chore: disable arm64 build for now

* feat: support hunyuan-embedding

* chore: improve implementation

---------

Co-authored-by: LUO Feng <luofeng@flowpp.com>
Co-authored-by: JustSong <quanpengsong@gmail.com>

.env.example

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

.github/workflows/ci.yml

@@ -0,0 +1,45 @@
+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:
+  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-en.yml

@@ -1,56 +0,0 @@
-name: Publish Docker image (amd64, English)
-
-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: Translate
-        run: |
-          python ./i18n/translate.py --repository_path . --json_file_path ./i18n/en.json
-      - name: Log in to Docker Hub
-        uses: docker/login-action@v2
-        with:
-          username: ${{ secrets.DOCKERHUB_USERNAME }}
-          password: ${{ secrets.DOCKERHUB_TOKEN }}
-
-      - name: Extract metadata (tags, labels) for Docker
-        id: meta
-        uses: docker/metadata-action@v4
-        with:
-          images: |
-            justsong/one-api-en
-
-      - 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-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.yml

@@ -1,10 +1,9 @@
-name: Publish Docker image (arm64)
+name: Publish Docker image
 
 on:
   push:
     tags:
       - 'v*.*.*'
-      - '!*-alpha*'
   workflow_dispatch:
     inputs:
       name:
@@ -63,7 +62,8 @@ jobs:
         uses: docker/build-push-action@v3
         with:
           context: .
-          platforms: linux/amd64,linux/arm64
+          #          platforms: linux/amd64,linux/arm64
+          platforms: linux/amd64 # TODO disable arm64 for now, because it cause error
           push: true
           tags: ${{ steps.meta.outputs.tags }}
           labels: ${{ steps.meta.outputs.labels }}

.gitignore

@@ -8,4 +8,6 @@ build
 logs
 data
 /web/node_modules
-cmd.md
+cmd.md
+.env
+/one-api

Dockerfile

@@ -1,42 +1,51 @@
-FROM node:16 as builder
+FROM --platform=$BUILDPLATFORM node:16 AS builder
 
 WORKDIR /web
 COPY ./VERSION .
 COPY ./web .
 
-WORKDIR /web/default
-RUN npm install
-RUN DISABLE_ESLINT_PLUGIN='true' REACT_APP_VERSION=$(cat VERSION) npm run build
+RUN npm install --prefix /web/default & \
+    npm install --prefix /web/berry & \
+    npm install --prefix /web/air & \
+    wait
 
-WORKDIR /web/berry
-RUN npm install
-RUN DISABLE_ESLINT_PLUGIN='true' REACT_APP_VERSION=$(cat VERSION) npm run build
-
-WORKDIR /web/air
-RUN npm install
-RUN DISABLE_ESLINT_PLUGIN='true' REACT_APP_VERSION=$(cat VERSION) npm run build
+RUN DISABLE_ESLINT_PLUGIN='true' REACT_APP_VERSION=$(cat /web/default/VERSION) npm run build --prefix /web/default & \
+    DISABLE_ESLINT_PLUGIN='true' REACT_APP_VERSION=$(cat /web/berry/VERSION) npm run build --prefix /web/berry & \
+    DISABLE_ESLINT_PLUGIN='true' REACT_APP_VERSION=$(cat /web/air/VERSION) npm run build --prefix /web/air & \
+    wait
 
 FROM golang AS builder2
 
+RUN apt-get update && apt-get install -y --no-install-recommends \
+    build-essential \
+    sqlite3 libsqlite3-dev \
+    && rm -rf /var/lib/apt/lists/*
+
 ENV GO111MODULE=on \
     CGO_ENABLED=1 \
-    GOOS=linux
+    GOOS=linux \
+    CGO_CFLAGS="-I/usr/include" \
+    CGO_LDFLAGS="-L/usr/lib"
 
 WORKDIR /build
+
 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
 
-FROM alpine
+RUN go build -trimpath -ldflags "-s -w -X 'github.com/songquanpeng/one-api/common.Version=$(cat VERSION)'" -o one-api
 
-RUN apk update \
-    && apk upgrade \
-    && apk add --no-cache ca-certificates tzdata \
-    && update-ca-certificates 2>/dev/null || true
+# Final runtime image
+FROM ubuntu:22.04
+
+RUN apt-get update && apt-get install -y --no-install-recommends \
+    ca-certificates tzdata bash \
+    && rm -rf /var/lib/apt/lists/*
 
 COPY --from=builder2 /build/one-api /
+
 EXPOSE 3000
 WORKDIR /data
 ENTRYPOINT ["/one-api"]

README.en.md

@@ -101,7 +101,7 @@ Nginx reference configuration:
 ```
 server{
    server_name openai.justsong.cn;  # Modify your domain name accordingly
-   
+
    location / {
           client_max_body_size  64m;
           proxy_http_version 1.1;
@@ -132,12 +132,12 @@ The initial account username is `root` and password is `123456`.
 1. Download the executable file from [GitHub Releases](https://github.com/songquanpeng/one-api/releases/latest) or compile from source:
    ```shell
    git clone https://github.com/songquanpeng/one-api.git
-   
+
    # Build the frontend
    cd one-api/web/default
    npm install
    npm run build
-   
+
    # Build the backend
    cd ../..
    go mod download
@@ -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

@@ -53,7 +53,7 @@ _✨ 通过标准的 OpenAI API 格式访问所有的大模型，开箱即用
 
 > [!NOTE]
 > 本项目为开源项目，使用者必须在遵循 OpenAI 的[使用条款](https://openai.com/policies/terms-of-use)以及**法律法规**的情况下使用，不得用于非法用途。
-> 
+>
 > 根据[《生成式人工智能服务管理暂行办法》](http://www.cac.gov.cn/2023-07/13/c_1690898327029107.htm)的要求，请勿对中国地区公众提供一切未经备案的生成式人工智能服务。
 
 > [!WARNING]
@@ -68,6 +68,7 @@ _✨ 通过标准的 OpenAI API 格式访问所有的大模型，开箱即用
    + [x] [Anthropic Claude 系列模型](https://anthropic.com) (支持 AWS Claude)
    + [x] [Google PaLM2/Gemini 系列模型](https://developers.generativeai.google)
    + [x] [Mistral 系列模型](https://mistral.ai/)
+   + [x] [字节跳动豆包大模型](https://console.volcengine.com/ark/region:ark+cn-beijing/model)
    + [x] [百度文心一言系列模型](https://cloud.baidu.com/doc/WENXINWORKSHOP/index.html)
    + [x] [阿里通义千问系列模型](https://help.aliyun.com/document_detail/2400395.html)
    + [x] [讯飞星火认知大模型](https://www.xfyun.cn/doc/spark/Web.html)
@@ -76,7 +77,6 @@ _✨ 通过标准的 OpenAI API 格式访问所有的大模型，开箱即用
    + [x] [腾讯混元大模型](https://cloud.tencent.com/document/product/1729)
    + [x] [Moonshot AI](https://platform.moonshot.cn/)
    + [x] [百川大模型](https://platform.baichuan-ai.com)
-   + [ ] [字节云雀大模型](https://www.volcengine.com/product/ark) (WIP)
    + [x] [MINIMAX](https://api.minimax.chat/)
    + [x] [Groq](https://wow.groq.com/)
    + [x] [Ollama](https://github.com/ollama/ollama)
@@ -87,6 +87,10 @@ _✨ 通过标准的 OpenAI API 格式访问所有的大模型，开箱即用
    + [x] [DeepSeek](https://www.deepseek.com/)
    + [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)
+   + [x] [xAI](https://x.ai/)
 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。
@@ -111,8 +115,8 @@ _✨ 通过标准的 OpenAI API 格式访问所有的大模型，开箱即用
 21. 支持 Cloudflare Turnstile 用户校验。
 22. 支持用户管理，支持**多种用户登录注册方式**：
     + 邮箱登录注册（支持注册邮箱白名单）以及通过邮箱进行密码重置。
-    + 支持使用飞书进行授权登录。
-    + [GitHub 开放授权](https://github.com/settings/applications/new)。
+    + 支持[飞书授权登录](https://open.feishu.cn/document/uAjLw4CM/ukTMukTMukTM/reference/authen-v1/authorize/get)（[这里有 One API 的实现细节阐述供参考](https://iamazing.cn/page/feishu-oauth-login)）。
+    + 支持 [GitHub 授权登录](https://github.com/settings/applications/new)。
     + 微信公众号授权（需要额外部署 [WeChat Server](https://github.com/songquanpeng/wechat-server)）。
 23. 支持主题切换，设置环境变量 `THEME` 即可，默认为 `default`，欢迎 PR 更多主题，具体参考[此处](./web/README.md)。
 24. 配合 [Message Pusher](https://github.com/songquanpeng/message-pusher) 可将报警信息推送到多种 App 上。
@@ -143,7 +147,7 @@ Nginx 的参考配置：
 ```
 server{
    server_name openai.justsong.cn;  # 请根据实际情况修改你的域名
-   
+
    location / {
           client_max_body_size  64m;
           proxy_http_version 1.1;
@@ -171,6 +175,10 @@ sudo service nginx restart
 
 初始账号用户名为 `root`，密码为 `123456`。
 
+### 通过宝塔面板进行一键部署
+1. 安装宝塔面板9.2.0及以上版本，前往 [宝塔面板](https://www.bt.cn/new/download.html?r=dk_oneapi) 官网，选择正式版的脚本下载安装；
+2. 安装后登录宝塔面板，在左侧菜单栏中点击 `Docker`，首次进入会提示安装 `Docker` 服务，点击立即安装，按提示完成安装；
+3. 安装完成后在应用商店中搜索 `One-API`，点击安装，配置域名等基本信息即可完成安装；
 
 ### 基于 Docker Compose 进行部署
 
@@ -188,12 +196,12 @@ docker-compose ps
 1. 从 [GitHub Releases](https://github.com/songquanpeng/one-api/releases/latest) 下载可执行文件或者从源码编译：
    ```shell
    git clone https://github.com/songquanpeng/one-api.git
-   
+
    # 构建前端
    cd one-api/web/default
    npm install
    npm run build
-   
+
    # 构建后端
    cd ../..
    go mod download
@@ -214,7 +222,7 @@ docker-compose ps
 3. 所有从服务器必须设置 `NODE_TYPE` 为 `slave`，不设置则默认为主服务器。
 4. 设置 `SYNC_FREQUENCY` 后服务器将定期从数据库同步配置，在使用远程数据库的情况下，推荐设置该项并启用 Redis，无论主从。
 5. 从服务器可以选择设置 `FRONTEND_BASE_URL`，以重定向页面请求到主服务器。
-6. 从服务器上**分别**装好 Redis，设置好 `REDIS_CONN_STRING`，这样可以做到在缓存未过期的情况下数据库零访问，可以减少延迟。
+6. 从服务器上**分别**装好 Redis，设置好 `REDIS_CONN_STRING`，这样可以做到在缓存未过期的情况下数据库零访问，可以减少延迟（Redis 集群或者哨兵模式的支持请参考环境变量说明）。
 7. 如果主服务器访问数据库延迟也比较高，则也需要启用 Redis，并设置 `SYNC_FREQUENCY`，以定期从数据库同步配置。
 
