctxkey/key.go

@@ -19,4 +19,5 @@ const (
 	TokenName         = "token_name"
 	BaseURL           = "base_url"
 	AvailableModels   = "available_models"
+	KeyRequestBody    = "key_request_body"
 )

common/gin.go

@@ -4,14 +4,13 @@ import (
 	"bytes"
 	"encoding/json"
 	"github.com/gin-gonic/gin"
+	"github.com/songquanpeng/one-api/common/ctxkey"
 	"io"
 	"strings"
 )
 
-const KeyRequestBody = "key_request_body"
-
 func GetRequestBody(c *gin.Context) ([]byte, error) {
-	requestBody, _ := c.Get(KeyRequestBody)
+	requestBody, _ := c.Get(ctxkey.KeyRequestBody)
 	if requestBody != nil {
 		return requestBody.([]byte), nil
 	}
@@ -20,7 +19,7 @@ func GetRequestBody(c *gin.Context) ([]byte, error) {
 		return nil, err
 	}
 	_ = c.Request.Body.Close()
-	c.Set(KeyRequestBody, requestBody)
+	c.Set(ctxkey.KeyRequestBody, requestBody)
 	return requestBody.([]byte), nil
 }
 

common/image/image_test.go

@@ -2,6 +2,7 @@ package image_test
 
 import (
 	"encoding/base64"
+	"github.com/songquanpeng/one-api/common/client"
 	"image"
 	_ "image/gif"
 	_ "image/jpeg"
@@ -44,6 +45,11 @@ var (
 	}
 )
 
+func TestMain(m *testing.M) {
+	client.Init()
+	m.Run()
+}
+
 func TestDecode(t *testing.T) {
 	// Bytes read: varies sometimes
 	// jpeg: 1063892

common/init.go

@@ -24,7 +24,7 @@ func printHelp() {
 	fmt.Println("Usage: one-api [--port <port>] [--log-dir <log directory>] [--version] [--help]")
 }
 
-func init() {
+func Init() {
 	flag.Parse()
 
 	if *PrintVersion {

common/logger/logger.go

@@ -27,7 +27,12 @@ var setupLogOnce sync.Once
 func SetupLogger() {
 	setupLogOnce.Do(func() {
 		if LogDir != "" {
-			logPath := filepath.Join(LogDir, fmt.Sprintf("oneapi-%s.log", time.Now().Format("20060102")))
+			var logPath string
+			if config.OnlyOneLogFile {
+				logPath = filepath.Join(LogDir, "oneapi.log")
+			} else {
+				logPath = filepath.Join(LogDir, fmt.Sprintf("oneapi-%s.log", time.Now().Format("20060102")))
+			}
 			fd, err := os.OpenFile(logPath, os.O_APPEND|os.O_CREATE|os.O_WRONLY, 0644)
 			if err != nil {
 				log.Fatal("failed to open log file")
@@ -43,11 +48,19 @@ 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)

common/message/email.go

@@ -6,11 +6,16 @@ import (
 	"encoding/base64"
 	"fmt"
 	"github.com/songquanpeng/one-api/common/config"
+	"net"
 	"net/smtp"
 	"strings"
 	"time"
 )
 
+func shouldAuth() bool {
+	return config.SMTPAccount != "" || config.SMTPToken != ""
+}
+
 func SendEmail(subject string, receiver string, content string) error {
 	if receiver == "" {
 		return fmt.Errorf("receiver is empty")
@@ -41,16 +46,24 @@ func SendEmail(subject string, receiver string, content string) error {
 		"Date: %s\r\n"+
 		"Content-Type: text/html; charset=UTF-8\r\n\r\n%s\r\n",
 		receiver, config.SystemName, config.SMTPFrom, encodedSubject, messageId, time.Now().Format(time.RFC1123Z), content))
+
 	auth := smtp.PlainAuth("", config.SMTPAccount, config.SMTPToken, config.SMTPServer)
 	addr := fmt.Sprintf("%s:%d", config.SMTPServer, config.SMTPPort)
 	to := strings.Split(receiver, ";")
 
+	if config.SMTPPort == 465 || !shouldAuth() {
+		// need advanced client
+		var conn net.Conn
+		var err error
 		if config.SMTPPort == 465 {
 			tlsConfig := &tls.Config{
 				InsecureSkipVerify: true,
 				ServerName:         config.SMTPServer,
 			}
-		conn, err := tls.Dial("tcp", fmt.Sprintf("%s:%d", config.SMTPServer, config.SMTPPort), tlsConfig)
+			conn, err = tls.Dial("tcp", fmt.Sprintf("%s:%d", config.SMTPServer, config.SMTPPort), tlsConfig)
+		} else {
+			conn, err = net.Dial("tcp", fmt.Sprintf("%s:%d", config.SMTPServer, config.SMTPPort))
+		}
 		if err != nil {
 			return err
 		}
@@ -59,9 +72,11 @@ func SendEmail(subject string, receiver string, content string) error {
 			return err
 		}
 		defer client.Close()
+		if shouldAuth() {
 			if err = client.Auth(auth); err != nil {
 				return err
 			}
+		}
 		if err = client.Mail(config.SMTPFrom); err != nil {
 			return err
 		}

common/render/render.go

@@ -0,0 +1,29 @@
+package render
+
+import (
+	"encoding/json"
+	"fmt"
+	"github.com/gin-gonic/gin"
+	"github.com/songquanpeng/one-api/common"
+	"strings"
+)
+
+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]")
+}

controller/channel-test.go

@@ -14,6 +14,7 @@ import (
 	"sync"
 	"time"
 
+	"github.com/gin-gonic/gin"
 	"github.com/songquanpeng/one-api/common/config"
 	"github.com/songquanpeng/one-api/common/ctxkey"
 	"github.com/songquanpeng/one-api/common/logger"
@@ -27,15 +28,15 @@ import (
 	"github.com/songquanpeng/one-api/relay/meta"
 	relaymodel "github.com/songquanpeng/one-api/relay/model"
 	"github.com/songquanpeng/one-api/relay/relaymode"
-
-	"github.com/gin-gonic/gin"
 )
 
-func buildTestRequest() *relaymodel.GeneralOpenAIRequest {
+func buildTestRequest(model string) *relaymodel.GeneralOpenAIRequest {
+	if model == "" {
+		model = "gpt-3.5-turbo"
+	}
 	testRequest := &relaymodel.GeneralOpenAIRequest{
 		MaxTokens: 2,
-		Stream:    false,
+		Model:     model,
-		Model:     "gpt-3.5-turbo",
 	}
 	testMessage := relaymodel.Message{
 		Role:    "user",
@@ -45,7 +46,7 @@ func buildTestRequest() *relaymodel.GeneralOpenAIRequest {
 	return testRequest
 }
 
-func testChannel(channel *model.Channel) (err error, openaiErr *relaymodel.Error) {
+func testChannel(channel *model.Channel, request *relaymodel.GeneralOpenAIRequest) (err error, openaiErr *relaymodel.Error) {
 	w := httptest.NewRecorder()
 	c, _ := gin.CreateTestContext(w)
 	c.Request = &http.Request{
@@ -68,12 +69,8 @@ func testChannel(channel *model.Channel) (err error, openaiErr *relaymodel.Error
 		return fmt.Errorf("invalid api type: %d, adaptor is nil", apiType), nil
 	}
 	adaptor.Init(meta)
-	var modelName string
+	modelName := request.Model
-	modelList := adaptor.GetModelList()
 	modelMap := channel.GetModelMapping()
-	if len(modelList) != 0 {
-		modelName = modelList[0]
-	}
 	if modelName == "" || !strings.Contains(channel.Models, modelName) {
 		modelNames := strings.Split(channel.Models, ",")
 		if len(modelNames) > 0 {
@@ -83,9 +80,8 @@ func testChannel(channel *model.Channel) (err error, openaiErr *relaymodel.Error
 			modelName = modelMap[modelName]
 		}
 	}
-	request := buildTestRequest()
+	meta.OriginModelName, meta.ActualModelName = request.Model, modelName
 	request.Model = modelName
-	meta.OriginModelName, meta.ActualModelName = modelName, modelName
 	convertedRequest, err := adaptor.ConvertRequest(c, relaymode.ChatCompletions, request)
 	if err != nil {
 		return err, nil
@@ -139,10 +135,15 @@ func TestChannel(c *gin.Context) {
 		})
 		return
 	}
+	model := c.Query("model")
+	testRequest := buildTestRequest(model)
 	tik := time.Now()
-	err, _ = testChannel(channel)
+	err, _ = testChannel(channel, testRequest)
 	tok := time.Now()
 	milliseconds := tok.Sub(tik).Milliseconds()
+	if err != nil {
+		milliseconds = 0
+	}
 	go channel.UpdateResponseTime(milliseconds)
 	consumedTime := float64(milliseconds) / 1000.0
 	if err != nil {
@@ -150,6 +151,7 @@ func TestChannel(c *gin.Context) {
 			"success": false,
 			"message": err.Error(),
 			"time":    consumedTime,
+			"model":   model,
 		})
 		return
 	}
@@ -157,6 +159,7 @@ func TestChannel(c *gin.Context) {
 		"success": true,
 		"message": "",
 		"time":    consumedTime,
+		"model":   model,
 	})
 	return
 }
@@ -187,11 +190,12 @@ func testChannels(notify bool, scope string) error {
 		for _, channel := range channels {
 			isChannelEnabled := channel.Status == model.ChannelStatusEnabled
 			tik := time.Now()
-			err, openaiErr := testChannel(channel)
+			testRequest := buildTestRequest("")
+			err, openaiErr := testChannel(channel, testRequest)
 			tok := time.Now()
 			milliseconds := tok.Sub(tik).Milliseconds()
 			if isChannelEnabled && milliseconds > disableThreshold {
-				err = errors.New(fmt.Sprintf("响应时间 %.2fs 超过阈值 %.2fs", float64(milliseconds)/1000.0, float64(disableThreshold)/1000.0))
+				err = fmt.Errorf("响应时间 %.2fs 超过阈值 %.2fs", float64(milliseconds)/1000.0, float64(disableThreshold)/1000.0)
 				if config.AutomaticDisableChannelEnabled {
 					monitor.DisableChannel(channel.Id, channel.Name, err.Error())
 				} else {

controller/relay.go

@@ -34,6 +34,8 @@ func relayHelper(c *gin.Context, relayMode int) *model.ErrorWithStatusCode {
 		fallthrough
 	case relaymode.AudioTranscription:
 		err = controller.RelayAudioHelper(c, relayMode)
+	case relaymode.Proxy:
+		err = controller.RelayProxyHelper(c, relayMode)
 	default:
 		err = controller.RelayTextHelper(c)
 	}
@@ -48,7 +50,7 @@ func Relay(c *gin.Context) {
 		logger.Debugf(ctx, "request body: %s", string(requestBody))
 	}
 	channelId := c.GetInt(ctxkey.ChannelId)
-	userId := c.GetInt("id")
+	userId := c.GetInt(ctxkey.Id)
 	bizErr := relayHelper(c, relayMode)
 	if bizErr == nil {
 		monitor.Emit(channelId, true)
@@ -85,12 +87,15 @@ func Relay(c *gin.Context) {
 		channelId := c.GetInt(ctxkey.ChannelId)
 		lastFailedChannelId = channelId
 		channelName := c.GetString(ctxkey.ChannelName)
+		// BUG: bizErr is in race condition
 		go processChannelRelayError(ctx, userId, channelId, channelName, bizErr)
 	}
 	if bizErr != nil {
 		if bizErr.StatusCode == http.StatusTooManyRequests {
 			bizErr.Error.Message = "当前分组上游负载已饱和，请稍后再试"
 		}
+
+		// BUG: bizErr is in race condition
 		bizErr.Error.Message = helper.MessageWithRequestId(bizErr.Error.Message, requestId)
 		c.JSON(bizErr.StatusCode, gin.H{
 			"error": bizErr.Error,

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": "",

go.mod

@@ -4,6 +4,7 @@ module github.com/songquanpeng/one-api
 go 1.20
 
 require (
+	cloud.google.com/go/iam v1.1.10
 	github.com/aws/aws-sdk-go-v2 v1.27.0
 	github.com/aws/aws-sdk-go-v2/credentials v1.17.15
 	github.com/aws/aws-sdk-go-v2/service/bedrockruntime v1.8.3
@@ -18,12 +19,15 @@ require (
 	github.com/google/uuid v1.6.0
 	github.com/gorilla/websocket v1.5.1
 	github.com/jinzhu/copier v0.4.0
+	github.com/joho/godotenv v1.5.1
+	github.com/patrickmn/go-cache v2.1.0+incompatible
 	github.com/pkg/errors v0.9.1
 	github.com/pkoukk/tiktoken-go v0.1.7
 	github.com/smartystreets/goconvey v1.8.1
 	github.com/stretchr/testify v1.9.0
-	golang.org/x/crypto v0.23.0
+	golang.org/x/crypto v0.24.0
-	golang.org/x/image v0.16.0
+	golang.org/x/image v0.18.0
+	google.golang.org/api v0.187.0
 	gorm.io/driver/mysql v1.5.6
 	gorm.io/driver/postgres v1.5.7
 	gorm.io/driver/sqlite v1.5.5
@@ -31,6 +35,9 @@ require (
 )
 
 require (
+	cloud.google.com/go/auth v0.6.1 // indirect
+	cloud.google.com/go/auth/oauth2adapt v0.2.2 // indirect
+	cloud.google.com/go/compute/metadata v0.3.0 // indirect
 	filippo.io/edwards25519 v1.1.0 // indirect
 	github.com/aws/aws-sdk-go-v2/aws/protocol/eventstream v1.6.2 // indirect
 	github.com/aws/aws-sdk-go-v2/internal/configsources v1.3.7 // indirect
@@ -44,13 +51,21 @@ require (
 	github.com/davecgh/go-spew v1.1.1 // indirect
 	github.com/dgryski/go-rendezvous v0.0.0-20200823014737-9f7001d12a5f // indirect
 	github.com/dlclark/regexp2 v1.11.0 // indirect
+	github.com/felixge/httpsnoop v1.0.4 // indirect
 	github.com/fsnotify/fsnotify v1.7.0 // indirect
 	github.com/gabriel-vasile/mimetype v1.4.3 // indirect
 	github.com/gin-contrib/sse v0.1.0 // indirect
+	github.com/go-logr/logr v1.4.1 // indirect
+	github.com/go-logr/stdr v1.2.2 // indirect
 	github.com/go-playground/locales v0.14.1 // indirect
 	github.com/go-playground/universal-translator v0.18.1 // indirect
 	github.com/go-sql-driver/mysql v1.8.1 // indirect
 	github.com/goccy/go-json v0.10.3 // indirect
+	github.com/golang/groupcache v0.0.0-20210331224755-41bb18bfe9da // indirect
+	github.com/golang/protobuf v1.5.4 // indirect
+	github.com/google/s2a-go v0.1.7 // indirect
+	github.com/googleapis/enterprise-certificate-proxy v0.3.2 // indirect
+	github.com/googleapis/gax-go/v2 v2.12.5 // indirect
 	github.com/gopherjs/gopherjs v1.17.2 // indirect
 	github.com/gorilla/context v1.1.2 // indirect
 	github.com/gorilla/securecookie v1.1.2 // indirect
@@ -67,7 +82,7 @@ require (
 	github.com/kr/text v0.2.0 // indirect
 	github.com/leodido/go-urn v1.4.0 // indirect
 	github.com/mattn/go-isatty v0.0.20 // indirect
-	github.com/mattn/go-sqlite3 v2.0.3+incompatible // indirect
+	github.com/mattn/go-sqlite3 v1.14.22 // indirect
 	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 v2.2.2 // indirect
@@ -75,11 +90,22 @@ require (
 	github.com/smarty/assertions v1.15.0 // indirect
 	github.com/twitchyliquid64/golang-asm v0.15.1 // indirect
 	github.com/ugorji/go/codec v1.2.12 // indirect
+	go.opencensus.io v0.24.0 // indirect
+	go.opentelemetry.io/contrib/instrumentation/google.golang.org/grpc/otelgrpc v0.49.0 // indirect
+	go.opentelemetry.io/contrib/instrumentation/net/http/otelhttp v0.49.0 // indirect
+	go.opentelemetry.io/otel v1.24.0 // indirect
+	go.opentelemetry.io/otel/metric v1.24.0 // indirect
+	go.opentelemetry.io/otel/trace v1.24.0 // indirect
 	golang.org/x/arch v0.8.0 // indirect
-	golang.org/x/net v0.25.0 // indirect
+	golang.org/x/net v0.26.0 // indirect
+	golang.org/x/oauth2 v0.21.0 // indirect
 	golang.org/x/sync v0.7.0 // indirect
-	golang.org/x/sys v0.20.0 // indirect
+	golang.org/x/sys v0.21.0 // indirect
-	golang.org/x/text v0.15.0 // indirect
+	golang.org/x/text v0.16.0 // indirect
-	google.golang.org/protobuf v1.34.1 // indirect
+	golang.org/x/time v0.5.0 // indirect
+	google.golang.org/genproto/googleapis/api v0.0.0-20240617180043-68d350f18fd4 // indirect
+	google.golang.org/genproto/googleapis/rpc v0.0.0-20240624140628-dc46fd24d27d // indirect
+	google.golang.org/grpc v1.64.1 // indirect
+	google.golang.org/protobuf v1.34.2 // indirect
 	gopkg.in/yaml.v3 v3.0.1 // indirect
 )