 环境变量的具体使用方法详见[此处](#环境变量)。
@@ -249,9 +257,9 @@ 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>
@@ -320,7 +328,7 @@ Render 可以直接部署 docker 镜像，不需要 fork 仓库：https://dashbo
 例如对于 OpenAI 的官方库：
 ```bash
 OPENAI_API_KEY="sk-xxxxxx"
-OPENAI_API_BASE="https://<HOST>:<PORT>/v1" 
+OPENAI_API_BASE="https://<HOST>:<PORT>/v1"
 ```
 
 ```mermaid
@@ -339,9 +347,15 @@ 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。
+   + 如果需要使用哨兵或者集群模式：
+     + 则需要把该环境变量设置为节点列表，例如：`localhost:49153,localhost:49154,localhost:49155`。
+     + 除此之外还需要设置以下环境变量：
+       + `REDIS_PASSWORD`：Redis 集群或者哨兵模式下的密码设置。
+       + `REDIS_MASTER_NAME`：Redis 哨兵模式下主节点的名称。
 2. `SESSION_SECRET`：设置之后将使用固定的会话密钥，这样系统重新启动后已登录用户的 cookie 将依旧有效。
    + 例子：`SESSION_SECRET=random_string`
 3. `SQL_DSN`：设置之后将使用指定数据库而非 SQLite，请使用 MySQL 或 PostgreSQL。
@@ -367,30 +381,36 @@ graph LR
    + 例子：`NODE_TYPE=slave`
 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`
-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。
+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。 
+   +例子：`CHANNEL_TEST_FREQUENCY=1440`
+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。
-18. `SQLITE_BUSY_TIMEOUT`：SQLite 锁等待超时设置，单位为毫秒，默认 `3000`。
-19. `GEMINI_SAFETY_SETTING`：Gemini 的安全设置，默认 `BLOCK_NONE`。
-20. `GEMINI_VERSION`：One API 所使用的 Gemini 版本，默认为 `v1`。
-21. `THEME`：系统的主题设置，默认为 `default`，具体可选值参考[此处](./web/README.md)。
-22. `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`。
-23. `METRIC_QUEUE_SIZE`：请求成功率统计队列大小，默认为 `10`。
-24. `METRIC_SUCCESS_RATE_THRESHOLD`：请求成功率阈值，默认为 `0.8`。
-25. `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。
+16. `RELAY_TIMEOUT`：中继超时设置，单位为秒，默认不设置超时时间。
+17. `RELAY_PROXY`：设置后使用该代理来请求 API。
+18. `USER_CONTENT_REQUEST_TIMEOUT`：用户上传内容下载超时时间，单位为秒。
+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。
+20. `SQLITE_BUSY_TIMEOUT`：SQLite 锁等待超时设置，单位为毫秒，默认 `3000`。
+21. `GEMINI_SAFETY_SETTING`：Gemini 的安全设置，默认 `BLOCK_NONE`。
+22. `GEMINI_VERSION`：One API 所使用的 Gemini 版本，默认为 `v1`。
+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)。
+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`。
+25. `METRIC_QUEUE_SIZE`：请求成功率统计队列大小，默认为 `10`。
+26. `METRIC_SUCCESS_RATE_THRESHOLD`：请求成功率阈值，默认为 `0.8`。
+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。
+29. `ENFORCE_INCLUDE_USAGE`：是否强制在 stream 模型下返回 usage，默认不开启，可选值为 `true` 和 `false`。
+30. `TEST_PROMPT`：测试模型时的用户 prompt，默认为 `Print your model name exactly and do not output without any other text.`。
 
 ### 命令行参数
 1. `--port <port_number>`: 指定服务器监听的端口号，默认为 `3000`。
@@ -443,6 +463,8 @@ 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/client/init.go

@@ -0,0 +1,60 @@
+package client
+
+import (
+	"fmt"
+	"github.com/songquanpeng/one-api/common/config"
+	"github.com/songquanpeng/one-api/common/logger"
+	"net/http"
+	"net/url"
+	"time"
+)
+
+var HTTPClient *http.Client
+var ImpatientHTTPClient *http.Client
+var UserContentRequestHTTPClient *http.Client
+
+func Init() {
+	if config.UserContentRequestProxy != "" {
+		logger.SysLog(fmt.Sprintf("using %s as proxy to fetch user content", config.UserContentRequestProxy))
+		proxyURL, err := url.Parse(config.UserContentRequestProxy)
+		if err != nil {
+			logger.FatalLog(fmt.Sprintf("USER_CONTENT_REQUEST_PROXY set but invalid: %s", config.UserContentRequestProxy))
+		}
+		transport := &http.Transport{
+			Proxy: http.ProxyURL(proxyURL),
+		}
+		UserContentRequestHTTPClient = &http.Client{
+			Transport: transport,
+			Timeout:   time.Second * time.Duration(config.UserContentRequestTimeout),
+		}
+	} else {
+		UserContentRequestHTTPClient = &http.Client{}
+	}
+	var transport http.RoundTripper
+	if config.RelayProxy != "" {
+		logger.SysLog(fmt.Sprintf("using %s as api relay proxy", config.RelayProxy))
+		proxyURL, err := url.Parse(config.RelayProxy)
+		if err != nil {
+			logger.FatalLog(fmt.Sprintf("USER_CONTENT_REQUEST_PROXY set but invalid: %s", config.UserContentRequestProxy))
+		}
+		transport = &http.Transport{
+			Proxy: http.ProxyURL(proxyURL),
+		}
+	}
+
+	if config.RelayTimeout == 0 {
+		HTTPClient = &http.Client{
+			Transport: transport,
+		}
+	} else {
+		HTTPClient = &http.Client{
+			Timeout:   time.Duration(config.RelayTimeout) * time.Second,
+			Transport: transport,
+		}
+	}
+
+	ImpatientHTTPClient = &http.Client{
+		Timeout:   5 * time.Second,
+		Transport: transport,
+	}
+}

common/config/config.go

@@ -1,13 +1,14 @@
 package config
 
 import (
-	"github.com/songquanpeng/one-api/common/env"
 	"os"
 	"strconv"
 	"strings"
 	"sync"
 	"time"
 
+	"github.com/songquanpeng/one-api/common/env"
+
 	"github.com/google/uuid"
 )
 
@@ -35,6 +36,7 @@ var PasswordLoginEnabled = true
 var PasswordRegisterEnabled = true
 var EmailVerificationEnabled = false
 var GitHubOAuthEnabled = false
+var OidcEnabled = false
 var WeChatAuthEnabled = false
 var TurnstileCheckEnabled = false
 var RegisterEnabled = true
@@ -70,6 +72,13 @@ var GitHubClientSecret = ""
 var LarkClientId = ""
 var LarkClientSecret = ""
 
+var OidcClientId = ""
+var OidcClientSecret = ""
+var OidcWellKnown = ""
+var OidcAuthorizationEndpoint = ""
+var OidcTokenEndpoint = ""
+var OidcUserinfoEndpoint = ""
+
 var WeChatServerAddress = ""
 var WeChatServerToken = ""
 var WeChatAccountQRCodeImageURL = ""
@@ -117,10 +126,10 @@ 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
@@ -143,4 +152,15 @@ 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)
+
+var EnforceIncludeUsage = env.Bool("ENFORCE_INCLUDE_USAGE", false)
+var TestPrompt = env.String("TEST_PROMPT", "Print your model name exactly and do not output without any other text.")

common/ctxkey/key.go

@@ -19,4 +19,6 @@ const (
 	TokenName         = "token_name"
 	BaseURL           = "base_url"
 	AvailableModels   = "available_models"
+	KeyRequestBody    = "key_request_body"
+	SystemPrompt      = "system_prompt"
 )

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
 }
 
@@ -32,15 +31,15 @@ func UnmarshalBodyReusable(c *gin.Context, v any) error {
 	contentType := c.Request.Header.Get("Content-Type")
 	if strings.HasPrefix(contentType, "application/json") {
 		err = json.Unmarshal(requestBody, &v)
+		c.Request.Body = io.NopCloser(bytes.NewBuffer(requestBody))
 	} else {
-		// skip for now
-		// TODO: someday non json request have variant model, we will need to implementation this
+		c.Request.Body = io.NopCloser(bytes.NewBuffer(requestBody))
+		err = c.ShouldBind(&v)
 	}
 	if err != nil {
 		return err
 	}
 	// Reset request body
-	c.Request.Body = io.NopCloser(bytes.NewBuffer(requestBody))
 	return nil
 }
 

common/helper/helper.go

@@ -1,9 +1,8 @@
 package helper
 
 import (
+	"context"
 	"fmt"
-	"github.com/gin-gonic/gin"
-	"github.com/songquanpeng/one-api/common/random"
 	"html/template"
 	"log"
 	"net"
@@ -11,6 +10,10 @@ import (
 	"runtime"
 	"strconv"
 	"strings"
+
+	"github.com/gin-gonic/gin"
+
+	"github.com/songquanpeng/one-api/common/random"
 )
 
 func OpenBrowser(url string) {
@@ -106,6 +109,18 @@ func GenRequestID() string {
 	return GetTimeString() + random.GetRandomNumberString(8)
 }
 
+func SetRequestID(ctx context.Context, id string) context.Context {
+	return context.WithValue(ctx, RequestIdKey, id)
+}
+
+func GetRequestID(ctx context.Context) string {
+	rawRequestId := ctx.Value(RequestIdKey)
+	if rawRequestId == nil {
+		return ""
+	}
+	return rawRequestId.(string)
+}
+
 func GetResponseID(c *gin.Context) string {
 	logID := c.GetString(RequestIdKey)
 	return fmt.Sprintf("chatcmpl-%s", logID)
@@ -137,3 +152,23 @@ func String2Int(str string) int {
 	}
 	return num
 }
+
+func Float64PtrMax(p *float64, maxValue float64) *float64 {
+	if p == nil {
+		return nil
+	}
+	if *p > maxValue {
+		return &maxValue
+	}
+	return p
+}
+
+func Float64PtrMin(p *float64, minValue float64) *float64 {
+	if p == nil {
+		return nil
+	}
+	if *p < minValue {
+		return &minValue
+	}
+	return p
+}

common/helper/time.go

@@ -13,3 +13,8 @@ func GetTimeString() string {
 	now := time.Now()
 	return fmt.Sprintf("%s%d", now.Format("20060102150405"), now.UnixNano()%1e9)
 }
+
+// CalcElapsedTime return the elapsed time in milliseconds (ms)
+func CalcElapsedTime(start time.Time) int64 {
+	return time.Now().Sub(start).Milliseconds()
+}

common/image/image.go

@@ -3,6 +3,7 @@ package image
 import (
 	"bytes"
 	"encoding/base64"
+	"github.com/songquanpeng/one-api/common/client"
 	"image"
 	_ "image/gif"
 	_ "image/jpeg"
@@ -19,7 +20,7 @@ import (
 var dataURLPattern = regexp.MustCompile(`data:image/([^;]+);base64,(.*)`)
 
 func IsImageUrl(url string) (bool, error) {
-	resp, err := http.Head(url)
+	resp, err := client.UserContentRequestHTTPClient.Head(url)
 	if err != nil {
 		return false, err
 	}
@@ -34,7 +35,7 @@ func GetImageSizeFromUrl(url string) (width int, height int, err error) {
 	if !isImage {
 		return
 	}
-	resp, err := http.Get(url)
+	resp, err := client.UserContentRequestHTTPClient.Get(url)
 	if err != nil {
 		return
 	}

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

@@ -7,19 +7,25 @@ import (
 	"log"
 	"os"
 	"path/filepath"
+	"runtime"
+	"strings"
 	"sync"
 	"time"
 
 	"github.com/gin-gonic/gin"
+
 	"github.com/songquanpeng/one-api/common/config"
 	"github.com/songquanpeng/one-api/common/helper"
 )
 
+type loggerLevel string
+
 const (
-	loggerDEBUG = "DEBUG"
-	loggerINFO  = "INFO"
-	loggerWarn  = "WARN"
-	loggerError = "ERR"
+	loggerDEBUG loggerLevel = "DEBUG"
+	loggerINFO  loggerLevel = "INFO"
+	loggerWarn  loggerLevel = "WARN"
+	loggerError loggerLevel = "ERROR"
+	loggerFatal loggerLevel = "FATAL"
 )
 
 var setupLogOnce sync.Once
@@ -27,7 +33,12 @@ 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")
@@ -39,19 +50,26 @@ func SetupLogger() {
 }
 
 func SysLog(s string) {
-	t := time.Now()
-	_, _ = fmt.Fprintf(gin.DefaultWriter, "[SYS] %v | %s \n", t.Format("2006/01/02 - 15:04:05"), s)
+	logHelper(nil, loggerINFO, s)
+}
+
+func SysLogf(format string, a ...any) {
+	logHelper(nil, loggerINFO, 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)
+	logHelper(nil, loggerError, s)
+}
+
+func SysErrorf(format string, a ...any) {
+	logHelper(nil, loggerError, fmt.Sprintf(format, a...))
 }
 
 func Debug(ctx context.Context, msg string) {
-	if config.DebugEnabled {
-		logHelper(ctx, loggerDEBUG, msg)
+	if !config.DebugEnabled {
+		return
 	}
+	logHelper(ctx, loggerDEBUG, msg)
 }
 
 func Info(ctx context.Context, msg string) {
@@ -67,37 +85,65 @@ func Error(ctx context.Context, msg string) {
 }
 
 func Debugf(ctx context.Context, format string, a ...any) {
-	Debug(ctx, fmt.Sprintf(format, a...))
+	logHelper(ctx, loggerDEBUG, fmt.Sprintf(format, a...))
 }
 
 func Infof(ctx context.Context, format string, a ...any) {
-	Info(ctx, fmt.Sprintf(format, a...))
+	logHelper(ctx, loggerINFO, fmt.Sprintf(format, a...))
 }
 
 func Warnf(ctx context.Context, format string, a ...any) {
-	Warn(ctx, fmt.Sprintf(format, a...))
+	logHelper(ctx, loggerWarn, fmt.Sprintf(format, a...))
 }
 
 func Errorf(ctx context.Context, format string, a ...any) {
-	Error(ctx, fmt.Sprintf(format, a...))
+	logHelper(ctx, loggerError, fmt.Sprintf(format, a...))
 }
 
-func logHelper(ctx context.Context, level string, msg string) {
+func FatalLog(s string) {
+	logHelper(nil, loggerFatal, s)
+}
+
+func FatalLogf(format string, a ...any) {
+	logHelper(nil, loggerFatal, fmt.Sprintf(format, a...))
+}
+
+func logHelper(ctx context.Context, level loggerLevel, msg string) {
 	writer := gin.DefaultErrorWriter
 	if level == loggerINFO {
 		writer = gin.DefaultWriter
 	}
-	id := ctx.Value(helper.RequestIdKey)
-	if id == nil {
-		id = helper.GenRequestID()
+	var requestId string
+	if ctx != nil {
+		rawRequestId := helper.GetRequestID(ctx)
+		if rawRequestId != "" {
+			requestId = fmt.Sprintf(" | %s", rawRequestId)
+		}
 	}
+	lineInfo, funcName := getLineInfo()
 	now := time.Now()
-	_, _ = fmt.Fprintf(writer, "[%s] %v | %s | %s \n", level, now.Format("2006/01/02 - 15:04:05"), id, msg)
+	_, _ = fmt.Fprintf(writer, "[%s] %v%s%s %s%s \n", level, now.Format("2006/01/02 - 15:04:05"), requestId, lineInfo, funcName, msg)
 	SetupLogger()
+	if level == loggerFatal {
+		os.Exit(1)
+	}
 }
 
-func FatalLog(v ...any) {
-	t := time.Now()
-	_, _ = fmt.Fprintf(gin.DefaultErrorWriter, "[FATAL] %v | %v \n", t.Format("2006/01/02 - 15:04:05"), v)
-	os.Exit(1)
+func getLineInfo() (string, string) {
+	funcName := "[unknown] "
+	pc, file, line, ok := runtime.Caller(3)
+	if ok {
+		if fn := runtime.FuncForPC(pc); fn != nil {
+			parts := strings.Split(fn.Name(), ".")
+			funcName = "[" + parts[len(parts)-1] + "] "
+		}
+	} else {
+		file = "unknown"
+		line = 0
+	}
+	parts := strings.Split(file, "one-api/")
+	if len(parts) > 1 {
+		file = parts[1]
+	}
+	return fmt.Sprintf(" | %s:%d", file, line), funcName
 }

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 {
-		tlsConfig := &tls.Config{
-			InsecureSkipVerify: true,
-			ServerName:         config.SMTPServer,
+	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)
+		} else {
+			conn, err = net.Dial("tcp", fmt.Sprintf("%s:%d", config.SMTPServer, config.SMTPPort))
 		}
-		conn, err := tls.Dial("tcp", fmt.Sprintf("%s:%d", config.SMTPServer, config.SMTPPort), tlsConfig)
 		if err != nil {
 			return err
 		}
@@ -59,8 +72,10 @@ func SendEmail(subject string, receiver string, content string) error {
 			return err
 		}
 		defer client.Close()
-		if err = client.Auth(auth); err != nil {
-			return err
+		if shouldAuth() {
+			if err = client.Auth(auth); err != nil {
+				return err
+			}
 		}
 		if err = client.Mail(config.SMTPFrom); err != nil {
 			return err

common/redis.go

@@ -2,13 +2,15 @@ package common
 
 import (
 	"context"
+	"os"
+	"strings"
+	"time"
+
 	"github.com/go-redis/redis/v8"
 	"github.com/songquanpeng/one-api/common/logger"
-	"os"
-	"time"
 )
 
-var RDB *redis.Client
+var RDB redis.Cmdable
 var RedisEnabled = true
 
 // InitRedisClient This function is called after init()
@@ -23,13 +25,23 @@ func InitRedisClient() (err error) {
 		logger.SysLog("SYNC_FREQUENCY not set, Redis is disabled")
 		return nil
 	}
-	logger.SysLog("Redis is enabled")
-	opt, err := redis.ParseURL(os.Getenv("REDIS_CONN_STRING"))
-	if err != nil {
-		logger.FatalLog("failed to parse Redis connection string: " + err.Error())
+	redisConnString := os.Getenv("REDIS_CONN_STRING")
+	if os.Getenv("REDIS_MASTER_NAME") == "" {
+		logger.SysLog("Redis is enabled")
+		opt, err := redis.ParseURL(redisConnString)
+		if err != nil {
+			logger.FatalLog("failed to parse Redis connection string: " + err.Error())
+		}
+		RDB = redis.NewClient(opt)
+	} else {
+		// cluster mode
+		logger.SysLog("Redis cluster mode enabled")
+		RDB = redis.NewUniversalClient(&redis.UniversalOptions{
+			Addrs:      strings.Split(redisConnString, ","),
+			Password:   os.Getenv("REDIS_PASSWORD"),
+			MasterName: os.Getenv("REDIS_MASTER_NAME"),
+		})
 	}
-	RDB = redis.NewClient(opt)
-
 	ctx, cancel := context.WithTimeout(context.Background(), 5*time.Second)
 	defer cancel()
 

common/render/render.go

@@ -0,0 +1,30 @@
+package render
+
+import (
+	"encoding/json"
+	"fmt"
+	"strings"
+
+	"github.com/gin-gonic/gin"
+	"github.com/songquanpeng/one-api/common"
+)
+
+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/auth/github.go

@@ -5,16 +5,18 @@ import (
 	"encoding/json"
 	"errors"
 	"fmt"
+	"net/http"
+	"strconv"
+	"time"
+
 	"github.com/gin-contrib/sessions"
 	"github.com/gin-gonic/gin"
+
 	"github.com/songquanpeng/one-api/common/config"
 	"github.com/songquanpeng/one-api/common/logger"
 	"github.com/songquanpeng/one-api/common/random"
 	"github.com/songquanpeng/one-api/controller"
 	"github.com/songquanpeng/one-api/model"
-	"net/http"
-	"strconv"
-	"time"
 )
 
 type GitHubOAuthResponse struct {
@@ -81,6 +83,7 @@ func getGitHubUserInfoByCode(code string) (*GitHubUser, error) {
 }
 
 func GitHubOAuth(c *gin.Context) {
+	ctx := c.Request.Context()
 	session := sessions.Default(c)
 	state := c.Query("state")
 	if state == "" || session.Get("oauth_state") == nil || state != session.Get("oauth_state").(string) {
@@ -136,7 +139,7 @@ func GitHubOAuth(c *gin.Context) {
 			user.Role = model.RoleCommonUser
 			user.Status = model.UserStatusEnabled
 
-			if err := user.Insert(0); err != nil {
+			if err := user.Insert(ctx, 0); err != nil {
 				c.JSON(http.StatusOK, gin.H{
 					"success": false,
 					"message": err.Error(),

controller/auth/lark.go

@@ -5,15 +5,17 @@ import (
 	"encoding/json"
 	"errors"
 	"fmt"
+	"net/http"
+	"strconv"
+	"time"
+
 	"github.com/gin-contrib/sessions"
 	"github.com/gin-gonic/gin"
+
 	"github.com/songquanpeng/one-api/common/config"
 	"github.com/songquanpeng/one-api/common/logger"
 	"github.com/songquanpeng/one-api/controller"
 	"github.com/songquanpeng/one-api/model"
-	"net/http"
-	"strconv"
-	"time"
 )
 
 type LarkOAuthResponse struct {
@@ -40,7 +42,7 @@ func getLarkUserInfoByCode(code string) (*LarkUser, error) {
 	if err != nil {
 		return nil, err
 	}
-	req, err := http.NewRequest("POST", "https://passport.feishu.cn/suite/passport/oauth/token", bytes.NewBuffer(jsonData))
+	req, err := http.NewRequest("POST", "https://open.feishu.cn/open-apis/authen/v2/oauth/token", bytes.NewBuffer(jsonData))
 	if err != nil {
 		return nil, err
 	}
@@ -79,6 +81,7 @@ func getLarkUserInfoByCode(code string) (*LarkUser, error) {
 }
 
 func LarkOAuth(c *gin.Context) {
+	ctx := c.Request.Context()
 	session := sessions.Default(c)
 	state := c.Query("state")
 	if state == "" || session.Get("oauth_state") == nil || state != session.Get("oauth_state").(string) {
@@ -125,7 +128,7 @@ func LarkOAuth(c *gin.Context) {
 			user.Role = model.RoleCommonUser
 			user.Status = model.UserStatusEnabled
 
-			if err := user.Insert(0); err != nil {
+			if err := user.Insert(ctx, 0); err != nil {
 				c.JSON(http.StatusOK, gin.H{
 					"success": false,
 					"message": err.Error(),

controller/auth/oidc.go

@@ -0,0 +1,228 @@
+package auth
+
+import (
+	"bytes"
+	"encoding/json"
+	"errors"
+	"fmt"
+	"net/http"
+	"strconv"
+	"time"
+
+	"github.com/gin-contrib/sessions"
+	"github.com/gin-gonic/gin"
+
+	"github.com/songquanpeng/one-api/common/config"
+	"github.com/songquanpeng/one-api/common/logger"
+	"github.com/songquanpeng/one-api/controller"
+	"github.com/songquanpeng/one-api/model"
+)
+
+type OidcResponse struct {
+	AccessToken  string `json:"access_token"`
+	IDToken      string `json:"id_token"`
+	RefreshToken string `json:"refresh_token"`
+	TokenType    string `json:"token_type"`
+	ExpiresIn    int    `json:"expires_in"`
+	Scope        string `json:"scope"`
+}
+
+type OidcUser struct {
+	OpenID            string `json:"sub"`
+	Email             string `json:"email"`
+	Name              string `json:"name"`
+	PreferredUsername string `json:"preferred_username"`
+	Picture           string `json:"picture"`
+}
+
+func getOidcUserInfoByCode(code string) (*OidcUser, error) {
+	if code == "" {
+		return nil, errors.New("无效的参数")
+	}
+	values := map[string]string{
+		"client_id":     config.OidcClientId,
+		"client_secret": config.OidcClientSecret,
+		"code":          code,
+		"grant_type":    "authorization_code",
+		"redirect_uri":  fmt.Sprintf("%s/oauth/oidc", config.ServerAddress),
+	}
+	jsonData, err := json.Marshal(values)
+	if err != nil {
+		return nil, err
+	}
+	req, err := http.NewRequest("POST", config.OidcTokenEndpoint, bytes.NewBuffer(jsonData))
+	if err != nil {
+		return nil, err
+	}
+	req.Header.Set("Content-Type", "application/json")
+	req.Header.Set("Accept", "application/json")
+	client := http.Client{
+		Timeout: 5 * time.Second,
+	}
+	res, err := client.Do(req)
+	if err != nil {
+		logger.SysLog(err.Error())
+		return nil, errors.New("无法连接至 OIDC 服务器，请稍后重试！")
+	}
+	defer res.Body.Close()
+	var oidcResponse OidcResponse
+	err = json.NewDecoder(res.Body).Decode(&oidcResponse)
+	if err != nil {
+		return nil, err
+	}
+	req, err = http.NewRequest("GET", config.OidcUserinfoEndpoint, nil)
+	if err != nil {
+		return nil, err
+	}
+	req.Header.Set("Authorization", "Bearer "+oidcResponse.AccessToken)
+	res2, err := client.Do(req)
+	if err != nil {
+		logger.SysLog(err.Error())
+		return nil, errors.New("无法连接至 OIDC 服务器，请稍后重试！")
+	}
+	var oidcUser OidcUser
+	err = json.NewDecoder(res2.Body).Decode(&oidcUser)
+	if err != nil {
+		return nil, err
+	}
+	return &oidcUser, nil
+}
+
+func OidcAuth(c *gin.Context) {
+	ctx := c.Request.Context()
+	session := sessions.Default(c)
+	state := c.Query("state")
+	if state == "" || session.Get("oauth_state") == nil || state != session.Get("oauth_state").(string) {
+		c.JSON(http.StatusForbidden, gin.H{
+			"success": false,
+			"message": "state is empty or not same",
+		})
+		return
+	}
+	username := session.Get("username")
+	if username != nil {
+		OidcBind(c)
+		return
+	}
+	if !config.OidcEnabled {
+		c.JSON(http.StatusOK, gin.H{
+			"success": false,
+			"message": "管理员未开启通过 OIDC 登录以及注册",
+		})
+		return
+	}
+	code := c.Query("code")
+	oidcUser, err := getOidcUserInfoByCode(code)
+	if err != nil {
+		c.JSON(http.StatusOK, gin.H{
+			"success": false,
+			"message": err.Error(),
+		})
+		return
+	}
+	user := model.User{
+		OidcId: oidcUser.OpenID,
+	}
+	if model.IsOidcIdAlreadyTaken(user.OidcId) {
+		err := user.FillUserByOidcId()
+		if err != nil {
+			c.JSON(http.StatusOK, gin.H{
+				"success": false,
+				"message": err.Error(),
+			})
+			return
+		}
+	} else {
+		if config.RegisterEnabled {
+			user.Email = oidcUser.Email
+			if oidcUser.PreferredUsername != "" {
+				user.Username = oidcUser.PreferredUsername
+			} else {
+				user.Username = "oidc_" + strconv.Itoa(model.GetMaxUserId()+1)
+			}
+			if oidcUser.Name != "" {
+				user.DisplayName = oidcUser.Name
+			} else {
+				user.DisplayName = "OIDC User"
+			}
+			err := user.Insert(ctx, 0)
+			if err != nil {
+				c.JSON(http.StatusOK, gin.H{
+					"success": false,
+					"message": err.Error(),
+				})
+				return
+			}
+		} else {
+			c.JSON(http.StatusOK, gin.H{
+				"success": false,
+				"message": "管理员关闭了新用户注册",
+			})
+			return
+		}
+	}
+
+	if user.Status != model.UserStatusEnabled {
+		c.JSON(http.StatusOK, gin.H{
+			"message": "用户已被封禁",
+			"success": false,
+		})
+		return
+	}
+	controller.SetupLogin(&user, c)
+}
+
+func OidcBind(c *gin.Context) {
+	if !config.OidcEnabled {
+		c.JSON(http.StatusOK, gin.H{
+			"success": false,
+			"message": "管理员未开启通过 OIDC 登录以及注册",
+		})
+		return
+	}
+	code := c.Query("code")
+	oidcUser, err := getOidcUserInfoByCode(code)
+	if err != nil {
+		c.JSON(http.StatusOK, gin.H{
+			"success": false,
+			"message": err.Error(),
+		})
+		return
+	}
+	user := model.User{
+		OidcId: oidcUser.OpenID,
+	}
+	if model.IsOidcIdAlreadyTaken(user.OidcId) {
+		c.JSON(http.StatusOK, gin.H{
+			"success": false,
+			"message": "该 OIDC 账户已被绑定",
+		})
+		return
+	}
+	session := sessions.Default(c)
+	id := session.Get("id")
+	// id := c.GetInt("id")  // critical bug!
+	user.Id = id.(int)
+	err = user.FillUserById()
+	if err != nil {
+		c.JSON(http.StatusOK, gin.H{
+			"success": false,
+			"message": err.Error(),
+		})
+		return
+	}
+	user.OidcId = oidcUser.OpenID
+	err = user.Update(false)
+	if err != nil {
+		c.JSON(http.StatusOK, gin.H{
+			"success": false,
+			"message": err.Error(),
+		})
+		return
+	}
+	c.JSON(http.StatusOK, gin.H{
+		"success": true,
+		"message": "bind",
+	})
+	return
+}

controller/auth/wechat.go

@@ -4,14 +4,16 @@ import (
 	"encoding/json"
 	"errors"
 	"fmt"
+	"net/http"
+	"strconv"
+	"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/controller"
 	"github.com/songquanpeng/one-api/model"
-	"net/http"
-	"strconv"
-	"time"
 )
 
 type wechatLoginResponse struct {
@@ -52,6 +54,7 @@ func getWeChatIdByCode(code string) (string, error) {
 }
 
 func WeChatAuth(c *gin.Context) {
+	ctx := c.Request.Context()
 	if !config.WeChatAuthEnabled {
 		c.JSON(http.StatusOK, gin.H{
 			"message": "管理员未开启通过微信登录以及注册",
@@ -87,7 +90,7 @@ func WeChatAuth(c *gin.Context) {
 			user.Role = model.RoleCommonUser
 			user.Status = model.UserStatusEnabled
 
-			if err := user.Insert(0); err != nil {
+			if err := user.Insert(ctx, 0); err != nil {
 				c.JSON(http.StatusOK, gin.H{
 					"success": false,
 					"message": err.Error(),

controller/billing.go

@@ -17,9 +17,11 @@ func GetSubscription(c *gin.Context) {
 	if config.DisplayTokenStatEnabled {
 		tokenId := c.GetInt(ctxkey.TokenId)
 		token, err = model.GetTokenById(tokenId)
-		expiredTime = token.ExpiredTime
-		remainQuota = token.RemainQuota
-		usedQuota = token.UsedQuota
+		if err == nil {
+			expiredTime = token.ExpiredTime
+			remainQuota = token.RemainQuota
+			usedQuota = token.UsedQuota
+		}
 	} else {
 		userId := c.GetInt(ctxkey.Id)
 		remainQuota, err = model.GetUserQuota(userId)

controller/channel-billing.go

@@ -4,16 +4,17 @@ import (
 	"encoding/json"
 	"errors"
 	"fmt"
+	"io"
+	"net/http"
+	"strconv"
+	"time"
+
+	"github.com/songquanpeng/one-api/common/client"
 	"github.com/songquanpeng/one-api/common/config"
 	"github.com/songquanpeng/one-api/common/logger"
 	"github.com/songquanpeng/one-api/model"
 	"github.com/songquanpeng/one-api/monitor"
 	"github.com/songquanpeng/one-api/relay/channeltype"
-	"github.com/songquanpeng/one-api/relay/client"
-	"io"
-	"net/http"
-	"strconv"
-	"time"
 
 	"github.com/gin-gonic/gin"
 )
@@ -81,6 +82,36 @@ type APGC2DGPTUsageResponse struct {
 	TotalUsed      float64 `json:"total_used"`
 }
 
+type SiliconFlowUsageResponse struct {
+	Code    int    `json:"code"`
+	Message string `json:"message"`
+	Status  bool   `json:"status"`
+	Data    struct {
+		ID            string `json:"id"`
+		Name          string `json:"name"`
+		Image         string `json:"image"`
+		Email         string `json:"email"`
+		IsAdmin       bool   `json:"isAdmin"`
+		Balance       string `json:"balance"`
+		Status        string `json:"status"`
+		Introduction  string `json:"introduction"`
+		Role          string `json:"role"`
+		ChargeBalance string `json:"chargeBalance"`
+		TotalBalance  string `json:"totalBalance"`
+		Category      string `json:"category"`
+	} `json:"data"`
+}
+
+type DeepSeekUsageResponse struct {
+	IsAvailable  bool `json:"is_available"`
+	BalanceInfos []struct {
+		Currency        string `json:"currency"`
+		TotalBalance    string `json:"total_balance"`
+		GrantedBalance  string `json:"granted_balance"`
+		ToppedUpBalance string `json:"topped_up_balance"`
+	} `json:"balance_infos"`
+}
+
 // GetAuthHeader get auth header
 func GetAuthHeader(token string) http.Header {
 	h := http.Header{}
@@ -203,6 +234,57 @@ func updateChannelAIGC2DBalance(channel *model.Channel) (float64, error) {
 	return response.TotalAvailable, nil
 }
 
+func updateChannelSiliconFlowBalance(channel *model.Channel) (float64, error) {
+	url := "https://api.siliconflow.cn/v1/user/info"
+	body, err := GetResponseBody("GET", url, channel, GetAuthHeader(channel.Key))
+	if err != nil {
+		return 0, err
+	}
+	response := SiliconFlowUsageResponse{}
+	err = json.Unmarshal(body, &response)
+	if err != nil {
+		return 0, err
+	}
+	if response.Code != 20000 {
+		return 0, fmt.Errorf("code: %d, message: %s", response.Code, response.Message)
+	}
+	balance, err := strconv.ParseFloat(response.Data.TotalBalance, 64)
+	if err != nil {
+		return 0, err
+	}
+	channel.UpdateBalance(balance)
+	return balance, nil
+}
+
+func updateChannelDeepSeekBalance(channel *model.Channel) (float64, error) {
+	url := "https://api.deepseek.com/user/balance"
+	body, err := GetResponseBody("GET", url, channel, GetAuthHeader(channel.Key))
+	if err != nil {
+		return 0, err
+	}
+	response := DeepSeekUsageResponse{}
+	err = json.Unmarshal(body, &response)
+	if err != nil {
+		return 0, err
+	}
+	index := -1
+	for i, balanceInfo := range response.BalanceInfos {
+		if balanceInfo.Currency == "CNY" {
+			index = i
+			break
+		}
+	}
+	if index == -1 {
+		return 0, errors.New("currency CNY not found")
+	}
+	balance, err := strconv.ParseFloat(response.BalanceInfos[index].TotalBalance, 64)
+	if err != nil {
+		return 0, err
+	}
+	channel.UpdateBalance(balance)
+	return balance, nil
+}
+
 func updateChannelBalance(channel *model.Channel) (float64, error) {
 	baseURL := channeltype.ChannelBaseURLs[channel.Type]
 	if channel.GetBaseURL() == "" {
@@ -227,6 +309,10 @@ func updateChannelBalance(channel *model.Channel) (float64, error) {
 		return updateChannelAPI2GPTBalance(channel)
 	case channeltype.AIGC2D:
 		return updateChannelAIGC2DBalance(channel)
+	case channeltype.SiliconFlow:
+		return updateChannelSiliconFlowBalance(channel)
+	case channeltype.DeepSeek:
+		return updateChannelDeepSeekBalance(channel)
 	default:
 		return 0, errors.New("尚未实现")
 	}

controller/channel-test.go

@@ -2,6 +2,7 @@ package controller
 
 import (
 	"bytes"
+	"context"
 	"encoding/json"
 	"errors"
 	"fmt"
@@ -14,38 +15,58 @@ 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/helper"
 	"github.com/songquanpeng/one-api/common/logger"
 	"github.com/songquanpeng/one-api/common/message"
 	"github.com/songquanpeng/one-api/middleware"
 	"github.com/songquanpeng/one-api/model"
 	"github.com/songquanpeng/one-api/monitor"
-	relay "github.com/songquanpeng/one-api/relay"
+	"github.com/songquanpeng/one-api/relay"
+	"github.com/songquanpeng/one-api/relay/adaptor/openai"
 	"github.com/songquanpeng/one-api/relay/channeltype"
 	"github.com/songquanpeng/one-api/relay/controller"
 	"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:     "gpt-3.5-turbo",
+		Model: model,
 	}
 	testMessage := relaymodel.Message{
 		Role:    "user",
-		Content: "hi",
+		Content: config.TestPrompt,
 	}
 	testRequest.Messages = append(testRequest.Messages, testMessage)
 	return testRequest
 }
 
-func testChannel(channel *model.Channel) (err error, openaiErr *relaymodel.Error) {
+func parseTestResponse(resp string) (*openai.TextResponse, string, error) {
+	var response openai.TextResponse
+	err := json.Unmarshal([]byte(resp), &response)
+	if err != nil {
+		return nil, "", err
+	}
+	if len(response.Choices) == 0 {
+		return nil, "", errors.New("response has no choices")
+	}
+	stringContent, ok := response.Choices[0].Content.(string)
+	if !ok {
+		return nil, "", errors.New("response content is not string")
+	}
+	return &response, stringContent, nil
+}
+
+func testChannel(ctx context.Context, channel *model.Channel, request *relaymodel.GeneralOpenAIRequest) (responseMessage string, err error, openaiErr *relaymodel.Error) {
+	startTime := time.Now()
 	w := httptest.NewRecorder()
 	c, _ := gin.CreateTestContext(w)
 	c.Request = &http.Request{
@@ -65,64 +86,87 @@ func testChannel(channel *model.Channel) (err error, openaiErr *relaymodel.Error
 	apiType := channeltype.ToAPIType(channel.Type)
 	adaptor := relay.GetAdaptor(apiType)
 	if adaptor == nil {
-		return fmt.Errorf("invalid api type: %d, adaptor is nil", apiType), nil
+		return "", fmt.Errorf("invalid api type: %d, adaptor is nil", apiType), nil
 	}
 	adaptor.Init(meta)
-	var modelName string
-	modelList := adaptor.GetModelList()
+	modelName := request.Model
 	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 {
 			modelName = modelNames[0]
 		}
-		if modelMap != nil && modelMap[modelName] != "" {
-			modelName = modelMap[modelName]
-		}
 	}
-	request := buildTestRequest()
+	if modelMap != nil && modelMap[modelName] != "" {
+		modelName = modelMap[modelName]
+	}
+	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
+		return "", err, nil
 	}
 	jsonData, err := json.Marshal(convertedRequest)
 	if err != nil {
-		return err, nil
+		return "", err, nil
 	}
+	defer func() {
+		logContent := fmt.Sprintf("渠道 %s 测试成功，响应：%s", channel.Name, responseMessage)
+		if err != nil || openaiErr != nil {
+			errorMessage := ""
+			if err != nil {
+				errorMessage = err.Error()
+			} else {
+				errorMessage = openaiErr.Message
+			}
+			logContent = fmt.Sprintf("渠道 %s 测试失败，错误：%s", channel.Name, errorMessage)
+		}
+		go model.RecordTestLog(ctx, &model.Log{
+			ChannelId:   channel.Id,
+			ModelName:   modelName,
+			Content:     logContent,
+			ElapsedTime: helper.CalcElapsedTime(startTime),
+		})
+	}()
 	logger.SysLog(string(jsonData))
 	requestBody := bytes.NewBuffer(jsonData)
 	c.Request.Body = io.NopCloser(requestBody)
 	resp, err := adaptor.DoRequest(c, meta, requestBody)
 	if err != nil {
-		return err, nil
+		return "", err, nil
 	}
 	if resp != nil && resp.StatusCode != http.StatusOK {
 		err := controller.RelayErrorHandler(resp)
-		return fmt.Errorf("status code %d: %s", resp.StatusCode, err.Error.Message), &err.Error
+		errorMessage := err.Error.Message
+		if errorMessage != "" {
+			errorMessage = ", error message: " + errorMessage
+		}
+		return "", fmt.Errorf("http status code: %d%s", resp.StatusCode, errorMessage), &err.Error
 	}
 	usage, respErr := adaptor.DoResponse(c, resp, meta)
 	if respErr != nil {
-		return fmt.Errorf("%s", respErr.Error.Message), &respErr.Error
+		return "", fmt.Errorf("%s", respErr.Error.Message), &respErr.Error
 	}
 	if usage == nil {
-		return errors.New("usage is nil"), nil
+		return "", errors.New("usage is nil"), nil
+	}
+	rawResponse := w.Body.String()
+	_, responseMessage, err = parseTestResponse(rawResponse)
+	if err != nil {
+		return "", err, nil
 	}
 	result := w.Result()
 	// print result.Body
 	respBody, err := io.ReadAll(result.Body)
 	if err != nil {
-		return err, nil
+		return "", err, nil
 	}
 	logger.SysLog(fmt.Sprintf("testing channel #%d, response: \n%s", channel.Id, string(respBody)))
-	return nil, nil
+	return responseMessage, nil, nil
 }
 
 func TestChannel(c *gin.Context) {
+	ctx := c.Request.Context()
 	id, err := strconv.Atoi(c.Param("id"))
 	if err != nil {
 		c.JSON(http.StatusOK, gin.H{
@@ -139,24 +183,31 @@ func TestChannel(c *gin.Context) {
 		})
 		return
 	}
+	modelName := c.Query("model")
+	testRequest := buildTestRequest(modelName)
 	tik := time.Now()
-	err, _ = testChannel(channel)
+	responseMessage, err, _ := testChannel(ctx, 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 {
 		c.JSON(http.StatusOK, gin.H{
-			"success": false,
-			"message": err.Error(),
-			"time":    consumedTime,
+			"success":   false,
+			"message":   err.Error(),
+			"time":      consumedTime,
+			"modelName": modelName,
 		})
 		return
 	}
 	c.JSON(http.StatusOK, gin.H{
-		"success": true,
-		"message": "",
-		"time":    consumedTime,
+		"success":   true,
+		"message":   responseMessage,
+		"time":      consumedTime,
+		"modelName": modelName,
 	})
 	return
 }
@@ -164,7 +215,7 @@ func TestChannel(c *gin.Context) {
 var testAllChannelsLock sync.Mutex
 var testAllChannelsRunning bool = false
 
-func testChannels(notify bool, scope string) error {
+func testChannels(ctx context.Context, notify bool, scope string) error {
 	if config.RootUserEmail == "" {
 		config.RootUserEmail = model.GetRootUserEmail()
 	}
@@ -187,11 +238,12 @@ 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(ctx, 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 {
@@ -221,11 +273,12 @@ func testChannels(notify bool, scope string) error {
 }
 
 func TestChannels(c *gin.Context) {
+	ctx := c.Request.Context()
 	scope := c.Query("scope")
 	if scope == "" {
 		scope = "all"
 	}
-	err := testChannels(true, scope)
+	err := testChannels(ctx, true, scope)
 	if err != nil {
 		c.JSON(http.StatusOK, gin.H{
 			"success": false,
@@ -241,10 +294,11 @@ func TestChannels(c *gin.Context) {
 }
 
 func AutomaticallyTestChannels(frequency int) {
+	ctx := context.Background()
 	for {
 		time.Sleep(time.Duration(frequency) * time.Minute)
 		logger.SysLog("testing all channels")
-		_ = testChannels(false, "all")
+		_ = testChannels(ctx, false, "all")
 		logger.SysLog("channel test finished")
 	}
 }

controller/misc.go

@@ -18,24 +18,30 @@ func GetStatus(c *gin.Context) {
 		"success": true,
 		"message": "",
 		"data": gin.H{
-			"version":             common.Version,
-			"start_time":          common.StartTime,
-			"email_verification":  config.EmailVerificationEnabled,
-			"github_oauth":        config.GitHubOAuthEnabled,
-			"github_client_id":    config.GitHubClientId,
-			"lark_client_id":      config.LarkClientId,
-			"system_name":         config.SystemName,
-			"logo":                config.Logo,
-			"footer_html":         config.Footer,
-			"wechat_qrcode":       config.WeChatAccountQRCodeImageURL,
-			"wechat_login":        config.WeChatAuthEnabled,
-			"server_address":      config.ServerAddress,
-			"turnstile_check":     config.TurnstileCheckEnabled,
-			"turnstile_site_key":  config.TurnstileSiteKey,
-			"top_up_link":         config.TopUpLink,
-			"chat_link":           config.ChatLink,
-			"quota_per_unit":      config.QuotaPerUnit,
-			"display_in_currency": config.DisplayInCurrencyEnabled,
+			"version":                     common.Version,
+			"start_time":                  common.StartTime,
+			"email_verification":          config.EmailVerificationEnabled,
+			"github_oauth":                config.GitHubOAuthEnabled,
+			"github_client_id":            config.GitHubClientId,
+			"lark_client_id":              config.LarkClientId,
+			"system_name":                 config.SystemName,
+			"logo":                        config.Logo,
+			"footer_html":                 config.Footer,
+			"wechat_qrcode":               config.WeChatAccountQRCodeImageURL,
+			"wechat_login":                config.WeChatAuthEnabled,
+			"server_address":              config.ServerAddress,
+			"turnstile_check":             config.TurnstileCheckEnabled,
+			"turnstile_site_key":          config.TurnstileSiteKey,
+			"top_up_link":                 config.TopUpLink,