go.sum

@@ -1,44 +1,42 @@
+cloud.google.com/go v0.26.0/go.mod h1:aQUYkXzVsufM+DwF1aE+0xfcU+56JwCaLick0ClmMTw=
+cloud.google.com/go/auth v0.6.1 h1:T0Zw1XM5c1GlpN2HYr2s+m3vr1p2wy+8VN+Z1FKxW38=
+cloud.google.com/go/auth v0.6.1/go.mod h1:eFHG7zDzbXHKmjJddFG/rBlcGp6t25SwRUiEQSlO4x4=
+cloud.google.com/go/auth/oauth2adapt v0.2.2 h1:+TTV8aXpjeChS9M+aTtN/TjdQnzJvmzKFt//oWu7HX4=
+cloud.google.com/go/auth/oauth2adapt v0.2.2/go.mod h1:wcYjgpZI9+Yu7LyYBg4pqSiaRkfEK3GQcpb7C/uyF1Q=
+cloud.google.com/go/compute/metadata v0.3.0 h1:Tz+eQXMEqDIKRsmY3cHTL6FVaynIjX2QxYC4trgAKZc=
+cloud.google.com/go/compute/metadata v0.3.0/go.mod h1:zFmK7XCadkQkj6TtorcaGlCW1hT1fIilQDwofLpJ20k=
+cloud.google.com/go/iam v1.1.10 h1:ZSAr64oEhQSClwBL670MsJAW5/RLiC6kfw3Bqmd5ZDI=
+cloud.google.com/go/iam v1.1.10/go.mod h1:iEgMq62sg8zx446GCaijmA2Miwg5o3UbO+nI47WHJps=
 filippo.io/edwards25519 v1.1.0 h1:FNf4tywRC1HmFuKW5xopWpigGjJKiJSV0Cqo0cJWDaA=
 filippo.io/edwards25519 v1.1.0/go.mod h1:BxyFTGdWcka3PhytdK4V28tE5sGfRvvvRV7EaN4VDT4=
-github.com/aws/aws-sdk-go-v2 v1.26.1 h1:5554eUqIYVWpU0YmeeYZ0wU64H2VLBs8TlhRB2L+EkA=
+github.com/BurntSushi/toml v0.3.1/go.mod h1:xHWCNGjB5oqiDr8zfno3MHue2Ht5sIBksp03qcyfWMU=
-github.com/aws/aws-sdk-go-v2 v1.26.1/go.mod h1:ffIFB97e2yNsv4aTSGkqtHnppsIJzw7G7BReUZ3jCXM=
 github.com/aws/aws-sdk-go-v2 v1.27.0 h1:7bZWKoXhzI+mMR/HjdMx8ZCC5+6fY0lS5tr0bbgiLlo=
 github.com/aws/aws-sdk-go-v2 v1.27.0/go.mod h1:ffIFB97e2yNsv4aTSGkqtHnppsIJzw7G7BReUZ3jCXM=
 github.com/aws/aws-sdk-go-v2/aws/protocol/eventstream v1.6.2 h1:x6xsQXGSmW6frevwDA+vi/wqhp1ct18mVXYN08/93to=
 github.com/aws/aws-sdk-go-v2/aws/protocol/eventstream v1.6.2/go.mod h1:lPprDr1e6cJdyYeGXnRaJoP4Md+cDBvi2eOj00BlGmg=
-github.com/aws/aws-sdk-go-v2/credentials v1.17.11 h1:YuIB1dJNf1Re822rriUOTxopaHHvIq0l/pX3fwO+Tzs=
-github.com/aws/aws-sdk-go-v2/credentials v1.17.11/go.mod h1:AQtFPsDH9bI2O+71anW6EKL+NcD7LG3dpKGMV4SShgo=
 github.com/aws/aws-sdk-go-v2/credentials v1.17.15 h1:YDexlvDRCA8ems2T5IP1xkMtOZ1uLJOCJdTr0igs5zo=
 github.com/aws/aws-sdk-go-v2/credentials v1.17.15/go.mod h1:vxHggqW6hFNaeNC0WyXS3VdyjcV0a4KMUY4dKJ96buU=
-github.com/aws/aws-sdk-go-v2/internal/configsources v1.3.5 h1:aw39xVGeRWlWx9EzGVnhOR4yOjQDHPQ6o6NmBlscyQg=
-github.com/aws/aws-sdk-go-v2/internal/configsources v1.3.5/go.mod h1:FSaRudD0dXiMPK2UjknVwwTYyZMRsHv3TtkabsZih5I=
 github.com/aws/aws-sdk-go-v2/internal/configsources v1.3.7 h1:lf/8VTF2cM+N4SLzaYJERKEWAXq8MOMpZfU6wEPWsPk=
 github.com/aws/aws-sdk-go-v2/internal/configsources v1.3.7/go.mod h1:4SjkU7QiqK2M9oozyMzfZ/23LmUY+h3oFqhdeP5OMiI=
-github.com/aws/aws-sdk-go-v2/internal/endpoints/v2 v2.6.5 h1:PG1F3OD1szkuQPzDw3CIQsRIrtTlUC3lP84taWzHlq0=
-github.com/aws/aws-sdk-go-v2/internal/endpoints/v2 v2.6.5/go.mod h1:jU1li6RFryMz+so64PpKtudI+QzbKoIEivqdf6LNpOc=
 github.com/aws/aws-sdk-go-v2/internal/endpoints/v2 v2.6.7 h1:4OYVp0705xu8yjdyoWix0r9wPIRXnIzzOoUpQVHIJ/g=
 github.com/aws/aws-sdk-go-v2/internal/endpoints/v2 v2.6.7/go.mod h1:vd7ESTEvI76T2Na050gODNmNU7+OyKrIKroYTu4ABiI=
-github.com/aws/aws-sdk-go-v2/service/bedrockruntime v1.7.4 h1:JgHnonzbnA3pbqj76wYsSZIZZQYBxkmMEjvL6GHy8XU=
-github.com/aws/aws-sdk-go-v2/service/bedrockruntime v1.7.4/go.mod h1:nZspkhg+9p8iApLFoyAqfyuMP0F38acy2Hm3r5r95Cg=
 github.com/aws/aws-sdk-go-v2/service/bedrockruntime v1.8.3 h1:Fihjyd6DeNjcawBEGLH9dkIEUi6AdhucDKPE9nJ4QiY=
 github.com/aws/aws-sdk-go-v2/service/bedrockruntime v1.8.3/go.mod h1:opvUj3ismqSCxYc+m4WIjPL0ewZGtvp0ess7cKvBPOQ=
 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/sonic v1.11.5 h1:G00FYjjqll5iQ1PYXynbg/hyzqBqavH8Mo9/oTopd9k=
-github.com/bytedance/sonic v1.11.5/go.mod h1:X2PC2giUdj/Cv2lliWFLk6c/DUQok5rViJSemeB0wDw=
 github.com/bytedance/sonic v1.11.6 h1:oUp34TzMlL+OY1OUWxHqsdkgC/Zfc85zGqw9siXjrc0=
 github.com/bytedance/sonic v1.11.6/go.mod h1:LysEHSvpvDySVdC2f87zGWf6CIKJcAvqab1ZaiQtds4=
-github.com/bytedance/sonic/loader v0.1.0/go.mod h1:UmRT+IRTGKz/DAkzcEGzyVqQFJ7H9BqwBO3pm9H/+HY=
 github.com/bytedance/sonic/loader v0.1.1 h1:c+e5Pt1k/cy5wMveRDyk2X4B9hF4g7an8N3zCYjJFNM=
 github.com/bytedance/sonic/loader v0.1.1/go.mod h1:ncP89zfokxS5LZrJxl5z0UJcsk4M4yY2JpfqGeCtNLU=
+github.com/census-instrumentation/opencensus-proto v0.2.1/go.mod h1:f6KPmirojxKA12rnyqOA5BBL4O983OfeGPqjHWSTneU=
 github.com/cespare/xxhash/v2 v2.3.0 h1:UL815xU9SqsFlibzuggzjXhog7bL6oX9BbNZnL2UFvs=
 github.com/cespare/xxhash/v2 v2.3.0/go.mod h1:VGX0DQ3Q6kWi7AoAeZDth3/j3BFtOZR5XLFGgcrjCOs=
-github.com/cloudwego/base64x v0.1.3 h1:b5J/l8xolB7dyDTTmhJP2oTs5LdrjyrUFuNxdfq5hAg=
+github.com/client9/misspell v0.3.4/go.mod h1:qj6jICC3Q7zFZvVWo7KLAzC3yx5G7kyvSDkc90ppPyw=
-github.com/cloudwego/base64x v0.1.3/go.mod h1:1+1K5BUHIQzyapgpF7LwvOGAEDicKtt1umPV+aN8pi8=
 github.com/cloudwego/base64x v0.1.4 h1:jwCgWpFanWmN8xoIUHa2rtzmkd5J2plF/dnLS6Xd/0Y=
 github.com/cloudwego/base64x v0.1.4/go.mod h1:0zlkT4Wn5C6NdauXdJRhSKRlJvmclQ1hhJgA0rcu/8w=
 github.com/cloudwego/iasm v0.2.0 h1:1KNIy1I1H9hNNFEEH3DVnI4UujN+1zjpuk6gwHLTssg=
 github.com/cloudwego/iasm v0.2.0/go.mod h1:8rXZaNYT2n95jn+zTI1sDr+IgcD2GVs0nlbbQPiEFhY=
+github.com/cncf/udpa/go v0.0.0-20191209042840-269d4d468f6f/go.mod h1:M8M6+tZqaGXZJjfX53e64911xZQV5JYwmTeXPW+k8Sc=
 github.com/creack/pty v1.1.9/go.mod h1:oKZEueFk5CKHvIhNR5MUki03XCEU+Q6VDXinZuGJ33E=
 github.com/davecgh/go-spew v1.1.0/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
 github.com/davecgh/go-spew v1.1.1 h1:vj9j/u1bqnvCEfJOwUhtlOARqs3+rkHYY13jYWTU97c=
@@ -47,39 +45,38 @@ github.com/dgryski/go-rendezvous v0.0.0-20200823014737-9f7001d12a5f h1:lO4WD4F/r
 github.com/dgryski/go-rendezvous v0.0.0-20200823014737-9f7001d12a5f/go.mod h1:cuUVRXasLTGF7a8hSLbxyZXjz+1KgoB3wDUb6vlszIc=
 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/envoyproxy/go-control-plane v0.9.0/go.mod h1:YTl/9mNaCwkRvm6d1a2C3ymFceY/DCBVvsKhRF0iEA4=
+github.com/envoyproxy/go-control-plane v0.9.1-0.20191026205805-5f8ba28d4473/go.mod h1:YTl/9mNaCwkRvm6d1a2C3ymFceY/DCBVvsKhRF0iEA4=
+github.com/envoyproxy/go-control-plane v0.9.4/go.mod h1:6rpuAdCZL397s3pYoYcLgu1mIlRU8Am5FuJP05cCM98=
+github.com/envoyproxy/protoc-gen-validate v0.1.0/go.mod h1:iSmxcyjqTsJpI2R4NaDN7+kN2VEUnK/pcBlmesArF7c=
+github.com/felixge/httpsnoop v1.0.4 h1:NFTV2Zj1bL4mc9sqWACXbQFVBBg2W3GPvqp8/ESS2Wg=
+github.com/felixge/httpsnoop v1.0.4/go.mod h1:m8KPJKqk1gH5J9DgRY2ASl2lWCfGKXixSwevea8zH2U=
 github.com/fsnotify/fsnotify v1.7.0 h1:8JEhPFa5W2WU7YfeZzPNqzMP6Lwt7L2715Ggo0nosvA=
 github.com/fsnotify/fsnotify v1.7.0/go.mod h1:40Bi/Hjc2AVfZrqy+aj+yEI+/bRxZnMJyTJwOpGvigM=
 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 v1.7.1 h1:s9SIppU/rk8enVvkzwiC2VK3UZ/0NNGsWfUKvV55rqs=
-github.com/gin-contrib/cors v1.7.1/go.mod h1:n/Zj7B4xyrgk/cX1WCX2dkzFfaNm/xJb6oIUk7WTtps=
 github.com/gin-contrib/cors v1.7.2 h1:oLDHxdg8W/XDoN/8zamqk/Drgt4oVZDvaV0YmvVICQw=
 github.com/gin-contrib/cors v1.7.2/go.mod h1:SUJVARKgQ40dmrzgXEVxj2m7Ig1v1qIboQkPDTQ9t2E=
-github.com/gin-contrib/gzip v1.0.0 h1:UKN586Po/92IDX6ie5CWLgMI81obiIp5nSP85T3wlTk=
-github.com/gin-contrib/gzip v1.0.0/go.mod h1:CtG7tQrPB3vIBo6Gat9FVUsis+1emjvQqd66ME5TdnE=
 github.com/gin-contrib/gzip v1.0.1 h1:HQ8ENHODeLY7a4g1Au/46Z92bdGFl74OhxcZble9WJE=
 github.com/gin-contrib/gzip v1.0.1/go.mod h1:njt428fdUNRvjuJf16tZMYZ2Yl+WQB53X5wmhDwXvC4=
-github.com/gin-contrib/sessions v1.0.0 h1:r5GLta4Oy5xo9rAwMHx8B4wLpeRGHMdz9NafzJAdP8Y=
-github.com/gin-contrib/sessions v1.0.0/go.mod h1:DN0f4bvpqMQElDdi+gNGScrP2QEI04IErRyMFyorUOI=
 github.com/gin-contrib/sessions v1.0.1 h1:3hsJyNs7v7N8OtelFmYXFrulAf6zSR7nW/putcPEHxI=
 github.com/gin-contrib/sessions v1.0.1/go.mod h1:ouxSFM24/OgIud5MJYQJLpy6AwxQ5EYO9yLhbtObGkM=
 github.com/gin-contrib/sse v0.1.0 h1:Y/yl/+YNO8GZSjAhjMsSuLt29uWRFHdHYUb5lYOV9qE=
 github.com/gin-contrib/sse v0.1.0/go.mod h1:RHrZQHXnP2xjPF+u1gW/2HnVO7nvIa9PG3Gm+fLHvGI=
-github.com/gin-contrib/static v1.1.1 h1:XEvBd4DDLG1HBlyPBQU1XO8NlTpw6mgdqcPteetYA5k=
-github.com/gin-contrib/static v1.1.1/go.mod h1:yRGmar7+JYvbMLRPIi4H5TVVSBwULfT9vetnVD0IO74=
 github.com/gin-contrib/static v1.1.2 h1:c3kT4bFkUJn2aoRU3s6XnMjJT8J6nNWJkR0NglqmlZ4=
 github.com/gin-contrib/static v1.1.2/go.mod h1:Fw90ozjHCmZBWbgrsqrDvO28YbhKEKzKp8GixhR4yLw=
-github.com/gin-gonic/gin v1.9.1 h1:4idEAncQnU5cB7BeOkPtxjfCSye0AAm1R0RVIqJ+Jmg=
-github.com/gin-gonic/gin v1.9.1/go.mod h1:hPrL7YrpYKXt5YId3A/Tnip5kqbEAP+KLuI3SUcPTeU=
 github.com/gin-gonic/gin v1.10.0 h1:nTuyha1TYqgedzytsKYqna+DfLos46nTv2ygFy86HFU=
 github.com/gin-gonic/gin v1.10.0/go.mod h1:4PMNQiOhvDRa013RKVbsiNwoyezlm2rm0uX/T7kzp5Y=
+github.com/go-logr/logr v1.2.2/go.mod h1:jdQByPbusPIv2/zmleS9BjJVeZ6kBagPoEUsqbVz/1A=
+github.com/go-logr/logr v1.4.1 h1:pKouT5E8xu9zeFC39JXRDukb6JFQPXM5p5I91188VAQ=
+github.com/go-logr/logr v1.4.1/go.mod h1:9T104GzyrTigFIr8wt5mBrctHMim0Nb2HLGrmQ40KvY=
+github.com/go-logr/stdr v1.2.2 h1:hSWxHoqTgW2S2qGc0LTAI563KZ5YKYRhT3MFKZMbjag=
+github.com/go-logr/stdr v1.2.2/go.mod h1:mMo/vtBO5dYbehREoey6XUKy/eSumjCCveDpRre4VKE=
 github.com/go-playground/assert/v2 v2.2.0 h1:JvknZsQTYeFEAhQwI4qEt9cyV5ONwRHC+lYKSsYSR8s=
 github.com/go-playground/locales v0.14.1 h1:EWaQ/wswjilfKLTECiXz7Rh+3BjFhfDFKv/oXslEjJA=
 github.com/go-playground/locales v0.14.1/go.mod h1:hxrqLVvrK65+Rwrd5Fc6F2O76J/NuW9t0sjnWqG1slY=
 github.com/go-playground/universal-translator v0.18.1 h1:Bcnm0ZwsGyWbCzImXv+pAJnYK9S473LQFuzCbDbfSFY=
 github.com/go-playground/universal-translator v0.18.1/go.mod h1:xekY+UJKNuX9WP91TpwSH2VMlDf28Uj24BCp08ZFTUY=
-github.com/go-playground/validator/v10 v10.19.0 h1:ol+5Fu+cSq9JD7SoSqe04GMI92cbn0+wvQ3bZ8b/AU4=
-github.com/go-playground/validator/v10 v10.19.0/go.mod h1:dbuPbCMFw/DrkbEynArYaCwl3amGuJotoKCe95atGMM=
 github.com/go-playground/validator/v10 v10.20.0 h1:K9ISHbSaI0lyB2eWMPJo+kOS/FBExVwjEviJTixqxL8=
 github.com/go-playground/validator/v10 v10.20.0/go.mod h1:dbuPbCMFw/DrkbEynArYaCwl3amGuJotoKCe95atGMM=
 github.com/go-redis/redis/v8 v8.11.5 h1:AcZZR7igkdvfVmQTPnu9WE37LRrO/YrBH5zWyjDC0oI=
@@ -87,17 +84,44 @@ github.com/go-redis/redis/v8 v8.11.5/go.mod h1:gREzHqY1hg6oD9ngVRbLStwAWKhA0FEgq
 github.com/go-sql-driver/mysql v1.7.0/go.mod h1:OXbVy3sEdcQ2Doequ6Z5BW6fXNQTmx+9S1MCJN5yJMI=
 github.com/go-sql-driver/mysql v1.8.1 h1:LedoTUt/eveggdHS9qUFC1EFSa8bU2+1pZjSRpvNJ1Y=
 github.com/go-sql-driver/mysql v1.8.1/go.mod h1:wEBSXgmK//2ZFJyE+qWnIsVGmvmEKlqwuVSjsCm7DZg=
-github.com/goccy/go-json v0.10.2 h1:CrxCmQqYDkv1z7lO7Wbh2HN93uovUHgrECaO5ZrCXAU=
-github.com/goccy/go-json v0.10.2/go.mod h1:6MelG93GURQebXPDq3khkgXZkazVtN9CRI+MGFi0w8I=
 github.com/goccy/go-json v0.10.3 h1:KZ5WoDbxAIgm2HNbYckL0se1fHD6rz5j4ywS6ebzDqA=
 github.com/goccy/go-json v0.10.3/go.mod h1:oq7eo15ShAhp70Anwd5lgX2pLfOS3QCiwU/PULtXL6M=
 github.com/golang-jwt/jwt v3.2.2+incompatible h1:IfV12K8xAKAnZqdXVzCZ+TOjboZ2keLg81eXfW3O+oY=
 github.com/golang-jwt/jwt v3.2.2+incompatible/go.mod h1:8pz2t5EyA70fFQQSrl6XZXzqecmYZeUEB8OUGHkxJ+I=
-github.com/google/go-cmp v0.5.6 h1:BKbKCqvP6I+rmFHt06ZmyQtvB8xAkWdhFyr0ZUNZcxQ=
+github.com/golang/glog v0.0.0-20160126235308-23def4e6c14b/go.mod h1:SBH7ygxi8pfUlaOkMMuAQtPIUF8ecWP5IEl/CR7VP2Q=
+github.com/golang/groupcache v0.0.0-20200121045136-8c9f03a8e57e/go.mod h1:cIg4eruTrX1D+g88fzRXU5OdNfaM+9IcxsU14FzY7Hc=
+github.com/golang/groupcache v0.0.0-20210331224755-41bb18bfe9da h1:oI5xCqsCo564l8iNU+DwB5epxmsaqB+rhGL0m5jtYqE=
+github.com/golang/groupcache v0.0.0-20210331224755-41bb18bfe9da/go.mod h1:cIg4eruTrX1D+g88fzRXU5OdNfaM+9IcxsU14FzY7Hc=
+github.com/golang/mock v1.1.1/go.mod h1:oTYuIxOrZwtPieC+H1uAHpcLFnEyAGVDL/k47Jfbm0A=
+github.com/golang/protobuf v1.2.0/go.mod h1:6lQm79b+lXiMfvg/cZm0SGofjICqVBUtrP5yJMmIC1U=
+github.com/golang/protobuf v1.3.2/go.mod h1:6lQm79b+lXiMfvg/cZm0SGofjICqVBUtrP5yJMmIC1U=
+github.com/golang/protobuf v1.4.0-rc.1/go.mod h1:ceaxUfeHdC40wWswd/P6IGgMaK3YpKi5j83Wpe3EHw8=
+github.com/golang/protobuf v1.4.0-rc.1.0.20200221234624-67d41d38c208/go.mod h1:xKAWHe0F5eneWXFV3EuXVDTCmh+JuBKY0li0aMyXATA=
+github.com/golang/protobuf v1.4.0-rc.2/go.mod h1:LlEzMj4AhA7rCAGe4KMBDvJI+AwstrUpVNzEA03Pprs=
+github.com/golang/protobuf v1.4.0-rc.4.0.20200313231945-b860323f09d0/go.mod h1:WU3c8KckQ9AFe+yFwt9sWVRKCVIyN9cPHBJSNnbL67w=
+github.com/golang/protobuf v1.4.0/go.mod h1:jodUvKwWbYaEsadDk5Fwe5c77LiNKVO9IDvqG2KuDX0=
+github.com/golang/protobuf v1.4.1/go.mod h1:U8fpvMrcmy5pZrNK1lt4xCsGvpyWQ/VVv6QDs8UjoX8=
+github.com/golang/protobuf v1.4.3/go.mod h1:oDoupMAO8OvCJWAcko0GGGIgR6R6ocIYbsSw735rRwI=
+github.com/golang/protobuf v1.5.4 h1:i7eJL8qZTpSEXOPTxNKhASYpMn+8e5Q6AdndVa1dWek=
+github.com/golang/protobuf v1.5.4/go.mod h1:lnTiLA8Wa4RWRcIUkrtSVa5nRhsEGBg48fD6rSs7xps=
+github.com/google/go-cmp v0.2.0/go.mod h1:oXzfMopK8JAjlY9xF4vHSVASa0yLyX7SntLO5aqRK0M=
+github.com/google/go-cmp v0.3.0/go.mod h1:8QqcDgzrUqlUb/G2PQTWiueGozuR1884gddMywk6iLU=
+github.com/google/go-cmp v0.3.1/go.mod h1:8QqcDgzrUqlUb/G2PQTWiueGozuR1884gddMywk6iLU=
+github.com/google/go-cmp v0.4.0/go.mod h1:v8dTdLbMG2kIc/vJvl+f65V22dbkXbowE6jgT/gNBxE=
+github.com/google/go-cmp v0.5.0/go.mod h1:v8dTdLbMG2kIc/vJvl+f65V22dbkXbowE6jgT/gNBxE=
+github.com/google/go-cmp v0.5.3/go.mod h1:v8dTdLbMG2kIc/vJvl+f65V22dbkXbowE6jgT/gNBxE=
+github.com/google/go-cmp v0.6.0 h1:ofyhxvXcZhMsU5ulbFiLKl/XBFqE1GSq7atu8tAmTRI=
 github.com/google/gofuzz v1.0.0/go.mod h1:dBl0BpW6vV/+mYPU4Po3pmUjxk6FQPldtuIdl/M65Eg=
 github.com/google/gofuzz v1.2.0 h1:xRy4A+RhZaiKjJ1bPfwQ8sedCA+YS2YcCHW6ec7JMi0=
+github.com/google/s2a-go v0.1.7 h1:60BLSyTrOV4/haCDW4zb1guZItoSq8foHCXrAnjBo/o=
+github.com/google/s2a-go v0.1.7/go.mod h1:50CgR4k1jNlWBu4UfS4AcfhVe1r6pdZPygJ3R8F0Qdw=
+github.com/google/uuid v1.1.2/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/googleapis/enterprise-certificate-proxy v0.3.2 h1:Vie5ybvEvT75RniqhfFxPRy3Bf7vr3h0cechB90XaQs=
+github.com/googleapis/enterprise-certificate-proxy v0.3.2/go.mod h1:VLSiSSBs/ksPL8kq3OBOQ6WRI2QnaFynd1DCjZ62+V0=
+github.com/googleapis/gax-go/v2 v2.12.5 h1:8gw9KZK8TiVKB6q3zHY3SBzLnrGp6HQjyfYBYGmXdxA=
+github.com/googleapis/gax-go/v2 v2.12.5/go.mod h1:BUDKcWo+RaKq5SC9vVYL0wLADa3VcfswbOMMRmB9H3E=
 github.com/gopherjs/gopherjs v1.17.2 h1:fQnZVsXk8uxXIStYb0N4bGk7jeyTalG/wsZjQ25dO0g=
 github.com/gopherjs/gopherjs v1.17.2/go.mod h1:pRRIvn/QzFLrKfvEz3qUuEhtE/zLCWfreZ6J5gM2i+k=
 github.com/gorilla/context v1.1.2 h1:WRkNAv2uoa03QNIc1A6u4O7DAGMUVoopZhkiXWA2V1o=
@@ -122,6 +146,8 @@ github.com/jinzhu/inflection v1.0.0 h1:K317FqzuhWc8YvSVlFMCCUb36O/S9MCKRDI7QkRKD
 github.com/jinzhu/inflection v1.0.0/go.mod h1:h+uFLlag+Qp1Va5pdKtLDYj+kHp5pxUVkryuEj+Srlc=
 github.com/jinzhu/now v1.1.5 h1:/o9tlHleP7gOFmsnYNz3RGnqzefHA47wQpKrrdTIwXQ=
 github.com/jinzhu/now v1.1.5/go.mod h1:d3SSVoowX0Lcu0IBviAWJpolVfI5UJVZZ7cO71lE/z8=
+github.com/joho/godotenv v1.5.1 h1:7eLL/+HRGLY0ldzfGMeQkb7vMd0as4CfYvUVzLqw0N0=
+github.com/joho/godotenv v1.5.1/go.mod h1:f4LDr5Voq0i2e/R5DDNOoa2zzDfwtkZa6DnEwAbqwq4=
 github.com/json-iterator/go v1.1.12 h1:PV8peI4a0ysnczrg+LtxykD8LfKY9ML6u2jnxaEnrnM=
 github.com/json-iterator/go v1.1.12/go.mod h1:e30LSqwooZae/UwlEbR2852Gd8hjQvJoHmT4TnhNGBo=
 github.com/jtolds/gls v4.20.0+incompatible h1:xdiiI2gbIgH/gLH7ADydsJ1uDOEzR8yvV7C0MuV77Wo=
@@ -137,8 +163,8 @@ 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/mattn/go-isatty v0.0.20 h1:xfD0iDuEKnDkl03q4limB+vH+GxLEtL/jb4xVJSWWEY=
 github.com/mattn/go-isatty v0.0.20/go.mod h1:W+V8PltTTMOvKvAeJH7IuucS94S2C6jfK/D7dTCTo3Y=
-github.com/mattn/go-sqlite3 v2.0.3+incompatible h1:gXHsfypPkaMZrKbD5209QV9jbUTJKjyR5WD3HYQSd+U=
+github.com/mattn/go-sqlite3 v1.14.22 h1:2gZY6PC6kBnID23Tichd1K+Z0oS6nE/XwU+Vz/5o4kU=
-github.com/mattn/go-sqlite3 v2.0.3+incompatible/go.mod h1:FPy6KqzDD04eiIsT53CuJW3U88zkxoIYsOqkbpncsNc=
+github.com/mattn/go-sqlite3 v1.14.22/go.mod h1:Uh1q+B4BYcTPb+yiD3kU8Ct7aC0hY9fxUwlHK0RXw+Y=
 github.com/modern-go/concurrent v0.0.0-20180228061459-e0a39a4cb421/go.mod h1:6dJC0mAP4ikYIbvyc7fijjWJddQyLn8Ig3JB5CqoB9Q=
 github.com/modern-go/concurrent v0.0.0-20180306012644-bacd9c7ef1dd h1:TRLaZ9cD/w8PVh93nsPXa1VrQ6jlwL5oN8l14QlcNfg=
 github.com/modern-go/concurrent v0.0.0-20180306012644-bacd9c7ef1dd/go.mod h1:6dJC0mAP4ikYIbvyc7fijjWJddQyLn8Ig3JB5CqoB9Q=
@@ -147,18 +173,17 @@ github.com/modern-go/reflect2 v1.0.2/go.mod h1:yWuevngMOJpCy52FWWMvUC8ws7m/LJsjY
 github.com/nxadm/tail v1.4.8 h1:nPr65rt6Y5JFSKQO7qToXr7pePgD6Gwiw05lkbyAQTE=
 github.com/onsi/ginkgo v1.16.5 h1:8xi0RTUf59SOSfEtZMvwTvXYMzG4gV23XVHOZiXNtnE=
 github.com/onsi/gomega v1.18.1 h1:M1GfJqGRrBrrGGsbxzV5dqM2U2ApXefZCQpkukxYRLE=
-github.com/pelletier/go-toml/v2 v2.2.1 h1:9TA9+T8+8CUCO2+WYnDLCgrYi9+omqKXyjDtosvtEhg=
+github.com/patrickmn/go-cache v2.1.0+incompatible h1:HRMgzkcYKYpi3C8ajMPV8OFXaaRUnok+kx1WdO15EQc=
-github.com/pelletier/go-toml/v2 v2.2.1/go.mod h1:1t835xjRzz80PqgE6HHgN2JOsmgYu/h4qDAS4n929Rs=
+github.com/patrickmn/go-cache v2.1.0+incompatible/go.mod h1:3Qf8kWWT7OJRJbdiICTKqZju1ZixQ/KpMGzzAfe6+WQ=
 github.com/pelletier/go-toml/v2 v2.2.2 h1:aYUidT7k73Pcl9nb2gScu7NSrKCSHIDE89b3+6Wq+LM=
 github.com/pelletier/go-toml/v2 v2.2.2/go.mod h1:1t835xjRzz80PqgE6HHgN2JOsmgYu/h4qDAS4n929Rs=
 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=
 github.com/pmezard/go-difflib v1.0.0/go.mod h1:iKH77koFhYxTK1pcRnkKkqfTogsbg7gZNVY4sRDYZ/4=
+github.com/prometheus/client_model v0.0.0-20190812154241-14fe0d1b01d4/go.mod h1:xMI15A0UPsDsEKsMN9yxemIoYk6Tm2C1GtYGdfGttqA=
 github.com/rogpeppe/go-internal v1.8.0 h1:FCbCCtXNOY3UtUuHUYaghJg4y7Fd14rXifAYUAtL9R8=
 github.com/smarty/assertions v1.15.0 h1:cR//PqUBUiQRakZWqBiFFQ9wb8emQGDb0HeGdqGByCY=
 github.com/smarty/assertions v1.15.0/go.mod h1:yABtdzeQs6l1brC900WlRNwj6ZR55d7B+E8C6HtKdec=
@@ -180,40 +205,96 @@ github.com/twitchyliquid64/golang-asm v0.15.1 h1:SU5vSMR7hnwNxj24w34ZyCi/FmDZTkS
 github.com/twitchyliquid64/golang-asm v0.15.1/go.mod h1:a1lVb/DtPvCB8fslRZhAngC2+aY1QWCk3Cedj/Gdt08=
 github.com/ugorji/go/codec v1.2.12 h1:9LC83zGrHhuUA9l16C9AHXAqEV/2wBQ4nkvumAE65EE=
 github.com/ugorji/go/codec v1.2.12/go.mod h1:UNopzCgEMSXjBc6AOMqYvWC1ktqTAfzJZUZgYf6w6lg=
+go.opencensus.io v0.24.0 h1:y73uSU6J157QMP2kn2r30vwW1A2W2WFwSCGnAVxeaD0=
+go.opencensus.io v0.24.0/go.mod h1:vNK8G9p7aAivkbmorf4v+7Hgx+Zs0yY+0fOtgBfjQKo=
+go.opentelemetry.io/contrib/instrumentation/google.golang.org/grpc/otelgrpc v0.49.0 h1:4Pp6oUg3+e/6M4C0A/3kJ2VYa++dsWVTtGgLVj5xtHg=
+go.opentelemetry.io/contrib/instrumentation/google.golang.org/grpc/otelgrpc v0.49.0/go.mod h1:Mjt1i1INqiaoZOMGR1RIUJN+i3ChKoFRqzrRQhlkbs0=
+go.opentelemetry.io/contrib/instrumentation/net/http/otelhttp v0.49.0 h1:jq9TW8u3so/bN+JPT166wjOI6/vQPF6Xe7nMNIltagk=
+go.opentelemetry.io/contrib/instrumentation/net/http/otelhttp v0.49.0/go.mod h1:p8pYQP+m5XfbZm9fxtSKAbM6oIllS7s2AfxrChvc7iw=
+go.opentelemetry.io/otel v1.24.0 h1:0LAOdjNmQeSTzGBzduGe/rU4tZhMwL5rWgtp9Ku5Jfo=
+go.opentelemetry.io/otel v1.24.0/go.mod h1:W7b9Ozg4nkF5tWI5zsXkaKKDjdVjpD4oAt9Qi/MArHo=
+go.opentelemetry.io/otel/metric v1.24.0 h1:6EhoGWWK28x1fbpA4tYTOWBkPefTDQnb8WSGXlc88kI=
+go.opentelemetry.io/otel/metric v1.24.0/go.mod h1:VYhLe1rFfxuTXLgj4CBiyz+9WYBA8pNGJgDcSFRKBco=
+go.opentelemetry.io/otel/trace v1.24.0 h1:CsKnnL4dUAr/0llH9FKuc698G04IrpWV0MQA/Y1YELI=
+go.opentelemetry.io/otel/trace v1.24.0/go.mod h1:HPc3Xr/cOApsBI154IU0OI0HJexz+aw5uPdbs3UCjNU=
 golang.org/x/arch v0.0.0-20210923205945-b76863e36670/go.mod h1:5om86z9Hs0C8fWVUuoMHwpExlXzs5Tkyp9hOrfG7pp8=
-golang.org/x/arch v0.7.0 h1:pskyeJh/3AmoQ8CPE95vxHLqp1G1GfGNXTmcl9NEKTc=
-golang.org/x/arch v0.7.0/go.mod h1:FEVrYAQjsQXMVJ1nsMoVVXPZg6p2JE2mx8psSWTDQys=
 golang.org/x/arch v0.8.0 h1:3wRIsP3pM4yUptoR96otTUOXI367OS0+c9eeRi9doIc=
 golang.org/x/arch v0.8.0/go.mod h1:FEVrYAQjsQXMVJ1nsMoVVXPZg6p2JE2mx8psSWTDQys=
-golang.org/x/crypto v0.22.0 h1:g1v0xeRhjcugydODzvb3mEM9SQ0HGp9s/nh3COQ/C30=
+golang.org/x/crypto v0.0.0-20190308221718-c2843e01d9a2/go.mod h1:djNgcEr1/C05ACkg1iLfiJU5Ep61QUkGW8qpdssI0+w=
-golang.org/x/crypto v0.22.0/go.mod h1:vr6Su+7cTlO45qkww3VDJlzDn0ctJvRgYbC2NvXHt+M=
+golang.org/x/crypto v0.0.0-20200622213623-75b288015ac9/go.mod h1:LzIPMQfyMNhhGPhUkYOs5KpL4U8rLKemX1yGLhDgUto=
-golang.org/x/crypto v0.23.0 h1:dIJU/v2J8Mdglj/8rJ6UUOM3Zc9zLZxVZwwxMooUSAI=
+golang.org/x/crypto v0.24.0 h1:mnl8DM0o513X8fdIkmyFE/5hTYxbwYOjDS/+rK6qpRI=
-golang.org/x/crypto v0.23.0/go.mod h1:CKFgDieR+mRhux2Lsu27y0fO304Db0wZe70UKqHu0v8=
+golang.org/x/crypto v0.24.0/go.mod h1:Z1PMYSOR5nyMcyAVAIQSKCDwalqy85Aqn1x3Ws4L5DM=
-golang.org/x/image v0.15.0 h1:kOELfmgrmJlw4Cdb7g/QGuB3CvDrXbqEIww/pNtNBm8=
+golang.org/x/exp v0.0.0-20190121172915-509febef88a4/go.mod h1:CJ0aWSM057203Lf6IL+f9T1iT9GByDxfZKAQTCR3kQA=
-golang.org/x/image v0.15.0/go.mod h1:HUYqC05R2ZcZ3ejNQsIHQDQiwWM4JBqmm6MKANTp4LE=
+golang.org/x/image v0.18.0 h1:jGzIakQa/ZXI1I0Fxvaa9W7yP25TqT6cHIHn+6CqvSQ=
-golang.org/x/image v0.16.0 h1:9kloLAKhUufZhA12l5fwnx2NZW39/we1UhBesW433jw=
+golang.org/x/image v0.18.0/go.mod h1:4yyo5vMFQjVjUcVk4jEQcU9MGy/rulF5WvUILseCM2E=
-golang.org/x/image v0.16.0/go.mod h1:ugSZItdV4nOxyqp56HmXwH0Ry0nBCpjnZdpDaIHdoPs=
+golang.org/x/lint v0.0.0-20181026193005-c67002cb31c3/go.mod h1:UVdnD1Gm6xHRNCYTkRU2/jEulfH38KcIWyp/GAMgvoE=
-golang.org/x/net v0.24.0 h1:1PcaxkF854Fu3+lvBIx5SYn9wRlBzzcnHZSiaFFAb0w=
+golang.org/x/lint v0.0.0-20190227174305-5b3e6a55c961/go.mod h1:wehouNa3lNwaWXcvxsM5YxQ5yQlVC4a0KAMCusXpPoU=
-golang.org/x/net v0.24.0/go.mod h1:2Q7sJY5mzlzWjKtYUEXSlBWCdyaioyXzRB2RtU8KVE8=
+golang.org/x/lint v0.0.0-20190313153728-d0100b6bd8b3/go.mod h1:6SW0HCj/g11FgYtHlgUYUwCkIfeOF89ocIRzGO/8vkc=
-golang.org/x/net v0.25.0 h1:d/OCCoBEUq33pjydKrGQhw7IlUPI2Oylr+8qLx49kac=
+golang.org/x/net v0.0.0-20180724234803-3673e40ba225/go.mod h1:mL1N/T3taQHkDXs73rZJwtUhF3w3ftmwwsq0BUmARs4=
-golang.org/x/net v0.25.0/go.mod h1:JkAGAh7GEvH74S6FOH42FLoXpXbE/aqXSrIQjXgsiwM=
+golang.org/x/net v0.0.0-20180826012351-8a410e7b638d/go.mod h1:mL1N/T3taQHkDXs73rZJwtUhF3w3ftmwwsq0BUmARs4=
+golang.org/x/net v0.0.0-20190213061140-3a22650c66bd/go.mod h1:mL1N/T3taQHkDXs73rZJwtUhF3w3ftmwwsq0BUmARs4=
+golang.org/x/net v0.0.0-20190311183353-d8887717615a/go.mod h1:t9HGtf8HONx5eT2rtn7q6eTqICYqUVnKs3thJo3Qplg=
+golang.org/x/net v0.0.0-20190404232315-eb5bcb51f2a3/go.mod h1:t9HGtf8HONx5eT2rtn7q6eTqICYqUVnKs3thJo3Qplg=
+golang.org/x/net v0.0.0-20201110031124-69a78807bb2b/go.mod h1:sp8m0HH+o8qH0wwXwYZr8TS3Oi6o0r6Gce1SSxlDquU=
+golang.org/x/net v0.26.0 h1:soB7SVo0PWrY4vPW/+ay0jKDNScG2X9wFeYlXIvJsOQ=
+golang.org/x/net v0.26.0/go.mod h1:5YKkiSynbBIh3p6iOc/vibscux0x38BZDkn8sCUPxHE=
+golang.org/x/oauth2 v0.0.0-20180821212333-d2e6202438be/go.mod h1:N/0e6XlmueqKjAGxoOufVs8QHGRruUQn6yWY3a++T0U=
+golang.org/x/oauth2 v0.21.0 h1:tsimM75w1tF/uws5rbeHzIWxEqElMehnc+iW793zsZs=
+golang.org/x/oauth2 v0.21.0/go.mod h1:XYTD2NtWslqkgxebSiOHnXEap4TF09sJSc7H1sXbhtI=
+golang.org/x/sync v0.0.0-20180314180146-1d60e4601c6f/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
+golang.org/x/sync v0.0.0-20181108010431-42b317875d0f/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
+golang.org/x/sync v0.0.0-20190423024810-112230192c58/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
 golang.org/x/sync v0.7.0 h1:YsImfSBoP9QPYL0xyKJPq0gcaJdG3rInoqxTWbfQu9M=
 golang.org/x/sync v0.7.0/go.mod h1:Czt+wKu1gCyEFDUtn0jG5QVvpJ6rzVqr5aXyt9drQfk=
+golang.org/x/sys v0.0.0-20180830151530-49385e6e1522/go.mod h1:STP8DvDyc/dI5b8T5hshtkjS+E42TnysNCUPdjciGhY=
+golang.org/x/sys v0.0.0-20190215142949-d0b11bdaac8a/go.mod h1:STP8DvDyc/dI5b8T5hshtkjS+E42TnysNCUPdjciGhY=
+golang.org/x/sys v0.0.0-20190412213103-97732733099d/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
+golang.org/x/sys v0.0.0-20200930185726-fdedc70b468f/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.5.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.6.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
-golang.org/x/sys v0.19.0 h1:q5f1RH2jigJ1MoAWp2KTp3gm5zAGFUTarQZ5U386+4o=
+golang.org/x/sys v0.21.0 h1:rF+pYz3DAGSQAxAu1CbC7catZg4ebC4UIeIhKxBZvws=
-golang.org/x/sys v0.19.0/go.mod h1:/VUhepiaJMQUp4+oa/7Zr1D23ma6VTLIYjOOTFZPUcA=
+golang.org/x/sys v0.21.0/go.mod h1:/VUhepiaJMQUp4+oa/7Zr1D23ma6VTLIYjOOTFZPUcA=
-golang.org/x/sys v0.20.0 h1:Od9JTbYCk261bKm4M/mw7AklTlFYIa0bIp9BgSm1S8Y=
+golang.org/x/text v0.3.0/go.mod h1:NqM8EUOU14njkJ3fqMW+pc6Ldnwhi/IjpwHt7yyuwOQ=
-golang.org/x/sys v0.20.0/go.mod h1:/VUhepiaJMQUp4+oa/7Zr1D23ma6VTLIYjOOTFZPUcA=
+golang.org/x/text v0.3.3/go.mod h1:5Zoc/QRtKVWzQhOtBMvqHzDpF6irO9z98xDceosuGiQ=
-golang.org/x/text v0.14.0 h1:ScX5w1eTa3QqT8oi6+ziP7dTV1S2+ALU0bI+0zXKWiQ=
+golang.org/x/text v0.16.0 h1:a94ExnEXNtEwYLGJSIUxnWoxoRz/ZcCsV63ROupILh4=
-golang.org/x/text v0.14.0/go.mod h1:18ZOQIKpY8NJVqYksKHtTdi31H5itFRjB5/qKTNYzSU=
+golang.org/x/text v0.16.0/go.mod h1:GhwF1Be+LQoKShO3cGOHzqOgRrGaYc9AvblQOmPVHnI=
-golang.org/x/text v0.15.0 h1:h1V/4gjBv8v9cjcR6+AR5+/cIYK5N/WAgiv4xlsEtAk=
+golang.org/x/time v0.5.0 h1:o7cqy6amK/52YcAKIPlM3a+Fpj35zvRj2TP+e1xFSfk=
-golang.org/x/text v0.15.0/go.mod h1:18ZOQIKpY8NJVqYksKHtTdi31H5itFRjB5/qKTNYzSU=
+golang.org/x/time v0.5.0/go.mod h1:3BpzKBy/shNhVucY/MWOyx10tF3SFh9QdLuxbVysPQM=
-golang.org/x/xerrors v0.0.0-20200804184101-5ec99f83aff1 h1:go1bK/D/BFZV2I8cIQd1NKEZ+0owSTG1fDTci4IqFcE=
+golang.org/x/tools v0.0.0-20180917221912-90fa682c2a6e/go.mod h1:n7NCudcB/nEzxVGmLbDWY5pfWTLqBcC2KZ6jyYvM4mQ=
-google.golang.org/protobuf v1.33.0 h1:uNO2rsAINq/JlFpSdYEKIZ0uKD/R9cpdv0T+yoGwGmI=
+golang.org/x/tools v0.0.0-20190114222345-bf090417da8b/go.mod h1:n7NCudcB/nEzxVGmLbDWY5pfWTLqBcC2KZ6jyYvM4mQ=
-google.golang.org/protobuf v1.33.0/go.mod h1:c6P6GXX6sHbq/GpV6MGZEdwhWPcYBgnhAHhKbcUYpos=
+golang.org/x/tools v0.0.0-20190226205152-f727befe758c/go.mod h1:9Yl7xja0Znq3iFh3HoIrodX9oNMXvdceNzlUR8zjMvY=
-google.golang.org/protobuf v1.34.1 h1:9ddQBjfCyZPOHPUiPxpYESBLc+T8P3E+Vo4IbKZgFWg=
+golang.org/x/tools v0.0.0-20190311212946-11955173bddd/go.mod h1:LCzVGOaR6xXOjkQ3onu1FJEFr0SW1gC7cKk1uF8kGRs=
-google.golang.org/protobuf v1.34.1/go.mod h1:c6P6GXX6sHbq/GpV6MGZEdwhWPcYBgnhAHhKbcUYpos=
+golang.org/x/tools v0.0.0-20190524140312-2c0ae7006135/go.mod h1:RgjU9mgBXZiqYHBnxXauZ1Gv1EHHAz9KjViQ78xBX0Q=
+golang.org/x/xerrors v0.0.0-20191204190536-9bdfabe68543/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=
+google.golang.org/api v0.187.0 h1:Mxs7VATVC2v7CY+7Xwm4ndkX71hpElcvx0D1Ji/p1eo=
+google.golang.org/api v0.187.0/go.mod h1:KIHlTc4x7N7gKKuVsdmfBXN13yEEWXWFURWY6SBp2gk=
+google.golang.org/appengine v1.1.0/go.mod h1:EbEs0AVv82hx2wNQdGPgUI5lhzA/G0D9YwlJXL52JkM=
+google.golang.org/appengine v1.4.0/go.mod h1:xpcJRLb0r/rnEns0DIKYYv+WjYCduHsrkT7/EB5XEv4=
+google.golang.org/genproto v0.0.0-20180817151627-c66870c02cf8/go.mod h1:JiN7NxoALGmiZfu7CAH4rXhgtRTLTxftemlI0sWmxmc=
+google.golang.org/genproto v0.0.0-20190819201941-24fa4b261c55/go.mod h1:DMBHOl98Agz4BDEuKkezgsaosCRResVns1a3J2ZsMNc=
+google.golang.org/genproto v0.0.0-20200526211855-cb27e3aa2013/go.mod h1:NbSheEEYHJ7i3ixzK3sjbqSGDJWnxyFXZblF3eUsNvo=
+google.golang.org/genproto/googleapis/api v0.0.0-20240617180043-68d350f18fd4 h1:MuYw1wJzT+ZkybKfaOXKp5hJiZDn2iHaXRw0mRYdHSc=
+google.golang.org/genproto/googleapis/api v0.0.0-20240617180043-68d350f18fd4/go.mod h1:px9SlOOZBg1wM1zdnr8jEL4CNGUBZ+ZKYtNPApNQc4c=
+google.golang.org/genproto/googleapis/rpc v0.0.0-20240624140628-dc46fd24d27d h1:k3zyW3BYYR30e8v3x0bTDdE9vpYFjZHK+HcyqkrppWk=
+google.golang.org/genproto/googleapis/rpc v0.0.0-20240624140628-dc46fd24d27d/go.mod h1:Ue6ibwXGpU+dqIcODieyLOcgj7z8+IcskoNIgZxtrFY=
+google.golang.org/grpc v1.19.0/go.mod h1:mqu4LbDTu4XGKhr4mRzUsmM4RtVoemTSY81AxZiDr8c=
+google.golang.org/grpc v1.23.0/go.mod h1:Y5yQAOtifL1yxbo5wqy6BxZv8vAUGQwXBOALyacEbxg=
+google.golang.org/grpc v1.25.1/go.mod h1:c3i+UQWmh7LiEpx4sFZnkU36qjEYZ0imhYfXVyQciAY=
+google.golang.org/grpc v1.27.0/go.mod h1:qbnxyOmOxrQa7FizSgH+ReBfzJrCY1pSN7KXBS8abTk=
+google.golang.org/grpc v1.33.2/go.mod h1:JMHMWHQWaTccqQQlmk3MJZS+GWXOdAesneDmEnv2fbc=
+google.golang.org/grpc v1.64.1 h1:LKtvyfbX3UGVPFcGqJ9ItpVWW6oN/2XqTxfAnwRRXiA=
+google.golang.org/grpc v1.64.1/go.mod h1:hiQF4LFZelK2WKaP6W0L92zGHtiQdZxk8CrSdvyjeP0=
+google.golang.org/protobuf v0.0.0-20200109180630-ec00e32a8dfd/go.mod h1:DFci5gLYBciE7Vtevhsrf46CRTquxDuWsQurQQe4oz8=
+google.golang.org/protobuf v0.0.0-20200221191635-4d8936d0db64/go.mod h1:kwYJMbMJ01Woi6D6+Kah6886xMZcty6N08ah7+eCXa0=
+google.golang.org/protobuf v0.0.0-20200228230310-ab0ca4ff8a60/go.mod h1:cfTl7dwQJ+fmap5saPgwCLgHXTUD7jkjRqWcaiX5VyM=
+google.golang.org/protobuf v1.20.1-0.20200309200217-e05f789c0967/go.mod h1:A+miEFZTKqfCUM6K7xSMQL9OKL/b6hQv+e19PK+JZNE=
+google.golang.org/protobuf v1.21.0/go.mod h1:47Nbq4nVaFHyn7ilMalzfO3qCViNmqZ2kzikPIcrTAo=
+google.golang.org/protobuf v1.22.0/go.mod h1:EGpADcykh3NcUnDUJcl1+ZksZNG86OlYog2l/sGQquU=
+google.golang.org/protobuf v1.23.0/go.mod h1:EGpADcykh3NcUnDUJcl1+ZksZNG86OlYog2l/sGQquU=
+google.golang.org/protobuf v1.23.1-0.20200526195155-81db48ad09cc/go.mod h1:EGpADcykh3NcUnDUJcl1+ZksZNG86OlYog2l/sGQquU=
+google.golang.org/protobuf v1.25.0/go.mod h1:9JNX74DMeImyA3h4bdi1ymwjUzf21/xIlbajtzgsN7c=
+google.golang.org/protobuf v1.34.2 h1:6xV6lTsCfpGD21XK49h7MhtcApnLqkfYgPcdHftf6hg=
+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 v1.0.0-20201130134442-10cb98267c6c h1:Hei/4ADfdWqJk1ZMxUNpqntNwaWcugrBjAiHlqqRiVk=
 gopkg.in/tomb.v1 v1.0.0-20141024135613-dd632973f1e7 h1:uRGJdciOHaEIrze2W8Q3AKkepLTh2hOroT7a+7czfdQ=
@@ -228,9 +309,9 @@ gorm.io/driver/postgres v1.5.7/go.mod h1:3e019WlBaYI5o5LIdNV+LyxCMNtLOQETBXL2h4c
 gorm.io/driver/sqlite v1.5.5 h1:7MDMtUZhV065SilG62E0MquljeArQZNfJnjd9i9gx3E=
 gorm.io/driver/sqlite v1.5.5/go.mod h1:6NgQ7sQWAIFsPrJJl1lSNSu2TABh0ZZ/zm5fosATavE=
 gorm.io/gorm v1.25.7/go.mod h1:hbnx/Oo0ChWMn1BIhpy1oYozzpM15i4YPuHDmfYtwg8=
-gorm.io/gorm v1.25.9 h1:wct0gxZIELDk8+ZqF/MVnHLkA1rvYlBWUMv2EdsK1g8=
-gorm.io/gorm v1.25.9/go.mod h1:hbnx/Oo0ChWMn1BIhpy1oYozzpM15i4YPuHDmfYtwg8=
 gorm.io/gorm v1.25.10 h1:dQpO+33KalOA+aFYGlK+EfxcI5MbO7EP2yYygwh9h+s=
 gorm.io/gorm v1.25.10/go.mod h1:hbnx/Oo0ChWMn1BIhpy1oYozzpM15i4YPuHDmfYtwg8=
+honnef.co/go/tools v0.0.0-20190102054323-c2f93a96b099/go.mod h1:rf3lG4BRIbNafJWhAfAdb/ePZxsR/4RtNHQocxwk9r4=
+honnef.co/go/tools v0.0.0-20190523083050-ea95bdfd59fc/go.mod h1:rf3lG4BRIbNafJWhAfAdb/ePZxsR/4RtNHQocxwk9r4=
 nullprogram.com/x/optparse v1.0.0/go.mod h1:KdyPE+Igbe0jQUrVfMqDMeJQIJZEuyV7pjYmp6pbG50=
 rsc.io/pdf v0.1.1/go.mod h1:n8OzWcQ6Sp37PL01nO98y4iUCRdTGarVfzxY20ICaU4=