+			"chat_link":                   config.ChatLink,
+			"quota_per_unit":              config.QuotaPerUnit,
+			"display_in_currency":         config.DisplayInCurrencyEnabled,
+			"oidc":                        config.OidcEnabled,
+			"oidc_client_id":              config.OidcClientId,
+			"oidc_well_known":             config.OidcWellKnown,
+			"oidc_authorization_endpoint": config.OidcAuthorizationEndpoint,
+			"oidc_token_endpoint":         config.OidcTokenEndpoint,
+			"oidc_userinfo_endpoint":      config.OidcUserinfoEndpoint,
 		},
 	})
 	return

controller/relay.go

@@ -4,6 +4,9 @@ import (
 	"bytes"
 	"context"
 	"fmt"
+	"io"
+	"net/http"
+
 	"github.com/gin-gonic/gin"
 	"github.com/songquanpeng/one-api/common"
 	"github.com/songquanpeng/one-api/common/config"
@@ -16,8 +19,6 @@ import (
 	"github.com/songquanpeng/one-api/relay/controller"
 	"github.com/songquanpeng/one-api/relay/model"
 	"github.com/songquanpeng/one-api/relay/relaymode"
-	"io"
-	"net/http"
 )
 
 // https://platform.openai.com/docs/api-reference/chat
@@ -33,6 +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)
 	}
@@ -47,6 +50,7 @@ func Relay(c *gin.Context) {
 		logger.Debugf(ctx, "request body: %s", string(requestBody))
 	}
 	channelId := c.GetInt(ctxkey.ChannelId)
+	userId := c.GetInt(ctxkey.Id)
 	bizErr := relayHelper(c, relayMode)
 	if bizErr == nil {
 		monitor.Emit(channelId, true)
@@ -56,7 +60,7 @@ func Relay(c *gin.Context) {
 	channelName := c.GetString(ctxkey.ChannelName)
 	group := c.GetString(ctxkey.Group)
 	originalModel := c.GetString(ctxkey.OriginalModel)
-	go processChannelRelayError(ctx, channelId, channelName, bizErr)
+	go processChannelRelayError(ctx, userId, channelId, channelName, *bizErr)
 	requestId := c.GetString(helper.RequestIdKey)
 	retryTimes := config.RetryTimes
 	if !shouldRetry(c, bizErr.StatusCode) {
@@ -83,12 +87,14 @@ func Relay(c *gin.Context) {
 		channelId := c.GetInt(ctxkey.ChannelId)
 		lastFailedChannelId = channelId
 		channelName := c.GetString(ctxkey.ChannelName)
-		go processChannelRelayError(ctx, channelId, channelName, bizErr)
+		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,
@@ -115,8 +121,8 @@ func shouldRetry(c *gin.Context, statusCode int) bool {
 	return true
 }
 
-func processChannelRelayError(ctx context.Context, channelId int, channelName string, err *model.ErrorWithStatusCode) {
-	logger.Errorf(ctx, "relay error (channel #%d): %s", channelId, err.Message)
+func processChannelRelayError(ctx context.Context, userId int, channelId int, channelName string, err model.ErrorWithStatusCode) {
+	logger.Errorf(ctx, "relay error (channel id %d, user id: %d): %s", channelId, userId, err.Message)
 	// https://platform.openai.com/docs/guides/error-codes/api-errors
 	if monitor.ShouldDisableChannel(&err.Error, err.StatusCode) {
 		monitor.DisableChannel(channelId, channelName, err.Message)

controller/user.go

@@ -109,6 +109,7 @@ 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",
@@ -166,13 +167,14 @@ func Register(c *gin.Context) {
 	if config.EmailVerificationEnabled {
 		cleanUser.Email = user.Email
 	}
-	if err := cleanUser.Insert(inviterId); err != nil {
+	if err := cleanUser.Insert(ctx, inviterId); err != nil {
 		c.JSON(http.StatusOK, gin.H{
 			"success": false,
 			"message": err.Error(),
 		})
 		return
 	}
+
 	c.JSON(http.StatusOK, gin.H{
 		"success": true,
 		"message": "",
@@ -361,6 +363,7 @@ func GetSelf(c *gin.Context) {
 }
 
 func UpdateUser(c *gin.Context) {
+	ctx := c.Request.Context()
 	var updatedUser model.User
 	err := json.NewDecoder(c.Request.Body).Decode(&updatedUser)
 	if err != nil || updatedUser.Id == 0 {
@@ -415,7 +418,7 @@ func UpdateUser(c *gin.Context) {
 		return
 	}
 	if originUser.Quota != updatedUser.Quota {
-		model.RecordLog(originUser.Id, model.LogTypeManage, fmt.Sprintf("管理员将用户额度从 %s修改为 %s", common.LogQuota(originUser.Quota), common.LogQuota(updatedUser.Quota)))
+		model.RecordLog(ctx, originUser.Id, model.LogTypeManage, fmt.Sprintf("管理员将用户额度从 %s修改为 %s", common.LogQuota(originUser.Quota), common.LogQuota(updatedUser.Quota)))
 	}
 	c.JSON(http.StatusOK, gin.H{
 		"success": true,
@@ -534,6 +537,7 @@ func DeleteSelf(c *gin.Context) {
 }
 
 func CreateUser(c *gin.Context) {
+	ctx := c.Request.Context()
 	var user model.User
 	err := json.NewDecoder(c.Request.Body).Decode(&user)
 	if err != nil || user.Username == "" || user.Password == "" {
@@ -567,7 +571,7 @@ func CreateUser(c *gin.Context) {
 		Password:    user.Password,
 		DisplayName: user.DisplayName,
 	}
-	if err := cleanUser.Insert(0); err != nil {
+	if err := cleanUser.Insert(ctx, 0); err != nil {
 		c.JSON(http.StatusOK, gin.H{
 			"success": false,
 			"message": err.Error(),
@@ -746,6 +750,7 @@ type topUpRequest struct {
 }
 
 func TopUp(c *gin.Context) {
+	ctx := c.Request.Context()
 	req := topUpRequest{}
 	err := c.ShouldBindJSON(&req)
 	if err != nil {
@@ -756,7 +761,7 @@ func TopUp(c *gin.Context) {
 		return
 	}
 	id := c.GetInt("id")
-	quota, err := model.Redeem(req.Key, id)
+	quota, err := model.Redeem(ctx, req.Key, id)
 	if err != nil {
 		c.JSON(http.StatusOK, gin.H{
 			"success": false,
@@ -779,6 +784,7 @@ type adminTopUpRequest struct {
 }
 
 func AdminTopUp(c *gin.Context) {
+	ctx := c.Request.Context()
 	req := adminTopUpRequest{}
 	err := c.ShouldBindJSON(&req)
 	if err != nil {
@@ -799,7 +805,7 @@ func AdminTopUp(c *gin.Context) {
 	if req.Remark == "" {
 		req.Remark = fmt.Sprintf("通过 API 充值 %s", common.LogQuota(int64(req.Quota)))
 	}
-	model.RecordTopupLog(req.UserId, req.Remark, req.Quota)
+	model.RecordTopupLog(ctx, req.UserId, req.Remark, req.Quota)
 	c.JSON(http.StatusOK, gin.H{
 		"success": true,
 		"message": "",

go.mod

@@ -1,56 +1,71 @@
 module github.com/songquanpeng/one-api
 
-// +heroku goVersion go1.18
 go 1.20
 
 require (
-	github.com/aws/aws-sdk-go-v2 v1.26.1
-	github.com/aws/aws-sdk-go-v2/credentials v1.17.11
-	github.com/aws/aws-sdk-go-v2/service/bedrockruntime v1.7.4
-	github.com/gin-contrib/cors v1.7.1
-	github.com/gin-contrib/gzip v1.0.0
-	github.com/gin-contrib/sessions v1.0.0
-	github.com/gin-contrib/static v1.1.1
-	github.com/gin-gonic/gin v1.9.1
-	github.com/go-playground/validator/v10 v10.19.0
+	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
+	github.com/gin-contrib/cors v1.7.2
+	github.com/gin-contrib/gzip v1.0.1
+	github.com/gin-contrib/sessions v1.0.1
+	github.com/gin-contrib/static v1.1.2
+	github.com/gin-gonic/gin v1.10.0
+	github.com/go-playground/validator/v10 v10.20.0
 	github.com/go-redis/redis/v8 v8.11.5
 	github.com/golang-jwt/jwt v3.2.2+incompatible
 	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 v0.1.6
+	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 v0.22.0
-	golang.org/x/image v0.15.0
+	golang.org/x/crypto v0.31.0
+	golang.org/x/image v0.18.0
+	golang.org/x/sync v0.10.0
+	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
-	gorm.io/gorm v1.25.9
+	gorm.io/driver/sqlite v1.5.1
+	gorm.io/gorm v1.25.10
 )
 
 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 v1.3.5 // indirect
-	github.com/aws/aws-sdk-go-v2/internal/endpoints/v2 v2.6.5 // indirect
+	github.com/aws/aws-sdk-go-v2/internal/configsources v1.3.7 // indirect
+	github.com/aws/aws-sdk-go-v2/internal/endpoints/v2 v2.6.7 // indirect
 	github.com/aws/smithy-go v1.20.2 // indirect
-	github.com/bytedance/sonic v1.11.5 // indirect
+	github.com/bytedance/sonic v1.11.6 // indirect
 	github.com/bytedance/sonic/loader v0.1.1 // indirect
 	github.com/cespare/xxhash/v2 v2.3.0 // indirect
-	github.com/cloudwego/base64x v0.1.3 // indirect
+	github.com/cloudwego/base64x v0.1.4 // indirect
 	github.com/cloudwego/iasm v0.2.0 // indirect
 	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 v0.10.2 // 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
@@ -67,19 +82,29 @@ 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 v1.14.24 // 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 v2.2.1 // indirect
+	github.com/pelletier/go-toml/v2 v2.2.2 // indirect
 	github.com/pmezard/go-difflib v1.0.0 // indirect
 	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
-	golang.org/x/arch v0.7.0 // indirect
-	golang.org/x/net v0.24.0 // indirect
-	golang.org/x/sync v0.7.0 // indirect
-	golang.org/x/sys v0.19.0 // indirect
-	golang.org/x/text v0.14.0 // indirect
-	google.golang.org/protobuf v1.33.0 // 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 v0.26.0 // indirect
+	golang.org/x/oauth2 v0.21.0 // indirect
+	golang.org/x/sys v0.28.0 // indirect
+	golang.org/x/text v0.21.0 // 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

@@ -1,30 +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/aws/aws-sdk-go-v2 v1.26.1/go.mod h1:ffIFB97e2yNsv4aTSGkqtHnppsIJzw7G7BReUZ3jCXM=
+github.com/BurntSushi/toml v0.3.1/go.mod h1:xHWCNGjB5oqiDr8zfno3MHue2Ht5sIBksp03qcyfWMU=
+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/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/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/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/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 v1.3.7 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 v2.6.7 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 v1.8.3 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/loader v0.1.0/go.mod h1:UmRT+IRTGKz/DAkzcEGzyVqQFJ7H9BqwBO3pm9H/+HY=
+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 v0.1.1 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/cloudwego/base64x v0.1.3/go.mod h1:1+1K5BUHIQzyapgpF7LwvOGAEDicKtt1umPV+aN8pi8=
+github.com/client9/misspell v0.3.4/go.mod h1:qj6jICC3Q7zFZvVWo7KLAzC3yx5G7kyvSDkc90ppPyw=
+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=
@@ -33,43 +45,83 @@ 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/gzip v1.0.0 h1:UKN586Po/92IDX6ie5CWLgMI81obiIp5nSP85T3wlTk=
-github.com/gin-contrib/gzip v1.0.0/go.mod h1:CtG7tQrPB3vIBo6Gat9FVUsis+1emjvQqd66ME5TdnE=
-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/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 v1.0.1 h1:HQ8ENHODeLY7a4g1Au/46Z92bdGFl74OhxcZble9WJE=
+github.com/gin-contrib/gzip v1.0.1/go.mod h1:njt428fdUNRvjuJf16tZMYZ2Yl+WQB53X5wmhDwXvC4=
+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-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-contrib/static v1.1.2 h1:c3kT4bFkUJn2aoRU3s6XnMjJT8J6nNWJkR0NglqmlZ4=
+github.com/gin-contrib/static v1.1.2/go.mod h1:Fw90ozjHCmZBWbgrsqrDvO28YbhKEKzKp8GixhR4yLw=
+github.com/gin-gonic/gin v1.10.0 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=
 github.com/go-redis/redis/v8 v8.11.5/go.mod h1:gREzHqY1hg6oD9ngVRbLStwAWKhA0FEgq8Jd4h5lpwo=
 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=
@@ -94,6 +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=
@@ -109,8 +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 v2.0.3+incompatible/go.mod h1:FPy6KqzDD04eiIsT53CuJW3U88zkxoIYsOqkbpncsNc=
+github.com/mattn/go-sqlite3 v1.14.24 h1:tpSp2G2KyMnnQu99ngJ47EIkWVmliIizyZBfPrBWDRM=
+github.com/mattn/go-sqlite3 v1.14.24/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=
@@ -119,14 +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/pelletier/go-toml/v2 v2.2.1/go.mod h1:1t835xjRzz80PqgE6HHgN2JOsmgYu/h4qDAS4n929Rs=
+github.com/patrickmn/go-cache v2.1.0+incompatible h1:HRMgzkcYKYpi3C8ajMPV8OFXaaRUnok+kx1WdO15EQc=
+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=
@@ -148,26 +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/crypto v0.22.0 h1:g1v0xeRhjcugydODzvb3mEM9SQ0HGp9s/nh3COQ/C30=
-golang.org/x/crypto v0.22.0/go.mod h1:vr6Su+7cTlO45qkww3VDJlzDn0ctJvRgYbC2NvXHt+M=
-golang.org/x/image v0.15.0 h1:kOELfmgrmJlw4Cdb7g/QGuB3CvDrXbqEIww/pNtNBm8=
-golang.org/x/image v0.15.0/go.mod h1:HUYqC05R2ZcZ3ejNQsIHQDQiwWM4JBqmm6MKANTp4LE=
-golang.org/x/net v0.24.0 h1:1PcaxkF854Fu3+lvBIx5SYn9wRlBzzcnHZSiaFFAb0w=
-golang.org/x/net v0.24.0/go.mod h1:2Q7sJY5mzlzWjKtYUEXSlBWCdyaioyXzRB2RtU8KVE8=
-golang.org/x/sync v0.7.0 h1:YsImfSBoP9QPYL0xyKJPq0gcaJdG3rInoqxTWbfQu9M=
-golang.org/x/sync v0.7.0/go.mod h1:Czt+wKu1gCyEFDUtn0jG5QVvpJ6rzVqr5aXyt9drQfk=
+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 v0.0.0-20190308221718-c2843e01d9a2/go.mod h1:djNgcEr1/C05ACkg1iLfiJU5Ep61QUkGW8qpdssI0+w=
+golang.org/x/crypto v0.0.0-20200622213623-75b288015ac9/go.mod h1:LzIPMQfyMNhhGPhUkYOs5KpL4U8rLKemX1yGLhDgUto=
+golang.org/x/crypto v0.31.0 h1:ihbySMvVjLAeSH1IbfcRTkD/iNscyz8rGzjF/E5hV6U=
+golang.org/x/crypto v0.31.0/go.mod h1:kDsLvtWBEx7MV9tJOj9bnXsPbxwJQ6csT/x4KIN4Ssk=
+golang.org/x/exp v0.0.0-20190121172915-509febef88a4/go.mod h1:CJ0aWSM057203Lf6IL+f9T1iT9GByDxfZKAQTCR3kQA=
+golang.org/x/image v0.18.0 h1:jGzIakQa/ZXI1I0Fxvaa9W7yP25TqT6cHIHn+6CqvSQ=
+golang.org/x/image v0.18.0/go.mod h1:4yyo5vMFQjVjUcVk4jEQcU9MGy/rulF5WvUILseCM2E=
+golang.org/x/lint v0.0.0-20181026193005-c67002cb31c3/go.mod h1:UVdnD1Gm6xHRNCYTkRU2/jEulfH38KcIWyp/GAMgvoE=
+golang.org/x/lint v0.0.0-20190227174305-5b3e6a55c961/go.mod h1:wehouNa3lNwaWXcvxsM5YxQ5yQlVC4a0KAMCusXpPoU=
+golang.org/x/lint v0.0.0-20190313153728-d0100b6bd8b3/go.mod h1:6SW0HCj/g11FgYtHlgUYUwCkIfeOF89ocIRzGO/8vkc=
+golang.org/x/net v0.0.0-20180724234803-3673e40ba225/go.mod h1:mL1N/T3taQHkDXs73rZJwtUhF3w3ftmwwsq0BUmARs4=
+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 v0.10.0 h1:3NQrjDixjgGwUOCaF8w2+VYHv0Ve/vGYSbdkTa98gmQ=
+golang.org/x/sync v0.10.0/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 v0.19.0/go.mod h1:/VUhepiaJMQUp4+oa/7Zr1D23ma6VTLIYjOOTFZPUcA=
-golang.org/x/text v0.14.0 h1:ScX5w1eTa3QqT8oi6+ziP7dTV1S2+ALU0bI+0zXKWiQ=
-golang.org/x/text v0.14.0/go.mod h1:18ZOQIKpY8NJVqYksKHtTdi31H5itFRjB5/qKTNYzSU=
-golang.org/x/xerrors v0.0.0-20200804184101-5ec99f83aff1 h1:go1bK/D/BFZV2I8cIQd1NKEZ+0owSTG1fDTci4IqFcE=
-google.golang.org/protobuf v1.33.0 h1:uNO2rsAINq/JlFpSdYEKIZ0uKD/R9cpdv0T+yoGwGmI=
-google.golang.org/protobuf v1.33.0/go.mod h1:c6P6GXX6sHbq/GpV6MGZEdwhWPcYBgnhAHhKbcUYpos=
+golang.org/x/sys v0.28.0 h1:Fksou7UEQUWlKvIdsqzJmUmCX3cZuD2+P3XyyzwMhlA=
+golang.org/x/sys v0.28.0/go.mod h1:/VUhepiaJMQUp4+oa/7Zr1D23ma6VTLIYjOOTFZPUcA=
+golang.org/x/text v0.3.0/go.mod h1:NqM8EUOU14njkJ3fqMW+pc6Ldnwhi/IjpwHt7yyuwOQ=
+golang.org/x/text v0.3.3/go.mod h1:5Zoc/QRtKVWzQhOtBMvqHzDpF6irO9z98xDceosuGiQ=
+golang.org/x/text v0.21.0 h1:zyQAAkrwaneQ066sspRyJaG9VNi/YJ1NfzcGB3hZ/qo=
+golang.org/x/text v0.21.0/go.mod h1:4IBbMaMmOPCJ8SecivzSH54+73PCFmPWxNTLm+vZkEQ=
+golang.org/x/time v0.5.0 h1:o7cqy6amK/52YcAKIPlM3a+Fpj35zvRj2TP+e1xFSfk=
+golang.org/x/time v0.5.0/go.mod h1:3BpzKBy/shNhVucY/MWOyx10tF3SFh9QdLuxbVysPQM=
+golang.org/x/tools v0.0.0-20180917221912-90fa682c2a6e/go.mod h1:n7NCudcB/nEzxVGmLbDWY5pfWTLqBcC2KZ6jyYvM4mQ=
+golang.org/x/tools v0.0.0-20190114222345-bf090417da8b/go.mod h1:n7NCudcB/nEzxVGmLbDWY5pfWTLqBcC2KZ6jyYvM4mQ=
+golang.org/x/tools v0.0.0-20190226205152-f727befe758c/go.mod h1:9Yl7xja0Znq3iFh3HoIrodX9oNMXvdceNzlUR8zjMvY=
+golang.org/x/tools v0.0.0-20190311212946-11955173bddd/go.mod h1:LCzVGOaR6xXOjkQ3onu1FJEFr0SW1gC7cKk1uF8kGRs=
+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=
@@ -179,10 +306,12 @@ gorm.io/driver/mysql v1.5.6 h1:Ld4mkIickM+EliaQZQx3uOJDJHtrd70MxAUqWqlx3Y8=
 gorm.io/driver/mysql v1.5.6/go.mod h1:sEtPWMiqiN1N1cMXoXmBbd8C6/l+TESwriotuRRpkDM=
 gorm.io/driver/postgres v1.5.7 h1:8ptbNJTDbEmhdr62uReG5BGkdQyeasu/FZHxI0IMGnM=
 gorm.io/driver/postgres v1.5.7/go.mod h1:3e019WlBaYI5o5LIdNV+LyxCMNtLOQETBXL2h4chKpA=
-gorm.io/driver/sqlite v1.5.5 h1:7MDMtUZhV065SilG62E0MquljeArQZNfJnjd9i9gx3E=
-gorm.io/driver/sqlite v1.5.5/go.mod h1:6NgQ7sQWAIFsPrJJl1lSNSu2TABh0ZZ/zm5fosATavE=
+gorm.io/driver/sqlite v1.5.1 h1:hYyrLkAWE71bcarJDPdZNTLWtr8XrSjOWyjUYI6xdL4=
+gorm.io/driver/sqlite v1.5.1/go.mod h1:7MZZ2Z8bqyfSQA1gYEV6MagQWj3cpUkJj9Z+d1HEMEQ=
 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

@@ -6,7 +6,9 @@ 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"
 	"github.com/songquanpeng/one-api/common/logger"
 	"github.com/songquanpeng/one-api/controller"
@@ -22,29 +24,22 @@ import (
 var buildFS embed.FS
 
 func main() {
+	common.Init()
 	logger.SetupLogger()
-	logger.SysLog(fmt.Sprintf("One API %s started", common.Version))
-	if os.Getenv("GIN_MODE") != "debug" {
+	logger.SysLogf("One API %s started", common.Version)
+
+	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")
-	if err != nil {
-		logger.FatalLog("failed to initialize database: " + 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
-	}
+	model.InitDB()
+	model.InitLogDB()
+
+	var err error
 	err = model.CreateRootAccountIfNeed()
 	if err != nil {
 		logger.FatalLog("database init error: " + err.Error())
@@ -94,6 +89,7 @@ func main() {
 		logger.SysLog("metric enabled, will disable channel if too much request failed")
 	}
 	openai.InitTokenEncoders()
+	client.Init()
 
 	// Initialize HTTP server
 	server := gin.New()
@@ -111,6 +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

@@ -2,21 +2,24 @@ package middleware
 
 import (
 	"fmt"
+	"net/http"
+	"strconv"
+
 	"github.com/gin-gonic/gin"
+
 	"github.com/songquanpeng/one-api/common/ctxkey"
 	"github.com/songquanpeng/one-api/common/logger"
 	"github.com/songquanpeng/one-api/model"
 	"github.com/songquanpeng/one-api/relay/channeltype"
-	"net/http"
-	"strconv"
 )
 
 type ModelRequest struct {
-	Model string `json:"model"`
+	Model string `json:"model" form:"model"`
 }
 
 func Distribute() func(c *gin.Context) {
 	return func(c *gin.Context) {
+		ctx := c.Request.Context()
 		userId := c.GetInt(ctxkey.Id)
 		userGroup, _ := model.CacheGetUserGroup(userId)
 		c.Set(ctxkey.Group, userGroup)
@@ -52,6 +55,7 @@ func Distribute() func(c *gin.Context) {
 				return
 			}
 		}
+		logger.Debugf(ctx, "user id %d, user group: %s, request model: %s, using channel #%d", userId, userGroup, requestModel, channel.Id)
 		SetupContextForSelectedChannel(c, channel, requestModel)
 		c.Next()
 	}
@@ -61,32 +65,37 @@ func SetupContextForSelectedChannel(c *gin.Context, channel *model.Channel, mode
 	c.Set(ctxkey.Channel, channel.Type)
 	c.Set(ctxkey.ChannelId, channel.Id)
 	c.Set(ctxkey.ChannelName, channel.Name)
+	if channel.SystemPrompt != nil && *channel.SystemPrompt != "" {
+		c.Set(ctxkey.SystemPrompt, *channel.SystemPrompt)
+	}
 	c.Set(ctxkey.ModelMapping, channel.GetModelMapping())
 	c.Set(ctxkey.OriginalModel, modelName) // for retry
 	c.Request.Header.Set("Authorization", fmt.Sprintf("Bearer %s", channel.Key))
 	c.Set(ctxkey.BaseURL, channel.GetBaseURL())
 	cfg, _ := channel.LoadConfig()
 	// this is for backward compatibility
-	switch channel.Type {
-	case channeltype.Azure:
-		if cfg.APIVersion == "" {
-			cfg.APIVersion = channel.Other
-		}
-	case channeltype.Xunfei:
-		if cfg.APIVersion == "" {
-			cfg.APIVersion = channel.Other
-		}
-	case channeltype.Gemini:
-		if cfg.APIVersion == "" {