main.go

@@ -6,6 +6,7 @@ import (
 	"github.com/gin-contrib/sessions"
 	"github.com/gin-contrib/sessions/cookie"
 	"github.com/gin-gonic/gin"
+	_ "github.com/joho/godotenv/autoload"
 	"github.com/songquanpeng/one-api/common"
 	"github.com/songquanpeng/one-api/common/client"
 	"github.com/songquanpeng/one-api/common/config"
@@ -23,29 +24,22 @@ import (
 var buildFS embed.FS
 
 func main() {
+	common.Init()
 	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" {
+
+	if os.Getenv("GIN_MODE") != gin.DebugMode {
 		gin.SetMode(gin.ReleaseMode)
 	}
 	if config.DebugEnabled {
 		logger.SysLog("running in debug mode")
 	}
-	var err error
+
 	// Initialize SQL Database
-	model.DB, err = model.InitDB("SQL_DSN")
+	model.InitDB()
-	if err != nil {
+	model.InitLogDB()
-		logger.FatalLog("failed to initialize database: " + err.Error())
+
-	}
+	var err error
-	if os.Getenv("LOG_SQL_DSN") != "" {
-		logger.SysLog("using secondary database for table logs")
-		model.LOG_DB, err = model.InitDB("LOG_SQL_DSN")
-		if err != nil {
-			logger.FatalLog("failed to initialize secondary database: " + err.Error())
-		}
-	} else {
-		model.LOG_DB = model.DB
-	}
 	err = model.CreateRootAccountIfNeed()
 	if err != nil {
 		logger.FatalLog("database init error: " + err.Error())
@@ -113,6 +107,7 @@ func main() {
 	if port == "" {
 		port = strconv.Itoa(*common.Port)
 	}
+	logger.SysLogf("server started on http://localhost:%s", port)
 	err = server.Run(":" + port)
 	if err != nil {
 		logger.FatalLog("failed to start HTTP server: " + err.Error())

middleware/auth.go

@@ -140,6 +140,12 @@ func TokenAuth() func(c *gin.Context) {
 				return
 			}
 		}
+
+		// set channel id for proxy relay
+		if channelId := c.Param("channelid"); channelId != "" {
+			c.Set(ctxkey.SpecificChannelId, channelId)
+		}
+
 		c.Next()
 	}
 }

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
+	if channel.Other != nil {
 		switch channel.Type {
 		case channeltype.Azure:
 			if cfg.APIVersion == "" {
-			cfg.APIVersion = channel.Other
+				cfg.APIVersion = *channel.Other
 			}
 		case channeltype.Xunfei:
 			if cfg.APIVersion == "" {
-			cfg.APIVersion = channel.Other
+				cfg.APIVersion = *channel.Other
 			}
 		case channeltype.Gemini:
 			if cfg.APIVersion == "" {
-			cfg.APIVersion = channel.Other
+				cfg.APIVersion = *channel.Other
 			}
 		case channeltype.AIProxyLibrary:
 			if cfg.LibraryID == "" {
-			cfg.LibraryID = channel.Other
+				cfg.LibraryID = *channel.Other
 			}
 		case channeltype.Ali:
 			if cfg.Plugin == "" {
-			cfg.Plugin = channel.Other
+				cfg.Plugin = *channel.Other
+			}
 		}
 	}
 	c.Set(ctxkey.Config, cfg)

middleware/rate-limit.go

@@ -3,11 +3,12 @@ package middleware
 import (
 	"context"
 	"fmt"
+	"net/http"
+	"time"
+
 	"github.com/gin-gonic/gin"
 	"github.com/songquanpeng/one-api/common"
 	"github.com/songquanpeng/one-api/common/config"
-	"net/http"
-	"time"
 )
 
 var timeFormat = "2006-01-02T15:04:05.000Z"
@@ -70,6 +71,11 @@ func memoryRateLimiter(c *gin.Context, maxRequestNum int, duration int64, mark s
 }
 
 func rateLimitFactory(maxRequestNum int, duration int64, mark string) func(c *gin.Context) {
+	if maxRequestNum == 0 {
+		return func(c *gin.Context) {
+			c.Next()
+		}
+	}
 	if common.RedisEnabled {
 		return func(c *gin.Context) {
 			redisRateLimiter(c, maxRequestNum, duration, mark)

model/channel.go

@@ -3,6 +3,7 @@ package model
 import (
 	"encoding/json"
 	"fmt"
+
 	"github.com/songquanpeng/one-api/common/config"
 	"github.com/songquanpeng/one-api/common/helper"
 	"github.com/songquanpeng/one-api/common/logger"
@@ -27,7 +28,7 @@ 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"`
@@ -46,6 +47,8 @@ type ChannelConfig struct {
 	APIVersion        string `json:"api_version,omitempty"`
 	LibraryID         string `json:"library_id,omitempty"`
 	Plugin            string `json:"plugin,omitempty"`
+	VertexAIProjectID string `json:"vertex_ai_project_id,omitempty"`
+	VertexAIADC       string `json:"vertex_ai_adc,omitempty"`
 }
 
 func GetAllChannels(startIdx int, num int, scope string) ([]*Channel, error) {

model/main.go

@@ -1,6 +1,7 @@
 package model
 
 import (
+	"database/sql"
 	"fmt"
 	"github.com/songquanpeng/one-api/common"
 	"github.com/songquanpeng/one-api/common/config"
@@ -29,13 +30,17 @@ func CreateRootAccountIfNeed() error {
 		if err != nil {
 			return err
 		}
+		accessToken := random.GetUUID()
+		if config.InitialRootAccessToken != "" {
+			accessToken = config.InitialRootAccessToken
+		}
 		rootUser := User{
 			Username:    "root",
 			Password:    hashedPassword,
 			Role:        RoleRootUser,
 			Status:      UserStatusEnabled,
 			DisplayName: "Root User",
-			AccessToken: random.GetUUID(),
+			AccessToken: accessToken,
 			Quota:       500000000000000,
 		}
 		DB.Create(&rootUser)
@@ -60,10 +65,22 @@ func CreateRootAccountIfNeed() error {
 }
 
 func chooseDB(envName string) (*gorm.DB, error) {
-	if os.Getenv(envName) != "" {
 	dsn := os.Getenv(envName)
-		if strings.HasPrefix(dsn, "postgres://") {
+
+	switch {
+	case strings.HasPrefix(dsn, "postgres://"):
 		// Use PostgreSQL
+		return openPostgreSQL(dsn)
+	case dsn != "":
+		// Use MySQL
+		return openMySQL(dsn)
+	default:
+		// Use SQLite
+		return openSQLite()
+	}
+}
+
+func openPostgreSQL(dsn string) (*gorm.DB, error) {
 	logger.SysLog("using PostgreSQL as database")
 	common.UsingPostgreSQL = true
 	return gorm.Open(postgres.New(postgres.Config{
@@ -72,78 +89,132 @@ func chooseDB(envName string) (*gorm.DB, error) {
 	}), &gorm.Config{
 		PrepareStmt: true, // precompile SQL
 	})
-		}
+}
-		// Use MySQL
+
+func openMySQL(dsn string) (*gorm.DB, error) {
 	logger.SysLog("using MySQL as database")
 	common.UsingMySQL = true
 	return gorm.Open(mysql.Open(dsn), &gorm.Config{
 		PrepareStmt: true, // precompile SQL
 	})
-	}
+}
-	// Use SQLite
+
+func openSQLite() (*gorm.DB, error) {
 	logger.SysLog("SQL_DSN not set, using SQLite as database")
 	common.UsingSQLite = true
-	config := fmt.Sprintf("?_busy_timeout=%d", common.SQLiteBusyTimeout)
+	dsn := fmt.Sprintf("%s?_busy_timeout=%d", common.SQLitePath, common.SQLiteBusyTimeout)
-	return gorm.Open(sqlite.Open(common.SQLitePath+config), &gorm.Config{
+	return gorm.Open(sqlite.Open(dsn), &gorm.Config{
 		PrepareStmt: true, // precompile SQL
 	})
 }
 
-func InitDB(envName string) (db *gorm.DB, err error) {
+func InitDB() {
-	db, err = chooseDB(envName)
+	var err error
-	if err == nil {
+	DB, err = chooseDB("SQL_DSN")
-		if config.DebugSQLEnabled {
-			db = db.Debug()
-		}
-		sqlDB, err := db.DB()
 	if err != nil {
-			return nil, err
+		logger.FatalLog("failed to initialize database: " + err.Error())
+		return
 	}
-		sqlDB.SetMaxIdleConns(env.Int("SQL_MAX_IDLE_CONNS", 100))
+
-		sqlDB.SetMaxOpenConns(env.Int("SQL_MAX_OPEN_CONNS", 1000))
+	sqlDB := setDBConns(DB)
-		sqlDB.SetConnMaxLifetime(time.Second * time.Duration(env.Int("SQL_MAX_LIFETIME", 60)))
 
 	if !config.IsMasterNode {
-			return db, err
+		return
 	}
+
 	if common.UsingMySQL {
 		_, _ = sqlDB.Exec("DROP INDEX idx_channels_key ON channels;") // TODO: delete this line when most users have upgraded
 	}
+
 	logger.SysLog("database migration started")
-		err = db.AutoMigrate(&Channel{})
+	if err = migrateDB(); err != nil {
-		if err != nil {
+		logger.FatalLog("failed to migrate database: " + err.Error())
-			return nil, err
+		return
-		}
-		err = db.AutoMigrate(&Token{})
-		if err != nil {
-			return nil, err
-		}
-		err = db.AutoMigrate(&User{})
-		if err != nil {
-			return nil, err
-		}
-		err = db.AutoMigrate(&Option{})
-		if err != nil {
-			return nil, err
-		}
-		err = db.AutoMigrate(&Redemption{})
-		if err != nil {
-			return nil, err
-		}
-		err = db.AutoMigrate(&Ability{})
-		if err != nil {
-			return nil, err
-		}
-		err = db.AutoMigrate(&Log{})
-		if err != nil {
-			return nil, err
 	}
 	logger.SysLog("database migrated")
-		return db, err
+}
-	} else {
+
-		logger.FatalLog(err)
+func migrateDB() error {
+	var err error
+	if err = DB.AutoMigrate(&Channel{}); err != nil {
+		return err
 	}
-	return db, err
+	if err = DB.AutoMigrate(&Token{}); err != nil {
+		return err
+	}
+	if err = DB.AutoMigrate(&User{}); err != nil {
+		return err
+	}
+	if err = DB.AutoMigrate(&Option{}); err != nil {
+		return err
+	}
+	if err = DB.AutoMigrate(&Redemption{}); err != nil {
+		return err
+	}
+	if err = DB.AutoMigrate(&Ability{}); err != nil {
+		return err
+	}
+	if err = DB.AutoMigrate(&Log{}); err != nil {
+		return err
+	}
+	if err = DB.AutoMigrate(&Channel{}); err != nil {
+		return err
+	}
+	return nil
+}
+
+func InitLogDB() {
+	if os.Getenv("LOG_SQL_DSN") == "" {
+		LOG_DB = DB
+		return
+	}
+
+	logger.SysLog("using secondary database for table logs")
+	var err error
+	LOG_DB, err = chooseDB("LOG_SQL_DSN")
+	if err != nil {
+		logger.FatalLog("failed to initialize secondary database: " + err.Error())
+		return
+	}
+
+	setDBConns(LOG_DB)
+
+	if !config.IsMasterNode {
+		return
+	}
+
+	logger.SysLog("secondary database migration started")
+	err = migrateLOGDB()
+	if err != nil {
+		logger.FatalLog("failed to migrate secondary database: " + err.Error())
+		return
+	}
+	logger.SysLog("secondary database migrated")
+}
+
+func migrateLOGDB() error {
+	var err error
+	if err = LOG_DB.AutoMigrate(&Log{}); err != nil {
+		return err
+	}
+	return nil
+}
+
+func setDBConns(db *gorm.DB) *sql.DB {
+	if config.DebugSQLEnabled {
+		db = db.Debug()
+	}
+
+	sqlDB, err := db.DB()
+	if err != nil {
+		logger.FatalLog("failed to connect database: " + err.Error())
+		return nil
+	}
+
+	sqlDB.SetMaxIdleConns(env.Int("SQL_MAX_IDLE_CONNS", 100))
+	sqlDB.SetMaxOpenConns(env.Int("SQL_MAX_OPEN_CONNS", 1000))
+	sqlDB.SetConnMaxLifetime(time.Second * time.Duration(env.Int("SQL_MAX_LIFETIME", 60)))
+	return sqlDB
 }
 
 func closeDB(db *gorm.DB) error {

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.go

@@ -15,7 +15,9 @@ import (
 	"github.com/songquanpeng/one-api/relay/adaptor/ollama"
 	"github.com/songquanpeng/one-api/relay/adaptor/openai"
 	"github.com/songquanpeng/one-api/relay/adaptor/palm"
+	"github.com/songquanpeng/one-api/relay/adaptor/proxy"
 	"github.com/songquanpeng/one-api/relay/adaptor/tencent"
+	"github.com/songquanpeng/one-api/relay/adaptor/vertexai"
 	"github.com/songquanpeng/one-api/relay/adaptor/xunfei"
 	"github.com/songquanpeng/one-api/relay/adaptor/zhipu"
 	"github.com/songquanpeng/one-api/relay/apitype"
@@ -55,6 +57,10 @@ func GetAdaptor(apiType int) adaptor.Adaptor {
 		return &cloudflare.Adaptor{}
 	case apitype.DeepL:
 		return &deepl.Adaptor{}
+	case apitype.VertexAI:
+		return &vertexai.Adaptor{}
+	case apitype.Proxy:
+		return &proxy.Adaptor{}
 	}
 	return nil
 }

relay/adaptor/aiproxy/main.go

@@ -4,6 +4,12 @@ import (
 	"bufio"
 	"encoding/json"
 	"fmt"
+	"github.com/songquanpeng/one-api/common/render"
+	"io"
+	"net/http"
+	"strconv"
+	"strings"
+
 	"github.com/gin-gonic/gin"
 	"github.com/songquanpeng/one-api/common"
 	"github.com/songquanpeng/one-api/common/helper"
@@ -12,10 +18,6 @@ import (
 	"github.com/songquanpeng/one-api/relay/adaptor/openai"
 	"github.com/songquanpeng/one-api/relay/constant"
 	"github.com/songquanpeng/one-api/relay/model"
-	"io"
-	"net/http"
-	"strconv"
-	"strings"
 )
 
 // https://docs.aiproxy.io/dev/library#使用已经定制好的知识库进行对话问答
@@ -89,6 +91,7 @@ func streamResponseAIProxyLibrary2OpenAI(response *LibraryStreamResponse) *opena
 
 func StreamHandler(c *gin.Context, resp *http.Response) (*model.ErrorWithStatusCode, *model.Usage) {
 	var usage model.Usage
+	var documents []LibraryDocument
 	scanner := bufio.NewScanner(resp.Body)
 	scanner.Split(func(data []byte, atEOF bool) (advance int, token []byte, err error) {
 		if atEOF && len(data) == 0 {
@@ -102,60 +105,48 @@ func StreamHandler(c *gin.Context, resp *http.Response) (*model.ErrorWithStatusC
 		}
 		return 0, nil, nil
 	})
-	dataChan := make(chan string)
+
-	stopChan := make(chan bool)
+	common.SetEventStreamHeaders(c)
-	go func() {
+
 	for scanner.Scan() {
 		data := scanner.Text()
-			if len(data) < 5 { // ignore blank line or wrong format
+		if len(data) < 5 || data[:5] != "data:" {
-				continue
-			}
-			if data[:5] != "data:" {
 			continue
 		}
 		data = data[5:]
-			dataChan <- data
+
-		}
-		stopChan <- true
-	}()
-	common.SetEventStreamHeaders(c)
-	var documents []LibraryDocument
-	c.Stream(func(w io.Writer) bool {
-		select {
-		case data := <-dataChan:
 		var AIProxyLibraryResponse LibraryStreamResponse
 		err := json.Unmarshal([]byte(data), &AIProxyLibraryResponse)
 		if err != nil {
 			logger.SysError("error unmarshalling stream response: " + err.Error())
-				return true
+			continue
 		}
 		if len(AIProxyLibraryResponse.Documents) != 0 {
 			documents = AIProxyLibraryResponse.Documents
 		}
 		response := streamResponseAIProxyLibrary2OpenAI(&AIProxyLibraryResponse)
-			jsonResponse, err := json.Marshal(response)
+		err = render.ObjectData(c, response)
 		if err != nil {
-				logger.SysError("error marshalling stream response: " + err.Error())
+			logger.SysError(err.Error())
-				return true
 		}
-			c.Render(-1, common.CustomEvent{Data: "data: " + string(jsonResponse)})
+	}
-			return true
+
-		case <-stopChan:
+	if err := scanner.Err(); err != nil {
+		logger.SysError("error reading stream: " + err.Error())
+	}
+
 	response := documentsAIProxyLibrary(documents)
-			jsonResponse, err := json.Marshal(response)
+	err := render.ObjectData(c, response)
 	if err != nil {
-				logger.SysError("error marshalling stream response: " + err.Error())
+		logger.SysError(err.Error())
-				return true
 	}
-			c.Render(-1, common.CustomEvent{Data: "data: " + string(jsonResponse)})
+	render.Done(c)
-			c.Render(-1, common.CustomEvent{Data: "data: [DONE]"})
+
-			return false
+	err = resp.Body.Close()
-		}
-	})
-	err := resp.Body.Close()
 	if err != nil {
 		return openai.ErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), nil
 	}
+
 	return nil, &usage
 }
 

relay/adaptor/ali/main.go

@@ -3,15 +3,17 @@ package ali
 import (
 	"bufio"
 	"encoding/json"
+	"github.com/songquanpeng/one-api/common/render"
+	"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/adaptor/openai"
 	"github.com/songquanpeng/one-api/relay/model"
-	"io"
-	"net/http"
-	"strings"
 )
 