-			cfg.APIVersion = channel.Other
-		}
-	case channeltype.AIProxyLibrary:
-		if cfg.LibraryID == "" {
-			cfg.LibraryID = channel.Other
-		}
-	case channeltype.Ali:
-		if cfg.Plugin == "" {
-			cfg.Plugin = channel.Other
+	if channel.Other != nil {
+		switch channel.Type {
+		case channeltype.Azure:
+			if cfg.APIVersion == "" {
+				cfg.APIVersion = *channel.Other
+			}
+		case channeltype.Xunfei:
+			if cfg.APIVersion == "" {
+				cfg.APIVersion = *channel.Other
+			}
+		case channeltype.Gemini:
+			if cfg.APIVersion == "" {
+				cfg.APIVersion = *channel.Other
+			}
+		case channeltype.AIProxyLibrary:
+			if cfg.LibraryID == "" {
+				cfg.LibraryID = *channel.Other
+			}
+		case channeltype.Ali:
+			if cfg.Plugin == "" {
+				cfg.Plugin = *channel.Other
+			}
 		}
 	}
 	c.Set(ctxkey.Config, cfg)

middleware/gzip.go

@@ -0,0 +1,27 @@
+package middleware
+
+import (
+	"compress/gzip"
+	"github.com/gin-gonic/gin"
+	"io"
+	"net/http"
+)
+
+func GzipDecodeMiddleware() gin.HandlerFunc {
+	return func(c *gin.Context) {
+		if c.GetHeader("Content-Encoding") == "gzip" {
+			gzipReader, err := gzip.NewReader(c.Request.Body)
+			if err != nil {
+				c.AbortWithStatus(http.StatusBadRequest)
+				return
+			}
+			defer gzipReader.Close()
+
+			// Replace the request body with the decompressed data
+			c.Request.Body = io.NopCloser(gzipReader)
+		}
+
+		// Continue processing the request
+		c.Next()
+	}
+}

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)

middleware/request-id.go

@@ -1,8 +1,8 @@
 package middleware
 
 import (
-	"context"
 	"github.com/gin-gonic/gin"
+
 	"github.com/songquanpeng/one-api/common/helper"
 )
 
@@ -10,7 +10,7 @@ func RequestId() func(c *gin.Context) {
 	return func(c *gin.Context) {
 		id := helper.GenRequestID()
 		c.Set(helper.RequestIdKey, id)
-		ctx := context.WithValue(c.Request.Context(), helper.RequestIdKey, id)
+		ctx := helper.SetRequestID(c.Request.Context(), id)
 		c.Request = c.Request.WithContext(ctx)
 		c.Header(helper.RequestIdKey, id)
 		c.Next()

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"`
@@ -36,16 +37,19 @@ type Channel struct {
 	ModelMapping       *string `json:"model_mapping" gorm:"type:varchar(1024);default:''"`
 	Priority           *int64  `json:"priority" gorm:"bigint;default:0"`
 	Config             string  `json:"config"`
+	SystemPrompt       *string `json:"system_prompt" gorm:"type:text"`
 }
 
 type ChannelConfig struct {
-	Region     string `json:"region,omitempty"`
-	SK         string `json:"sk,omitempty"`
-	AK         string `json:"ak,omitempty"`
-	UserID     string `json:"user_id,omitempty"`
-	APIVersion string `json:"api_version,omitempty"`
-	LibraryID  string `json:"library_id,omitempty"`
-	Plugin     string `json:"plugin,omitempty"`
+	Region            string `json:"region,omitempty"`
+	SK                string `json:"sk,omitempty"`
+	AK                string `json:"ak,omitempty"`
+	UserID            string `json:"user_id,omitempty"`
+	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/log.go

@@ -3,26 +3,32 @@ package model
 import (
 	"context"
 	"fmt"
+
+	"gorm.io/gorm"
+
 	"github.com/songquanpeng/one-api/common"
 	"github.com/songquanpeng/one-api/common/config"
 	"github.com/songquanpeng/one-api/common/helper"
 	"github.com/songquanpeng/one-api/common/logger"
-	"gorm.io/gorm"
 )
 
 type Log struct {
-	Id               int    `json:"id"`
-	UserId           int    `json:"user_id" gorm:"index"`
-	CreatedAt        int64  `json:"created_at" gorm:"bigint;index:idx_created_at_type"`
-	Type             int    `json:"type" gorm:"index:idx_created_at_type"`
-	Content          string `json:"content"`
-	Username         string `json:"username" gorm:"index:index_username_model_name,priority:2;default:''"`
-	TokenName        string `json:"token_name" gorm:"index;default:''"`
-	ModelName        string `json:"model_name" gorm:"index;index:index_username_model_name,priority:1;default:''"`
-	Quota            int    `json:"quota" gorm:"default:0"`
-	PromptTokens     int    `json:"prompt_tokens" gorm:"default:0"`
-	CompletionTokens int    `json:"completion_tokens" gorm:"default:0"`
-	ChannelId        int    `json:"channel" gorm:"index"`
+	Id                int    `json:"id"`
+	UserId            int    `json:"user_id" gorm:"index"`
+	CreatedAt         int64  `json:"created_at" gorm:"bigint;index:idx_created_at_type"`
+	Type              int    `json:"type" gorm:"index:idx_created_at_type"`
+	Content           string `json:"content"`
+	Username          string `json:"username" gorm:"index:index_username_model_name,priority:2;default:''"`
+	TokenName         string `json:"token_name" gorm:"index;default:''"`
+	ModelName         string `json:"model_name" gorm:"index;index:index_username_model_name,priority:1;default:''"`
+	Quota             int    `json:"quota" gorm:"default:0"`
+	PromptTokens      int    `json:"prompt_tokens" gorm:"default:0"`
+	CompletionTokens  int    `json:"completion_tokens" gorm:"default:0"`
+	ChannelId         int    `json:"channel" gorm:"index"`
+	RequestId         string `json:"request_id" gorm:"default:''"`
+	ElapsedTime       int64  `json:"elapsed_time" gorm:"default:0"` // unit is ms
+	IsStream          bool   `json:"is_stream" gorm:"default:false"`
+	SystemPromptReset bool   `json:"system_prompt_reset" gorm:"default:false"`
 }
 
 const (
@@ -31,9 +37,21 @@ const (
 	LogTypeConsume
 	LogTypeManage
 	LogTypeSystem
+	LogTypeTest
 )
 
-func RecordLog(userId int, logType int, content string) {
+func recordLogHelper(ctx context.Context, log *Log) {
+	requestId := helper.GetRequestID(ctx)
+	log.RequestId = requestId
+	err := LOG_DB.Create(log).Error
+	if err != nil {
+		logger.Error(ctx, "failed to record log: "+err.Error())
+		return
+	}
+	logger.Infof(ctx, "record log: %+v", log)
+}
+
+func RecordLog(ctx context.Context, userId int, logType int, content string) {
 	if logType == LogTypeConsume && !config.LogConsumeEnabled {
 		return
 	}
@@ -44,13 +62,10 @@ func RecordLog(userId int, logType int, content string) {
 		Type:      logType,
 		Content:   content,
 	}
-	err := LOG_DB.Create(log).Error
-	if err != nil {
-		logger.SysError("failed to record log: " + err.Error())
-	}
+	recordLogHelper(ctx, log)
 }
 
-func RecordTopupLog(userId int, content string, quota int) {
+func RecordTopupLog(ctx context.Context, userId int, content string, quota int) {
 	log := &Log{
 		UserId:    userId,
 		Username:  GetUsernameById(userId),
@@ -59,34 +74,23 @@ func RecordTopupLog(userId int, content string, quota int) {
 		Content:   content,
 		Quota:     quota,
 	}
-	err := LOG_DB.Create(log).Error
-	if err != nil {
-		logger.SysError("failed to record log: " + err.Error())
-	}
+	recordLogHelper(ctx, log)
 }
 
-func RecordConsumeLog(ctx context.Context, userId int, channelId int, promptTokens int, completionTokens int, modelName string, tokenName string, quota int64, content string) {
-	logger.Info(ctx, fmt.Sprintf("record consume log: userId=%d, channelId=%d, promptTokens=%d, completionTokens=%d, modelName=%s, tokenName=%s, quota=%d, content=%s", userId, channelId, promptTokens, completionTokens, modelName, tokenName, quota, content))
+func RecordConsumeLog(ctx context.Context, log *Log) {
 	if !config.LogConsumeEnabled {
 		return
 	}
-	log := &Log{
-		UserId:           userId,
-		Username:         GetUsernameById(userId),
-		CreatedAt:        helper.GetTimestamp(),
-		Type:             LogTypeConsume,
-		Content:          content,
-		PromptTokens:     promptTokens,
-		CompletionTokens: completionTokens,
-		TokenName:        tokenName,
-		ModelName:        modelName,
-		Quota:            int(quota),
-		ChannelId:        channelId,
-	}
-	err := LOG_DB.Create(log).Error
-	if err != nil {
-		logger.Error(ctx, "failed to record log: "+err.Error())
-	}
+	log.Username = GetUsernameById(log.UserId)
+	log.CreatedAt = helper.GetTimestamp()
+	log.Type = LogTypeConsume
+	recordLogHelper(ctx, log)
+}
+
+func RecordTestLog(ctx context.Context, log *Log) {
+	log.CreatedAt = helper.GetTimestamp()
+	log.Type = LogTypeTest
+	recordLogHelper(ctx, log)
 }
 
 func GetAllLogs(logType int, startTimestamp int64, endTimestamp int64, modelName string, username string, tokenName string, startIdx int, num int, channel int) (logs []*Log, err error) {
@@ -152,7 +156,11 @@ func SearchUserLogs(userId int, keyword string) (logs []*Log, err error) {
 }
 
 func SumUsedQuota(logType int, startTimestamp int64, endTimestamp int64, modelName string, username string, tokenName string, channel int) (quota int64) {
-	tx := LOG_DB.Table("logs").Select("ifnull(sum(quota),0)")
+	ifnull := "ifnull"
+	if common.UsingPostgreSQL {
+		ifnull = "COALESCE"
+	}
+	tx := LOG_DB.Table("logs").Select(fmt.Sprintf("%s(sum(quota),0)", ifnull))
 	if username != "" {
 		tx = tx.Where("username = ?", username)
 	}
@@ -176,7 +184,11 @@ func SumUsedQuota(logType int, startTimestamp int64, endTimestamp int64, modelNa
 }
 
 func SumUsedToken(logType int, startTimestamp int64, endTimestamp int64, modelName string, username string, tokenName string) (token int) {
-	tx := LOG_DB.Table("logs").Select("ifnull(sum(prompt_tokens),0) + ifnull(sum(completion_tokens),0)")
+	ifnull := "ifnull"
+	if common.UsingPostgreSQL {
+		ifnull = "COALESCE"
+	}
+	tx := LOG_DB.Table("logs").Select(fmt.Sprintf("%s(sum(prompt_tokens),0) + %s(sum(completion_tokens),0)", ifnull, ifnull))
 	if username != "" {
 		tx = tx.Where("username = ?", username)
 	}

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)
@@ -60,90 +65,156 @@ func CreateRootAccountIfNeed() error {
 }
 
 func chooseDB(envName string) (*gorm.DB, error) {
-	if os.Getenv(envName) != "" {
-		dsn := os.Getenv(envName)
-		if strings.HasPrefix(dsn, "postgres://") {
-			// Use PostgreSQL
-			logger.SysLog("using PostgreSQL as database")
-			common.UsingPostgreSQL = true
-			return gorm.Open(postgres.New(postgres.Config{
-				DSN:                  dsn,
-				PreferSimpleProtocol: true, // disables implicit prepared statement usage
-			}), &gorm.Config{
-				PrepareStmt: true, // precompile SQL
-			})
-		}
+	dsn := os.Getenv(envName)
+
+	switch {
+	case strings.HasPrefix(dsn, "postgres://"):
+		// Use PostgreSQL
+		return openPostgreSQL(dsn)
+	case dsn != "":
 		// Use MySQL
-		logger.SysLog("using MySQL as database")
-		common.UsingMySQL = true
-		return gorm.Open(mysql.Open(dsn), &gorm.Config{
-			PrepareStmt: true, // precompile SQL
-		})
+		return openMySQL(dsn)
+	default:
+		// Use SQLite
+		return openSQLite()
 	}
-	// Use SQLite
-	logger.SysLog("SQL_DSN not set, using SQLite as database")
-	common.UsingSQLite = true
-	config := fmt.Sprintf("?_busy_timeout=%d", common.SQLiteBusyTimeout)
-	return gorm.Open(sqlite.Open(common.SQLitePath+config), &gorm.Config{
+}
+
+func openPostgreSQL(dsn string) (*gorm.DB, error) {
+	logger.SysLog("using PostgreSQL as database")
+	common.UsingPostgreSQL = true
+	return gorm.Open(postgres.New(postgres.Config{
+		DSN:                  dsn,
+		PreferSimpleProtocol: true, // disables implicit prepared statement usage
+	}), &gorm.Config{
 		PrepareStmt: true, // precompile SQL
 	})
 }
 
-func InitDB(envName string) (db *gorm.DB, err error) {
-	db, err = chooseDB(envName)
-	if err == nil {
-		if config.DebugSQLEnabled {
-			db = db.Debug()
-		}
-		sqlDB, err := db.DB()
-		if err != nil {
-			return nil, err
-		}
-		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)))
+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
+	})
+}
 
-		if !config.IsMasterNode {
-			return db, err
-		}
-		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 != nil {
-			return nil, err
-		}
-		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 openSQLite() (*gorm.DB, error) {
+	logger.SysLog("SQL_DSN not set, using SQLite as database")
+	common.UsingSQLite = true
+	dsn := fmt.Sprintf("%s?_busy_timeout=%d", common.SQLitePath, common.SQLiteBusyTimeout)
+	return gorm.Open(sqlite.Open(dsn), &gorm.Config{
+		PrepareStmt: true, // precompile SQL
+	})
+}
+
+func InitDB() {
+	var err error
+	DB, err = chooseDB("SQL_DSN")
+	if err != nil {
+		logger.FatalLog("failed to initialize database: " + err.Error())
+		return
 	}
-	return db, err
+
+	sqlDB := setDBConns(DB)
+
+	if !config.IsMasterNode {
+		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")
+	if err = migrateDB(); err != nil {
+		logger.FatalLog("failed to migrate database: " + err.Error())
+		return
+	}
+	logger.SysLog("database migrated")
+}
+
+func migrateDB() error {
+	var err error
+	if err = DB.AutoMigrate(&Channel{}); err != nil {
+		return 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/option.go

@@ -28,6 +28,7 @@ func InitOptionMap() {
 	config.OptionMap["PasswordRegisterEnabled"] = strconv.FormatBool(config.PasswordRegisterEnabled)
 	config.OptionMap["EmailVerificationEnabled"] = strconv.FormatBool(config.EmailVerificationEnabled)
 	config.OptionMap["GitHubOAuthEnabled"] = strconv.FormatBool(config.GitHubOAuthEnabled)
+	config.OptionMap["OidcEnabled"] = strconv.FormatBool(config.OidcEnabled)
 	config.OptionMap["WeChatAuthEnabled"] = strconv.FormatBool(config.WeChatAuthEnabled)
 	config.OptionMap["TurnstileCheckEnabled"] = strconv.FormatBool(config.TurnstileCheckEnabled)
 	config.OptionMap["RegisterEnabled"] = strconv.FormatBool(config.RegisterEnabled)
@@ -130,6 +131,8 @@ func updateOptionMap(key string, value string) (err error) {
 			config.EmailVerificationEnabled = boolValue
 		case "GitHubOAuthEnabled":
 			config.GitHubOAuthEnabled = boolValue
+		case "OidcEnabled":
+			config.OidcEnabled = boolValue
 		case "WeChatAuthEnabled":
 			config.WeChatAuthEnabled = boolValue
 		case "TurnstileCheckEnabled":
@@ -176,6 +179,18 @@ func updateOptionMap(key string, value string) (err error) {
 		config.LarkClientId = value
 	case "LarkClientSecret":
 		config.LarkClientSecret = value
+	case "OidcClientId":
+		config.OidcClientId = value
+	case "OidcClientSecret":
+		config.OidcClientSecret = value
+	case "OidcWellKnown":
+		config.OidcWellKnown = value
+	case "OidcAuthorizationEndpoint":
+		config.OidcAuthorizationEndpoint = value
+	case "OidcTokenEndpoint":
+		config.OidcTokenEndpoint = value
+	case "OidcUserinfoEndpoint":
+		config.OidcUserinfoEndpoint = value
 	case "Footer":
 		config.Footer = value
 	case "SystemName":

model/redemption.go

@@ -1,11 +1,14 @@
 package model
 
 import (
+	"context"
 	"errors"
 	"fmt"
+
+	"gorm.io/gorm"
+
 	"github.com/songquanpeng/one-api/common"
 	"github.com/songquanpeng/one-api/common/helper"
-	"gorm.io/gorm"
 )
 
 const (
@@ -48,7 +51,7 @@ func GetRedemptionById(id int) (*Redemption, error) {
 	return &redemption, err
 }
 
-func Redeem(key string, userId int) (quota int64, err error) {
+func Redeem(ctx context.Context, key string, userId int) (quota int64, err error) {
 	if key == "" {
 		return 0, errors.New("未提供兑换码")
 	}
@@ -82,7 +85,7 @@ func Redeem(key string, userId int) (quota int64, err error) {
 	if err != nil {
 		return 0, errors.New("兑换失败，" + err.Error())
 	}
-	RecordLog(userId, LogTypeTopup, fmt.Sprintf("通过兑换码充值 %s", common.LogQuota(redemption.Quota)))
+	RecordLog(ctx, userId, LogTypeTopup, fmt.Sprintf("通过兑换码充值 %s", common.LogQuota(redemption.Quota)))
 	return redemption.Quota, nil
 }
 

model/token.go

@@ -30,7 +30,7 @@ type Token struct {
 	RemainQuota    int64   `json:"remain_quota" gorm:"bigint;default:0"`
 	UnlimitedQuota bool    `json:"unlimited_quota" gorm:"default:false"`
 	UsedQuota      int64   `json:"used_quota" gorm:"bigint;default:0"` // used quota
-	Models         *string `json:"models" gorm:"default:''"`           // allowed models
+	Models         *string `json:"models" gorm:"type:text"`            // allowed models
 	Subnet         *string `json:"subnet" gorm:"default:''"`           // allowed subnet
 }
 
@@ -121,30 +121,40 @@ func GetTokenById(id int) (*Token, error) {
 	return &token, err
 }
 
-func (token *Token) Insert() error {
+func (t *Token) Insert() error {
 	var err error
-	err = DB.Create(token).Error
+	err = DB.Create(t).Error
 	return err
 }
 
 // Update Make sure your token's fields is completed, because this will update non-zero values
-func (token *Token) Update() error {
+func (t *Token) Update() error {
 	var err error
-	err = DB.Model(token).Select("name", "status", "expired_time", "remain_quota", "unlimited_quota", "models", "subnet").Updates(token).Error
+	err = DB.Model(t).Select("name", "status", "expired_time", "remain_quota", "unlimited_quota", "models", "subnet").Updates(t).Error
 	return err
 }
 
-func (token *Token) SelectUpdate() error {
+func (t *Token) SelectUpdate() error {
 	// This can update zero values
-	return DB.Model(token).Select("accessed_time", "status").Updates(token).Error
+	return DB.Model(t).Select("accessed_time", "status").Updates(t).Error
 }
 
-func (token *Token) Delete() error {
+func (t *Token) Delete() error {
 	var err error
-	err = DB.Delete(token).Error
+	err = DB.Delete(t).Error
 	return err
 }
 
+func (t *Token) GetModels() string {
+	if t == nil {
+		return ""
+	}
+	if t.Models == nil {
+		return ""
+	}
+	return *t.Models
+}
+
 func DeleteTokenById(id int, userId int) (err error) {
 	// Why we need userId here? In case user want to delete other's token.
 	if id == 0 || userId == 0 {
@@ -254,14 +264,14 @@ func PreConsumeTokenQuota(tokenId int, quota int64) (err error) {
 
 func PostConsumeTokenQuota(tokenId int, quota int64) (err error) {
 	token, err := GetTokenById(tokenId)
+	if err != nil {
+		return err
+	}
 	if quota > 0 {
 		err = DecreaseUserQuota(token.UserId, quota)
 	} else {
 		err = IncreaseUserQuota(token.UserId, -quota)
 	}
-	if err != nil {
-		return err
-	}
 	if !token.UnlimitedQuota {
 		if quota > 0 {
 			err = DecreaseTokenQuota(tokenId, quota)

model/user.go

@@ -1,15 +1,19 @@
 package model
 
 import (
+	"context"
 	"errors"
 	"fmt"
+	"strings"
+
+	"gorm.io/gorm"
+
 	"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"
-	"strings"
 )
 
 const (
@@ -38,6 +42,7 @@ type User struct {
 	GitHubId         string `json:"github_id" gorm:"column:github_id;index"`
 	WeChatId         string `json:"wechat_id" gorm:"column:wechat_id;index"`
 	LarkId           string `json:"lark_id" gorm:"column:lark_id;index"`
+	OidcId           string `json:"oidc_id" gorm:"column:oidc_id;index"`
 	VerificationCode string `json:"verification_code" gorm:"-:all"`                                    // this field is only for Email verification, don't save it to database!
 	AccessToken      string `json:"access_token" gorm:"type:char(32);column:access_token;uniqueIndex"` // this token is for system management
 	Quota            int64  `json:"quota" gorm:"bigint;default:0"`
@@ -90,7 +95,7 @@ func GetUserById(id int, selectAll bool) (*User, error) {
 	if selectAll {
 		err = DB.First(&user, "id = ?", id).Error
 	} else {
-		err = DB.Omit("password").First(&user, "id = ?", id).Error
+		err = DB.Omit("password", "access_token").First(&user, "id = ?", id).Error
 	}
 	return &user, err
 }
@@ -112,7 +117,7 @@ func DeleteUserById(id int) (err error) {
 	return user.Delete()
 }
 
-func (user *User) Insert(inviterId int) error {
+func (user *User) Insert(ctx context.Context, inviterId int) error {
 	var err error
 	if user.Password != "" {
 		user.Password, err = common.Password2Hash(user.Password)
@@ -128,18 +133,34 @@ func (user *User) Insert(inviterId int) error {
 		return result.Error
 	}
 	if config.QuotaForNewUser > 0 {
-		RecordLog(user.Id, LogTypeSystem, fmt.Sprintf("新用户注册赠送 %s", common.LogQuota(config.QuotaForNewUser)))
+		RecordLog(ctx, user.Id, LogTypeSystem, fmt.Sprintf("新用户注册赠送 %s", common.LogQuota(config.QuotaForNewUser)))
 	}
 	if inviterId != 0 {
 		if config.QuotaForInvitee > 0 {
 			_ = IncreaseUserQuota(user.Id, config.QuotaForInvitee)
-			RecordLog(user.Id, LogTypeSystem, fmt.Sprintf("使用邀请码赠送 %s", common.LogQuota(config.QuotaForInvitee)))
+			RecordLog(ctx, user.Id, LogTypeSystem, fmt.Sprintf("使用邀请码赠送 %s", common.LogQuota(config.QuotaForInvitee)))
 		}
 		if config.QuotaForInviter > 0 {
 			_ = IncreaseUserQuota(inviterId, config.QuotaForInviter)
-			RecordLog(inviterId, LogTypeSystem, fmt.Sprintf("邀请用户赠送 %s", common.LogQuota(config.QuotaForInviter)))
+			RecordLog(ctx, 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
 }
 
@@ -228,6 +249,14 @@ func (user *User) FillUserByLarkId() error {
 	return nil
 }
 
+func (user *User) FillUserByOidcId() error {
+	if user.OidcId == "" {
+		return errors.New("oidc id 为空！")
+	}
+	DB.Where(User{OidcId: user.OidcId}).First(user)
+	return nil
+}
+
 func (user *User) FillUserByWeChatId() error {
 	if user.WeChatId == "" {
 		return errors.New("WeChat id 为空！")
@@ -260,6 +289,10 @@ func IsLarkIdAlreadyTaken(githubId string) bool {
 	return DB.Where("lark_id = ?", githubId).Find(&User{}).RowsAffected == 1
 }
 
+func IsOidcIdAlreadyTaken(oidcId string) bool {
+	return DB.Where("oidc_id = ?", oidcId).Find(&User{}).RowsAffected == 1
+}
+
 func IsUsernameAlreadyTaken(username string) bool {
 	return DB.Where("username = ?", username).Find(&User{}).RowsAffected == 1
 }

monitor/manage.go

@@ -1,10 +1,11 @@
 package monitor
 
 import (
-	"github.com/songquanpeng/one-api/common/config"
-	"github.com/songquanpeng/one-api/relay/model"
 	"net/http"
 	"strings"
+
+	"github.com/songquanpeng/one-api/common/config"
+	"github.com/songquanpeng/one-api/relay/model"
 )
 
 func ShouldDisableChannel(err *model.Error, statusCode int) bool {
@@ -18,31 +19,23 @@ func ShouldDisableChannel(err *model.Error, statusCode int) bool {
 		return true
 	}
 	switch err.Type {
-	case "insufficient_quota":
-		return true
-	// https://docs.anthropic.com/claude/reference/errors
-	case "authentication_error":
-		return true
-	case "permission_error":
-		return true
-	case "forbidden":
+	case "insufficient_quota", "authentication_error", "permission_error", "forbidden":
 		return true
 	}
 	if err.Code == "invalid_api_key" || err.Code == "account_deactivated" {
 		return true
 	}
-	if strings.HasPrefix(err.Message, "Your credit balance is too low") { // anthropic
-		return true
-	} else if strings.HasPrefix(err.Message, "This organization has been disabled.") {
-		return true
-	}
-	//if strings.Contains(err.Message, "quota") {
-	//	return true
-	//}
-	if strings.Contains(err.Message, "credit") {
-		return true
-	}
-	if strings.Contains(err.Message, "balance") {
+
+	lowerMessage := strings.ToLower(err.Message)
+	if strings.Contains(lowerMessage, "your access was terminated") ||
+		strings.Contains(lowerMessage, "violation of our policies") ||
+		strings.Contains(lowerMessage, "your credit balance is too low") ||
+		strings.Contains(lowerMessage, "organization has been disabled") ||
+		strings.Contains(lowerMessage, "credit") ||
+		strings.Contains(lowerMessage, "balance") ||
+		strings.Contains(lowerMessage, "permission denied") ||
+		strings.Contains(lowerMessage, "organization has been restricted") || // groq
+		strings.Contains(lowerMessage, "已欠费") {
 		return true
 	}
 	return false

relay/adaptor.go

@@ -15,7 +15,10 @@ 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/replicate"
 	"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 +58,12 @@ 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{}
+	case apitype.Replicate:
+		return &replicate.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)
-	go func() {
-		for scanner.Scan() {
-			data := scanner.Text()
-			if len(data) < 5 { // ignore blank line or wrong format
-				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
-			}
-			if len(AIProxyLibraryResponse.Documents) != 0 {
-				documents = AIProxyLibraryResponse.Documents
-			}
-			response := streamResponseAIProxyLibrary2OpenAI(&AIProxyLibraryResponse)
-			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)})
-			return true
-		case <-stopChan:
-			response := documentsAIProxyLibrary(documents)
-			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)})
-			c.Render(-1, common.CustomEvent{Data: "data: [DONE]"})
-			return false
+
+	for scanner.Scan() {
+		data := scanner.Text()
+		if len(data) < 5 || data[:5] != "data:" {
+			continue
 		}
-	})
-	err := resp.Body.Close()
+		data = data[5:]
+
+		var AIProxyLibraryResponse LibraryStreamResponse
+		err := json.Unmarshal([]byte(data), &AIProxyLibraryResponse)
+		if err != nil {
+			logger.SysError("error unmarshalling stream response: " + err.Error())
+			continue
+		}
+		if len(AIProxyLibraryResponse.Documents) != 0 {
+			documents = AIProxyLibraryResponse.Documents
+		}
+		response := streamResponseAIProxyLibrary2OpenAI(&AIProxyLibraryResponse)
+		err = render.ObjectData(c, response)
+		if err != nil {
+			logger.SysError(err.Error())
+		}
+	}
+
+	if err := scanner.Err(); err != nil {
+		logger.SysError("error reading stream: " + err.Error())
+	}
+
+	response := documentsAIProxyLibrary(documents)
+	err := render.ObjectData(c, response)
+	if err != nil {
+		logger.SysError(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/ali/constants.go

@@ -1,7 +1,23 @@
 package ali
 
 var ModelList = []string{
-	"qwen-turbo", "qwen-plus", "qwen-max", "qwen-max-longcontext",
-	"text-embedding-v1",
+	"qwen-turbo", "qwen-turbo-latest",
+	"qwen-plus", "qwen-plus-latest",
+	"qwen-max", "qwen-max-latest",
+	"qwen-max-longcontext",
+	"qwen-vl-max", "qwen-vl-max-latest", "qwen-vl-plus", "qwen-vl-plus-latest",
+	"qwen-vl-ocr", "qwen-vl-ocr-latest",
+	"qwen-audio-turbo",
+	"qwen-math-plus", "qwen-math-plus-latest", "qwen-math-turbo", "qwen-math-turbo-latest",
+	"qwen-coder-plus", "qwen-coder-plus-latest", "qwen-coder-turbo", "qwen-coder-turbo-latest",
+	"qwq-32b-preview", "qwen2.5-72b-instruct", "qwen2.5-32b-instruct", "qwen2.5-14b-instruct", "qwen2.5-7b-instruct", "qwen2.5-3b-instruct", "qwen2.5-1.5b-instruct", "qwen2.5-0.5b-instruct",
+	"qwen2-72b-instruct", "qwen2-57b-a14b-instruct", "qwen2-7b-instruct", "qwen2-1.5b-instruct", "qwen2-0.5b-instruct",
+	"qwen1.5-110b-chat", "qwen1.5-72b-chat", "qwen1.5-32b-chat", "qwen1.5-14b-chat", "qwen1.5-7b-chat", "qwen1.5-1.8b-chat", "qwen1.5-0.5b-chat",
+	"qwen-72b-chat", "qwen-14b-chat", "qwen-7b-chat", "qwen-1.8b-chat", "qwen-1.8b-longcontext-chat",
+	"qwen2-vl-7b-instruct", "qwen2-vl-2b-instruct", "qwen-vl-v1", "qwen-vl-chat-v1",
+	"qwen2-audio-instruct", "qwen-audio-chat",
+	"qwen2.5-math-72b-instruct", "qwen2.5-math-7b-instruct", "qwen2.5-math-1.5b-instruct", "qwen2-math-72b-instruct", "qwen2-math-7b-instruct", "qwen2-math-1.5b-instruct",
+	"qwen2.5-coder-32b-instruct", "qwen2.5-coder-14b-instruct", "qwen2.5-coder-7b-instruct", "qwen2.5-coder-3b-instruct", "qwen2.5-coder-1.5b-instruct", "qwen2.5-coder-0.5b-instruct",
+	"text-embedding-v1", "text-embedding-v3", "text-embedding-v2", "text-embedding-async-v2", "text-embedding-async-v1",
 	"ali-stable-diffusion-xl", "ali-stable-diffusion-v1.5", "wanx-v1",
 }

relay/adaptor/ali/main.go

@@ -3,15 +3,18 @@ package ali
 import (
 	"bufio"
 	"encoding/json"
+	"github.com/songquanpeng/one-api/common/ctxkey"
+	"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
@@ -33,9 +36,7 @@ func ConvertRequest(request model.GeneralOpenAIRequest) *ChatRequest {
 		enableSearch = true
 		aliModel = strings.TrimSuffix(aliModel, EnableSearchModelSuffix)
 	}
-	if request.TopP >= 1 {
-		request.TopP = 0.9999
-	}
+	request.TopP = helper.Float64PtrMax(request.TopP, 0.9999)
 	return &ChatRequest{
 		Model: aliModel,
 		Input: Input{
@@ -57,7 +58,7 @@ func ConvertRequest(request model.GeneralOpenAIRequest) *ChatRequest {
 
 func ConvertEmbeddingRequest(request model.GeneralOpenAIRequest) *EmbeddingRequest {
 	return &EmbeddingRequest{
-		Model: "text-embedding-v1",
+		Model: request.Model,
 		Input: struct {
 			Texts []string `json:"texts"`
 		}{
@@ -100,8 +101,9 @@ func EmbeddingHandler(c *gin.Context, resp *http.Response) (*model.ErrorWithStat
 			StatusCode: resp.StatusCode,
 		}, nil
 	}
-
+	requestModel := c.GetString(ctxkey.RequestModel)
 	fullTextResponse := embeddingResponseAli2OpenAI(&aliResponse)
+	fullTextResponse.Model = requestModel
 	jsonResponse, err := json.Marshal(fullTextResponse)
 	if err != nil {
 		return openai.ErrorWrapper(err, "marshal_response_body_failed", http.StatusInternalServerError), nil
@@ -181,56 +183,43 @@ 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) < 5 { // ignore blank line or wrong format
-				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
-			}
-			if aliResponse.Usage.OutputTokens != 0 {
-				usage.PromptTokens = aliResponse.Usage.InputTokens
-				usage.CompletionTokens = aliResponse.Usage.OutputTokens
-				usage.TotalTokens = aliResponse.Usage.InputTokens + aliResponse.Usage.OutputTokens
-			}
-			response := streamResponseAli2OpenAI(&aliResponse)
-			if response == nil {
-				return true
-			}
-			//response.Choices[0].Delta.Content = strings.TrimPrefix(response.Choices[0].Delta.Content, lastResponseText)
-			//lastResponseText = aliResponse.Output.Text
-			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)})
-			return true
-		case <-stopChan:
-			c.Render(-1, common.CustomEvent{Data: "data: [DONE]"})
-			return false
+
+	for scanner.Scan() {
+		data := scanner.Text()
+		if len(data) < 5 || data[:5] != "data:" {
+			continue
 		}
-	})
+		data = data[5:]
+
+		var aliResponse ChatResponse
+		err := json.Unmarshal([]byte(data), &aliResponse)
+		if err != nil {
+			logger.SysError("error unmarshalling stream response: " + err.Error())
+			continue
+		}
+		if aliResponse.Usage.OutputTokens != 0 {
+			usage.PromptTokens = aliResponse.Usage.InputTokens
+			usage.CompletionTokens = aliResponse.Usage.OutputTokens
+			usage.TotalTokens = aliResponse.Usage.InputTokens + aliResponse.Usage.OutputTokens
+		}
+		response := streamResponseAli2OpenAI(&aliResponse)
+		if response == nil {
+			continue
+		}
+		err = render.ObjectData(c, response)
+		if err != nil {
+			logger.SysError(err.Error())
+		}
+	}
+
+	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/ali/model.go

@@ -16,13 +16,13 @@ type Input struct {
 }
 
 type Parameters struct {
-	TopP              float64      `json:"top_p,omitempty"`
+	TopP              *float64     `json:"top_p,omitempty"`
 	TopK              int          `json:"top_k,omitempty"`
 	Seed              uint64       `json:"seed,omitempty"`
 	EnableSearch      bool         `json:"enable_search,omitempty"`
 	IncrementalOutput bool         `json:"incremental_output,omitempty"`
 	MaxTokens         int          `json:"max_tokens,omitempty"`
-	Temperature       float64      `json:"temperature,omitempty"`
+	Temperature       *float64     `json:"temperature,omitempty"`
 	ResultFormat      string       `json:"result_format,omitempty"`
 	Tools             []model.Tool `json:"tools,omitempty"`
 }

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

@@ -3,6 +3,10 @@ package anthropic
 var ModelList = []string{
 	"claude-instant-1.2", "claude-2.0", "claude-2.1",
 	"claude-3-haiku-20240307",
+	"claude-3-5-haiku-20241022",
 	"claude-3-sonnet-20240229",
 	"claude-3-opus-20240229",
+	"claude-3-5-sonnet-20240620",
+	"claude-3-5-sonnet-20241022",
+	"claude-3-5-sonnet-latest",
 }

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,
+			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 {
-		select {
-		case data := <-dataChan:
-			// some implementations may add \r at the end of 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
-			}
-			response, meta := StreamResponseClaude2OpenAI(&claudeResponse)
-			if meta != nil {
-				usage.PromptTokens += meta.Usage.InputTokens
-				usage.CompletionTokens += meta.Usage.OutputTokens
+	var lastToolCallChoice openai.ChatCompletionsStreamResponseChoice
+
+	for scanner.Scan() {
+		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())
+			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
-			}
-			response.Id = id
-			response.Model = modelName
-			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 <-stopChan:
-			c.Render(-1, common.CustomEvent{Data: "data: [DONE]"})
-			return false
 		}
-	})
-	_ = resp.Body.Close()
+		if response == nil {
+			continue
+		}
+
+		response.Id = id
+		response.Model = modelName
+		response.Created = createdTime
+
+		for _, choice := range response.Choices {
+			if len(choice.Delta.ToolCalls) > 0 {
+				lastToolCallChoice = choice
+			}
+		}
+		err = render.ObjectData(c, response)
+		if err != nil {
+			logger.SysError(err.Error())
+		}
+	}
+
+	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"`
@@ -30,9 +48,11 @@ type Request struct {
 	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"`
+	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"
-	"github.com/aws/aws-sdk-go-v2/credentials"
-	"github.com/aws/aws-sdk-go-v2/service/bedrockruntime"
-	"github.com/songquanpeng/one-api/common/ctxkey"
+	"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/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
-	awsClient *bedrockruntime.Client
+	awsAdapter utils.AwsAdapter
+
+	Meta      *meta.Meta
+	AwsClient *bedrockruntime.Client
 }
 
 func (a *Adaptor) Init(meta *meta.Meta) {
-	a.meta = meta
-	a.awsClient = bedrockruntime.New(bedrockruntime.Options{
+	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) 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 +15,31 @@ 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{
-	"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-opus-20240229":   "anthropic.claude-3-opus-20240229-v1:0",
-	"claude-3-haiku-20240307":  "anthropic.claude-3-haiku-20240307-v1:0",
+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-haiku-20240307":    "anthropic.claude-3-haiku-20240307-v1:0",
+	"claude-3-sonnet-20240229":   "anthropic.claude-3-sonnet-20240229-v1:0",
+	"claude-3-opus-20240229":     "anthropic.claude-3-opus-20240229-v1:0",
+	"claude-3-5-sonnet-20240620": "anthropic.claude-3-5-sonnet-20240620-v1:0",
+	"claude-3-5-sonnet-20241022": "anthropic.claude-3-5-sonnet-20241022-v2:0",
+	"claude-3-5-sonnet-latest":   "anthropic.claude-3-5-sonnet-20241022-v2:0",
+	"claude-3-5-haiku-20241022":  "anthropic.claude-3-5-haiku-20241022-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 +49,7 @@ 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 +60,30 @@ 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 +103,7 @@ 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 +114,7 @@ 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 +122,16 @@ 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 +139,8 @@ 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 +161,19 @@ func StreamHandler(c *gin.Context, awsCli *bedrockruntime.Client) (*relaymodel.E
 			if meta != nil {
 				usage.PromptTokens += meta.Usage.InputTokens
 				usage.CompletionTokens += meta.Usage.OutputTokens
-				id = fmt.Sprintf("chatcmpl-%s", meta.Id)
-				return true
+				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 +181,12 @@ 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"`
+	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

@@ -5,18 +5,20 @@ import (
 	"encoding/json"
 	"errors"
 	"fmt"
-	"github.com/gin-gonic/gin"
-	"github.com/songquanpeng/one-api/common"
-	"github.com/songquanpeng/one-api/common/logger"
-	"github.com/songquanpeng/one-api/relay/adaptor/openai"
-	"github.com/songquanpeng/one-api/relay/client"
-	"github.com/songquanpeng/one-api/relay/constant"
-	"github.com/songquanpeng/one-api/relay/model"
+	"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"
+	"github.com/songquanpeng/one-api/common/logger"
+	"github.com/songquanpeng/one-api/relay/adaptor/openai"
+	"github.com/songquanpeng/one-api/relay/constant"
+	"github.com/songquanpeng/one-api/relay/model"
 )
 
 // https://cloud.baidu.com/doc/WENXINWORKSHOP/s/flfmc9do2
@@ -33,9 +35,9 @@ type Message struct {
 
 type ChatRequest struct {
 	Messages        []Message `json:"messages"`
-	Temperature     float64   `json:"temperature,omitempty"`
-	TopP            float64   `json:"top_p,omitempty"`
-	PenaltyScore    float64   `json:"penalty_score,omitempty"`
+	Temperature     *float64  `json:"temperature,omitempty"`
+	TopP            *float64  `json:"top_p,omitempty"`
+	PenaltyScore    *float64  `json:"penalty_score,omitempty"`
 	Stream          bool      `json:"stream,omitempty"`
 	System          string    `json:"system,omitempty"`
 	DisableSearch   bool      `json:"disable_search,omitempty"`
@@ -137,59 +139,41 @@ 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) {
-		if atEOF && len(data) == 0 {
-			return 0, nil, nil
-		}
-		if i := strings.Index(string(data), "\n"); i >= 0 {
-			return i + 1, data[0:i], nil
-		}
-		if atEOF {
-			return len(data), data, nil
-		}
-		return 0, nil, nil
-	})
-	dataChan := make(chan string)
-	stopChan := make(chan bool)
-	go func() {
-		for scanner.Scan() {
-			data := scanner.Text()
-			if len(data) < 6 { // ignore blank line or wrong format
-				continue
-			}
-			data = data[6:]
-			dataChan <- data
-		}
-		stopChan <- true
-	}()
+	scanner.Split(bufio.ScanLines)
+
 	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
-			}
-			if baiduResponse.Usage.TotalTokens != 0 {
-				usage.TotalTokens = baiduResponse.Usage.TotalTokens
-				usage.PromptTokens = baiduResponse.Usage.PromptTokens
-				usage.CompletionTokens = baiduResponse.Usage.TotalTokens - baiduResponse.Usage.PromptTokens
-			}
-			response := streamResponseBaidu2OpenAI(&baiduResponse)
-			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)})
-			return true
-		case <-stopChan:
-			c.Render(-1, common.CustomEvent{Data: "data: [DONE]"})
-			return false
+
+	for scanner.Scan() {
+		data := scanner.Text()