 // https://help.aliyun.com/document_detail/613695.html?spm=a2c4g.2399480.0.0.1adb778fAdzP9w#341800c0f8w0r
@@ -181,32 +183,21 @@ func StreamHandler(c *gin.Context, resp *http.Response) (*model.ErrorWithStatusC
 		}
 		return 0, nil, nil
 	})
-	dataChan := make(chan string)
+
-	stopChan := make(chan bool)
+	common.SetEventStreamHeaders(c)
-	go func() {
+
 	for scanner.Scan() {
 		data := scanner.Text()
-			if len(data) < 5 { // ignore blank line or wrong format
+		if len(data) < 5 || data[:5] != "data:" {
-				continue
-			}
-			if data[:5] != "data:" {
 			continue
 		}
 		data = data[5:]
-			dataChan <- data
+
-		}
-		stopChan <- true
-	}()
-	common.SetEventStreamHeaders(c)
-	//lastResponseText := ""
-	c.Stream(func(w io.Writer) bool {
-		select {
-		case data := <-dataChan:
 		var aliResponse ChatResponse
 		err := json.Unmarshal([]byte(data), &aliResponse)
 		if err != nil {
 			logger.SysError("error unmarshalling stream response: " + err.Error())
-				return true
+			continue
 		}
 		if aliResponse.Usage.OutputTokens != 0 {
 			usage.PromptTokens = aliResponse.Usage.InputTokens
@@ -215,22 +206,20 @@ func StreamHandler(c *gin.Context, resp *http.Response) (*model.ErrorWithStatusC
 		}
 		response := streamResponseAli2OpenAI(&aliResponse)
 		if response == nil {
-				return true
+			continue
 		}
-			//response.Choices[0].Delta.Content = strings.TrimPrefix(response.Choices[0].Delta.Content, lastResponseText)
+		err = render.ObjectData(c, response)
-			//lastResponseText = aliResponse.Output.Text
-			jsonResponse, err := json.Marshal(response)
 		if err != nil {
-				logger.SysError("error marshalling stream response: " + err.Error())
+			logger.SysError(err.Error())
-				return true
 		}
-			c.Render(-1, common.CustomEvent{Data: "data: " + string(jsonResponse)})
-			return true
-		case <-stopChan:
-			c.Render(-1, common.CustomEvent{Data: "data: [DONE]"})
-			return false
 	}
-	})
+
+	if err := scanner.Err(); err != nil {
+		logger.SysError("error reading stream: " + err.Error())
+	}
+
+	render.Done(c)
+
 	err := resp.Body.Close()
 	if err != nil {
 		return openai.ErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), nil

relay/adaptor/anthropic/adaptor.go

@@ -3,12 +3,14 @@ package anthropic
 import (
 	"errors"
 	"fmt"
+	"io"
+	"net/http"
+	"strings"
+
 	"github.com/gin-gonic/gin"
 	"github.com/songquanpeng/one-api/relay/adaptor"
 	"github.com/songquanpeng/one-api/relay/meta"
 	"github.com/songquanpeng/one-api/relay/model"
-	"io"
-	"net/http"
 )
 
 type Adaptor struct {
@@ -31,6 +33,13 @@ func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, meta *me
 	}
 	req.Header.Set("anthropic-version", anthropicVersion)
 	req.Header.Set("anthropic-beta", "messages-2023-12-15")
+
+	// https://x.com/alexalbert__/status/1812921642143900036
+	// claude-3-5-sonnet can support 8k context
+	if strings.HasPrefix(meta.ActualModelName, "claude-3-5-sonnet") {
+		req.Header.Set("anthropic-beta", "max-tokens-3-5-sonnet-2024-07-15")
+	}
+
 	return nil
 }
 

relay/adaptor/anthropic/constants.go

@@ -5,4 +5,5 @@ var ModelList = []string{
 	"claude-3-haiku-20240307",
 	"claude-3-sonnet-20240229",
 	"claude-3-opus-20240229",
+	"claude-3-5-sonnet-20240620",
 }

relay/adaptor/anthropic/main.go

@@ -4,6 +4,7 @@ import (
 	"bufio"
 	"encoding/json"
 	"fmt"
+	"github.com/songquanpeng/one-api/common/render"
 	"io"
 	"net/http"
 	"strings"
@@ -28,12 +29,30 @@ func stopReasonClaude2OpenAI(reason *string) string {
 		return "stop"
 	case "max_tokens":
 		return "length"
+	case "tool_use":
+		return "tool_calls"
 	default:
 		return *reason
 	}
 }
 
 func ConvertRequest(textRequest model.GeneralOpenAIRequest) *Request {
+	claudeTools := make([]Tool, 0, len(textRequest.Tools))
+
+	for _, tool := range textRequest.Tools {
+		if params, ok := tool.Function.Parameters.(map[string]any); ok {
+			claudeTools = append(claudeTools, Tool{
+				Name:        tool.Function.Name,
+				Description: tool.Function.Description,
+				InputSchema: InputSchema{
+					Type:       params["type"].(string),
+					Properties: params["properties"],
+					Required:   params["required"],
+				},
+			})
+		}
+	}
+
 	claudeRequest := Request{
 		Model:       textRequest.Model,
 		MaxTokens:   textRequest.MaxTokens,
@@ -41,6 +60,24 @@ func ConvertRequest(textRequest model.GeneralOpenAIRequest) *Request {
 		TopP:        textRequest.TopP,
 		TopK:        textRequest.TopK,
 		Stream:      textRequest.Stream,
+		Tools:       claudeTools,
+	}
+	if len(claudeTools) > 0 {
+		claudeToolChoice := struct {
+			Type string `json:"type"`
+			Name string `json:"name,omitempty"`
+		}{Type: "auto"} // default value https://docs.anthropic.com/en/docs/build-with-claude/tool-use#controlling-claudes-output
+		if choice, ok := textRequest.ToolChoice.(map[string]any); ok {
+			if function, ok := choice["function"].(map[string]any); ok {
+				claudeToolChoice.Type = "tool"
+				claudeToolChoice.Name = function["name"].(string)
+			}
+		} else if toolChoiceType, ok := textRequest.ToolChoice.(string); ok {
+			if toolChoiceType == "any" {
+				claudeToolChoice.Type = toolChoiceType
+			}
+		}
+		claudeRequest.ToolChoice = claudeToolChoice
 	}
 	if claudeRequest.MaxTokens == 0 {
 		claudeRequest.MaxTokens = 4096
@@ -63,7 +100,24 @@ func ConvertRequest(textRequest model.GeneralOpenAIRequest) *Request {
 		if message.IsStringContent() {
 			content.Type = "text"
 			content.Text = message.StringContent()
+			if message.Role == "tool" {
+				claudeMessage.Role = "user"
+				content.Type = "tool_result"
+				content.Content = content.Text
+				content.Text = ""
+				content.ToolUseId = message.ToolCallId
+			}
 			claudeMessage.Content = append(claudeMessage.Content, content)
+			for i := range message.ToolCalls {
+				inputParam := make(map[string]any)
+				_ = json.Unmarshal([]byte(message.ToolCalls[i].Function.Arguments.(string)), &inputParam)
+				claudeMessage.Content = append(claudeMessage.Content, Content{
+					Type:  "tool_use",
+					Id:    message.ToolCalls[i].Id,
+					Name:  message.ToolCalls[i].Function.Name,
+					Input: inputParam,
+				})
+			}
 			claudeRequest.Messages = append(claudeRequest.Messages, claudeMessage)
 			continue
 		}
@@ -96,16 +150,35 @@ func StreamResponseClaude2OpenAI(claudeResponse *StreamResponse) (*openai.ChatCo
 	var response *Response
 	var responseText string
 	var stopReason string
+	tools := make([]model.Tool, 0)
+
 	switch claudeResponse.Type {
 	case "message_start":
 		return nil, claudeResponse.Message
 	case "content_block_start":
 		if claudeResponse.ContentBlock != nil {
 			responseText = claudeResponse.ContentBlock.Text
+			if claudeResponse.ContentBlock.Type == "tool_use" {
+				tools = append(tools, model.Tool{
+					Id:   claudeResponse.ContentBlock.Id,
+					Type: "function",
+					Function: model.Function{
+						Name:      claudeResponse.ContentBlock.Name,
+						Arguments: "",
+					},
+				})
+			}
 		}
 	case "content_block_delta":
 		if claudeResponse.Delta != nil {
 			responseText = claudeResponse.Delta.Text
+			if claudeResponse.Delta.Type == "input_json_delta" {
+				tools = append(tools, model.Tool{
+					Function: model.Function{
+						Arguments: claudeResponse.Delta.PartialJson,
+					},
+				})
+			}
 		}
 	case "message_delta":
 		if claudeResponse.Usage != nil {
@@ -119,6 +192,10 @@ func StreamResponseClaude2OpenAI(claudeResponse *StreamResponse) (*openai.ChatCo
 	}
 	var choice openai.ChatCompletionsStreamResponseChoice
 	choice.Delta.Content = responseText
+	if len(tools) > 0 {
+		choice.Delta.Content = nil // compatible with other OpenAI derivative applications, like LobeOpenAICompatibleFactory ...
+		choice.Delta.ToolCalls = tools
+	}
 	choice.Delta.Role = "assistant"
 	finishReason := stopReasonClaude2OpenAI(&stopReason)
 	if finishReason != "null" {
@@ -135,12 +212,27 @@ func ResponseClaude2OpenAI(claudeResponse *Response) *openai.TextResponse {
 	if len(claudeResponse.Content) > 0 {
 		responseText = claudeResponse.Content[0].Text
 	}
+	tools := make([]model.Tool, 0)
+	for _, v := range claudeResponse.Content {
+		if v.Type == "tool_use" {
+			args, _ := json.Marshal(v.Input)
+			tools = append(tools, model.Tool{
+				Id:   v.Id,
+				Type: "function", // compatible with other OpenAI derivative applications
+				Function: model.Function{
+					Name:      v.Name,
+					Arguments: string(args),
+				},
+			})
+		}
+	}
 	choice := openai.TextResponseChoice{
 		Index: 0,
 		Message: model.Message{
 			Role:      "assistant",
 			Content:   responseText,
 			Name:      nil,
+			ToolCalls: tools,
 		},
 		FinishReason: stopReasonClaude2OpenAI(claudeResponse.StopReason),
 	}
@@ -169,64 +261,77 @@ func StreamHandler(c *gin.Context, resp *http.Response) (*model.ErrorWithStatusC
 		}
 		return 0, nil, nil
 	})
-	dataChan := make(chan string)
+
-	stopChan := make(chan bool)
-	go func() {
-		for scanner.Scan() {
-			data := scanner.Text()
-			if len(data) < 6 {
-				continue
-			}
-			if !strings.HasPrefix(data, "data:") {
-				continue
-			}
-			data = strings.TrimPrefix(data, "data:")
-			dataChan <- data
-		}
-		stopChan <- true
-	}()
 	common.SetEventStreamHeaders(c)
+
 	var usage model.Usage
 	var modelName string
 	var id string
-	c.Stream(func(w io.Writer) bool {
+	var lastToolCallChoice openai.ChatCompletionsStreamResponseChoice
-		select {
+
-		case data := <-dataChan:
+	for scanner.Scan() {
-			// some implementations may add \r at the end of data
+		data := scanner.Text()
+		if len(data) < 6 || !strings.HasPrefix(data, "data:") {
+			continue
+		}
+		data = strings.TrimPrefix(data, "data:")
 		data = strings.TrimSpace(data)
+
 		var claudeResponse StreamResponse
 		err := json.Unmarshal([]byte(data), &claudeResponse)
 		if err != nil {
 			logger.SysError("error unmarshalling stream response: " + err.Error())
-				return true
+			continue
 		}
+
 		response, meta := StreamResponseClaude2OpenAI(&claudeResponse)
 		if meta != nil {
 			usage.PromptTokens += meta.Usage.InputTokens
 			usage.CompletionTokens += meta.Usage.OutputTokens
+			if len(meta.Id) > 0 { // only message_start has an id, otherwise it's a finish_reason event.
 				modelName = meta.Model
 				id = fmt.Sprintf("chatcmpl-%s", meta.Id)
-				return true
+				continue
+			} else { // finish_reason case
+				if len(lastToolCallChoice.Delta.ToolCalls) > 0 {
+					lastArgs := &lastToolCallChoice.Delta.ToolCalls[len(lastToolCallChoice.Delta.ToolCalls)-1].Function
+					if len(lastArgs.Arguments.(string)) == 0 { // compatible with OpenAI sending an empty object `{}` when no arguments.
+						lastArgs.Arguments = "{}"
+						response.Choices[len(response.Choices)-1].Delta.Content = nil
+						response.Choices[len(response.Choices)-1].Delta.ToolCalls = lastToolCallChoice.Delta.ToolCalls
+					}
+				}
+			}
 		}
 		if response == nil {
-				return true
+			continue
 		}
+
 		response.Id = id
 		response.Model = modelName
 		response.Created = createdTime
-			jsonStr, err := json.Marshal(response)
+
+		for _, choice := range response.Choices {
+			if len(choice.Delta.ToolCalls) > 0 {
+				lastToolCallChoice = choice
+			}
+		}
+		err = render.ObjectData(c, response)
 		if err != nil {
-				logger.SysError("error marshalling stream response: " + err.Error())
+			logger.SysError(err.Error())
-				return true
 		}
-			c.Render(-1, common.CustomEvent{Data: "data: " + string(jsonStr)})
-			return true
-		case <-stopChan:
-			c.Render(-1, common.CustomEvent{Data: "data: [DONE]"})
-			return false
 	}
-	})
+
-	_ = resp.Body.Close()
+	if err := scanner.Err(); err != nil {
+		logger.SysError("error reading stream: " + err.Error())
+	}
+
+	render.Done(c)
+
+	err := resp.Body.Close()
+	if err != nil {
+		return openai.ErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), nil
+	}
 	return nil, &usage
 }
 

relay/adaptor/anthropic/model.go

@@ -16,6 +16,12 @@ type Content struct {
 	Type   string       `json:"type"`
 	Text   string       `json:"text,omitempty"`
 	Source *ImageSource `json:"source,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 Message struct {
@@ -23,6 +29,18 @@ type Message struct {
 	Content []Content `json:"content"`
 }
 
+type Tool struct {
+	Name        string      `json:"name"`
+	Description string      `json:"description,omitempty"`
+	InputSchema InputSchema `json:"input_schema"`
+}
+
+type InputSchema struct {
+	Type       string `json:"type"`
+	Properties any    `json:"properties,omitempty"`
+	Required   any    `json:"required,omitempty"`
+}
+
 type Request struct {
 	Model         string    `json:"model"`
 	Messages      []Message `json:"messages"`
@@ -33,6 +51,8 @@ type Request struct {
 	Temperature   float64   `json:"temperature,omitempty"`
 	TopP          float64   `json:"top_p,omitempty"`
 	TopK          int       `json:"top_k,omitempty"`
+	Tools         []Tool    `json:"tools,omitempty"`
+	ToolChoice    any       `json:"tool_choice,omitempty"`
 	//Metadata    `json:"metadata,omitempty"`
 }
 
@@ -61,6 +81,7 @@ type Response struct {
 type Delta struct {
 	Type         string  `json:"type"`
 	Text         string  `json:"text"`
+	PartialJson  string  `json:"partial_json,omitempty"`
 	StopReason   *string `json:"stop_reason"`
 	StopSequence *string `json:"stop_sequence"`
 }

relay/adaptor/aws/adaptor.go

@@ -1,17 +1,16 @@
 package aws
 
 import (
-	"github.com/aws/aws-sdk-go-v2/aws"
+	"errors"
-	"github.com/aws/aws-sdk-go-v2/credentials"
-	"github.com/aws/aws-sdk-go-v2/service/bedrockruntime"
-	"github.com/songquanpeng/one-api/common/ctxkey"
 	"io"
 	"net/http"
 
+	"github.com/aws/aws-sdk-go-v2/aws"
+	"github.com/aws/aws-sdk-go-v2/credentials"
+	"github.com/aws/aws-sdk-go-v2/service/bedrockruntime"
 	"github.com/gin-gonic/gin"
-	"github.com/pkg/errors"
 	"github.com/songquanpeng/one-api/relay/adaptor"
-	"github.com/songquanpeng/one-api/relay/adaptor/anthropic"
+	"github.com/songquanpeng/one-api/relay/adaptor/aws/utils"
 	"github.com/songquanpeng/one-api/relay/meta"
 	"github.com/songquanpeng/one-api/relay/model"
 )
@@ -19,18 +18,52 @@ import (
 var _ adaptor.Adaptor = new(Adaptor)
 
 type Adaptor struct {
-	meta      *meta.Meta
+	awsAdapter utils.AwsAdapter
-	awsClient *bedrockruntime.Client
+
+	Meta      *meta.Meta
+	AwsClient *bedrockruntime.Client
 }
 
 func (a *Adaptor) Init(meta *meta.Meta) {
-	a.meta = meta
+	a.Meta = meta
-	a.awsClient = bedrockruntime.New(bedrockruntime.Options{
+	a.AwsClient = bedrockruntime.New(bedrockruntime.Options{
 		Region:      meta.Config.Region,
 		Credentials: aws.NewCredentialsCache(credentials.NewStaticCredentialsProvider(meta.Config.AK, meta.Config.SK, "")),
 	})
 }
 
+func (a *Adaptor) ConvertRequest(c *gin.Context, relayMode int, request *model.GeneralOpenAIRequest) (any, error) {
+	if request == nil {
+		return nil, errors.New("request is nil")
+	}
+
+	adaptor := GetAdaptor(request.Model)
+	if adaptor == nil {
+		return nil, errors.New("adaptor not found")
+	}
+
+	a.awsAdapter = adaptor
+	return adaptor.ConvertRequest(c, relayMode, request)
+}
+
+func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, meta *meta.Meta) (usage *model.Usage, err *model.ErrorWithStatusCode) {
+	if a.awsAdapter == nil {
+		return nil, utils.WrapErr(errors.New("awsAdapter is nil"))
+	}
+	return a.awsAdapter.DoResponse(c, a.AwsClient, meta)
+}
+
+func (a *Adaptor) GetModelList() (models []string) {
+	for model := range adaptors {
+		models = append(models, model)
+	}
+	return
+}
+
+func (a *Adaptor) GetChannelName() string {
+	return "aws"
+}
+
 func (a *Adaptor) GetRequestURL(meta *meta.Meta) (string, error) {
 	return "", nil
 }
@@ -39,17 +72,6 @@ func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, meta *me
 	return nil
 }
 
-func (a *Adaptor) ConvertRequest(c *gin.Context, relayMode int, request *model.GeneralOpenAIRequest) (any, error) {
-	if request == nil {
-		return nil, errors.New("request is nil")
-	}
-
-	claudeReq := anthropic.ConvertRequest(*request)
-	c.Set(ctxkey.RequestModel, request.Model)
-	c.Set(ctxkey.ConvertedRequest, claudeReq)
-	return claudeReq, nil
-}
-
 func (a *Adaptor) ConvertImageRequest(request *model.ImageRequest) (any, error) {
 	if request == nil {
 		return nil, errors.New("request is nil")
@@ -60,23 +82,3 @@ func (a *Adaptor) ConvertImageRequest(request *model.ImageRequest) (any, error)
 func (a *Adaptor) DoRequest(c *gin.Context, meta *meta.Meta, requestBody io.Reader) (*http.Response, error) {
 	return nil, nil
 }
-
-func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, meta *meta.Meta) (usage *model.Usage, err *model.ErrorWithStatusCode) {
-	if meta.IsStream {
-		err, usage = StreamHandler(c, a.awsClient)
-	} else {
-		err, usage = Handler(c, a.awsClient, meta.ActualModelName)
-	}
-	return
-}
-
-func (a *Adaptor) GetModelList() (models []string) {
-	for n := range awsModelIDMap {
-		models = append(models, n)
-	}
-	return
-}
-
-func (a *Adaptor) GetChannelName() string {
-	return "aws"
-}

relay/adaptor/aws/claude/adapter.go

@@ -0,0 +1,37 @@
+package aws
+
+import (
+	"github.com/aws/aws-sdk-go-v2/service/bedrockruntime"
+	"github.com/gin-gonic/gin"
+	"github.com/pkg/errors"
+	"github.com/songquanpeng/one-api/common/ctxkey"
+	"github.com/songquanpeng/one-api/relay/adaptor/anthropic"
+	"github.com/songquanpeng/one-api/relay/adaptor/aws/utils"
+	"github.com/songquanpeng/one-api/relay/meta"
+	"github.com/songquanpeng/one-api/relay/model"
+)
+
+var _ utils.AwsAdapter = new(Adaptor)
+
+type Adaptor struct {
+}
+
+func (a *Adaptor) ConvertRequest(c *gin.Context, relayMode int, request *model.GeneralOpenAIRequest) (any, error) {
+	if request == nil {
+		return nil, errors.New("request is nil")
+	}
+
+	claudeReq := anthropic.ConvertRequest(*request)
+	c.Set(ctxkey.RequestModel, request.Model)
+	c.Set(ctxkey.ConvertedRequest, claudeReq)
+	return claudeReq, nil
+}
+
+func (a *Adaptor) DoResponse(c *gin.Context, awsCli *bedrockruntime.Client, meta *meta.Meta) (usage *model.Usage, err *model.ErrorWithStatusCode) {
+	if meta.IsStream {
+		err, usage = StreamHandler(c, awsCli)
+	} else {
+		err, usage = Handler(c, awsCli, meta.ActualModelName)
+	}
+	return
+}

relay/adaptor/aws/claude/main.go

@@ -5,7 +5,6 @@ import (
 	"bytes"
 	"encoding/json"
 	"fmt"
-	"github.com/songquanpeng/one-api/common/ctxkey"
 	"io"
 	"net/http"
 
@@ -16,33 +15,28 @@ import (
 	"github.com/jinzhu/copier"
 	"github.com/pkg/errors"
 	"github.com/songquanpeng/one-api/common"
+	"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/relay/adaptor/anthropic"
+	"github.com/songquanpeng/one-api/relay/adaptor/aws/utils"
+	"github.com/songquanpeng/one-api/relay/adaptor/openai"
 	relaymodel "github.com/songquanpeng/one-api/relay/model"
 )
 
-func wrapErr(err error) *relaymodel.ErrorWithStatusCode {
-	return &relaymodel.ErrorWithStatusCode{
-		StatusCode: http.StatusInternalServerError,
-		Error: relaymodel.Error{
-			Message: fmt.Sprintf("%s", err.Error()),
-		},
-	}
-}
-
 // https://docs.aws.amazon.com/bedrock/latest/userguide/model-ids.html
-var awsModelIDMap = map[string]string{
+var AwsModelIDMap = map[string]string{
 	"claude-instant-1.2":         "anthropic.claude-instant-v1",
 	"claude-2.0":                 "anthropic.claude-v2",
 	"claude-2.1":                 "anthropic.claude-v2:1",
 	"claude-3-sonnet-20240229":   "anthropic.claude-3-sonnet-20240229-v1:0",
+	"claude-3-5-sonnet-20240620": "anthropic.claude-3-5-sonnet-20240620-v1:0",
 	"claude-3-opus-20240229":     "anthropic.claude-3-opus-20240229-v1:0",
 	"claude-3-haiku-20240307":    "anthropic.claude-3-haiku-20240307-v1:0",
 }
 
 func awsModelID(requestModel string) (string, error) {
-	if awsModelID, ok := awsModelIDMap[requestModel]; ok {
+	if awsModelID, ok := AwsModelIDMap[requestModel]; ok {
 		return awsModelID, nil
 	}
 
@@ -52,7 +46,7 @@ func awsModelID(requestModel string) (string, error) {
 func Handler(c *gin.Context, awsCli *bedrockruntime.Client, modelName string) (*relaymodel.ErrorWithStatusCode, *relaymodel.Usage) {
 	awsModelId, err := awsModelID(c.GetString(ctxkey.RequestModel))
 	if err != nil {
-		return wrapErr(errors.Wrap(err, "awsModelID")), nil
+		return utils.WrapErr(errors.Wrap(err, "awsModelID")), nil
 	}
 
 	awsReq := &bedrockruntime.InvokeModelInput{
@@ -63,30 +57,30 @@ func Handler(c *gin.Context, awsCli *bedrockruntime.Client, modelName string) (*
 
 	claudeReq_, ok := c.Get(ctxkey.ConvertedRequest)
 	if !ok {
-		return wrapErr(errors.New("request not found")), nil
+		return utils.WrapErr(errors.New("request not found")), nil
 	}
 	claudeReq := claudeReq_.(*anthropic.Request)
 	awsClaudeReq := &Request{
 		AnthropicVersion: "bedrock-2023-05-31",
 	}
 	if err = copier.Copy(awsClaudeReq, claudeReq); err != nil {
-		return wrapErr(errors.Wrap(err, "copy request")), nil
+		return utils.WrapErr(errors.Wrap(err, "copy request")), nil
 	}
 
 	awsReq.Body, err = json.Marshal(awsClaudeReq)
 	if err != nil {
-		return wrapErr(errors.Wrap(err, "marshal request")), nil
+		return utils.WrapErr(errors.Wrap(err, "marshal request")), nil
 	}
 
 	awsResp, err := awsCli.InvokeModel(c.Request.Context(), awsReq)
 	if err != nil {
-		return wrapErr(errors.Wrap(err, "InvokeModel")), nil
+		return utils.WrapErr(errors.Wrap(err, "InvokeModel")), nil
 	}
 
 	claudeResponse := new(anthropic.Response)
 	err = json.Unmarshal(awsResp.Body, claudeResponse)
 	if err != nil {
-		return wrapErr(errors.Wrap(err, "unmarshal response")), nil
+		return utils.WrapErr(errors.Wrap(err, "unmarshal response")), nil
 	}
 
 	openaiResp := anthropic.ResponseClaude2OpenAI(claudeResponse)
@@ -106,7 +100,7 @@ func StreamHandler(c *gin.Context, awsCli *bedrockruntime.Client) (*relaymodel.E
 	createdTime := helper.GetTimestamp()
 	awsModelId, err := awsModelID(c.GetString(ctxkey.RequestModel))
 	if err != nil {
-		return wrapErr(errors.Wrap(err, "awsModelID")), nil
+		return utils.WrapErr(errors.Wrap(err, "awsModelID")), nil
 	}
 
 	awsReq := &bedrockruntime.InvokeModelWithResponseStreamInput{
@@ -117,7 +111,7 @@ func StreamHandler(c *gin.Context, awsCli *bedrockruntime.Client) (*relaymodel.E
 
 	claudeReq_, ok := c.Get(ctxkey.ConvertedRequest)
 	if !ok {
-		return wrapErr(errors.New("request not found")), nil
+		return utils.WrapErr(errors.New("request not found")), nil
 	}
 	claudeReq := claudeReq_.(*anthropic.Request)
 
@@ -125,16 +119,16 @@ func StreamHandler(c *gin.Context, awsCli *bedrockruntime.Client) (*relaymodel.E
 		AnthropicVersion: "bedrock-2023-05-31",
 	}
 	if err = copier.Copy(awsClaudeReq, claudeReq); err != nil {
-		return wrapErr(errors.Wrap(err, "copy request")), nil
+		return utils.WrapErr(errors.Wrap(err, "copy request")), nil
 	}
 	awsReq.Body, err = json.Marshal(awsClaudeReq)
 	if err != nil {
-		return wrapErr(errors.Wrap(err, "marshal request")), nil
+		return utils.WrapErr(errors.Wrap(err, "marshal request")), nil
 	}
 
 	awsResp, err := awsCli.InvokeModelWithResponseStream(c.Request.Context(), awsReq)
 	if err != nil {
-		return wrapErr(errors.Wrap(err, "InvokeModelWithResponseStream")), nil
+		return utils.WrapErr(errors.Wrap(err, "InvokeModelWithResponseStream")), nil
 	}
 	stream := awsResp.GetStream()
 	defer stream.Close()
@@ -142,6 +136,8 @@ func StreamHandler(c *gin.Context, awsCli *bedrockruntime.Client) (*relaymodel.E
 	c.Writer.Header().Set("Content-Type", "text/event-stream")
 	var usage relaymodel.Usage
 	var id string
+	var lastToolCallChoice openai.ChatCompletionsStreamResponseChoice
+
 	c.Stream(func(w io.Writer) bool {
 		event, ok := <-stream.Events()
 		if !ok {
@@ -162,8 +158,19 @@ func StreamHandler(c *gin.Context, awsCli *bedrockruntime.Client) (*relaymodel.E
 			if meta != nil {
 				usage.PromptTokens += meta.Usage.InputTokens
 				usage.CompletionTokens += meta.Usage.OutputTokens
+				if len(meta.Id) > 0 { // only message_start has an id, otherwise it's a finish_reason event.
 					id = fmt.Sprintf("chatcmpl-%s", meta.Id)
 					return true
+				} else { // finish_reason case
+					if len(lastToolCallChoice.Delta.ToolCalls) > 0 {
+						lastArgs := &lastToolCallChoice.Delta.ToolCalls[len(lastToolCallChoice.Delta.ToolCalls)-1].Function
+						if len(lastArgs.Arguments.(string)) == 0 { // compatible with OpenAI sending an empty object `{}` when no arguments.
+							lastArgs.Arguments = "{}"
+							response.Choices[len(response.Choices)-1].Delta.Content = nil
+							response.Choices[len(response.Choices)-1].Delta.ToolCalls = lastToolCallChoice.Delta.ToolCalls
+						}
+					}
+				}
 			}
 			if response == nil {
 				return true
@@ -171,6 +178,12 @@ func StreamHandler(c *gin.Context, awsCli *bedrockruntime.Client) (*relaymodel.E
 			response.Id = id
 			response.Model = c.GetString(ctxkey.OriginalModel)
 			response.Created = createdTime
+
+			for _, choice := range response.Choices {
+				if len(choice.Delta.ToolCalls) > 0 {
+					lastToolCallChoice = choice
+				}
+			}
 			jsonStr, err := json.Marshal(response)
 			if err != nil {
 				logger.SysError("error marshalling stream response: " + err.Error())

relay/adaptor/aws/claude/model.go

@@ -9,9 +9,12 @@ type Request struct {
 	// AnthropicVersion should be "bedrock-2023-05-31"
 	AnthropicVersion string              `json:"anthropic_version"`
 	Messages         []anthropic.Message `json:"messages"`
+	System           string              `json:"system,omitempty"`
 	MaxTokens        int                 `json:"max_tokens,omitempty"`
 	Temperature      float64             `json:"temperature,omitempty"`
 	TopP             float64             `json:"top_p,omitempty"`
 	TopK             int                 `json:"top_k,omitempty"`
 	StopSequences    []string            `json:"stop_sequences,omitempty"`
+	Tools            []anthropic.Tool    `json:"tools,omitempty"`
+	ToolChoice       any                 `json:"tool_choice,omitempty"`
 }

relay/adaptor/aws/llama3/adapter.go

@@ -0,0 +1,37 @@
+package aws
+
+import (
+	"github.com/aws/aws-sdk-go-v2/service/bedrockruntime"
+	"github.com/songquanpeng/one-api/common/ctxkey"
+
+	"github.com/gin-gonic/gin"
+	"github.com/pkg/errors"
+	"github.com/songquanpeng/one-api/relay/adaptor/aws/utils"
+	"github.com/songquanpeng/one-api/relay/meta"
+	"github.com/songquanpeng/one-api/relay/model"
+)
+
+var _ utils.AwsAdapter = new(Adaptor)
+
+type Adaptor struct {
+}
+
+func (a *Adaptor) ConvertRequest(c *gin.Context, relayMode int, request *model.GeneralOpenAIRequest) (any, error) {
+	if request == nil {
+		return nil, errors.New("request is nil")
+	}
+
+	llamaReq := ConvertRequest(*request)
+	c.Set(ctxkey.RequestModel, request.Model)
+	c.Set(ctxkey.ConvertedRequest, llamaReq)
+	return llamaReq, nil
+}
+
+func (a *Adaptor) DoResponse(c *gin.Context, awsCli *bedrockruntime.Client, meta *meta.Meta) (usage *model.Usage, err *model.ErrorWithStatusCode) {
+	if meta.IsStream {
+		err, usage = StreamHandler(c, awsCli)
+	} else {
+		err, usage = Handler(c, awsCli, meta.ActualModelName)
+	}
+	return
+}

relay/adaptor/aws/llama3/main.go

@@ -0,0 +1,231 @@
+// Package aws provides the AWS adaptor for the relay service.
+package aws
+
+import (
+	"bytes"
+	"encoding/json"
+	"fmt"
+	"io"
+	"net/http"
+	"text/template"
+
+	"github.com/songquanpeng/one-api/common/ctxkey"
+	"github.com/songquanpeng/one-api/common/random"
+
+	"github.com/aws/aws-sdk-go-v2/aws"
+	"github.com/aws/aws-sdk-go-v2/service/bedrockruntime"
+	"github.com/aws/aws-sdk-go-v2/service/bedrockruntime/types"
+	"github.com/gin-gonic/gin"
+	"github.com/pkg/errors"
+	"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/adaptor/aws/utils"
+	"github.com/songquanpeng/one-api/relay/adaptor/openai"
+	relaymodel "github.com/songquanpeng/one-api/relay/model"
+)
+
+// Only support llama-3-8b and llama-3-70b instruction models
+// https://docs.aws.amazon.com/bedrock/latest/userguide/model-ids.html
+var AwsModelIDMap = map[string]string{
+	"llama3-8b-8192":  "meta.llama3-8b-instruct-v1:0",
+	"llama3-70b-8192": "meta.llama3-70b-instruct-v1:0",
+}
+
+func awsModelID(requestModel string) (string, error) {
+	if awsModelID, ok := AwsModelIDMap[requestModel]; ok {
+		return awsModelID, nil
+	}
+
+	return "", errors.Errorf("model %s not found", requestModel)
+}
+
+// promptTemplate with range
+const promptTemplate = `<|begin_of_text|>{{range .Messages}}<|start_header_id|>{{.Role}}<|end_header_id|>{{.StringContent}}<|eot_id|>{{end}}<|start_header_id|>assistant<|end_header_id|>
+`
+
+var promptTpl = template.Must(template.New("llama3-chat").Parse(promptTemplate))
+
+func RenderPrompt(messages []relaymodel.Message) string {
+	var buf bytes.Buffer
+	err := promptTpl.Execute(&buf, struct{ Messages []relaymodel.Message }{messages})
+	if err != nil {
+		logger.SysError("error rendering prompt messages: " + err.Error())
+	}
+	return buf.String()
+}
+
+func ConvertRequest(textRequest relaymodel.GeneralOpenAIRequest) *Request {
+	llamaRequest := Request{
+		MaxGenLen:   textRequest.MaxTokens,
+		Temperature: textRequest.Temperature,
+		TopP:        textRequest.TopP,
+	}
+	if llamaRequest.MaxGenLen == 0 {
+		llamaRequest.MaxGenLen = 2048
+	}
+	prompt := RenderPrompt(textRequest.Messages)
+	llamaRequest.Prompt = prompt
+	return &llamaRequest
+}
+
+func Handler(c *gin.Context, awsCli *bedrockruntime.Client, modelName string) (*relaymodel.ErrorWithStatusCode, *relaymodel.Usage) {
+	awsModelId, err := awsModelID(c.GetString(ctxkey.RequestModel))
+	if err != nil {
+		return utils.WrapErr(errors.Wrap(err, "awsModelID")), nil
+	}
+
+	awsReq := &bedrockruntime.InvokeModelInput{
+		ModelId:     aws.String(awsModelId),
+		Accept:      aws.String("application/json"),
+		ContentType: aws.String("application/json"),
+	}
+
+	llamaReq, ok := c.Get(ctxkey.ConvertedRequest)
+	if !ok {
+		return utils.WrapErr(errors.New("request not found")), nil
+	}
+
+	awsReq.Body, err = json.Marshal(llamaReq)
+	if err != nil {
+		return utils.WrapErr(errors.Wrap(err, "marshal request")), nil
+	}
+
+	awsResp, err := awsCli.InvokeModel(c.Request.Context(), awsReq)
+	if err != nil {
+		return utils.WrapErr(errors.Wrap(err, "InvokeModel")), nil
+	}
+
+	var llamaResponse Response
+	err = json.Unmarshal(awsResp.Body, &llamaResponse)
+	if err != nil {
+		return utils.WrapErr(errors.Wrap(err, "unmarshal response")), nil
+	}
+
+	openaiResp := ResponseLlama2OpenAI(&llamaResponse)
+	openaiResp.Model = modelName
+	usage := relaymodel.Usage{
+		PromptTokens:     llamaResponse.PromptTokenCount,
+		CompletionTokens: llamaResponse.GenerationTokenCount,
+		TotalTokens:      llamaResponse.PromptTokenCount + llamaResponse.GenerationTokenCount,
+	}
+	openaiResp.Usage = usage
+
+	c.JSON(http.StatusOK, openaiResp)
+	return nil, &usage
+}
+
+func ResponseLlama2OpenAI(llamaResponse *Response) *openai.TextResponse {
+	var responseText string
+	if len(llamaResponse.Generation) > 0 {
+		responseText = llamaResponse.Generation
+	}
+	choice := openai.TextResponseChoice{
+		Index: 0,
+		Message: relaymodel.Message{
+			Role:    "assistant",
+			Content: responseText,
+			Name:    nil,
+		},
+		FinishReason: llamaResponse.StopReason,
+	}
+	fullTextResponse := openai.TextResponse{
+		Id:      fmt.Sprintf("chatcmpl-%s", random.GetUUID()),
+		Object:  "chat.completion",
+		Created: helper.GetTimestamp(),
+		Choices: []openai.TextResponseChoice{choice},
+	}
+	return &fullTextResponse
+}
+
+func StreamHandler(c *gin.Context, awsCli *bedrockruntime.Client) (*relaymodel.ErrorWithStatusCode, *relaymodel.Usage) {
+	createdTime := helper.GetTimestamp()
+	awsModelId, err := awsModelID(c.GetString(ctxkey.RequestModel))
+	if err != nil {
+		return utils.WrapErr(errors.Wrap(err, "awsModelID")), nil
+	}
+
+	awsReq := &bedrockruntime.InvokeModelWithResponseStreamInput{
+		ModelId:     aws.String(awsModelId),
+		Accept:      aws.String("application/json"),
+		ContentType: aws.String("application/json"),
+	}
+
+	llamaReq, ok := c.Get(ctxkey.ConvertedRequest)
+	if !ok {
+		return utils.WrapErr(errors.New("request not found")), nil
+	}
+
+	awsReq.Body, err = json.Marshal(llamaReq)
+	if err != nil {
+		return utils.WrapErr(errors.Wrap(err, "marshal request")), nil
+	}
+
+	awsResp, err := awsCli.InvokeModelWithResponseStream(c.Request.Context(), awsReq)
+	if err != nil {
+		return utils.WrapErr(errors.Wrap(err, "InvokeModelWithResponseStream")), nil
+	}
+	stream := awsResp.GetStream()
+	defer stream.Close()
+
+	c.Writer.Header().Set("Content-Type", "text/event-stream")
+	var usage relaymodel.Usage
+	c.Stream(func(w io.Writer) bool {
+		event, ok := <-stream.Events()
+		if !ok {
+			c.Render(-1, common.CustomEvent{Data: "data: [DONE]"})
+			return false
+		}
+
+		switch v := event.(type) {
+		case *types.ResponseStreamMemberChunk:
+			var llamaResp StreamResponse
+			err := json.NewDecoder(bytes.NewReader(v.Value.Bytes)).Decode(&llamaResp)
+			if err != nil {
+				logger.SysError("error unmarshalling stream response: " + err.Error())
+				return false
+			}
+
+			if llamaResp.PromptTokenCount > 0 {
+				usage.PromptTokens = llamaResp.PromptTokenCount
+			}
+			if llamaResp.StopReason == "stop" {
+				usage.CompletionTokens = llamaResp.GenerationTokenCount
+				usage.TotalTokens = usage.PromptTokens + usage.CompletionTokens
+			}
+			response := StreamResponseLlama2OpenAI(&llamaResp)
+			response.Id = fmt.Sprintf("chatcmpl-%s", random.GetUUID())
+			response.Model = c.GetString(ctxkey.OriginalModel)
+			response.Created = createdTime
+			jsonStr, err := json.Marshal(response)
+			if err != nil {
+				logger.SysError("error marshalling stream response: " + err.Error())
+				return true
+			}
+			c.Render(-1, common.CustomEvent{Data: "data: " + string(jsonStr)})
+			return true
+		case *types.UnknownUnionMember:
+			fmt.Println("unknown tag:", v.Tag)
+			return false
+		default:
+			fmt.Println("union is nil or unknown type")
+			return false
+		}
+	})
+
+	return nil, &usage
+}
+
+func StreamResponseLlama2OpenAI(llamaResponse *StreamResponse) *openai.ChatCompletionsStreamResponse {
+	var choice openai.ChatCompletionsStreamResponseChoice
+	choice.Delta.Content = llamaResponse.Generation
+	choice.Delta.Role = "assistant"
+	finishReason := llamaResponse.StopReason
+	if finishReason != "null" {
+		choice.FinishReason = &finishReason
+	}
+	var openaiResponse openai.ChatCompletionsStreamResponse
+	openaiResponse.Object = "chat.completion.chunk"
+	openaiResponse.Choices = []openai.ChatCompletionsStreamResponseChoice{choice}
+	return &openaiResponse
+}

relay/adaptor/aws/llama3/main_test.go

@@ -0,0 +1,45 @@
+package aws_test
+
+import (
+	"testing"
+
+	aws "github.com/songquanpeng/one-api/relay/adaptor/aws/llama3"
+	relaymodel "github.com/songquanpeng/one-api/relay/model"
+	"github.com/stretchr/testify/assert"
+)
+
+func TestRenderPrompt(t *testing.T) {
+	messages := []relaymodel.Message{
+		{
+			Role:    "user",
+			Content: "What's your name?",
+		},
+	}
+	prompt := aws.RenderPrompt(messages)
+	expected := `<|begin_of_text|><|start_header_id|>user<|end_header_id|>What's your name?<|eot_id|><|start_header_id|>assistant<|end_header_id|>
+`
+	assert.Equal(t, expected, prompt)
+
+	messages = []relaymodel.Message{
+		{
+			Role:    "system",
+			Content: "Your name is Kat. You are a detective.",
+		},
+		{
+			Role:    "user",
+			Content: "What's your name?",
+		},
+		{
+			Role:    "assistant",
+			Content: "Kat",
+		},
+		{
+			Role:    "user",
+			Content: "What's your job?",
+		},
+	}
+	prompt = aws.RenderPrompt(messages)
+	expected = `<|begin_of_text|><|start_header_id|>system<|end_header_id|>Your name is Kat. You are a detective.<|eot_id|><|start_header_id|>user<|end_header_id|>What's your name?<|eot_id|><|start_header_id|>assistant<|end_header_id|>Kat<|eot_id|><|start_header_id|>user<|end_header_id|>What's your job?<|eot_id|><|start_header_id|>assistant<|end_header_id|>
+`
+	assert.Equal(t, expected, prompt)
+}

relay/adaptor/aws/llama3/model.go

@@ -0,0 +1,29 @@
+package aws
+
+// Request is the request to AWS Llama3
+//
+// https://docs.aws.amazon.com/bedrock/latest/userguide/model-parameters-meta.html
+type Request struct {
+	Prompt      string  `json:"prompt"`
+	MaxGenLen   int     `json:"max_gen_len,omitempty"`
+	Temperature float64 `json:"temperature,omitempty"`
+	TopP        float64 `json:"top_p,omitempty"`
+}
+
+// Response is the response from AWS Llama3
+//
+// https://docs.aws.amazon.com/bedrock/latest/userguide/model-parameters-meta.html
+type Response struct {
+	Generation           string `json:"generation"`
+	PromptTokenCount     int    `json:"prompt_token_count"`
+	GenerationTokenCount int    `json:"generation_token_count"`
+	StopReason           string `json:"stop_reason"`
+}
+
+// {'generation': 'Hi', 'prompt_token_count': 15, 'generation_token_count': 1, 'stop_reason': None}
+type StreamResponse struct {
+	Generation           string `json:"generation"`
+	PromptTokenCount     int    `json:"prompt_token_count"`
+	GenerationTokenCount int    `json:"generation_token_count"`
+	StopReason           string `json:"stop_reason"`
+}

relay/adaptor/aws/registry.go

@@ -0,0 +1,39 @@
+package aws
+
+import (
+	claude "github.com/songquanpeng/one-api/relay/adaptor/aws/claude"
+	llama3 "github.com/songquanpeng/one-api/relay/adaptor/aws/llama3"
+	"github.com/songquanpeng/one-api/relay/adaptor/aws/utils"
+)
+
+type AwsModelType int
+
+const (
+	AwsClaude AwsModelType = iota + 1
+	AwsLlama3
+)
+
+var (
+	adaptors = map[string]AwsModelType{}
+)
+
+func init() {
+	for model := range claude.AwsModelIDMap {
+		adaptors[model] = AwsClaude
+	}
+	for model := range llama3.AwsModelIDMap {
+		adaptors[model] = AwsLlama3
+	}
+}
+
+func GetAdaptor(model string) utils.AwsAdapter {
+	adaptorType := adaptors[model]
+	switch adaptorType {
+	case AwsClaude:
+		return &claude.Adaptor{}
+	case AwsLlama3:
+		return &llama3.Adaptor{}
+	default:
+		return nil
+	}
+}

relay/adaptor/aws/utils/adaptor.go

@@ -0,0 +1,51 @@
+package utils
+
+import (
+	"errors"
+	"io"
+	"net/http"
+
+	"github.com/aws/aws-sdk-go-v2/aws"
+	"github.com/aws/aws-sdk-go-v2/credentials"
+	"github.com/aws/aws-sdk-go-v2/service/bedrockruntime"
+	"github.com/gin-gonic/gin"
+	"github.com/songquanpeng/one-api/relay/meta"
+	"github.com/songquanpeng/one-api/relay/model"
+)
+
+type AwsAdapter interface {
+	ConvertRequest(c *gin.Context, relayMode int, request *model.GeneralOpenAIRequest) (any, error)
+	DoResponse(c *gin.Context, awsCli *bedrockruntime.Client, meta *meta.Meta) (usage *model.Usage, err *model.ErrorWithStatusCode)
+}
+
+type Adaptor struct {
+	Meta      *meta.Meta
+	AwsClient *bedrockruntime.Client
+}
+
+func (a *Adaptor) Init(meta *meta.Meta) {
+	a.Meta = meta
+	a.AwsClient = bedrockruntime.New(bedrockruntime.Options{
+		Region:      meta.Config.Region,
+		Credentials: aws.NewCredentialsCache(credentials.NewStaticCredentialsProvider(meta.Config.AK, meta.Config.SK, "")),
+	})
+}
+
+func (a *Adaptor) GetRequestURL(meta *meta.Meta) (string, error) {
+	return "", nil
+}
+
+func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, meta *meta.Meta) error {
+	return nil
+}
+
+func (a *Adaptor) ConvertImageRequest(request *model.ImageRequest) (any, error) {
+	if request == nil {
+		return nil, errors.New("request is nil")
+	}
+	return request, nil
+}
+
+func (a *Adaptor) DoRequest(c *gin.Context, meta *meta.Meta, requestBody io.Reader) (*http.Response, error) {
+	return nil, nil
+}

relay/adaptor/aws/utils/utils.go

@@ -0,0 +1,16 @@
+package utils
+
+import (
+	"net/http"
+
+	relaymodel "github.com/songquanpeng/one-api/relay/model"
+)
+
+func WrapErr(err error) *relaymodel.ErrorWithStatusCode {
+	return &relaymodel.ErrorWithStatusCode{
+		StatusCode: http.StatusInternalServerError,
+		Error: relaymodel.Error{
+			Message: err.Error(),
+		},
+	}
+}

relay/adaptor/baidu/main.go

@@ -5,6 +5,13 @@ import (
 	"encoding/json"
 	"errors"
 	"fmt"
+	"github.com/songquanpeng/one-api/common/render"
+	"io"
+	"net/http"
+	"strings"
+	"sync"
+	"time"
+
 	"github.com/gin-gonic/gin"
 	"github.com/songquanpeng/one-api/common"
 	"github.com/songquanpeng/one-api/common/client"
@@ -12,11 +19,6 @@ import (
 	"github.com/songquanpeng/one-api/relay/adaptor/openai"
 	"github.com/songquanpeng/one-api/relay/constant"
 	"github.com/songquanpeng/one-api/relay/model"
-	"io"
-	"net/http"
-	"strings"
-	"sync"
-	"time"
 )
 
 // https://cloud.baidu.com/doc/WENXINWORKSHOP/s/flfmc9do2
@@ -137,40 +139,22 @@ func embeddingResponseBaidu2OpenAI(response *EmbeddingResponse) *openai.Embeddin
 func StreamHandler(c *gin.Context, resp *http.Response) (*model.ErrorWithStatusCode, *model.Usage) {
 	var usage model.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
+	common.SetEventStreamHeaders(c)
-		}
+
-		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() {
 	for scanner.Scan() {
 		data := scanner.Text()
-			if len(data) < 6 { // ignore blank line or wrong format
+		if len(data) < 6 {
 			continue
 		}
 		data = data[6:]
-			dataChan <- data
+
-		}
-		stopChan <- true
-	}()
-	common.SetEventStreamHeaders(c)
-	c.Stream(func(w io.Writer) bool {
-		select {
-		case data := <-dataChan:
 		var baiduResponse ChatStreamResponse
 		err := json.Unmarshal([]byte(data), &baiduResponse)
 		if err != nil {
 			logger.SysError("error unmarshalling stream response: " + err.Error())
-				return true
+			continue
 		}
 		if baiduResponse.Usage.TotalTokens != 0 {
 			usage.TotalTokens = baiduResponse.Usage.TotalTokens
@@ -178,18 +162,18 @@ func StreamHandler(c *gin.Context, resp *http.Response) (*model.ErrorWithStatusC
 			usage.CompletionTokens = baiduResponse.Usage.TotalTokens - baiduResponse.Usage.PromptTokens
 		}
 		response := streamResponseBaidu2OpenAI(&baiduResponse)
-			jsonResponse, err := json.Marshal(response)
+		err = render.ObjectData(c, response)
 		if err != nil {
-				logger.SysError("error marshalling stream response: " + err.Error())
+			logger.SysError(err.Error())
-				return true
 		}
-			c.Render(-1, common.CustomEvent{Data: "data: " + string(jsonResponse)})
-			return true
-		case <-stopChan:
-			c.Render(-1, common.CustomEvent{Data: "data: [DONE]"})
-			return false
 	}
-	})
+
+	if err := scanner.Err(); err != nil {
+		logger.SysError("error reading stream: " + err.Error())
+	}
+
+	render.Done(c)
+
 	err := resp.Body.Close()
 	if err != nil {
 		return openai.ErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), nil

relay/adaptor/cloudflare/adaptor.go

@@ -5,11 +5,13 @@ import (
 	"fmt"
 	"io"
 	"net/http"
+	"strings"
 
 	"github.com/gin-gonic/gin"
 	"github.com/songquanpeng/one-api/relay/adaptor"
 	"github.com/songquanpeng/one-api/relay/meta"
 	"github.com/songquanpeng/one-api/relay/model"
+	"github.com/songquanpeng/one-api/relay/relaymode"
 )
 
 type Adaptor struct {
@@ -27,8 +29,33 @@ func (a *Adaptor) Init(meta *meta.Meta) {
 	a.meta = meta
 }
 
+// WorkerAI cannot be used across accounts with AIGateWay
+// https://developers.cloudflare.com/ai-gateway/providers/workersai/#openai-compatible-endpoints
+// https://gateway.ai.cloudflare.com/v1/{account_id}/{gateway_id}/workers-ai
+func (a *Adaptor) isAIGateWay(baseURL string) bool {
+	return strings.HasPrefix(baseURL, "https://gateway.ai.cloudflare.com") && strings.HasSuffix(baseURL, "/workers-ai")
+}
+
 func (a *Adaptor) GetRequestURL(meta *meta.Meta) (string, error) {
-	return fmt.Sprintf("%s/client/v4/accounts/%s/ai/run/%s", meta.BaseURL, meta.Config.UserID, meta.ActualModelName), nil
+	isAIGateWay := a.isAIGateWay(meta.BaseURL)
+	var urlPrefix string
+	if isAIGateWay {
+		urlPrefix = meta.BaseURL
+	} else {
+		urlPrefix = fmt.Sprintf("%s/client/v4/accounts/%s/ai", meta.BaseURL, meta.Config.UserID)
+	}
+
+	switch meta.Mode {
+	case relaymode.ChatCompletions:
+		return fmt.Sprintf("%s/v1/chat/completions", urlPrefix), nil
+	case relaymode.Embeddings:
+		return fmt.Sprintf("%s/v1/embeddings", urlPrefix), nil
+	default:
+		if isAIGateWay {
+			return fmt.Sprintf("%s/%s", urlPrefix, meta.ActualModelName), nil
+		}
+		return fmt.Sprintf("%s/run/%s", urlPrefix, meta.ActualModelName), nil
+	}
 }
 
 func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, meta *meta.Meta) error {
@@ -41,7 +68,14 @@ func (a *Adaptor) ConvertRequest(c *gin.Context, relayMode int, request *model.G
 	if request == nil {
 		return nil, errors.New("request is nil")
 	}
-	return ConvertRequest(*request), nil
+	switch relayMode {
+	case relaymode.Completions:
+		return ConvertCompletionsRequest(*request), nil
+	case relaymode.ChatCompletions, relaymode.Embeddings:
+		return request, nil
+	default:
+		return nil, errors.New("not implemented")
+	}
 }
 
 func (a *Adaptor) DoRequest(c *gin.Context, meta *meta.Meta, requestBody io.Reader) (*http.Response, error) {

relay/adaptor/cloudflare/constant.go

@@ -1,6 +1,7 @@
 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",

relay/adaptor/cloudflare/main.go

@@ -2,12 +2,14 @@ package cloudflare
 
 import (
 	"bufio"
-	"bytes"
 	"encoding/json"
 	"io"
 	"net/http"
 	"strings"
 
+	"github.com/songquanpeng/one-api/common/ctxkey"
+	"github.com/songquanpeng/one-api/common/render"
+
 	"github.com/gin-gonic/gin"
 	"github.com/songquanpeng/one-api/common"
 	"github.com/songquanpeng/one-api/common/helper"
@@ -16,114 +18,66 @@ import (
 	"github.com/songquanpeng/one-api/relay/model"
 )
 
-func ConvertRequest(textRequest model.GeneralOpenAIRequest) *Request {
+func ConvertCompletionsRequest(textRequest model.GeneralOpenAIRequest) *Request {
-    var promptBuilder strings.Builder
+	p, _ := textRequest.Prompt.(string)
-    for _, message := range textRequest.Messages {
-        promptBuilder.WriteString(message.StringContent())
-        promptBuilder.WriteString("\n")  // 添加换行符来分隔每个消息
-    }
-
 	return &Request{
+		Prompt:      p,
 		MaxTokens:   textRequest.MaxTokens,
-        Prompt:      promptBuilder.String(),
 		Stream:      textRequest.Stream,
 		Temperature: textRequest.Temperature,
 	}
 }
 
-
-func ResponseCloudflare2OpenAI(cloudflareResponse *Response) *openai.TextResponse {
-	choice := openai.TextResponseChoice{
-		Index: 0,
-		Message: model.Message{
-			Role:    "assistant",
-			Content: cloudflareResponse.Result.Response,
-		},
-		FinishReason: "stop",
-	}
-	fullTextResponse := openai.TextResponse{
-		Object:  "chat.completion",
-		Created: helper.GetTimestamp(),
-		Choices: []openai.TextResponseChoice{choice},
-	}
-	return &fullTextResponse
-}
-
-func StreamResponseCloudflare2OpenAI(cloudflareResponse *StreamResponse) *openai.ChatCompletionsStreamResponse {
-	var choice openai.ChatCompletionsStreamResponseChoice
-	choice.Delta.Content = cloudflareResponse.Response
-	choice.Delta.Role = "assistant"
-	openaiResponse := openai.ChatCompletionsStreamResponse{
-		Object:  "chat.completion.chunk",
-		Choices: []openai.ChatCompletionsStreamResponseChoice{choice},
-		Created: helper.GetTimestamp(),
-	}
-	return &openaiResponse
-}
-
 func StreamHandler(c *gin.Context, resp *http.Response, promptTokens int, modelName string) (*model.ErrorWithStatusCode, *model.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
+	common.SetEventStreamHeaders(c)
-		}
+	id := helper.GetResponseID(c)
-		if i := bytes.IndexByte(data, '\n'); i >= 0 {
+	responseModel := c.GetString(ctxkey.OriginalModel)
-			return i + 1, data[0:i], nil
+	var responseText string
-		}
-		if atEOF {
-			return len(data), data, nil
-		}
-		return 0, nil, nil
-	})
 