+		if len(data) < 6 {
+			continue
 		}
-	})
+		data = data[6:]
+
+		var baiduResponse ChatStreamResponse
+		err := json.Unmarshal([]byte(data), &baiduResponse)
+		if err != nil {
+			logger.SysError("error unmarshalling stream response: " + err.Error())
+			continue
+		}
+		if baiduResponse.Usage.TotalTokens != 0 {
+			usage.TotalTokens = baiduResponse.Usage.TotalTokens
+			usage.PromptTokens = baiduResponse.Usage.PromptTokens
+			usage.CompletionTokens = baiduResponse.Usage.TotalTokens - baiduResponse.Usage.PromptTokens
+		}
+		response := streamResponseBaidu2OpenAI(&baiduResponse)
+		err = render.ObjectData(c, response)
+		if err != nil {
+			logger.SysError(err.Error())
+		}
+	}
+
+	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"
@@ -16,108 +18,66 @@ import (
 	"github.com/songquanpeng/one-api/relay/model"
 )
 
-func ConvertRequest(textRequest model.GeneralOpenAIRequest) *Request {
-	lastMessage := textRequest.Messages[len(textRequest.Messages)-1]
+func ConvertCompletionsRequest(textRequest model.GeneralOpenAIRequest) *Request {
+	p, _ := textRequest.Prompt.(string)
 	return &Request{
+		Prompt:      p,
 		MaxTokens:   textRequest.MaxTokens,
-		Prompt:      lastMessage.StringContent(),
 		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) {
-		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
-	})
+	scanner.Split(bufio.ScanLines)
 
-	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")
+	responseModel := c.GetString(ctxkey.OriginalModel)
 	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 err != nil {
-				logger.SysError("error unmarshalling stream response: " + err.Error())
-				return true
-			}
-			response := StreamResponseCloudflare2OpenAI(&cloudflareResponse)
-			if response == nil {
-				return true
-			}
-			responseText += cloudflareResponse.Response
-			response.Id = id
-			response.Model = responseModel
-			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 <-stopChan:
-			c.Render(-1, common.CustomEvent{Data: "data: [DONE]"})
-			return false
+
+	for scanner.Scan() {
+		data := scanner.Text()
+		if len(data) < len("data: ") {
+			continue
 		}
-	})
-	_ = resp.Body.Close()
+		data = strings.TrimPrefix(data, "data: ")
+		data = strings.TrimSuffix(data, "\r")
+
+		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())
+			continue
+		}
+		for _, v := range response.Choices {
+			v.Delta.Role = "assistant"
+			responseText += v.Delta.StringContent()
+		}
+		response.Id = id
+		response.Model = modelName
+		err = render.ObjectData(c, response)
+		if err != nil {
+			logger.SysError(err.Error())
+		}
+	}
+
+	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
 }
@@ -131,22 +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
-	err = json.Unmarshal(responseBody, &cloudflareResponse)
+	var response openai.TextResponse
+	err = json.Unmarshal(responseBody, &response)
 	if err != nil {
 		return openai.ErrorWrapper(err, "unmarshal_response_body_failed", http.StatusInternalServerError), nil
 	}
-	fullTextResponse := ResponseCloudflare2OpenAI(&cloudflareResponse)
-	fullTextResponse.Model = modelName
-	usage := openai.ResponseText2Usage(cloudflareResponse.Result.Response, modelName, promptTokens)
-	fullTextResponse.Usage = *usage
-	fullTextResponse.Id = helper.GetResponseID(c)
-	jsonResponse, err := json.Marshal(fullTextResponse)
+	response.Model = modelName
+	var responseText string
+	for _, v := range response.Choices {
+		responseText += v.Message.Content.(string)
+	}
+	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

@@ -1,25 +1,13 @@
 package cloudflare
 
+import "github.com/songquanpeng/one-api/relay/model"
+
 type Request struct {
-	Lora        string  `json:"lora,omitempty"`
-	MaxTokens   int     `json:"max_tokens,omitempty"`
-	Prompt      string  `json:"prompt,omitempty"`
-	Raw         bool    `json:"raw,omitempty"`
-	Stream      bool    `json:"stream,omitempty"`
-	Temperature float64 `json:"temperature,omitempty"`
-}
-
-type 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"`
+	Messages    []model.Message `json:"messages,omitempty"`
+	Lora        string          `json:"lora,omitempty"`
+	MaxTokens   int             `json:"max_tokens,omitempty"`
+	Prompt      string          `json:"prompt,omitempty"`
+	Raw         bool            `json:"raw,omitempty"`
+	Stream      bool            `json:"stream,omitempty"`
+	Temperature *float64        `json:"temperature,omitempty"`
 }

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"
@@ -43,7 +43,7 @@ func ConvertRequest(textRequest model.GeneralOpenAIRequest) *Request {
 		K:                textRequest.TopK,
 		Stream:           textRequest.Stream,
 		FrequencyPenalty: textRequest.FrequencyPenalty,
-		PresencePenalty:  textRequest.FrequencyPenalty,
+		PresencePenalty:  textRequest.PresencePenalty,
 		Seed:             int(textRequest.Seed),
 	}
 	if cohereRequest.Model == "" {
@@ -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) {
-		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
-	})
+	scanner.Split(bufio.ScanLines)
 
-	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 {
-		case data := <-dataChan:
-			// 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
-			}
-			response, meta := StreamResponseCohere2OpenAI(&cohereResponse)
-			if meta != nil {
-				usage.PromptTokens += meta.Meta.Tokens.InputTokens
-				usage.CompletionTokens += meta.Meta.Tokens.OutputTokens
-				return true
-			}
-			if response == nil {
-				return true
-			}
-			response.Id = fmt.Sprintf("chatcmpl-%d", createdTime)
-			response.Model = c.GetString("original_model")
-			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 <-stopChan:
-			c.Render(-1, common.CustomEvent{Data: "data: [DONE]"})
-			return false
+
+	for scanner.Scan() {
+		data := scanner.Text()
+		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())
+			continue
 		}
-	})
-	_ = resp.Body.Close()
+
+		response, meta := StreamResponseCohere2OpenAI(&cohereResponse)
+		if meta != nil {
+			usage.PromptTokens += meta.Meta.Tokens.InputTokens
+			usage.CompletionTokens += meta.Meta.Tokens.OutputTokens
+			continue
+		}
+		if response == nil {
+			continue
+		}
+
+		response.Id = fmt.Sprintf("chatcmpl-%d", createdTime)
+		response.Model = c.GetString("original_model")
+		response.Created = createdTime
+
+		err = render.ObjectData(c, response)
+		if err != nil {
+			logger.SysError(err.Error())
+		}
+	}
+
+	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/cohere/model.go

@@ -10,15 +10,15 @@ type Request struct {
 	PromptTruncation string        `json:"prompt_truncation,omitempty"` // 默认值为"AUTO"
 	Connectors       []Connector   `json:"connectors,omitempty"`
 	Documents        []Document    `json:"documents,omitempty"`
-	Temperature      float64       `json:"temperature,omitempty"` // 默认值为0.3
+	Temperature      *float64      `json:"temperature,omitempty"` // 默认值为0.3
 	MaxTokens        int           `json:"max_tokens,omitempty"`
 	MaxInputTokens   int           `json:"max_input_tokens,omitempty"`
 	K                int           `json:"k,omitempty"` // 默认值为0
-	P                float64       `json:"p,omitempty"` // 默认值为0.75
+	P                *float64      `json:"p,omitempty"` // 默认值为0.75
 	Seed             int           `json:"seed,omitempty"`
 	StopSequences    []string      `json:"stop_sequences,omitempty"`
-	FrequencyPenalty float64       `json:"frequency_penalty,omitempty"` // 默认值为0.0
-	PresencePenalty  float64       `json:"presence_penalty,omitempty"`  // 默认值为0.0
+	FrequencyPenalty *float64      `json:"frequency_penalty,omitempty"` // 默认值为0.0
+	PresencePenalty  *float64      `json:"presence_penalty,omitempty"`  // 默认值为0.0
 	Tools            []Tool        `json:"tools,omitempty"`
 	ToolResults      []ToolResult  `json:"tool_results,omitempty"`
 }

relay/adaptor/common.go

@@ -4,7 +4,7 @@ import (
 	"errors"
 	"fmt"
 	"github.com/gin-gonic/gin"
-	"github.com/songquanpeng/one-api/relay/client"
+	"github.com/songquanpeng/one-api/common/client"
 	"github.com/songquanpeng/one-api/relay/meta"
 	"io"
 	"net/http"

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) {
-		if atEOF && len(data) == 0 {
-			return 0, nil, nil
-		}
-		if i := strings.Index(string(data), "\n"); i >= 0 {
-			return i + 1, data[0:i], nil
-		}
-		if atEOF {
-			return len(data), data, nil
-		}
-		return 0, nil, nil
-	})
-	dataChan := make(chan string)
-	stopChan := make(chan bool)
-	go func() {
-		for scanner.Scan() {
-			data := scanner.Text()
-			if len(data) < 5 {
-				continue
-			}
-			if !strings.HasPrefix(data, "data:") {
-				continue
-			}
-			data = strings.TrimPrefix(data, "data:")
-			dataChan <- data
-		}
-		stopChan <- true
-	}()
+	scanner.Split(bufio.ScanLines)
+
 	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
-			}
-			response, _ := StreamResponseCoze2OpenAI(&cozeResponse)
-			if response == nil {
-				return true
-			}
-			for _, choice := range response.Choices {
-				responseText += conv.AsString(choice.Delta.Content)
-			}
-			response.Model = modelName
-			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 <-stopChan:
-			c.Render(-1, common.CustomEvent{Data: "data: [DONE]"})
-			return false
+
+	for scanner.Scan() {
+		data := scanner.Text()
+		if len(data) < 5 || !strings.HasPrefix(data, "data:") {
+			continue
 		}
-	})
-	_ = resp.Body.Close()
+		data = strings.TrimPrefix(data, "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())
+			continue
+		}
+
+		response, _ := StreamResponseCoze2OpenAI(&cozeResponse)
+		if response == nil {
+			continue
+		}
+
+		for _, choice := range response.Choices {
+			responseText += conv.AsString(choice.Delta.Content)
+		}
+		response.Model = modelName
+		response.Created = createdTime
+
+		err = render.ObjectData(c, response)
+		if err != nil {
+			logger.SysError(err.Error())
+		}
+	}
+
+	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/constants.go

@@ -0,0 +1,13 @@
+package doubao
+
+// https://console.volcengine.com/ark/region:ark+cn-beijing/model
+
+var ModelList = []string{
+	"Doubao-pro-128k",
+	"Doubao-pro-32k",
+	"Doubao-pro-4k",
+	"Doubao-lite-128k",
+	"Doubao-lite-32k",
+	"Doubao-lite-4k",
+	"Doubao-embedding",
+}

relay/adaptor/doubao/main.go

@@ -0,0 +1,18 @@
+package doubao
+
+import (
+	"fmt"
+	"github.com/songquanpeng/one-api/relay/meta"
+	"github.com/songquanpeng/one-api/relay/relaymode"
+)
+
+func GetRequestURL(meta *meta.Meta) (string, error) {
+	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/adaptor.go

@@ -7,12 +7,12 @@ import (
 	"net/http"
 
 	"github.com/gin-gonic/gin"
-	"github.com/songquanpeng/one-api/common/config"
 	"github.com/songquanpeng/one-api/common/helper"
 	channelhelper "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"
+	"github.com/songquanpeng/one-api/relay/relaymode"
 )
 
 type Adaptor struct {
@@ -23,11 +23,27 @@ func (a *Adaptor) Init(meta *meta.Meta) {
 }
 
 func (a *Adaptor) GetRequestURL(meta *meta.Meta) (string, error) {
-	version := helper.AssignOrDefault(meta.Config.APIVersion, config.GeminiVersion)
-	action := "generateContent"
+	var defaultVersion string
+	switch meta.ActualModelName {
+	case "gemini-2.0-flash-exp",
+		"gemini-2.0-flash-thinking-exp",
+		"gemini-2.0-flash-thinking-exp-01-21":
+		defaultVersion = "v1beta"
+	}
+
+	version := helper.AssignOrDefault(meta.Config.APIVersion, defaultVersion)
+	action := ""
+	switch meta.Mode {
+	case relaymode.Embeddings:
+		action = "batchEmbedContents"
+	default:
+		action = "generateContent"
+	}
+
 	if meta.IsStream {
 		action = "streamGenerateContent?alt=sse"
 	}
+
 	return fmt.Sprintf("%s/%s/models/%s:%s", meta.BaseURL, version, meta.ActualModelName, action), nil
 }
 
@@ -41,7 +57,14 @@ 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.Embeddings:
+		geminiEmbeddingRequest := ConvertEmbeddingRequest(*request)
+		return geminiEmbeddingRequest, nil
+	default:
+		geminiRequest := ConvertRequest(*request)
+		return geminiRequest, nil
+	}
 }
 
 func (a *Adaptor) ConvertImageRequest(request *model.ImageRequest) (any, error) {
@@ -61,7 +84,12 @@ func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, meta *meta.Met
 		err, responseText = StreamHandler(c, resp)
 		usage = openai.ResponseText2Usage(responseText, meta.ActualModelName, meta.PromptTokens)
 	} else {
-		err, usage = Handler(c, resp, meta.PromptTokens, meta.ActualModelName)
+		switch meta.Mode {
+		case relaymode.Embeddings:
+			err, usage = EmbeddingHandler(c, resp)
+		default:
+			err, usage = Handler(c, resp, meta.PromptTokens, meta.ActualModelName)
+		}
 	}
 	return
 }

relay/adaptor/gemini/constants.go

@@ -3,6 +3,9 @@ package gemini
 // https://ai.google.dev/models/gemini
 
 var ModelList = []string{
-	"gemini-pro", "gemini-1.0-pro-001", "gemini-1.5-pro",
-	"gemini-pro-vision", "gemini-1.0-pro-vision-001",
+	"gemini-pro", "gemini-1.0-pro",
+	"gemini-1.5-flash", "gemini-1.5-pro",
+	"text-embedding-004", "aqa",
+	"gemini-2.0-flash-exp",
+	"gemini-2.0-flash-thinking-exp", "gemini-2.0-flash-thinking-exp-01-21",
 }

relay/adaptor/gemini/main.go

@@ -8,6 +8,8 @@ import (
 	"net/http"
 	"strings"
 
+	"github.com/songquanpeng/one-api/common/render"
+
 	"github.com/songquanpeng/one-api/common"
 	"github.com/songquanpeng/one-api/common/config"
 	"github.com/songquanpeng/one-api/common/helper"
@@ -27,6 +29,11 @@ const (
 	VisionMaxImageNum = 16
 )
 
+var mimeTypeMap = map[string]string{
+	"json_object": "application/json",
+	"text":        "text/plain",
+}
+
 // Setting safety to the lowest possible values since Gemini is already powerless enough
 func ConvertRequest(textRequest model.GeneralOpenAIRequest) *ChatRequest {
 	geminiRequest := ChatRequest{
@@ -48,6 +55,10 @@ func ConvertRequest(textRequest model.GeneralOpenAIRequest) *ChatRequest {
 				Category:  "HARM_CATEGORY_DANGEROUS_CONTENT",
 				Threshold: config.GeminiSafetySetting,
 			},
+			{
+				Category:  "HARM_CATEGORY_CIVIC_INTEGRITY",
+				Threshold: config.GeminiSafetySetting,
+			},
 		},
 		GenerationConfig: ChatGenerationConfig{
 			Temperature:     textRequest.Temperature,
@@ -55,6 +66,15 @@ func ConvertRequest(textRequest model.GeneralOpenAIRequest) *ChatRequest {
 			MaxOutputTokens: textRequest.MaxTokens,
 		},
 	}
+	if textRequest.ResponseFormat != nil {
+		if mimeType, ok := mimeTypeMap[textRequest.ResponseFormat.Type]; ok {
+			geminiRequest.GenerationConfig.ResponseMimeType = mimeType
+		}
+		if textRequest.ResponseFormat.JsonSchema != nil {
+			geminiRequest.GenerationConfig.ResponseSchema = textRequest.ResponseFormat.JsonSchema.Schema
+			geminiRequest.GenerationConfig.ResponseMimeType = mimeTypeMap["json_object"]
+		}
+	}
 	if textRequest.Tools != nil {
 		functions := make([]model.Function, 0, len(textRequest.Tools))
 		for _, tool := range textRequest.Tools {
@@ -134,6 +154,29 @@ func ConvertRequest(textRequest model.GeneralOpenAIRequest) *ChatRequest {
 	return &geminiRequest
 }
 
+func ConvertEmbeddingRequest(request model.GeneralOpenAIRequest) *BatchEmbeddingRequest {
+	inputs := request.ParseInput()
+	requests := make([]EmbeddingRequest, len(inputs))
+	model := fmt.Sprintf("models/%s", request.Model)
+
+	for i, input := range inputs {
+		requests[i] = EmbeddingRequest{
+			Model: model,
+			Content: ChatContent{
+				Parts: []Part{
+					{
+						Text: input,
+					},
+				},
+			},
+		}
+	}
+
+	return &BatchEmbeddingRequest{
+		Requests: requests,
+	}
+}
+
 type ChatResponse struct {
 	Candidates     []ChatCandidate    `json:"candidates"`
 	PromptFeedback ChatPromptFeedback `json:"promptFeedback"`
@@ -208,7 +251,14 @@ func responseGeminiChat2OpenAI(response *ChatResponse) *openai.TextResponse {
 			if candidate.Content.Parts[0].FunctionCall != nil {
 				choice.Message.ToolCalls = getToolCalls(&candidate)
 			} else {
-				choice.Message.Content = candidate.Content.Parts[0].Text
+				var builder strings.Builder
+				for _, part := range candidate.Content.Parts {
+					if i > 0 {
+						builder.WriteString("\n")
+					}
+					builder.WriteString(part.Text)
+				}
+				choice.Message.Content = builder.String()
 			}
 		} else {
 			choice.Message.Content = ""
@@ -222,75 +272,80 @@ 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}
 	return &response
 }
 
+func embeddingResponseGemini2OpenAI(response *EmbeddingResponse) *openai.EmbeddingResponse {
+	openAIEmbeddingResponse := openai.EmbeddingResponse{
+		Object: "list",
+		Data:   make([]openai.EmbeddingResponseItem, 0, len(response.Embeddings)),
+		Model:  "gemini-embedding",
+		Usage:  model.Usage{TotalTokens: 0},
+	}
+	for _, item := range response.Embeddings {
+		openAIEmbeddingResponse.Data = append(openAIEmbeddingResponse.Data, openai.EmbeddingResponseItem{
+			Object:    `embedding`,
+			Index:     0,
+			Embedding: item.Values,
+		})
+	}
+	return &openAIEmbeddingResponse
+}
+
 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) {
-		if atEOF && len(data) == 0 {
-			return 0, nil, nil
-		}
-		if i := strings.Index(string(data), "\n"); i >= 0 {
-			return i + 1, data[0:i], nil
-		}
-		if atEOF {
-			return len(data), data, nil
-		}
-		return 0, nil, nil
-	})
-	dataChan := make(chan string)
-	stopChan := make(chan bool)
-	go func() {
-		for scanner.Scan() {
-			data := scanner.Text()
-			data = strings.TrimSpace(data)
-			if !strings.HasPrefix(data, "data: ") {
-				continue
-			}
-			data = strings.TrimPrefix(data, "data: ")
-			data = strings.TrimSuffix(data, "\"")
-			dataChan <- data
-		}
-		stopChan <- true
-	}()
+	scanner.Split(bufio.ScanLines)
+
 	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
-			}
-			response := streamResponseGeminiChat2OpenAI(&geminiResponse)
-			if response == nil {
-				return true
-			}
-			responseText += response.Choices[0].Delta.StringContent()
-			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)})
-			return true
-		case <-stopChan:
-			c.Render(-1, common.CustomEvent{Data: "data: [DONE]"})
-			return false
+
+	for scanner.Scan() {
+		data := scanner.Text()
+		data = strings.TrimSpace(data)
+		if !strings.HasPrefix(data, "data: ") {
+			continue
 		}
-	})
+		data = strings.TrimPrefix(data, "data: ")
+		data = strings.TrimSuffix(data, "\"")
+
+		var geminiResponse ChatResponse
+		err := json.Unmarshal([]byte(data), &geminiResponse)
+		if err != nil {
+			logger.SysError("error unmarshalling stream response: " + err.Error())
+			continue
+		}
+
+		response := streamResponseGeminiChat2OpenAI(&geminiResponse)
+		if response == nil {
+			continue
+		}
+
+		responseText += response.Choices[0].Delta.StringContent()
+
+		err = render.ObjectData(c, response)
+		if err != nil {
+			logger.SysError(err.Error())
+		}
+	}
+
+	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
 }
 
@@ -337,3 +392,39 @@ func Handler(c *gin.Context, resp *http.Response, promptTokens int, modelName st