-	dataChan := make(chan string)
-	stopChan := make(chan bool)
-	go func() {
 	for scanner.Scan() {
 		data := scanner.Text()
 		if len(data) < len("data: ") {
 			continue
 		}
 		data = strings.TrimPrefix(data, "data: ")
-			dataChan <- data
-		}
-		stopChan <- true
-	}()
-	common.SetEventStreamHeaders(c)
-	id := helper.GetResponseID(c)
-	responseModel := c.GetString("original_model")
-	var responseText string
-	c.Stream(func(w io.Writer) bool {
-		select {
-		case data := <-dataChan:
-			// some implementations may add \r at the end of data
 		data = strings.TrimSuffix(data, "\r")
-			var cloudflareResponse StreamResponse
+
-			err := json.Unmarshal([]byte(data), &cloudflareResponse)
+		if data == "[DONE]" {
+			break
+		}
+
+		var response openai.ChatCompletionsStreamResponse
+		err := json.Unmarshal([]byte(data), &response)
 		if err != nil {
 			logger.SysError("error unmarshalling stream response: " + err.Error())
-				return true
+			continue
 		}
-			response := StreamResponseCloudflare2OpenAI(&cloudflareResponse)
+		for _, v := range response.Choices {
-			if response == nil {
+			v.Delta.Role = "assistant"
-				return true
+			responseText += v.Delta.StringContent()
 		}
-			responseText += cloudflareResponse.Response
 		response.Id = id
-			response.Model = responseModel
+		response.Model = modelName
-			jsonStr, err := json.Marshal(response)
+		err = render.ObjectData(c, response)
 		if err != nil {
-				logger.SysError("error marshalling stream response: " + err.Error())
+			logger.SysError(err.Error())
-				return true
 		}
-			c.Render(-1, common.CustomEvent{Data: "data: " + string(jsonStr)})
-			return true
-		case <-stopChan:
-			c.Render(-1, common.CustomEvent{Data: "data: [DONE]"})
-			return false
 	}
-	})
+
-	_ = resp.Body.Close()
+	if err := scanner.Err(); err != nil {
+		logger.SysError("error reading stream: " + err.Error())
+	}
+
+	render.Done(c)
+
+	err := resp.Body.Close()
+	if err != nil {
+		return openai.ErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), nil
+	}
+
 	usage := openai.ResponseText2Usage(responseText, responseModel, promptTokens)
 	return nil, usage
 }
@@ -137,22 +91,25 @@ func Handler(c *gin.Context, resp *http.Response, promptTokens int, modelName st
 	if err != nil {
 		return openai.ErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), nil
 	}
-	var cloudflareResponse Response
+	var response openai.TextResponse
-	err = json.Unmarshal(responseBody, &cloudflareResponse)
+	err = json.Unmarshal(responseBody, &response)
 	if err != nil {
 		return openai.ErrorWrapper(err, "unmarshal_response_body_failed", http.StatusInternalServerError), nil
 	}
-	fullTextResponse := ResponseCloudflare2OpenAI(&cloudflareResponse)
+	response.Model = modelName
-	fullTextResponse.Model = modelName
+	var responseText string
-	usage := openai.ResponseText2Usage(cloudflareResponse.Result.Response, modelName, promptTokens)
+	for _, v := range response.Choices {
-	fullTextResponse.Usage = *usage
+		responseText += v.Message.Content.(string)
-	fullTextResponse.Id = helper.GetResponseID(c)
+	}
-	jsonResponse, err := json.Marshal(fullTextResponse)
+	usage := openai.ResponseText2Usage(responseText, modelName, promptTokens)
+	response.Usage = *usage
+	response.Id = helper.GetResponseID(c)
+	jsonResponse, err := json.Marshal(response)
 	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)
+	_, _ = c.Writer.Write(jsonResponse)
 	return nil, usage
 }

relay/adaptor/cloudflare/model.go

@@ -1,6 +1,9 @@
 package cloudflare
 
+import "github.com/songquanpeng/one-api/relay/model"
+
 type Request struct {
+	Messages    []model.Message `json:"messages,omitempty"`
 	Lora        string          `json:"lora,omitempty"`
 	MaxTokens   int             `json:"max_tokens,omitempty"`
 	Prompt      string          `json:"prompt,omitempty"`
@@ -8,18 +11,3 @@ type Request struct {
 	Stream      bool            `json:"stream,omitempty"`
 	Temperature float64         `json:"temperature,omitempty"`
 }
-
-type Result struct {
-	Response string `json:"response"`
-}
-
-type Response struct {
-	Result   Result   `json:"result"`
-	Success  bool     `json:"success"`
-	Errors   []string `json:"errors"`
-	Messages []string `json:"messages"`
-}
-
-type StreamResponse struct {
-	Response string `json:"response"`
-}

relay/adaptor/cohere/main.go

@@ -2,9 +2,9 @@ package cohere
 
 import (
 	"bufio"
-	"bytes"
 	"encoding/json"
 	"fmt"
+	"github.com/songquanpeng/one-api/common/render"
 	"io"
 	"net/http"
 	"strings"
@@ -134,66 +134,53 @@ func ResponseCohere2OpenAI(cohereResponse *Response) *openai.TextResponse {
 func StreamHandler(c *gin.Context, resp *http.Response) (*model.ErrorWithStatusCode, *model.Usage) {
 	createdTime := helper.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 {
-			return 0, nil, nil
-		}
-		if i := bytes.IndexByte(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() {
-		for scanner.Scan() {
-			data := scanner.Text()
-			dataChan <- data
-		}
-		stopChan <- true
-	}()
 	common.SetEventStreamHeaders(c)
 	var usage model.Usage
-	c.Stream(func(w io.Writer) bool {
+
-		select {
+	for scanner.Scan() {
-		case data := <-dataChan:
+		data := scanner.Text()
-			// some implementations may add \r at the end of data
 		data = strings.TrimSuffix(data, "\r")
+
 		var cohereResponse StreamResponse
 		err := json.Unmarshal([]byte(data), &cohereResponse)
 		if err != nil {
 			logger.SysError("error unmarshalling stream response: " + err.Error())
-				return true
+			continue
 		}
+
 		response, meta := StreamResponseCohere2OpenAI(&cohereResponse)
 		if meta != nil {
 			usage.PromptTokens += meta.Meta.Tokens.InputTokens
 			usage.CompletionTokens += meta.Meta.Tokens.OutputTokens
-				return true
+			continue
 		}
 		if response == nil {
-				return true
+			continue
 		}
+
 		response.Id = fmt.Sprintf("chatcmpl-%d", createdTime)
 		response.Model = c.GetString("original_model")
 		response.Created = createdTime
-			jsonStr, err := json.Marshal(response)
+
+		err = render.ObjectData(c, response)
 		if err != nil {
-				logger.SysError("error marshalling stream response: " + err.Error())
+			logger.SysError(err.Error())
-				return true
 		}
-			c.Render(-1, common.CustomEvent{Data: "data: " + string(jsonStr)})
-			return true
-		case <-stopChan:
-			c.Render(-1, common.CustomEvent{Data: "data: [DONE]"})
-			return false
 	}
-	})
+
-	_ = resp.Body.Close()
+	if err := scanner.Err(); err != nil {
+		logger.SysError("error reading stream: " + err.Error())
+	}
+
+	render.Done(c)
+
+	err := resp.Body.Close()
+	if err != nil {
+		return openai.ErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), nil
+	}
+
 	return nil, &usage
 }
 

relay/adaptor/coze/main.go

@@ -4,6 +4,11 @@ import (
 	"bufio"
 	"encoding/json"
 	"fmt"
+	"github.com/songquanpeng/one-api/common/render"
+	"io"
+	"net/http"
+	"strings"
+
 	"github.com/gin-gonic/gin"
 	"github.com/songquanpeng/one-api/common"
 	"github.com/songquanpeng/one-api/common/conv"
@@ -12,9 +17,6 @@ import (
 	"github.com/songquanpeng/one-api/relay/adaptor/coze/constant/messagetype"
 	"github.com/songquanpeng/one-api/relay/adaptor/openai"
 	"github.com/songquanpeng/one-api/relay/model"
-	"io"
-	"net/http"
-	"strings"
 )
 
 // https://www.coze.com/open
@@ -109,69 +111,54 @@ func StreamHandler(c *gin.Context, resp *http.Response) (*model.ErrorWithStatusC
 	var responseText string
 	createdTime := helper.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 {
+
-			return 0, nil, nil
+	common.SetEventStreamHeaders(c)
-		}
+	var modelName string
-		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() {
 	for scanner.Scan() {
 		data := scanner.Text()
-			if len(data) < 5 {
+		if len(data) < 5 || !strings.HasPrefix(data, "data:") {
-				continue
-			}
-			if !strings.HasPrefix(data, "data:") {
 			continue
 		}
 		data = strings.TrimPrefix(data, "data:")
-			dataChan <- data
-		}
-		stopChan <- true
-	}()
-	common.SetEventStreamHeaders(c)
-	var modelName string
-	c.Stream(func(w io.Writer) bool {
-		select {
-		case data := <-dataChan:
-			// some implementations may add \r at the end of data
 		data = strings.TrimSuffix(data, "\r")
+
 		var cozeResponse StreamResponse
 		err := json.Unmarshal([]byte(data), &cozeResponse)
 		if err != nil {
 			logger.SysError("error unmarshalling stream response: " + err.Error())
-				return true
+			continue
 		}
+
 		response, _ := StreamResponseCoze2OpenAI(&cozeResponse)
 		if response == nil {
-				return true
+			continue
 		}
+
 		for _, choice := range response.Choices {
 			responseText += conv.AsString(choice.Delta.Content)
 		}
 		response.Model = modelName
 		response.Created = createdTime
-			jsonStr, err := json.Marshal(response)
+
+		err = render.ObjectData(c, response)
 		if err != nil {
-				logger.SysError("error marshalling stream response: " + err.Error())
+			logger.SysError(err.Error())
-				return true
 		}
-			c.Render(-1, common.CustomEvent{Data: "data: " + string(jsonStr)})
-			return true
-		case <-stopChan:
-			c.Render(-1, common.CustomEvent{Data: "data: [DONE]"})
-			return false
 	}
-	})
+
-	_ = resp.Body.Close()
+	if err := scanner.Err(); err != nil {
+		logger.SysError("error reading stream: " + err.Error())
+	}
+
+	render.Done(c)
+
+	err := resp.Body.Close()
+	if err != nil {
+		return openai.ErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), nil
+	}
+
 	return nil, &responseText
 }
 

relay/adaptor/doubao/main.go

@@ -7,8 +7,12 @@ import (
 )
 
 func GetRequestURL(meta *meta.Meta) (string, error) {
-	if meta.Mode == relaymode.ChatCompletions {
+	switch meta.Mode {
+	case relaymode.ChatCompletions:
 		return fmt.Sprintf("%s/api/v3/chat/completions", meta.BaseURL), nil
+	case relaymode.Embeddings:
+		return fmt.Sprintf("%s/api/v3/embeddings", meta.BaseURL), nil
+	default:
 	}
 	return "", fmt.Errorf("unsupported relay mode %d for doubao", meta.Mode)
 }

relay/adaptor/gemini/constants.go

@@ -3,6 +3,5 @@ package gemini
 // https://ai.google.dev/models/gemini
 
 var ModelList = []string{
-	"gemini-pro", "gemini-1.0-pro-001", "gemini-1.5-pro",
+	"gemini-pro", "gemini-1.0-pro", "gemini-1.5-flash", "gemini-1.5-pro", "text-embedding-004", "aqa",
-	"gemini-pro-vision", "gemini-1.0-pro-vision-001", "embedding-001", "text-embedding-004",
 }

relay/adaptor/gemini/main.go

@@ -4,6 +4,7 @@ import (
 	"bufio"
 	"encoding/json"
 	"fmt"
+	"github.com/songquanpeng/one-api/common/render"
 	"io"
 	"net/http"
 	"strings"
@@ -245,8 +246,10 @@ 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}
@@ -273,21 +276,10 @@ func embeddingResponseGemini2OpenAI(response *EmbeddingResponse) *openai.Embeddi
 func StreamHandler(c *gin.Context, resp *http.Response) (*model.ErrorWithStatusCode, string) {
 	responseText := ""
 	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
+	common.SetEventStreamHeaders(c)
-		}
+
-		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() {
 	for scanner.Scan() {
 		data := scanner.Text()
 		data = strings.TrimSpace(data)
@@ -296,41 +288,38 @@ func StreamHandler(c *gin.Context, resp *http.Response) (*model.ErrorWithStatusC
 		}
 		data = strings.TrimPrefix(data, "data: ")
 		data = strings.TrimSuffix(data, "\"")
-			dataChan <- data
+
-		}
-		stopChan <- true
-	}()
-	common.SetEventStreamHeaders(c)
-	c.Stream(func(w io.Writer) bool {
-		select {
-		case data := <-dataChan:
 		var geminiResponse ChatResponse
 		err := json.Unmarshal([]byte(data), &geminiResponse)
 		if err != nil {
 			logger.SysError("error unmarshalling stream response: " + err.Error())
-				return true
+			continue
 		}
+
 		response := streamResponseGeminiChat2OpenAI(&geminiResponse)
 		if response == nil {
-				return true
+			continue
 		}
+
 		responseText += response.Choices[0].Delta.StringContent()
-			jsonResponse, err := json.Marshal(response)
+
+		err = render.ObjectData(c, response)
 		if err != nil {
-				logger.SysError("error marshalling stream response: " + err.Error())
+			logger.SysError(err.Error())
-				return true
 		}
-			c.Render(-1, common.CustomEvent{Data: "data: " + string(jsonResponse)})
-			return true
-		case <-stopChan:
-			c.Render(-1, common.CustomEvent{Data: "data: [DONE]"})
-			return false
 	}
-	})
+
+	if err := scanner.Err(); err != nil {
+		logger.SysError("error reading stream: " + err.Error())
+	}
+
+	render.Done(c)
+
 	err := resp.Body.Close()
 	if err != nil {
 		return openai.ErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), ""
 	}
+
 	return nil, responseText
 }
 

relay/adaptor/groq/constants.go

@@ -4,9 +4,14 @@ package groq
 
 var ModelList = []string{
 	"gemma-7b-it",
-	"llama2-7b-2048",
-	"llama2-70b-4096",
 	"mixtral-8x7b-32768",
 	"llama3-8b-8192",
 	"llama3-70b-8192",
+	"gemma2-9b-it",
+	"llama-3.1-405b-reasoning",
+	"llama-3.1-70b-versatile",
+	"llama-3.1-8b-instant",
+	"llama3-groq-70b-8192-tool-use-preview",
+	"llama3-groq-8b-8192-tool-use-preview",
+	"whisper-large-v3",
 }

relay/adaptor/novita/constants.go

@@ -0,0 +1,19 @@
+package novita
+
+// https://novita.ai/llm-api
+
+var ModelList = []string{
+	"meta-llama/llama-3-8b-instruct",
+	"meta-llama/llama-3-70b-instruct",
+	"nousresearch/hermes-2-pro-llama-3-8b",
+	"nousresearch/nous-hermes-llama2-13b",
+	"mistralai/mistral-7b-instruct",
+	"cognitivecomputations/dolphin-mixtral-8x22b",
+	"sao10k/l3-70b-euryale-v2.1",
+	"sophosympatheia/midnight-rose-70b",
+	"gryphe/mythomax-l2-13b",
+	"Nous-Hermes-2-Mixtral-8x7B-DPO",
+	"lzlv_70b",
+	"teknium/openhermes-2.5-mistral-7b",
+	"microsoft/wizardlm-2-8x22b",
+}

relay/adaptor/novita/main.go

@@ -0,0 +1,15 @@
+package novita
+
+import (
+	"fmt"
+
+	"github.com/songquanpeng/one-api/relay/meta"
+	"github.com/songquanpeng/one-api/relay/relaymode"
+)
+
+func GetRequestURL(meta *meta.Meta) (string, error) {
+	if meta.Mode == relaymode.ChatCompletions {
+		return fmt.Sprintf("%s/chat/completions", meta.BaseURL), nil
+	}
+	return "", fmt.Errorf("unsupported relay mode %d for novita", meta.Mode)
+}

relay/adaptor/ollama/adaptor.go

@@ -24,7 +24,7 @@ func (a *Adaptor) GetRequestURL(meta *meta.Meta) (string, error) {
 	// https://github.com/ollama/ollama/blob/main/docs/api.md
 	fullRequestURL := fmt.Sprintf("%s/api/chat", meta.BaseURL)
 	if meta.Mode == relaymode.Embeddings {
-		fullRequestURL = fmt.Sprintf("%s/api/embeddings", meta.BaseURL)
+		fullRequestURL = fmt.Sprintf("%s/api/embed", meta.BaseURL)
 	}
 	return fullRequestURL, nil
 }

relay/adaptor/ollama/main.go

@@ -5,12 +5,14 @@ import (
 	"context"
 	"encoding/json"
 	"fmt"
-	"github.com/songquanpeng/one-api/common/helper"
+	"github.com/songquanpeng/one-api/common/render"
-	"github.com/songquanpeng/one-api/common/random"
 	"io"
 	"net/http"
 	"strings"
 
+	"github.com/songquanpeng/one-api/common/helper"
+	"github.com/songquanpeng/one-api/common/random"
+
 	"github.com/gin-gonic/gin"
 	"github.com/songquanpeng/one-api/common"
 	"github.com/songquanpeng/one-api/common/image"
@@ -29,6 +31,8 @@ func ConvertRequest(request model.GeneralOpenAIRequest) *ChatRequest {
 			TopP:             request.TopP,
 			FrequencyPenalty: request.FrequencyPenalty,
 			PresencePenalty:  request.PresencePenalty,
+			NumPredict:  	  request.MaxTokens,
+			NumCtx:  	  request.NumCtx,
 		},
 		Stream: request.Stream,
 	}
@@ -105,61 +109,67 @@ func StreamHandler(c *gin.Context, resp *http.Response) (*model.ErrorWithStatusC
 			return 0, nil, nil
 		}
 		if i := strings.Index(string(data), "}\n"); i >= 0 {
-			return i + 2, data[0:i], nil
+			return i + 2, data[0 : i+1], nil
 		}
 		if atEOF {
 			return len(data), data, nil
 		}
 		return 0, nil, nil
 	})
-	dataChan := make(chan string)
+
-	stopChan := make(chan bool)
-	go func() {
-		for scanner.Scan() {
-			data := strings.TrimPrefix(scanner.Text(), "}")
-			dataChan <- data + "}"
-		}
-		stopChan <- true
-	}()
 	common.SetEventStreamHeaders(c)
-	c.Stream(func(w io.Writer) bool {
+
-		select {
+	for scanner.Scan() {
-		case data := <-dataChan:
+		data := scanner.Text()
+		if strings.HasPrefix(data, "}") {
+		    data = strings.TrimPrefix(data, "}") + "}"
+		}
+
 		var ollamaResponse ChatResponse
 		err := json.Unmarshal([]byte(data), &ollamaResponse)
 		if err != nil {
 			logger.SysError("error unmarshalling stream response: " + err.Error())
-				return true
+			continue
 		}
+
 		if ollamaResponse.EvalCount != 0 {
 			usage.PromptTokens = ollamaResponse.PromptEvalCount
 			usage.CompletionTokens = ollamaResponse.EvalCount
 			usage.TotalTokens = ollamaResponse.PromptEvalCount + ollamaResponse.EvalCount
 		}
+
 		response := streamResponseOllama2OpenAI(&ollamaResponse)
-			jsonResponse, err := json.Marshal(response)
+		err = render.ObjectData(c, response)
 		if err != nil {
-				logger.SysError("error marshalling stream response: " + err.Error())
+			logger.SysError(err.Error())
-				return true
 		}
-			c.Render(-1, common.CustomEvent{Data: "data: " + string(jsonResponse)})
-			return true
-		case <-stopChan:
-			c.Render(-1, common.CustomEvent{Data: "data: [DONE]"})
-			return false
 	}
-	})
+
+	if err := scanner.Err(); err != nil {
+		logger.SysError("error reading stream: " + err.Error())
+	}
+
+	render.Done(c)
+
 	err := resp.Body.Close()
 	if err != nil {
 		return openai.ErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), nil
 	}
+
 	return nil, &usage
 }
 
 func ConvertEmbeddingRequest(request model.GeneralOpenAIRequest) *EmbeddingRequest {
 	return &EmbeddingRequest{
 		Model: request.Model,
-		Prompt: strings.Join(request.ParseInput(), " "),
+		Input: request.ParseInput(),
+		Options: &Options{
+			Seed:             int(request.Seed),
+			Temperature:      request.Temperature,
+			TopP:             request.TopP,
+			FrequencyPenalty: request.FrequencyPenalty,
+			PresencePenalty:  request.PresencePenalty,
+		},
 	}
 }
 
@@ -202,15 +212,17 @@ func embeddingResponseOllama2OpenAI(response *EmbeddingResponse) *openai.Embeddi
 	openAIEmbeddingResponse := openai.EmbeddingResponse{
 		Object: "list",
 		Data:   make([]openai.EmbeddingResponseItem, 0, 1),
-		Model:  "text-embedding-v1",
+		Model:  response.Model,
 		Usage:  model.Usage{TotalTokens: 0},
 	}
 
+	for i, embedding := range response.Embeddings {
 		openAIEmbeddingResponse.Data = append(openAIEmbeddingResponse.Data, openai.EmbeddingResponseItem{
 			Object:    `embedding`,
-		Index:     0,
+			Index:     i,
-		Embedding: response.Embedding,
+			Embedding: embedding,
 		})
+	}
 	return &openAIEmbeddingResponse
 }
 

relay/adaptor/ollama/model.go

@@ -7,6 +7,8 @@ type Options struct {
 	TopP             float64 `json:"top_p,omitempty"`
 	FrequencyPenalty float64 `json:"frequency_penalty,omitempty"`
 	PresencePenalty  float64 `json:"presence_penalty,omitempty"`
+	NumPredict  	 int 	 `json:"num_predict,omitempty"`
+	NumCtx  	 int 	 `json:"num_ctx,omitempty"`
 }
 
 type Message struct {
@@ -38,10 +40,14 @@ type ChatResponse struct {
 
 type EmbeddingRequest struct {
 	Model string   `json:"model"`
-	Prompt string `json:"prompt"`
+	Input []string `json:"input"`
+	// Truncate  bool     `json:"truncate,omitempty"`
+	Options *Options `json:"options,omitempty"`
+	// KeepAlive string   `json:"keep_alive,omitempty"`
 }
 
 type EmbeddingResponse struct {
 	Error      string      `json:"error,omitempty"`
-	Embedding []float64 `json:"embedding,omitempty"`
+	Model      string      `json:"model"`
+	Embeddings [][]float64 `json:"embeddings"`
 }

relay/adaptor/openai/adaptor.go

@@ -3,17 +3,19 @@ package openai
 import (
 	"errors"
 	"fmt"
+	"io"
+	"net/http"
+	"strings"
+
 	"github.com/gin-gonic/gin"
 	"github.com/songquanpeng/one-api/relay/adaptor"
 	"github.com/songquanpeng/one-api/relay/adaptor/doubao"
 	"github.com/songquanpeng/one-api/relay/adaptor/minimax"
+	"github.com/songquanpeng/one-api/relay/adaptor/novita"
 	"github.com/songquanpeng/one-api/relay/channeltype"
 	"github.com/songquanpeng/one-api/relay/meta"
 	"github.com/songquanpeng/one-api/relay/model"
 	"github.com/songquanpeng/one-api/relay/relaymode"
-	"io"
-	"net/http"
-	"strings"
 )
 
 type Adaptor struct {
@@ -53,6 +55,8 @@ func (a *Adaptor) GetRequestURL(meta *meta.Meta) (string, error) {
 			return GetFullRequestURL("https://apivip.aiproxy.io", meta.RequestURLPath, meta.ChannelType), nil
 		}
 		fallthrough
+	case channeltype.Novita:
+		return novita.GetRequestURL(meta)
 	default:
 		return GetFullRequestURL(meta.BaseURL, meta.RequestURLPath, meta.ChannelType), nil
 	}

relay/adaptor/openai/compatible.go

@@ -10,8 +10,10 @@ import (
 	"github.com/songquanpeng/one-api/relay/adaptor/minimax"
 	"github.com/songquanpeng/one-api/relay/adaptor/mistral"
 	"github.com/songquanpeng/one-api/relay/adaptor/moonshot"
+	"github.com/songquanpeng/one-api/relay/adaptor/novita"
 	"github.com/songquanpeng/one-api/relay/adaptor/stepfun"
 	"github.com/songquanpeng/one-api/relay/adaptor/togetherai"
+	"github.com/songquanpeng/one-api/relay/adaptor/siliconflow"
 	"github.com/songquanpeng/one-api/relay/channeltype"
 )
 
@@ -28,6 +30,8 @@ var CompatibleChannels = []int{
 	channeltype.StepFun,
 	channeltype.DeepSeek,
 	channeltype.TogetherAI,
+	channeltype.Novita,
+	channeltype.SiliconFlow,
 }
 
 func GetCompatibleChannelMeta(channelType int) (string, []string) {
@@ -56,6 +60,10 @@ func GetCompatibleChannelMeta(channelType int) (string, []string) {
 		return "together.ai", togetherai.ModelList
 	case channeltype.Doubao:
 		return "doubao", doubao.ModelList
+	case channeltype.Novita:
+		return "novita", novita.ModelList
+	case channeltype.SiliconFlow:
+		return "siliconflow", siliconflow.ModelList
 	default:
 		return "openai", ModelList
 	}

relay/adaptor/openai/constants.go

@@ -8,6 +8,7 @@ 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-4o", "gpt-4o-2024-05-13",
+	"gpt-4o-mini", "gpt-4o-mini-2024-07-18",
 	"gpt-4-vision-preview",
 	"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",

relay/adaptor/openai/main.go

@@ -4,15 +4,18 @@ import (
 	"bufio"
 	"bytes"
 	"encoding/json"
+	"io"
+	"net/http"
+	"strings"
+
+	"github.com/songquanpeng/one-api/common/render"
+
 	"github.com/gin-gonic/gin"
 	"github.com/songquanpeng/one-api/common"
 	"github.com/songquanpeng/one-api/common/conv"
 	"github.com/songquanpeng/one-api/common/logger"
 	"github.com/songquanpeng/one-api/relay/model"
 	"github.com/songquanpeng/one-api/relay/relaymode"
-	"io"
-	"net/http"
-	"strings"
 )
 
 const (
@@ -24,22 +27,12 @@ const (
 func StreamHandler(c *gin.Context, resp *http.Response, relayMode int) (*model.ErrorWithStatusCode, string, *model.Usage) {
 	responseText := ""
 	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)
 	var usage *model.Usage
-	go func() {
+
+	common.SetEventStreamHeaders(c)
+
+	doneRendered := false
 	for scanner.Scan() {
 		data := scanner.Text()
 		if len(data) < dataPrefixLength { // ignore blank line or wrong format
@@ -49,7 +42,8 @@ func StreamHandler(c *gin.Context, resp *http.Response, relayMode int) (*model.E
 			continue
 		}
 		if strings.HasPrefix(data[dataPrefixLength:], done) {
-				dataChan <- data
+			render.StringData(c, data)
+			doneRendered = true
 			continue
 		}
 		switch relayMode {
@@ -58,14 +52,14 @@ func StreamHandler(c *gin.Context, resp *http.Response, relayMode int) (*model.E
 			err := json.Unmarshal([]byte(data[dataPrefixLength:]), &streamResponse)
 			if err != nil {
 				logger.SysError("error unmarshalling stream response: " + err.Error())
-					dataChan <- data // if error happened, pass the data to client
+				render.StringData(c, data) // if error happened, pass the data to client
 				continue                   // just ignore the error
 			}
 			if len(streamResponse.Choices) == 0 {
 				// but for empty choice, we should not pass it to client, this is for azure
 				continue // just ignore empty choice
 			}
-				dataChan <- data
+			render.StringData(c, data)
 			for _, choice := range streamResponse.Choices {
 				responseText += conv.AsString(choice.Delta.Content)
 			}
@@ -73,7 +67,7 @@ func StreamHandler(c *gin.Context, resp *http.Response, relayMode int) (*model.E
 				usage = streamResponse.Usage
 			}
 		case relaymode.Completions:
-				dataChan <- data
+			render.StringData(c, data)
 			var streamResponse CompletionsStreamResponse
 			err := json.Unmarshal([]byte(data[dataPrefixLength:]), &streamResponse)
 			if err != nil {
@@ -85,27 +79,20 @@ func StreamHandler(c *gin.Context, resp *http.Response, relayMode int) (*model.E
 			}
 		}
 	}
-		stopChan <- true
+
-	}()
+	if err := scanner.Err(); err != nil {
-	common.SetEventStreamHeaders(c)
+		logger.SysError("error reading stream: " + err.Error())
-	c.Stream(func(w io.Writer) bool {
-		select {
-		case data := <-dataChan:
-			if strings.HasPrefix(data, "data: [DONE]") {
-				data = data[:12]
 	}
-			// some implementations may add \r at the end of data
+
-			data = strings.TrimSuffix(data, "\r")
+	if !doneRendered {
-			c.Render(-1, common.CustomEvent{Data: data})
+		render.Done(c)
-			return true
-		case <-stopChan:
-			return false
 	}
-	})
+
 	err := resp.Body.Close()
 	if err != nil {
 		return ErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), "", nil
 	}
+
 	return nil, responseText, usage
 }
 
@@ -149,7 +136,7 @@ func Handler(c *gin.Context, resp *http.Response, promptTokens int, modelName st
 		return ErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), nil
 	}
 
-	if textResponse.Usage.TotalTokens == 0 {
+	if textResponse.Usage.TotalTokens == 0 || (textResponse.Usage.PromptTokens == 0 && textResponse.Usage.CompletionTokens == 0) {
 		completionTokens := 0
 		for _, choice := range textResponse.Choices {
 			completionTokens += CountTokenText(choice.Message.StringContent(), modelName)

relay/adaptor/openai/token.go

@@ -97,7 +97,11 @@ func CountTokenMessages(messages []model.Message, model string) int {
 				m := it.(map[string]any)
 				switch m["type"] {
 				case "text":
-					tokenNum += getTokenNum(tokenEncoder, m["text"].(string))
+					if textValue, ok := m["text"]; ok {
+						if textString, ok := textValue.(string); ok {
+							tokenNum += getTokenNum(tokenEncoder, textString)
+						}
+					}
 				case "image_url":
 					imageUrl, ok := m["image_url"].(map[string]any)
 					if ok {
@@ -106,7 +110,7 @@ func CountTokenMessages(messages []model.Message, model string) int {
 						if imageUrl["detail"] != nil {
 							detail = imageUrl["detail"].(string)
 						}
-						imageTokens, err := countImageTokens(url, detail)
+						imageTokens, err := countImageTokens(url, detail, model)
 						if err != nil {
 							logger.SysError("error counting image tokens: " + err.Error())
 						} else {
@@ -130,11 +134,15 @@ const (
 	lowDetailCost         = 85
 	highDetailCostPerTile = 170
 	additionalCost        = 85
+	// gpt-4o-mini cost higher than other model
+	gpt4oMiniLowDetailCost  = 2833
+	gpt4oMiniHighDetailCost = 5667
+	gpt4oMiniAdditionalCost = 2833
 )
 
 // https://platform.openai.com/docs/guides/vision/calculating-costs
 // https://github.com/openai/openai-cookbook/blob/05e3f9be4c7a2ae7ecf029a7c32065b024730ebe/examples/How_to_count_tokens_with_tiktoken.ipynb
-func countImageTokens(url string, detail string) (_ int, err error) {
+func countImageTokens(url string, detail string, model string) (_ int, err error) {
 	var fetchSize = true
 	var width, height int
 	// Reference: https://platform.openai.com/docs/guides/vision/low-or-high-fidelity-image-understanding
@@ -168,6 +176,9 @@ func countImageTokens(url string, detail string) (_ int, err error) {
 	}
 	switch detail {
 	case "low":
+		if strings.HasPrefix(model, "gpt-4o-mini") {
+			return gpt4oMiniLowDetailCost, nil
+		}
 		return lowDetailCost, nil
 	case "high":
 		if fetchSize {
@@ -187,6 +198,9 @@ func countImageTokens(url string, detail string) (_ int, err error) {
 			height = int(float64(height) * ratio)
 		}
 		numSquares := int(math.Ceil(float64(width)/512) * math.Ceil(float64(height)/512))
+		if strings.HasPrefix(model, "gpt-4o-mini") {
+			return numSquares*gpt4oMiniHighDetailCost + gpt4oMiniAdditionalCost, nil
+		}
 		result := numSquares*highDetailCostPerTile + additionalCost
 		return result, nil
 	default:

relay/adaptor/palm/palm.go

@@ -3,6 +3,10 @@ package palm
 import (
 	"encoding/json"
 	"fmt"
+	"github.com/songquanpeng/one-api/common/render"
+	"io"
+	"net/http"
+
 	"github.com/gin-gonic/gin"
 	"github.com/songquanpeng/one-api/common"
 	"github.com/songquanpeng/one-api/common/helper"
@@ -11,8 +15,6 @@ import (
 	"github.com/songquanpeng/one-api/relay/adaptor/openai"
 	"github.com/songquanpeng/one-api/relay/constant"
 	"github.com/songquanpeng/one-api/relay/model"
-	"io"
-	"net/http"
 )
 
 // https://developers.generativeai.google/api/rest/generativelanguage/models/generateMessage#request-body
@@ -77,58 +79,51 @@ func StreamHandler(c *gin.Context, resp *http.Response) (*model.ErrorWithStatusC
 	responseText := ""
 	responseId := fmt.Sprintf("chatcmpl-%s", random.GetUUID())
 	createdTime := helper.GetTimestamp()
-	dataChan := make(chan string)
+
-	stopChan := make(chan bool)
+	common.SetEventStreamHeaders(c)
-	go func() {
+
 	responseBody, err := io.ReadAll(resp.Body)
 	if err != nil {
 		logger.SysError("error reading stream response: " + err.Error())
-			stopChan <- true
+		err := resp.Body.Close()
-			return
+		if err != nil {
+			return openai.ErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), ""
 		}
+		return openai.ErrorWrapper(err, "read_response_body_failed", http.StatusInternalServerError), ""
+	}
+
 	err = resp.Body.Close()
 	if err != nil {
-			logger.SysError("error closing stream response: " + err.Error())
+		return openai.ErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), ""
-			stopChan <- true
-			return
 	}
+
 	var palmResponse ChatResponse
 	err = json.Unmarshal(responseBody, &palmResponse)
 	if err != nil {
 		logger.SysError("error unmarshalling stream response: " + err.Error())
-			stopChan <- true
+		return openai.ErrorWrapper(err, "unmarshal_response_body_failed", http.StatusInternalServerError), ""
-			return
 	}
+
 	fullTextResponse := streamResponsePaLM2OpenAI(&palmResponse)
 	fullTextResponse.Id = responseId
 	fullTextResponse.Created = createdTime
 	if len(palmResponse.Candidates) > 0 {
 		responseText = palmResponse.Candidates[0].Content
 	}
+
 	jsonResponse, err := json.Marshal(fullTextResponse)
 	if err != nil {
 		logger.SysError("error marshalling stream response: " + err.Error())
-			stopChan <- true
+		return openai.ErrorWrapper(err, "marshal_response_body_failed", http.StatusInternalServerError), ""
-			return
 	}
-		dataChan <- string(jsonResponse)
+
-		stopChan <- true
+	err = render.ObjectData(c, string(jsonResponse))
-	}()
-	common.SetEventStreamHeaders(c)
-	c.Stream(func(w io.Writer) bool {
-		select {
-		case data := <-dataChan:
-			c.Render(-1, common.CustomEvent{Data: "data: " + data})
-			return true
-		case <-stopChan:
-			c.Render(-1, common.CustomEvent{Data: "data: [DONE]"})
-			return false
-		}
-	})
-	err := resp.Body.Close()
 	if err != nil {
-		return openai.ErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), ""
+		logger.SysError(err.Error())
 	}
+
+	render.Done(c)
+
 	return nil, responseText
 }
 

relay/adaptor/proxy/adaptor.go

@@ -0,0 +1,89 @@
+package proxy
+
+import (
+	"fmt"
+	"io"
+	"net/http"
+	"strings"
+
+	"github.com/gin-gonic/gin"
+	"github.com/pkg/errors"
+	"github.com/songquanpeng/one-api/relay/adaptor"
+	channelhelper "github.com/songquanpeng/one-api/relay/adaptor"
+	"github.com/songquanpeng/one-api/relay/meta"
+	"github.com/songquanpeng/one-api/relay/model"
+	relaymodel "github.com/songquanpeng/one-api/relay/model"
+)
+
+var _ adaptor.Adaptor = new(Adaptor)
+
+const channelName = "proxy"
+
+type Adaptor struct{}
+
+func (a *Adaptor) Init(meta *meta.Meta) {
+}
+
+func (a *Adaptor) ConvertRequest(c *gin.Context, relayMode int, request *model.GeneralOpenAIRequest) (any, error) {
+	return nil, errors.New("notimplement")
+}
+
+func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, meta *meta.Meta) (usage *model.Usage, err *model.ErrorWithStatusCode) {
+	for k, v := range resp.Header {
+		for _, vv := range v {
+			c.Writer.Header().Set(k, vv)
+		}
+	}
+
+	c.Writer.WriteHeader(resp.StatusCode)
+	if _, gerr := io.Copy(c.Writer, resp.Body); gerr != nil {
+		return nil, &relaymodel.ErrorWithStatusCode{
+			StatusCode: http.StatusInternalServerError,
+			Error: relaymodel.Error{
+				Message: gerr.Error(),
+			},
+		}
+	}
+
+	return nil, nil
+}
+
+func (a *Adaptor) GetModelList() (models []string) {
+	return nil
+}
+
+func (a *Adaptor) GetChannelName() string {
+	return channelName
+}
+
+// GetRequestURL remove static prefix, and return the real request url to the upstream service
+func (a *Adaptor) GetRequestURL(meta *meta.Meta) (string, error) {
+	prefix := fmt.Sprintf("/v1/oneapi/proxy/%d", meta.ChannelId)
+	return meta.BaseURL + strings.TrimPrefix(meta.RequestURLPath, prefix), nil
+
+}
+
+func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, meta *meta.Meta) error {
+	for k, v := range c.Request.Header {
+		req.Header.Set(k, v[0])
+	}
+
+	// remove unnecessary headers
+	req.Header.Del("Host")
+	req.Header.Del("Content-Length")
+	req.Header.Del("Accept-Encoding")
+	req.Header.Del("Connection")
+
+	// set authorization header
+	req.Header.Set("Authorization", meta.APIKey)
+
+	return nil
+}
+
+func (a *Adaptor) ConvertImageRequest(request *model.ImageRequest) (any, error) {
+	return nil, errors.Errorf("not implement")
+}
+
+func (a *Adaptor) DoRequest(c *gin.Context, meta *meta.Meta, requestBody io.Reader) (*http.Response, error) {
+	return channelhelper.DoRequestHelper(a, c, meta, requestBody)
+}

relay/adaptor/siliconflow/constants.go

@@ -0,0 +1,36 @@
+package siliconflow
+
+// https://docs.siliconflow.cn/docs/getting-started
+
+var ModelList = []string{
+	"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",
+	"THUDM/glm-4-9b-chat",
+	"Qwen/Qwen2-72B-Instruct",
+	"Qwen/Qwen2-7B-Instruct",
+	"Qwen/Qwen2-57B-A14B-Instruct",
+	"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",
+}

relay/adaptor/tencent/adaptor.go

@@ -2,14 +2,15 @@ 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"
 )
 
@@ -17,20 +18,27 @@ import (
 
 type Adaptor struct {
 	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

@@ -3,11 +3,18 @@ package tencent
 import (
 	"bufio"
 	"crypto/hmac"
-	"crypto/sha1"
+	"crypto/sha256"
-	"encoding/base64"
+	"encoding/hex"
 	"encoding/json"
 	"errors"
 	"fmt"
+	"github.com/songquanpeng/one-api/common/render"
+	"io"
+	"net/http"
+	"strconv"
+	"strings"
+	"time"
+
 	"github.com/gin-gonic/gin"
 	"github.com/songquanpeng/one-api/common"
 	"github.com/songquanpeng/one-api/common/conv"
@@ -17,36 +24,23 @@ import (
 	"github.com/songquanpeng/one-api/relay/adaptor/openai"
 	"github.com/songquanpeng/one-api/relay/constant"
 	"github.com/songquanpeng/one-api/relay/model"
-	"io"
-	"net/http"
-	"sort"
-	"strconv"
-	"strings"
 )
 
-// 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 +48,11 @@ 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{
@@ -91,69 +89,52 @@ func streamResponseTencent2OpenAI(TencentResponse *ChatResponse) *openai.ChatCom
 func StreamHandler(c *gin.Context, resp *http.Response) (*model.ErrorWithStatusCode, string) {
 	var responseText string
 	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
+	common.SetEventStreamHeaders(c)
-		}
+
-		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() {
 	for scanner.Scan() {
 		data := scanner.Text()
-			if len(data) < 5 { // ignore blank line or wrong format
+		if len(data) < 5 || !strings.HasPrefix(data, "data:") {
 			continue
 		}
-			if data[:5] != "data:" {
+		data = strings.TrimPrefix(data, "data:")
-				continue
+
-			}
+		var tencentResponse ChatResponse
-			data = data[5:]
+		err := json.Unmarshal([]byte(data), &tencentResponse)
-			dataChan <- data
-		}
-		stopChan <- true
-	}()
-	common.SetEventStreamHeaders(c)
-	c.Stream(func(w io.Writer) bool {
-		select {
-		case data := <-dataChan:
-			var TencentResponse ChatResponse
-			err := json.Unmarshal([]byte(data), &TencentResponse)
 		if err != nil {
 			logger.SysError("error unmarshalling stream response: " + err.Error())
-				return true
+			continue
 		}
-			response := streamResponseTencent2OpenAI(&TencentResponse)
+
+		response := streamResponseTencent2OpenAI(&tencentResponse)
 		if len(response.Choices) != 0 {
 			responseText += conv.AsString(response.Choices[0].Delta.Content)
 		}
-			jsonResponse, err := json.Marshal(response)
+
+		err = render.ObjectData(c, response)
 		if err != nil {
-				logger.SysError("error marshalling stream response: " + err.Error())
+			logger.SysError(err.Error())
-				return true
 		}
-			c.Render(-1, common.CustomEvent{Data: "data: " + string(jsonResponse)})
-			return true
-		case <-stopChan:
-			c.Render(-1, common.CustomEvent{Data: "data: [DONE]"})
-			return false
 	}
-	})
+
+	if err := scanner.Err(); err != nil {
+		logger.SysError("error reading stream: " + err.Error())
+	}
+
+	render.Done(c)
+
 	err := resp.Body.Close()
 	if err != nil {
 		return openai.ErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), ""
 	}
+
 	return nil, responseText
 }
 
 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 +143,11 @@ 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 +184,62 @@ 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/vertexai/adaptor.go

@@ -0,0 +1,117 @@
+package vertexai
+
+import (
+	"errors"
+	"fmt"
+	"io"
+	"net/http"
+	"strings"
+
+	"github.com/gin-gonic/gin"
+	"github.com/songquanpeng/one-api/relay/adaptor"
+	channelhelper "github.com/songquanpeng/one-api/relay/adaptor"
+	"github.com/songquanpeng/one-api/relay/meta"
+	"github.com/songquanpeng/one-api/relay/model"
+	relaymodel "github.com/songquanpeng/one-api/relay/model"
+)
+
+var _ adaptor.Adaptor = new(Adaptor)
+
+const channelName = "vertexai"
+
+type Adaptor struct{}
+
+func (a *Adaptor) Init(meta *meta.Meta) {
+}
+
+func (a *Adaptor) ConvertRequest(c *gin.Context, relayMode int, request *model.GeneralOpenAIRequest) (any, error) {
+	if request == nil {
+		return nil, errors.New("request is nil")
+	}
+
+	adaptor := GetAdaptor(request.Model)
+	if adaptor == nil {
+		return nil, errors.New("adaptor not found")
+	}
+
+	return adaptor.ConvertRequest(c, relayMode, request)
+}
+
+func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, meta *meta.Meta) (usage *model.Usage, err *model.ErrorWithStatusCode) {
+	adaptor := GetAdaptor(meta.ActualModelName)
+	if adaptor == nil {
+		return nil, &relaymodel.ErrorWithStatusCode{
+			StatusCode: http.StatusInternalServerError,
+			Error: relaymodel.Error{
+				Message: "adaptor not found",
+			},
+		}
+	}
+	return adaptor.DoResponse(c, resp, meta)
+}
+
+func (a *Adaptor) GetModelList() (models []string) {
+	models = modelList
+	return
+}
+
+func (a *Adaptor) GetChannelName() string {
+	return channelName
+}
+
+func (a *Adaptor) GetRequestURL(meta *meta.Meta) (string, error) {
+	suffix := ""
+	if strings.HasPrefix(meta.ActualModelName, "gemini") {
+		if meta.IsStream {
+			suffix = "streamGenerateContent?alt=sse"
+		} else {
+			suffix = "generateContent"
+		}
+	} else {
+		if meta.IsStream {
+			suffix = "streamRawPredict?alt=sse"
+		} else {
+			suffix = "rawPredict"
+		}
+	}
+
+	if meta.BaseURL != "" {
+		return fmt.Sprintf(
+			"%s/v1/projects/%s/locations/%s/publishers/google/models/%s:%s",
+			meta.BaseURL,
+			meta.Config.VertexAIProjectID,
+			meta.Config.Region,
+			meta.ActualModelName,
+			suffix,
+		), nil
+	}
+	return fmt.Sprintf(
+		"https://%s-aiplatform.googleapis.com/v1/projects/%s/locations/%s/publishers/google/models/%s:%s",
+		meta.Config.Region,
+		meta.Config.VertexAIProjectID,
+		meta.Config.Region,
+		meta.ActualModelName,
+		suffix,
+	), nil
+}
+
+func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, meta *meta.Meta) error {
+	adaptor.SetupCommonRequestHeader(c, req, meta)
+	token, err := getToken(c, meta.ChannelId, meta.Config.VertexAIADC)
+	if err != nil {
+		return err
+	}
+	req.Header.Set("Authorization", "Bearer "+token)
+	return nil
+}
+
+func (a *Adaptor) ConvertImageRequest(request *model.ImageRequest) (any, error) {
+	if request == nil {
+		return nil, errors.New("request is nil")
+	}
+	return request, nil
+}
+
+func (a *Adaptor) DoRequest(c *gin.Context, meta *meta.Meta, requestBody io.Reader) (*http.Response, error) {
+	return channelhelper.DoRequestHelper(a, c, meta, requestBody)
+}

relay/adaptor/vertexai/claude/adapter.go

@@ -0,0 +1,55 @@
+package vertexai
+
+import (
+	"net/http"
+
+	"github.com/gin-gonic/gin"
+	"github.com/pkg/errors"
+	"github.com/songquanpeng/one-api/common/ctxkey"
+	"github.com/songquanpeng/one-api/relay/adaptor/anthropic"
+
+	"github.com/songquanpeng/one-api/relay/meta"
+	"github.com/songquanpeng/one-api/relay/model"
+)
+
+var ModelList = []string{
+	"claude-3-haiku@20240307", "claude-3-opus@20240229", "claude-3-5-sonnet@20240620", "claude-3-sonnet@20240229",
+}
+
+const anthropicVersion = "vertex-2023-10-16"
+
+type Adaptor struct {
+}
+
+func (a *Adaptor) ConvertRequest(c *gin.Context, relayMode int, request *model.GeneralOpenAIRequest) (any, error) {
+	if request == nil {
+		return nil, errors.New("request is nil")
+	}
+
+	claudeReq := anthropic.ConvertRequest(*request)
+	req := Request{
+		AnthropicVersion: anthropicVersion,
+		// Model:            claudeReq.Model,
+		Messages:    claudeReq.Messages,
+		System:      claudeReq.System,
+		MaxTokens:   claudeReq.MaxTokens,
+		Temperature: claudeReq.Temperature,
+		TopP:        claudeReq.TopP,
+		TopK:        claudeReq.TopK,
+		Stream:      claudeReq.Stream,
+		Tools:       claudeReq.Tools,
+	}
+
+	c.Set(ctxkey.RequestModel, request.Model)
+	c.Set(ctxkey.ConvertedRequest, req)
+	return req, nil
+}
+
+func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, meta *meta.Meta) (usage *model.Usage, err *model.ErrorWithStatusCode) {
+	if meta.IsStream {
+		err, usage = anthropic.StreamHandler(c, resp)
+	} else {
+		err, usage = anthropic.Handler(c, resp, meta.PromptTokens, meta.ActualModelName)
+	}
+	return
+}

relay/adaptor/vertexai/claude/model.go

@@ -0,0 +1,19 @@
+package vertexai
+
+import "github.com/songquanpeng/one-api/relay/adaptor/anthropic"
+
+type Request struct {
+	// AnthropicVersion must be "vertex-2023-10-16"
+	AnthropicVersion string `json:"anthropic_version"`
+	// Model            string              `json:"model"`
+	Messages      []anthropic.Message `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         []anthropic.Tool    `json:"tools,omitempty"`
+	ToolChoice    any                 `json:"tool_choice,omitempty"`
+}

relay/adaptor/vertexai/gemini/adapter.go

@@ -0,0 +1,49 @@
+package vertexai
+
+import (
+	"net/http"
+
+	"github.com/gin-gonic/gin"
+	"github.com/pkg/errors"
+	"github.com/songquanpeng/one-api/common/ctxkey"
+	"github.com/songquanpeng/one-api/relay/adaptor/gemini"
+	"github.com/songquanpeng/one-api/relay/adaptor/openai"
+	"github.com/songquanpeng/one-api/relay/relaymode"
+
+	"github.com/songquanpeng/one-api/relay/meta"
+	"github.com/songquanpeng/one-api/relay/model"
+)
+
+var ModelList = []string{
+	"gemini-1.5-pro-001", "gemini-1.5-flash-001", "gemini-pro", "gemini-pro-vision",
+}
+
+type Adaptor struct {
+}
+
+func (a *Adaptor) ConvertRequest(c *gin.Context, relayMode int, request *model.GeneralOpenAIRequest) (any, error) {
+	if request == nil {
+		return nil, errors.New("request is nil")
+	}
+
+	geminiRequest := gemini.ConvertRequest(*request)
+	c.Set(ctxkey.RequestModel, request.Model)
+	c.Set(ctxkey.ConvertedRequest, geminiRequest)
+	return geminiRequest, nil
+}
+
+func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, meta *meta.Meta) (usage *model.Usage, err *model.ErrorWithStatusCode) {
+	if meta.IsStream {
+		var responseText string
+		err, responseText = gemini.StreamHandler(c, resp)
+		usage = openai.ResponseText2Usage(responseText, meta.ActualModelName, meta.PromptTokens)
+	} else {
+		switch meta.Mode {
+		case relaymode.Embeddings:
+			err, usage = gemini.EmbeddingHandler(c, resp)
+		default:
+			err, usage = gemini.Handler(c, resp, meta.PromptTokens, meta.ActualModelName)
+		}
+	}
+	return
+}

relay/adaptor/vertexai/registry.go

@@ -0,0 +1,50 @@
+package vertexai
+
+import (
+	"net/http"
+
+	"github.com/gin-gonic/gin"
+	claude "github.com/songquanpeng/one-api/relay/adaptor/vertexai/claude"
+	gemini "github.com/songquanpeng/one-api/relay/adaptor/vertexai/gemini"
+	"github.com/songquanpeng/one-api/relay/meta"
+	"github.com/songquanpeng/one-api/relay/model"
+)
+
+type VertexAIModelType int
+
+const (
+	VerterAIClaude VertexAIModelType = iota + 1
+	VerterAIGemini
+)
+
+var modelMapping = map[string]VertexAIModelType{}
+var modelList = []string{}
+
+func init() {
+	modelList = append(modelList, claude.ModelList...)
+	for _, model := range claude.ModelList {
+		modelMapping[model] = VerterAIClaude
+	}
+
+	modelList = append(modelList, gemini.ModelList...)
+	for _, model := range gemini.ModelList {
+		modelMapping[model] = VerterAIGemini
+	}
+}
+
+type innerAIAdapter interface {
+	ConvertRequest(c *gin.Context, relayMode int, request *model.GeneralOpenAIRequest) (any, error)
+	DoResponse(c *gin.Context, resp *http.Response, meta *meta.Meta) (usage *model.Usage, err *model.ErrorWithStatusCode)
+}
+
+func GetAdaptor(model string) innerAIAdapter {
+	adaptorType := modelMapping[model]
+	switch adaptorType {
+	case VerterAIClaude:
+		return &claude.Adaptor{}
+	case VerterAIGemini:
+		return &gemini.Adaptor{}
+	default:
+		return nil
+	}
+}

relay/adaptor/vertexai/token.go

@@ -0,0 +1,62 @@
+package vertexai
+
+import (
+	"context"
+	"encoding/json"
+	"fmt"
+	"time"
+
+	credentials "cloud.google.com/go/iam/credentials/apiv1"
+	"cloud.google.com/go/iam/credentials/apiv1/credentialspb"
+	"github.com/patrickmn/go-cache"
+	"google.golang.org/api/option"
+)
+
+type ApplicationDefaultCredentials struct {
+	Type                    string `json:"type"`
+	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"`
+	AuthURI                 string `json:"auth_uri"`
+	TokenURI                string `json:"token_uri"`
+	AuthProviderX509CertURL string `json:"auth_provider_x509_cert_url"`
+	ClientX509CertURL       string `json:"client_x509_cert_url"`
+	UniverseDomain          string `json:"universe_domain"`
+}
+
+var Cache = cache.New(50*time.Minute, 55*time.Minute)
+
+const defaultScope = "https://www.googleapis.com/auth/cloud-platform"
+
+func getToken(ctx context.Context, channelId int, adcJson string) (string, error) {
+	cacheKey := fmt.Sprintf("vertexai-token-%d", channelId)
+	if token, found := Cache.Get(cacheKey); found {
+		return token.(string), nil
+	}
+	adc := &ApplicationDefaultCredentials{}
+	if err := json.Unmarshal([]byte(adcJson), adc); err != nil {
+		return "", fmt.Errorf("Failed to decode credentials file: %w", err)
+	}
+
+	c, err := credentials.NewIamCredentialsClient(ctx, option.WithCredentialsJSON([]byte(adcJson)))
+	if err != nil {
+		return "", fmt.Errorf("Failed to create client: %w", err)
+	}
+	defer c.Close()
+
+	req := &credentialspb.GenerateAccessTokenRequest{
+		// See https://pkg.go.dev/cloud.google.com/go/iam/credentials/apiv1/credentialspb#GenerateAccessTokenRequest.
+		Name:  fmt.Sprintf("projects/-/serviceAccounts/%s", adc.ClientEmail),
+		Scope: []string{defaultScope},
+	}
+	resp, err := c.GenerateAccessToken(ctx, req)
+	if err != nil {
+		return "", fmt.Errorf("Failed to generate access token: %w", err)
+	}
+	_ = resp
+
+	Cache.Set(cacheKey, resp.AccessToken, cache.DefaultExpiration)
+	return resp.AccessToken, nil
+}

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/constants.go

@@ -6,4 +6,5 @@ var ModelList = []string{
 	"SparkDesk-v2.1",
 	"SparkDesk-v3.1",
 	"SparkDesk-v3.5",
+	"SparkDesk-v4.0",
 }

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 +43,14 @@ 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") || domain == "4.0Ultra" {
-			xunfeiRequest.Payload.Functions.Text = append(xunfeiRequest.Payload.Functions.Text, toolCall.Function)
+		functions := make([]model.Function, len(request.Tools))
+		for i, tool := range request.Tools {
+			functions[i] = tool.Function
+		}
+		xunfeiRequest.Payload.Functions = &Functions{
+			Text: functions,
 		}
 	}
 
@@ -203,7 +206,7 @@ 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
 
@@ -287,6 +290,8 @@ func apiVersion2domain(apiVersion string) string {
 		return "generalv3"
 	case "v3.5":
 		return "generalv3.5"
+	case "v4.0":
+		return "4.0Ultra"
 	}
 	return "general" + apiVersion
 }

relay/adaptor/xunfei/model.go

@@ -9,6 +9,10 @@ type Message struct {
 	Content string `json:"content"`
 }
 
+type Functions struct {
+	Text []model.Function `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/adaptor/zhipu/main.go

@@ -3,6 +3,13 @@ package zhipu
 import (
 	"bufio"
 	"encoding/json"
+	"github.com/songquanpeng/one-api/common/render"
+	"io"
+	"net/http"
+	"strings"
+	"sync"
+	"time"
+
 	"github.com/gin-gonic/gin"
 	"github.com/golang-jwt/jwt"
 	"github.com/songquanpeng/one-api/common"
@@ -11,11 +18,6 @@ import (
 	"github.com/songquanpeng/one-api/relay/adaptor/openai"
 	"github.com/songquanpeng/one-api/relay/constant"
 	"github.com/songquanpeng/one-api/relay/model"
-	"io"
-	"net/http"
-	"strings"
-	"sync"
-	"time"
 )
 
 // https://open.bigmodel.cn/doc/api#chatglm_std
@@ -155,10 +157,9 @@ func StreamHandler(c *gin.Context, resp *http.Response) (*model.ErrorWithStatusC
 		}
 		return 0, nil, nil
 	})
-	dataChan := make(chan string)
+
-	metaChan := make(chan string)
+	common.SetEventStreamHeaders(c)
-	stopChan := make(chan bool)
+
-	go func() {
 	for scanner.Scan() {
 		data := scanner.Text()
 		lines := strings.Split(data, "\n")
@@ -166,55 +167,45 @@ func StreamHandler(c *gin.Context, resp *http.Response) (*model.ErrorWithStatusC
 			if len(line) < 5 {
 				continue
 			}
-				if line[:5] == "data:" {
+			if strings.HasPrefix(line, "data:") {
-					dataChan <- line[5:]
+				dataSegment := line[5:]
 				if i != len(lines)-1 {
-						dataChan <- "\n"
+					dataSegment += "\n"
 				}
-				} else if line[:5] == "meta:" {
+				response := streamResponseZhipu2OpenAI(dataSegment)
-					metaChan <- line[5:]
+				err := render.ObjectData(c, response)
-				}
-			}
-		}
-		stopChan <- true
-	}()
-	common.SetEventStreamHeaders(c)
-	c.Stream(func(w io.Writer) bool {
-		select {
-		case data := <-dataChan:
-			response := streamResponseZhipu2OpenAI(data)
-			jsonResponse, err := json.Marshal(response)
 				if err != nil {
 					logger.SysError("error marshalling stream response: " + err.Error())
-				return true
 				}
-			c.Render(-1, common.CustomEvent{Data: "data: " + string(jsonResponse)})
+			} else if strings.HasPrefix(line, "meta:") {
-			return true
+				metaSegment := line[5:]
-		case data := <-metaChan:
 				var zhipuResponse StreamMetaResponse
-			err := json.Unmarshal([]byte(data), &zhipuResponse)
+				err := json.Unmarshal([]byte(metaSegment), &zhipuResponse)
 				if err != nil {
 					logger.SysError("error unmarshalling stream response: " + err.Error())
-				return true
+					continue
 				}
 				response, zhipuUsage := streamMetaResponseZhipu2OpenAI(&zhipuResponse)
-			jsonResponse, err := json.Marshal(response)
+				err = render.ObjectData(c, response)
 				if err != nil {
 					logger.SysError("error marshalling stream response: " + err.Error())
-				return true
 				}
 				usage = zhipuUsage
-			c.Render(-1, common.CustomEvent{Data: "data: " + string(jsonResponse)})
-			return true
-		case <-stopChan:
-			c.Render(-1, common.CustomEvent{Data: "data: [DONE]"})
-			return false
 			}
-	})
+		}
+	}
+
+	if err := scanner.Err(); err != nil {
+		logger.SysError("error reading stream: " + err.Error())
+	}
+
+	render.Done(c)
+
 	err := resp.Body.Close()
 	if err != nil {
 		return openai.ErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), nil
 	}
+
 	return nil, usage
 }
 

relay/adaptor_test.go

@@ -0,0 +1,16 @@
+package relay
+
+import (
+	. "github.com/smartystreets/goconvey/convey"
+	"github.com/songquanpeng/one-api/relay/apitype"
+	"testing"
+)
+
+func TestGetAdaptor(t *testing.T) {
+	Convey("get adaptor", t, func() {
+		for i := 0; i < apitype.Dummy; i++ {
+			a := GetAdaptor(i)
+			So(a, ShouldNotBeNil)
+		}
+	})
+}

relay/apitype/define.go

@@ -17,6 +17,8 @@ const (
 	Cohere
 	Cloudflare
 	DeepL
+	VertexAI
+	Proxy
 
 	Dummy // this one is only for count, do not add any channel after this
 )

relay/billing/ratio/model.go

@@ -2,6 +2,7 @@ package ratio
 
 import (
 	"encoding/json"
+	"fmt"
 	"strings"
 