 	_, err = c.Writer.Write(jsonResponse)
 	return nil, &usage
 }
+
+func EmbeddingHandler(c *gin.Context, resp *http.Response) (*model.ErrorWithStatusCode, *model.Usage) {
+	var geminiEmbeddingResponse EmbeddingResponse
+	responseBody, err := io.ReadAll(resp.Body)
+	if err != nil {
+		return openai.ErrorWrapper(err, "read_response_body_failed", http.StatusInternalServerError), nil
+	}
+	err = resp.Body.Close()
+	if err != nil {
+		return openai.ErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), nil
+	}
+	err = json.Unmarshal(responseBody, &geminiEmbeddingResponse)
+	if err != nil {
+		return openai.ErrorWrapper(err, "unmarshal_response_body_failed", http.StatusInternalServerError), nil
+	}
+	if geminiEmbeddingResponse.Error != nil {
+		return &model.ErrorWithStatusCode{
+			Error: model.Error{
+				Message: geminiEmbeddingResponse.Error.Message,
+				Type:    "gemini_error",
+				Param:   "",
+				Code:    geminiEmbeddingResponse.Error.Code,
+			},
+			StatusCode: resp.StatusCode,
+		}, nil
+	}
+	fullTextResponse := embeddingResponseGemini2OpenAI(&geminiEmbeddingResponse)
+	jsonResponse, err := json.Marshal(fullTextResponse)
+	if err != nil {
+		return openai.ErrorWrapper(err, "marshal_response_body_failed", http.StatusInternalServerError), nil
+	}
+	c.Writer.Header().Set("Content-Type", "application/json")
+	c.Writer.WriteHeader(resp.StatusCode)
+	_, err = c.Writer.Write(jsonResponse)
+	return nil, &fullTextResponse.Usage
+}

relay/adaptor/gemini/model.go

@@ -7,6 +7,33 @@ type ChatRequest struct {
 	Tools            []ChatTools          `json:"tools,omitempty"`
 }
 
+type EmbeddingRequest struct {
+	Model                string      `json:"model"`
+	Content              ChatContent `json:"content"`
+	TaskType             string      `json:"taskType,omitempty"`
+	Title                string      `json:"title,omitempty"`
+	OutputDimensionality int         `json:"outputDimensionality,omitempty"`
+}
+
+type BatchEmbeddingRequest struct {
+	Requests []EmbeddingRequest `json:"requests"`
+}
+
+type EmbeddingData struct {
+	Values []float64 `json:"values"`
+}
+
+type EmbeddingResponse struct {
+	Embeddings []EmbeddingData `json:"embeddings"`
+	Error      *Error          `json:"error,omitempty"`
+}
+
+type Error struct {
+	Code    int    `json:"code,omitempty"`
+	Message string `json:"message,omitempty"`
+	Status  string `json:"status,omitempty"`
+}
+
 type InlineData struct {
 	MimeType string `json:"mimeType"`
 	Data     string `json:"data"`
@@ -38,10 +65,12 @@ type ChatTools struct {
 }
 
 type ChatGenerationConfig struct {
-	Temperature     float64  `json:"temperature,omitempty"`
-	TopP            float64  `json:"topP,omitempty"`
-	TopK            float64  `json:"topK,omitempty"`
-	MaxOutputTokens int      `json:"maxOutputTokens,omitempty"`
-	CandidateCount  int      `json:"candidateCount,omitempty"`
-	StopSequences   []string `json:"stopSequences,omitempty"`
+	ResponseMimeType string   `json:"responseMimeType,omitempty"`
+	ResponseSchema   any      `json:"responseSchema,omitempty"`
+	Temperature      *float64 `json:"temperature,omitempty"`
+	TopP             *float64 `json:"topP,omitempty"`
+	TopK             float64  `json:"topK,omitempty"`
+	MaxOutputTokens  int      `json:"maxOutputTokens,omitempty"`
+	CandidateCount   int      `json:"candidateCount,omitempty"`
+	StopSequences    []string `json:"stopSequences,omitempty"`
 }

relay/adaptor/groq/constants.go

@@ -4,9 +4,24 @@ package groq
 
 var ModelList = []string{
 	"gemma-7b-it",
-	"llama2-7b-2048",
-	"llama2-70b-4096",
-	"mixtral-8x7b-32768",
-	"llama3-8b-8192",
+	"gemma2-9b-it",
+	"llama-3.1-70b-versatile",
+	"llama-3.1-8b-instant",
+	"llama-3.2-11b-text-preview",
+	"llama-3.2-11b-vision-preview",
+	"llama-3.2-1b-preview",
+	"llama-3.2-3b-preview",
+	"llama-3.2-11b-vision-preview",
+	"llama-3.2-90b-text-preview",
+	"llama-3.2-90b-vision-preview",
+	"llama-guard-3-8b",
 	"llama3-70b-8192",
+	"llama3-8b-8192",
+	"llama3-groq-70b-8192-tool-use-preview",
+	"llama3-groq-8b-8192-tool-use-preview",
+	"llava-v1.5-7b-4096-preview",
+	"mixtral-8x7b-32768",
+	"distil-whisper-large-v3-en",
+	"whisper-large-v3",
+	"whisper-large-v3-turbo",
 }

relay/adaptor/minimax/constants.go

@@ -1,7 +1,11 @@
 package minimax
 
+// https://www.minimaxi.com/document/guides/chat-model/V2?id=65e0736ab2845de20908e2dd
+
 var ModelList = []string{
-	"abab5.5s-chat",
-	"abab5.5-chat",
+	"abab6.5-chat",
+	"abab6.5s-chat",
 	"abab6-chat",
+	"abab5.5-chat",
+	"abab5.5s-chat",
 }

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/random"
+	"github.com/songquanpeng/one-api/common/render"
 	"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 {
-		case data := <-dataChan:
-			var ollamaResponse ChatResponse
-			err := json.Unmarshal([]byte(data), &ollamaResponse)
-			if err != nil {
-				logger.SysError("error unmarshalling stream response: " + err.Error())
-				return true
-			}
-			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)
-			if err != nil {
-				logger.SysError("error marshalling stream response: " + 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
+
+	for scanner.Scan() {
+		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())
+			continue
+		}
+
+		if ollamaResponse.EvalCount != 0 {
+			usage.PromptTokens = ollamaResponse.PromptEvalCount
+			usage.CompletionTokens = ollamaResponse.EvalCount
+			usage.TotalTokens = ollamaResponse.PromptEvalCount + ollamaResponse.EvalCount
+		}
+
+		response := streamResponseOllama2OpenAI(&ollamaResponse)
+		err = render.ObjectData(c, response)
+		if err != nil {
+			logger.SysError(err.Error())
+		}
+	}
+
+	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(), " "),
+		Model: request.Model,
+		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},
 	}
 
-	openAIEmbeddingResponse.Data = append(openAIEmbeddingResponse.Data, openai.EmbeddingResponseItem{
-		Object:    `embedding`,
-		Index:     0,
-		Embedding: response.Embedding,
-	})
+	for i, embedding := range response.Embeddings {
+		openAIEmbeddingResponse.Data = append(openAIEmbeddingResponse.Data, openai.EmbeddingResponseItem{
+			Object:    `embedding`,
+			Index:     i,
+			Embedding: embedding,
+		})
+	}
 	return &openAIEmbeddingResponse
 }
 

relay/adaptor/ollama/model.go

@@ -1,12 +1,14 @@
 package ollama
 
 type Options struct {
-	Seed             int     `json:"seed,omitempty"`
-	Temperature      float64 `json:"temperature,omitempty"`
-	TopK             int     `json:"top_k,omitempty"`
-	TopP             float64 `json:"top_p,omitempty"`
-	FrequencyPenalty float64 `json:"frequency_penalty,omitempty"`
-	PresencePenalty  float64 `json:"presence_penalty,omitempty"`
+	Seed             int      `json:"seed,omitempty"`
+	Temperature      *float64 `json:"temperature,omitempty"`
+	TopK             int      `json:"top_k,omitempty"`
+	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 {
@@ -37,11 +39,15 @@ type ChatResponse struct {
 }
 
 type EmbeddingRequest struct {
-	Model  string `json:"model"`
-	Prompt string `json:"prompt"`
+	Model string   `json:"model"`
+	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"`
+	Error      string      `json:"error,omitempty"`
+	Model      string      `json:"model"`
+	Embeddings [][]float64 `json:"embeddings"`
 }

relay/adaptor/openai/adaptor.go

@@ -3,16 +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 {
@@ -45,6 +48,10 @@ func (a *Adaptor) GetRequestURL(meta *meta.Meta) (string, error) {
 		return GetFullRequestURL(meta.BaseURL, requestURL, meta.ChannelType), nil
 	case channeltype.Minimax:
 		return minimax.GetRequestURL(meta)
+	case channeltype.Doubao:
+		return doubao.GetRequestURL(meta)
+	case channeltype.Novita:
+		return novita.GetRequestURL(meta)
 	default:
 		return GetFullRequestURL(meta.BaseURL, meta.RequestURLPath, meta.ChannelType), nil
 	}
@@ -68,6 +75,13 @@ func (a *Adaptor) ConvertRequest(c *gin.Context, relayMode int, request *model.G
 	if request == nil {
 		return nil, errors.New("request is nil")
 	}
+	if request.Stream {
+		// always return usage in stream mode
+		if request.StreamOptions == nil {
+			request.StreamOptions = &model.StreamOptions{}
+		}
+		request.StreamOptions.IncludeUsage = true
+	}
 	return request, nil
 }
 

relay/adaptor/openai/compatible.go

@@ -4,12 +4,17 @@ import (
 	"github.com/songquanpeng/one-api/relay/adaptor/ai360"
 	"github.com/songquanpeng/one-api/relay/adaptor/baichuan"
 	"github.com/songquanpeng/one-api/relay/adaptor/deepseek"
+	"github.com/songquanpeng/one-api/relay/adaptor/doubao"
 	"github.com/songquanpeng/one-api/relay/adaptor/groq"
 	"github.com/songquanpeng/one-api/relay/adaptor/lingyiwanwu"
 	"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/siliconflow"
 	"github.com/songquanpeng/one-api/relay/adaptor/stepfun"
+	"github.com/songquanpeng/one-api/relay/adaptor/togetherai"
+	"github.com/songquanpeng/one-api/relay/adaptor/xai"
 	"github.com/songquanpeng/one-api/relay/channeltype"
 )
 
@@ -19,11 +24,16 @@ var CompatibleChannels = []int{
 	channeltype.Moonshot,
 	channeltype.Baichuan,
 	channeltype.Minimax,
+	channeltype.Doubao,
 	channeltype.Mistral,
 	channeltype.Groq,
 	channeltype.LingYiWanWu,
 	channeltype.StepFun,
 	channeltype.DeepSeek,
+	channeltype.TogetherAI,
+	channeltype.Novita,
+	channeltype.SiliconFlow,
+	channeltype.XAI,
 }
 
 func GetCompatibleChannelMeta(channelType int) (string, []string) {
@@ -48,6 +58,16 @@ func GetCompatibleChannelMeta(channelType int) (string, []string) {
 		return "stepfun", stepfun.ModelList
 	case channeltype.DeepSeek:
 		return "deepseek", deepseek.ModelList
+	case channeltype.TogetherAI:
+		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
+	case channeltype.XAI:
+		return "xai", xai.ModelList
 	default:
 		return "openai", ModelList
 	}

relay/adaptor/openai/constants.go

@@ -7,6 +7,11 @@ var ModelList = []string{
 	"gpt-4", "gpt-4-0314", "gpt-4-0613", "gpt-4-1106-preview", "gpt-4-0125-preview",
 	"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-2024-08-06",
+	"gpt-4o-2024-11-20",
+	"chatgpt-4o-latest",
+	"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",
@@ -16,4 +21,7 @@ var ModelList = []string{
 	"dall-e-2", "dall-e-3",
 	"whisper-1",
 	"tts-1", "tts-1-1106", "tts-1-hd", "tts-1-hd-1106",
+	"o1", "o1-2024-12-17",
+	"o1-preview", "o1-preview-2024-09-12",
+	"o1-mini", "o1-mini-2024-09-12",
 }

relay/adaptor/openai/helper.go

@@ -2,15 +2,16 @@ package openai
 
 import (
 	"fmt"
+	"strings"
+
 	"github.com/songquanpeng/one-api/relay/channeltype"
 	"github.com/songquanpeng/one-api/relay/model"
-	"strings"
 )
 
-func ResponseText2Usage(responseText string, modeName string, promptTokens int) *model.Usage {
+func ResponseText2Usage(responseText string, modelName string, promptTokens int) *model.Usage {
 	usage := &model.Usage{}
 	usage.PromptTokens = promptTokens
-	usage.CompletionTokens = CountTokenText(responseText, modeName)
+	usage.CompletionTokens = CountTokenText(responseText, modelName)
 	usage.TotalTokens = usage.PromptTokens + usage.CompletionTokens
 	return usage
 }

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 (
@@ -24,88 +27,72 @@ 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) {
-		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)
+	scanner.Split(bufio.ScanLines)
 	var usage *model.Usage
-	go func() {
-		for scanner.Scan() {
-			data := scanner.Text()
-			if len(data) < dataPrefixLength { // ignore blank line or wrong format
-				continue
-			}
-			if data[:dataPrefixLength] != dataPrefix && data[:dataPrefixLength] != done {
-				continue
-			}
-			if strings.HasPrefix(data[dataPrefixLength:], done) {
-				dataChan <- data
-				continue
-			}
-			switch relayMode {
-			case relaymode.ChatCompletions:
-				var streamResponse ChatCompletionsStreamResponse
-				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
-					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
-				for _, choice := range streamResponse.Choices {
-					responseText += conv.AsString(choice.Delta.Content)
-				}
-				if streamResponse.Usage != nil {
-					usage = streamResponse.Usage
-				}
-			case relaymode.Completions:
-				dataChan <- data
-				var streamResponse CompletionsStreamResponse
-				err := json.Unmarshal([]byte(data[dataPrefixLength:]), &streamResponse)
-				if err != nil {
-					logger.SysError("error unmarshalling stream response: " + err.Error())
-					continue
-				}
-				for _, choice := range streamResponse.Choices {
-					responseText += choice.Text
-				}
-			}
-		}
-		stopChan <- true
-	}()
+
 	common.SetEventStreamHeaders(c)
-	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")
-			c.Render(-1, common.CustomEvent{Data: data})
-			return true
-		case <-stopChan:
-			return false
+
+	doneRendered := false
+	for scanner.Scan() {
+		data := scanner.Text()
+		if len(data) < dataPrefixLength { // ignore blank line or wrong format
+			continue
 		}
-	})
+		if data[:dataPrefixLength] != dataPrefix && data[:dataPrefixLength] != done {
+			continue
+		}
+		if strings.HasPrefix(data[dataPrefixLength:], done) {
+			render.StringData(c, data)
+			doneRendered = true
+			continue
+		}
+		switch relayMode {
+		case relaymode.ChatCompletions:
+			var streamResponse ChatCompletionsStreamResponse
+			err := json.Unmarshal([]byte(data[dataPrefixLength:]), &streamResponse)
+			if err != nil {
+				logger.SysError("error unmarshalling stream response: " + err.Error())
+				render.StringData(c, data) // if error happened, pass the data to client
+				continue                   // just ignore the error
+			}
+			if len(streamResponse.Choices) == 0 && streamResponse.Usage == nil {
+				// but for empty choice and no usage, we should not pass it to client, this is for azure
+				continue // just ignore empty choice
+			}
+			render.StringData(c, data)
+			for _, choice := range streamResponse.Choices {
+				responseText += conv.AsString(choice.Delta.Content)
+			}
+			if streamResponse.Usage != nil {
+				usage = streamResponse.Usage
+			}
+		case relaymode.Completions:
+			render.StringData(c, data)
+			var streamResponse CompletionsStreamResponse
+			err := json.Unmarshal([]byte(data[dataPrefixLength:]), &streamResponse)
+			if err != nil {
+				logger.SysError("error unmarshalling stream response: " + err.Error())
+				continue
+			}
+			for _, choice := range streamResponse.Choices {
+				responseText += choice.Text
+			}
+		}
+	}
+
+	if err := scanner.Err(); err != nil {
+		logger.SysError("error reading stream: " + err.Error())
+	}
+
+	if !doneRendered {
+		render.Done(c)
+	}
+
 	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

@@ -24,6 +24,10 @@ func InitTokenEncoders() {
 		logger.FatalLog(fmt.Sprintf("failed to get gpt-3.5-turbo token encoder: %s", err.Error()))
 	}
 	defaultTokenEncoder = gpt35TokenEncoder
+	gpt4oTokenEncoder, err := tiktoken.EncodingForModel("gpt-4o")
+	if err != nil {
+		logger.FatalLog(fmt.Sprintf("failed to get gpt-4o token encoder: %s", err.Error()))
+	}
 	gpt4TokenEncoder, err := tiktoken.EncodingForModel("gpt-4")
 	if err != nil {
 		logger.FatalLog(fmt.Sprintf("failed to get gpt-4 token encoder: %s", err.Error()))
@@ -31,6 +35,8 @@ func InitTokenEncoders() {
 	for model := range billingratio.ModelRatio {
 		if strings.HasPrefix(model, "gpt-3.5") {
 			tokenEncoderMap[model] = gpt35TokenEncoder
+		} else if strings.HasPrefix(model, "gpt-4o") {
+			tokenEncoderMap[model] = gpt4oTokenEncoder
 		} else if strings.HasPrefix(model, "gpt-4") {
 			tokenEncoderMap[model] = gpt4TokenEncoder
 		} else {
@@ -91,7 +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 {
@@ -100,7 +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 {
@@ -124,11 +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
@@ -162,6 +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 {
@@ -181,6 +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/openai/util.go

@@ -1,8 +1,16 @@
 package openai
 
-import "github.com/songquanpeng/one-api/relay/model"
+import (
+	"context"
+	"fmt"
+
+	"github.com/songquanpeng/one-api/common/logger"
+	"github.com/songquanpeng/one-api/relay/model"
+)
 
 func ErrorWrapper(err error, code string, statusCode int) *model.ErrorWithStatusCode {
+	logger.Error(context.TODO(), fmt.Sprintf("[%s]%+v", code, err))
+
 	Error := model.Error{
 		Message: err.Error(),
 		Type:    "one_api_error",

relay/adaptor/palm/model.go

@@ -19,11 +19,11 @@ type Prompt struct {
 }
 
 type ChatRequest struct {
-	Prompt         Prompt  `json:"prompt"`
-	Temperature    float64 `json:"temperature,omitempty"`
-	CandidateCount int     `json:"candidateCount,omitempty"`
-	TopP           float64 `json:"topP,omitempty"`
-	TopK           int     `json:"topK,omitempty"`
+	Prompt         Prompt   `json:"prompt"`
+	Temperature    *float64 `json:"temperature,omitempty"`
+	CandidateCount int      `json:"candidateCount,omitempty"`
+	TopP           *float64 `json:"topP,omitempty"`
+	TopK           int      `json:"topK,omitempty"`
 }
 
 type Error struct {

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)
-	go func() {
-		responseBody, err := io.ReadAll(resp.Body)
-		if err != nil {
-			logger.SysError("error reading stream response: " + err.Error())
-			stopChan <- true
-			return
-		}
-		err = resp.Body.Close()
-		if err != nil {
-			logger.SysError("error closing stream response: " + err.Error())
-			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
-		}
-		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
-		}
-		dataChan <- string(jsonResponse)
-		stopChan <- true
-	}()
+
 	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
+
+	responseBody, err := io.ReadAll(resp.Body)
+	if err != nil {
+		logger.SysError("error reading stream response: " + err.Error())
+		err := resp.Body.Close()
+		if err != nil {
+			return openai.ErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), ""
 		}
-	})
-	err := resp.Body.Close()
+		return openai.ErrorWrapper(err, "read_response_body_failed", http.StatusInternalServerError), ""
+	}
+
+	err = resp.Body.Close()
 	if err != nil {
 		return openai.ErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), ""
 	}
+
+	var palmResponse ChatResponse
+	err = json.Unmarshal(responseBody, &palmResponse)
+	if err != nil {
+		logger.SysError("error unmarshalling stream response: " + err.Error())
+		return openai.ErrorWrapper(err, "unmarshal_response_body_failed", http.StatusInternalServerError), ""
+	}
+
+	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())
+		return openai.ErrorWrapper(err, "marshal_response_body_failed", http.StatusInternalServerError), ""
+	}
+
+	err = render.ObjectData(c, string(jsonResponse))
+	if err != nil {
+		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)
+}