 	"github.com/songquanpeng/one-api/common/logger"
@@ -34,6 +35,8 @@ var ModelRatio = map[string]float64{
 	"gpt-4-turbo-2024-04-09":  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-mini":             0.075, // $0.00015 / 1K tokens
+	"gpt-4o-mini-2024-07-18":  0.075, // $0.00015 / 1K tokens
 	"gpt-4-vision-preview":    5,     // $0.01 / 1K tokens
 	"gpt-3.5-turbo":           0.25,  // $0.0005 / 1K tokens
 	"gpt-3.5-turbo-0301":      0.75,
@@ -75,6 +78,7 @@ var ModelRatio = map[string]float64{
 	"claude-2.1":                 8.0 / 1000 * USD,
 	"claude-3-haiku-20240307":    0.25 / 1000 * USD,
 	"claude-3-sonnet-20240229":   3.0 / 1000 * USD,
+	"claude-3-5-sonnet-20240620": 3.0 / 1000 * USD,
 	"claude-3-opus-20240229":     15.0 / 1000 * USD,
 	// https://cloud.baidu.com/doc/WENXINWORKSHOP/s/hlrk4akp7
 	"ERNIE-4.0-8K":       0.120 * RMB,
@@ -94,12 +98,11 @@ var ModelRatio = map[string]float64{
 	"bge-large-en":       0.002 * RMB,
 	"tao-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
+	"gemini-1.0-pro":   1,
-	"gemini-1.0-pro-vision-001": 1,
+	"gemini-1.5-flash": 1,
-	"gemini-1.0-pro-001":        1,
 	"gemini-1.5-pro":   1,
+	"aqa":              1,
 	// https://open.bigmodel.cn/pricing
 	"glm-4":         0.1 * RMB,
 	"glm-4v":        0.1 * RMB,
@@ -124,6 +127,7 @@ var ModelRatio = map[string]float64{
 	"SparkDesk-v2.1":            1.2858, // ￥0.018 / 1k tokens
 	"SparkDesk-v3.1":            1.2858, // ￥0.018 / 1k tokens
 	"SparkDesk-v3.5":            1.2858, // ￥0.018 / 1k tokens
+	"SparkDesk-v4.0":            1.2858, // ￥0.018 / 1k tokens
 	"360GPT_S2_V9":              0.8572, // ¥0.012 / 1k tokens
 	"embedding-bert-512-v1":     0.0715, // ¥0.001 / 1k tokens
 	"embedding_s1_v1":           0.0715, // ¥0.001 / 1k tokens
@@ -153,12 +157,16 @@ var ModelRatio = map[string]float64{
 	"mistral-large-latest":  8.0 / 1000 * USD,
 	"mistral-embed":         0.1 / 1000 * USD,
 	// https://wow.groq.com/#:~:text=inquiries%C2%A0here.-,Model,-Current%20Speed
-	"llama3-70b-8192":    0.59 / 1000 * USD,
+	"gemma-7b-it":                           0.07 / 1000000 * USD,
-	"mixtral-8x7b-32768": 0.27 / 1000 * USD,
+	"mixtral-8x7b-32768":                    0.24 / 1000000 * USD,
-	"llama3-8b-8192":     0.05 / 1000 * USD,
+	"llama3-8b-8192":                        0.05 / 1000000 * USD,
-	"gemma-7b-it":        0.1 / 1000 * USD,
+	"llama3-70b-8192":                       0.59 / 1000000 * USD,
-	"llama2-70b-4096":    0.64 / 1000 * USD,
+	"gemma2-9b-it":                          0.20 / 1000000 * USD,
-	"llama2-7b-2048":     0.1 / 1000 * USD,
+	"llama-3.1-405b-reasoning":              0.89 / 1000000 * USD,
+	"llama-3.1-70b-versatile":               0.59 / 1000000 * USD,
+	"llama-3.1-8b-instant":                  0.05 / 1000000 * USD,
+	"llama3-groq-70b-8192-tool-use-preview": 0.89 / 1000000 * USD,
+	"llama3-groq-8b-8192-tool-use-preview":  0.19 / 1000000 * USD,
 	// https://platform.lingyiwanwu.com/docs#-计费单元
 	"yi-34b-chat-0205": 2.5 / 1000 * RMB,
 	"yi-34b-chat-200k": 12.0 / 1000 * RMB,
@@ -167,6 +175,9 @@ var ModelRatio = map[string]float64{
 	"step-1v-32k": 0.024 * RMB,
 	"step-1-32k":  0.024 * RMB,
 	"step-1-200k": 0.15 * RMB,
+	// aws llama3 https://aws.amazon.com/cn/bedrock/pricing/
+	"llama3-8b-8192(33)":  0.0003 / 0.002,  // $0.0003 / 1K tokens
+	"llama3-70b-8192(33)": 0.00265 / 0.002, // $0.00265 / 1K tokens
 	// https://cohere.com/pricing
 	"command":               0.5,
 	"command-nightly":       0.5,
@@ -183,7 +194,11 @@ var ModelRatio = map[string]float64{
 	"deepl-ja": 25.0 / 1000 * USD,
 }
 
-var CompletionRatio = map[string]float64{}
+var CompletionRatio = map[string]float64{
+	// aws llama3
+	"llama3-8b-8192(33)":  0.0006 / 0.0003,
+	"llama3-70b-8192(33)": 0.0035 / 0.00265,
+}
 
 var DefaultModelRatio map[string]float64
 var DefaultCompletionRatio map[string]float64
@@ -232,22 +247,28 @@ func UpdateModelRatioByJSONString(jsonStr string) error {
 	return json.Unmarshal([]byte(jsonStr), &ModelRatio)
 }
 
-func GetModelRatio(name string) float64 {
+func GetModelRatio(name string, channelType int) float64 {
 	if strings.HasPrefix(name, "qwen-") && strings.HasSuffix(name, "-internet") {
 		name = strings.TrimSuffix(name, "-internet")
 	}
 	if strings.HasPrefix(name, "command-") && strings.HasSuffix(name, "-internet") {
 		name = strings.TrimSuffix(name, "-internet")
 	}
-	ratio, ok := ModelRatio[name]
+	model := fmt.Sprintf("%s(%d)", name, channelType)
-	if !ok {
+	if ratio, ok := ModelRatio[model]; ok {
-		ratio, ok = DefaultModelRatio[name]
+		return ratio
+	}
+	if ratio, ok := DefaultModelRatio[model]; ok {
+		return ratio
+	}
+	if ratio, ok := ModelRatio[name]; ok {
+		return ratio
+	}
+	if ratio, ok := DefaultModelRatio[name]; ok {
+		return ratio
 	}
-	if !ok {
 	logger.SysError("model ratio not found: " + name)
 	return 30
-	}
-	return ratio
 }
 
 func CompletionRatio2JSONString() string {
@@ -263,7 +284,17 @@ func UpdateCompletionRatioByJSONString(jsonStr string) error {
 	return json.Unmarshal([]byte(jsonStr), &CompletionRatio)
 }
 
-func GetCompletionRatio(name string) float64 {
+func GetCompletionRatio(name string, channelType int) float64 {
+	if strings.HasPrefix(name, "qwen-") && strings.HasSuffix(name, "-internet") {
+		name = strings.TrimSuffix(name, "-internet")
+	}
+	model := fmt.Sprintf("%s(%d)", name, channelType)
+	if ratio, ok := CompletionRatio[model]; ok {
+		return ratio
+	}
+	if ratio, ok := DefaultCompletionRatio[model]; ok {
+		return ratio
+	}
 	if ratio, ok := CompletionRatio[name]; ok {
 		return ratio
 	}
@@ -282,6 +313,9 @@ func GetCompletionRatio(name string) float64 {
 		return 4.0 / 3.0
 	}
 	if strings.HasPrefix(name, "gpt-4") {
+		if strings.HasPrefix(name, "gpt-4o-mini") {
+			return 4
+		}
 		if strings.HasPrefix(name, "gpt-4-turbo") ||
 			strings.HasPrefix(name, "gpt-4o") ||
 			strings.HasSuffix(name, "preview") {

relay/channeltype/define.go

@@ -42,5 +42,9 @@ const (
 	DeepL
 	TogetherAI
 	Doubao
+	Novita
+	VertextAI
+	Proxy
+	SiliconFlow
 	Dummy
 )

relay/channeltype/helper.go

@@ -35,6 +35,10 @@ func ToAPIType(channelType int) int {
 		apiType = apitype.Cloudflare
 	case DeepL:
 		apiType = apitype.DeepL
+	case VertextAI:
+		apiType = apitype.VertexAI
+	case Proxy:
+		apiType = apitype.Proxy
 	}
 
 	return apiType

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
@@ -42,6 +42,10 @@ var ChannelBaseURLs = []string{
 	"https://api-free.deepl.com",                // 38
 	"https://api.together.xyz",                  // 39
 	"https://ark.cn-beijing.volces.com",         // 40
+	"https://api.novita.ai/v3/openai",           // 41
+	"",                                          // 42
+	"",                                          // 43
+	"https://api.siliconflow.cn",                 // 44
 }
 
 func init() {

relay/channeltype/url_test.go

@@ -0,0 +1,12 @@
+package channeltype
+
+import (
+	. "github.com/smartystreets/goconvey/convey"
+	"testing"
+)
+
+func TestChannelBaseURLs(t *testing.T) {
+	Convey("channel base urls", t, func() {
+		So(len(ChannelBaseURLs), ShouldEqual, Dummy)
+	})
+}

relay/controller/audio.go

@@ -7,6 +7,10 @@ import (
 	"encoding/json"
 	"errors"
 	"fmt"
+	"io"
+	"net/http"
+	"strings"
+
 	"github.com/gin-gonic/gin"
 	"github.com/songquanpeng/one-api/common"
 	"github.com/songquanpeng/one-api/common/client"
@@ -21,9 +25,6 @@ import (
 	"github.com/songquanpeng/one-api/relay/meta"
 	relaymodel "github.com/songquanpeng/one-api/relay/model"
 	"github.com/songquanpeng/one-api/relay/relaymode"
-	"io"
-	"net/http"
-	"strings"
 )
 
 func RelayAudioHelper(c *gin.Context, relayMode int) *relaymodel.ErrorWithStatusCode {
@@ -53,7 +54,7 @@ func RelayAudioHelper(c *gin.Context, relayMode int) *relaymodel.ErrorWithStatus
 		}
 	}
 
-	modelRatio := billingratio.GetModelRatio(audioModel)
+	modelRatio := billingratio.GetModelRatio(audioModel, channelType)
 	groupRatio := billingratio.GetGroupRatio(group)
 	ratio := modelRatio * groupRatio
 	var quota int64

relay/controller/helper.go

@@ -4,6 +4,10 @@ import (
 	"context"
 	"errors"
 	"fmt"
+	"math"
+	"net/http"
+	"strings"
+
 	"github.com/gin-gonic/gin"
 	"github.com/songquanpeng/one-api/common"
 	"github.com/songquanpeng/one-api/common/config"
@@ -16,9 +20,6 @@ import (
 	"github.com/songquanpeng/one-api/relay/meta"
 	relaymodel "github.com/songquanpeng/one-api/relay/model"
 	"github.com/songquanpeng/one-api/relay/relaymode"
-	"math"
-	"net/http"
-	"strings"
 )
 
 func getAndValidateTextRequest(c *gin.Context, relayMode int) (*relaymodel.GeneralOpenAIRequest, error) {
@@ -40,78 +41,6 @@ func getAndValidateTextRequest(c *gin.Context, relayMode int) (*relaymodel.Gener
 	return textRequest, nil
 }
 
-func getImageRequest(c *gin.Context, relayMode int) (*relaymodel.ImageRequest, error) {
-	imageRequest := &relaymodel.ImageRequest{}
-	err := common.UnmarshalBodyReusable(c, imageRequest)
-	if err != nil {
-		return nil, err
-	}
-	if imageRequest.N == 0 {
-		imageRequest.N = 1
-	}
-	if imageRequest.Size == "" {
-		imageRequest.Size = "1024x1024"
-	}
-	if imageRequest.Model == "" {
-		imageRequest.Model = "dall-e-2"
-	}
-	return imageRequest, nil
-}
-
-func isValidImageSize(model string, size string) bool {
-	if model == "cogview-3" {
-		return true
-	}
-	_, ok := billingratio.ImageSizeRatios[model][size]
-	return ok
-}
-
-func getImageSizeRatio(model string, size string) float64 {
-	ratio, ok := billingratio.ImageSizeRatios[model][size]
-	if !ok {
-		return 1
-	}
-	return ratio
-}
-
-func validateImageRequest(imageRequest *relaymodel.ImageRequest, meta *meta.Meta) *relaymodel.ErrorWithStatusCode {
-	// model validation
-	hasValidSize := isValidImageSize(imageRequest.Model, imageRequest.Size)
-	if !hasValidSize {
-		return openai.ErrorWrapper(errors.New("size not supported for this image model"), "size_not_supported", http.StatusBadRequest)
-	}
-	// check prompt length
-	if imageRequest.Prompt == "" {
-		return openai.ErrorWrapper(errors.New("prompt is required"), "prompt_missing", http.StatusBadRequest)
-	}
-	if len(imageRequest.Prompt) > billingratio.ImagePromptLengthLimitations[imageRequest.Model] {
-		return openai.ErrorWrapper(errors.New("prompt is too long"), "prompt_too_long", http.StatusBadRequest)
-	}
-	// Number of generated images validation
-	if !isWithinRange(imageRequest.Model, imageRequest.N) {
-		// channel not azure
-		if meta.ChannelType != channeltype.Azure {
-			return openai.ErrorWrapper(errors.New("invalid value of n"), "n_not_within_range", http.StatusBadRequest)
-		}
-	}
-	return nil
-}
-
-func getImageCostRatio(imageRequest *relaymodel.ImageRequest) (float64, error) {
-	if imageRequest == nil {
-		return 0, errors.New("imageRequest is nil")
-	}
-	imageCostRatio := getImageSizeRatio(imageRequest.Model, imageRequest.Size)
-	if imageRequest.Quality == "hd" && imageRequest.Model == "dall-e-3" {
-		if imageRequest.Size == "1024x1024" {
-			imageCostRatio *= 2
-		} else {
-			imageCostRatio *= 1.5
-		}
-	}
-	return imageCostRatio, nil
-}
-
 func getPromptTokens(textRequest *relaymodel.GeneralOpenAIRequest, relayMode int) int {
 	switch relayMode {
 	case relaymode.ChatCompletions:
@@ -167,7 +96,7 @@ func postConsumeQuota(ctx context.Context, usage *relaymodel.Usage, meta *meta.M
 		return
 	}
 	var quota int64
-	completionRatio := billingratio.GetCompletionRatio(textRequest.Model)
+	completionRatio := billingratio.GetCompletionRatio(textRequest.Model, meta.ChannelType)
 	promptTokens := usage.PromptTokens
 	completionTokens := usage.CompletionTokens
 	quota = int64(math.Ceil((float64(promptTokens) + float64(completionTokens)*completionRatio) * ratio))

relay/controller/image.go

@@ -6,7 +6,11 @@ import (
 	"encoding/json"
 	"errors"
 	"fmt"
+	"io"
+	"net/http"
+
 	"github.com/gin-gonic/gin"
+	"github.com/songquanpeng/one-api/common"
 	"github.com/songquanpeng/one-api/common/ctxkey"
 	"github.com/songquanpeng/one-api/common/logger"
 	"github.com/songquanpeng/one-api/model"
@@ -16,17 +20,86 @@ import (
 	"github.com/songquanpeng/one-api/relay/channeltype"
 	"github.com/songquanpeng/one-api/relay/meta"
 	relaymodel "github.com/songquanpeng/one-api/relay/model"
-	"io"
-	"net/http"
 )
 
-func isWithinRange(element string, value int) bool {
+func getImageRequest(c *gin.Context, relayMode int) (*relaymodel.ImageRequest, error) {
-	if _, ok := billingratio.ImageGenerationAmounts[element]; !ok {
+	imageRequest := &relaymodel.ImageRequest{}
-		return false
+	err := common.UnmarshalBodyReusable(c, imageRequest)
+	if err != nil {
+		return nil, err
 	}
-	min := billingratio.ImageGenerationAmounts[element][0]
+	if imageRequest.N == 0 {
-	max := billingratio.ImageGenerationAmounts[element][1]
+		imageRequest.N = 1
-	return value >= min && value <= max
+	}
+	if imageRequest.Size == "" {
+		imageRequest.Size = "1024x1024"
+	}
+	if imageRequest.Model == "" {
+		imageRequest.Model = "dall-e-2"
+	}
+	return imageRequest, nil
+}
+
+func isValidImageSize(model string, size string) bool {
+	if model == "cogview-3" || billingratio.ImageSizeRatios[model] == nil {
+		return true
+	}
+	_, ok := billingratio.ImageSizeRatios[model][size]
+	return ok
+}
+
+func isValidImagePromptLength(model string, promptLength int) bool {
+	maxPromptLength, ok := billingratio.ImagePromptLengthLimitations[model]
+	return !ok || promptLength <= maxPromptLength
+}
+
+func isWithinRange(element string, value int) bool {
+	amounts, ok := billingratio.ImageGenerationAmounts[element]
+	return !ok || (value >= amounts[0] && value <= amounts[1])
+}
+
+func getImageSizeRatio(model string, size string) float64 {
+	if ratio, ok := billingratio.ImageSizeRatios[model][size]; ok {
+		return ratio
+	}
+	return 1
+}
+
+func validateImageRequest(imageRequest *relaymodel.ImageRequest, meta *meta.Meta) *relaymodel.ErrorWithStatusCode {
+	// check prompt length
+	if imageRequest.Prompt == "" {
+		return openai.ErrorWrapper(errors.New("prompt is required"), "prompt_missing", http.StatusBadRequest)
+	}
+
+	// model validation
+	if !isValidImageSize(imageRequest.Model, imageRequest.Size) {
+		return openai.ErrorWrapper(errors.New("size not supported for this image model"), "size_not_supported", http.StatusBadRequest)
+	}
+
+	if !isValidImagePromptLength(imageRequest.Model, len(imageRequest.Prompt)) {
+		return openai.ErrorWrapper(errors.New("prompt is too long"), "prompt_too_long", http.StatusBadRequest)
+	}
+
+	// Number of generated images validation
+	if !isWithinRange(imageRequest.Model, imageRequest.N) {
+		return openai.ErrorWrapper(errors.New("invalid value of n"), "n_not_within_range", http.StatusBadRequest)
+	}
+	return nil
+}
+
+func getImageCostRatio(imageRequest *relaymodel.ImageRequest) (float64, error) {
+	if imageRequest == nil {
+		return 0, errors.New("imageRequest is nil")
+	}
+	imageCostRatio := getImageSizeRatio(imageRequest.Model, imageRequest.Size)
+	if imageRequest.Quality == "hd" && imageRequest.Model == "dall-e-3" {
+		if imageRequest.Size == "1024x1024" {
+			imageCostRatio *= 2
+		} else {
+			imageCostRatio *= 1.5
+		}
+	}
+	return imageCostRatio, nil
 }
 
 func RelayImageHelper(c *gin.Context, relayMode int) *relaymodel.ErrorWithStatusCode {
@@ -95,7 +168,7 @@ func RelayImageHelper(c *gin.Context, relayMode int) *relaymodel.ErrorWithStatus
 		requestBody = bytes.NewBuffer(jsonStr)
 	}
 
-	modelRatio := billingratio.GetModelRatio(imageModel)
+	modelRatio := billingratio.GetModelRatio(imageModel, meta.ChannelType)
 	groupRatio := billingratio.GetGroupRatio(meta.Group)
 	ratio := modelRatio * groupRatio
 	userQuota, err := model.CacheGetUserQuota(ctx, meta.UserId)

relay/controller/proxy.go

@@ -0,0 +1,41 @@
+// Package controller is a package for handling the relay controller
+package controller
+
+import (
+	"fmt"
+	"net/http"
+
+	"github.com/gin-gonic/gin"
+	"github.com/songquanpeng/one-api/common/logger"
+	"github.com/songquanpeng/one-api/relay"
+	"github.com/songquanpeng/one-api/relay/adaptor/openai"
+	"github.com/songquanpeng/one-api/relay/meta"
+	relaymodel "github.com/songquanpeng/one-api/relay/model"
+)
+
+// RelayProxyHelper is a helper function to proxy the request to the upstream service
+func RelayProxyHelper(c *gin.Context, relayMode int) *relaymodel.ErrorWithStatusCode {
+	ctx := c.Request.Context()
+	meta := meta.GetByContext(c)
+
+	adaptor := relay.GetAdaptor(meta.APIType)
+	if adaptor == nil {
+		return openai.ErrorWrapper(fmt.Errorf("invalid api type: %d", meta.APIType), "invalid_api_type", http.StatusBadRequest)
+	}
+	adaptor.Init(meta)
+
+	resp, err := adaptor.DoRequest(c, meta, c.Request.Body)
+	if err != nil {
+		logger.Errorf(ctx, "DoRequest failed: %s", err.Error())
+		return openai.ErrorWrapper(err, "do_request_failed", http.StatusInternalServerError)
+	}
+
+	// do response
+	_, respErr := adaptor.DoResponse(c, resp, meta)
+	if respErr != nil {
+		logger.Errorf(ctx, "respErr is not nil: %+v", respErr)
+		return respErr
+	}
+
+	return nil
+}

relay/controller/text.go

@@ -4,6 +4,10 @@ import (
 	"bytes"
 	"encoding/json"
 	"fmt"
+	"github.com/songquanpeng/one-api/relay/adaptor"
+	"io"
+	"net/http"
+
 	"github.com/gin-gonic/gin"
 	"github.com/songquanpeng/one-api/common/logger"
 	"github.com/songquanpeng/one-api/relay"
@@ -14,15 +18,8 @@ import (
 	"github.com/songquanpeng/one-api/relay/channeltype"
 	"github.com/songquanpeng/one-api/relay/meta"
 	"github.com/songquanpeng/one-api/relay/model"
-	"io"
-	"math"
-	"net/http"
-	"os"
-	"strings"
 )
 
-var fixAIProxyGpt4oMaxTokens = os.Getenv("FIX_AI_PROXY_GPT4O_MAX_TOKENS") == "1"
-
 func RelayTextHelper(c *gin.Context) *model.ErrorWithStatusCode {
 	ctx := c.Request.Context()
 	meta := meta.GetByContext(c)
@@ -35,12 +32,11 @@ func RelayTextHelper(c *gin.Context) *model.ErrorWithStatusCode {
 	meta.IsStream = textRequest.Stream
 
 	// map model name
-	var isModelMapped bool
 	meta.OriginModelName = textRequest.Model
-	textRequest.Model, isModelMapped = getMappedModelName(textRequest.Model, meta.ModelMapping)
+	textRequest.Model, _ = getMappedModelName(textRequest.Model, meta.ModelMapping)
 	meta.ActualModelName = textRequest.Model
 	// get model ratio & group ratio
-	modelRatio := billingratio.GetModelRatio(textRequest.Model)
+	modelRatio := billingratio.GetModelRatio(textRequest.Model, meta.ChannelType)
 	groupRatio := billingratio.GetGroupRatio(meta.Group)
 	ratio := modelRatio * groupRatio
 	// pre-consume quota
@@ -58,42 +54,10 @@ func RelayTextHelper(c *gin.Context) *model.ErrorWithStatusCode {
 	}
 	adaptor.Init(meta)
 
-	// get request body
+	requestBody, err := getRequestBody(c, meta, textRequest, adaptor)
-	var requestBody io.Reader
-	if meta.APIType == apitype.OpenAI {
-		// no need to convert request for openai
-		shouldResetRequestBody := isModelMapped ||
-			meta.ChannelType == channeltype.Baichuan /*frequency_penalty 0 is not acceptable for baichuan*/ ||
-			(meta.ChannelType == channeltype.AIProxy && fixAIProxyGpt4oMaxTokens && strings.HasPrefix(textRequest.Model, "gpt-4o"))
-		if shouldResetRequestBody {
-			if meta.ChannelType == channeltype.AIProxy {
-				maxTokens := textRequest.MaxTokens
-				maxTokens = int(math.Min(float64(maxTokens), 4096))
-				if maxTokens == 0 {
-					maxTokens = 4096
-				}
-				textRequest.MaxTokens = maxTokens
-			}
-			jsonStr, err := json.Marshal(textRequest)
-			if err != nil {
-				return openai.ErrorWrapper(err, "json_marshal_failed", http.StatusInternalServerError)
-			}
-			requestBody = bytes.NewBuffer(jsonStr)
-		} else {
-			requestBody = c.Request.Body
-		}
-	} else {
-		convertedRequest, err := adaptor.ConvertRequest(c, meta.Mode, textRequest)
 	if err != nil {
 		return openai.ErrorWrapper(err, "convert_request_failed", http.StatusInternalServerError)
 	}
-		jsonData, err := json.Marshal(convertedRequest)
-		if err != nil {
-			return openai.ErrorWrapper(err, "json_marshal_failed", http.StatusInternalServerError)
-		}
-		logger.Debugf(ctx, "converted request: \n%s", string(jsonData))
-		requestBody = bytes.NewBuffer(jsonData)
-	}
 
 	// do request
 	resp, err := adaptor.DoRequest(c, meta, requestBody)
@@ -117,3 +81,26 @@ func RelayTextHelper(c *gin.Context) *model.ErrorWithStatusCode {
 	go postConsumeQuota(ctx, usage, meta, textRequest, ratio, preConsumedQuota, modelRatio, groupRatio)
 	return nil
 }
+
+func getRequestBody(c *gin.Context, meta *meta.Meta, textRequest *model.GeneralOpenAIRequest, adaptor adaptor.Adaptor) (io.Reader, error) {
+	if meta.APIType == apitype.OpenAI && meta.OriginModelName == meta.ActualModelName && meta.ChannelType != channeltype.Baichuan {
+		// no need to convert request for openai
+		return c.Request.Body, nil
+	}
+
+	// get request body
+	var requestBody io.Reader
+	convertedRequest, err := adaptor.ConvertRequest(c, meta.Mode, textRequest)
+	if err != nil {
+		logger.Debugf(c.Request.Context(), "converted request failed: %s\n", err.Error())
+		return nil, err
+	}
+	jsonData, err := json.Marshal(convertedRequest)
+	if err != nil {
+		logger.Debugf(c.Request.Context(), "converted request json_marshal_failed: %s\n", err.Error())
+		return nil, err
+	}
+	logger.Debugf(c.Request.Context(), "converted request: \n%s", string(jsonData))
+	requestBody = bytes.NewBuffer(jsonData)
+	return requestBody, nil
+}

relay/meta/relay_meta.go

@@ -18,12 +18,15 @@ type Meta struct {
 	UserId       int
 	Group        string
 	ModelMapping map[string]string
+	// BaseURL is the proxy url set in the channel config
 	BaseURL  string
 	APIKey   string
 	APIType  int
 	Config   model.ChannelConfig
 	IsStream bool
+	// OriginModelName is the model name from the raw user request
 	OriginModelName string
+	// ActualModelName is the model name after mapping
 	ActualModelName string
 	RequestURLPath  string
 	PromptTokens    int // only for DoResponse

relay/model/general.go

@@ -13,6 +13,7 @@ type GeneralOpenAIRequest struct {
 	PresencePenalty  float64         `json:"presence_penalty,omitempty"`
 	ResponseFormat   *ResponseFormat `json:"response_format,omitempty"`
 	Seed             float64         `json:"seed,omitempty"`
+	Stop             any             `json:"stop,omitempty"`
 	Stream           bool            `json:"stream,omitempty"`
 	Temperature      float64         `json:"temperature,omitempty"`
 	TopP             float64         `json:"top_p,omitempty"`
@@ -28,6 +29,7 @@ type GeneralOpenAIRequest struct {
 	Dimensions       int             `json:"dimensions,omitempty"`
 	Instruction      string          `json:"instruction,omitempty"`
 	Size             string          `json:"size,omitempty"`
+	NumCtx           int         	 `json:"num_ctx,omitempty"`
 }
 
 func (r GeneralOpenAIRequest) ParseInput() []string {