From 50889e83e3ae02ca6e7173151014643d5c0b6325 Mon Sep 17 00:00:00 2001 From: xiaoweii Date: Fri, 15 Nov 2024 16:11:46 +0800 Subject: [PATCH] feat: added README_CN file. change default model to Claude 3.5 Sonnet --- .github/workflows/release.yml | 10 +- README.md | 74 ++++++++++----- README_CN.md | 167 ++++++++++++++++++++++++++++++++++ src/storage/StorageUtils.ts | 4 +- 4 files changed, 227 insertions(+), 28 deletions(-) create mode 100644 README_CN.md diff --git a/.github/workflows/release.yml b/.github/workflows/release.yml index e583fa8..07fdf90 100644 --- a/.github/workflows/release.yml +++ b/.github/workflows/release.yml @@ -1,10 +1,12 @@ name: Release Template on: - push: - branches: [ "main" ] - pull_request: - branches: [ "main" ] + workflow_dispatch: + inputs: + reason: + description: 'the reason for triggering this workflow' + required: false + default: 'manually publish the ecr images and templates' jobs: release_template: runs-on: ubuntu-latest diff --git a/README.md b/README.md index 477baaf..c96b8e8 100644 --- a/README.md +++ b/README.md @@ -1,3 +1,5 @@ +[中文](/README_CN.md) + # SwiftChat - A Cross-platform AI Chat App SwiftChat is a fast and responsive AI chat application built with React Native and powered by Amazon Bedrock. Featuring @@ -24,7 +26,7 @@ generation across Android, iOS, and macOS platforms. By default, we use AWS App Runner, which is commonly used to host Python FastAPI servers, offering high performance, scalability and low latency. -Alternatively, we provide the option to replace App Runner with AWS Lambda using Function URLs for a more cost-effective +Alternatively, we provide the option to replace App Runner with AWS Lambda using Function URL for a more cost-effective solution, as shown in this [example](https://github.com/awslabs/aws-lambda-web-adapter/tree/main/examples/fastapi-response-streaming). @@ -32,14 +34,16 @@ this [example](https://github.com/awslabs/aws-lambda-web-adapter/tree/main/examp ### Step 1: Set up your API Key -1. Click Parameter Store to open your AWS Console. +1. Right-click [Parameter Store](https://console.aws.amazon.com/systems-manager/parameters/) to open your AWS Console in + new window. 2. Check whether you are in the [supported region](#supported-region), then click on the "Create parameter" button. -3. Name: Enter a descriptive name for your parameter (e.g., "SwiftChatAPIKey"). +3. Name: Enter a descriptive name for your parameter (e.g., "SwiftChatAPIKey", this is `ApiKeyParam` you will fill + in [Step2](#step-2-deploy-stack-and-get-your-api-url)). 4. Tier: Select **Standard**. 5. Type: Select **SecureString**. -6. Value: Any string without spaces (This value is your API Key which should fill in your App's Settings page). +6. Value: Any string without spaces (This is the API Key you'll need to configure your App + in [Step 3](#step-3-download-the-app-and-setup-with-api-url-and-api-key)). 7. Click "Create parameter". -8. Make a note of the parameter name you used (e.g., "SwiftChatAPIKey"). You'll need this in the next step. ### Step 2: Deploy stack and get your API URL @@ -48,7 +52,7 @@ this [example](https://github.com/awslabs/aws-lambda-web-adapter/tree/main/examp [![Launch Stack](images/launch-stack.png)](https://console.aws.amazon.com/cloudformation/home#/stacks/create/template?stackName=SwiftChatAPI&templateURL=https://aws-gcr-solutions.s3.amazonaws.com/swift-chat/latest/SwiftChatAppRunner.template) - - Lambda (You need to config your Lambda Function URL manually) + - Lambda (Requires manual setup of Function URL with NONE auth) [![Launch Stack](images/launch-stack.png)](https://console.aws.amazon.com/cloudformation/home#/stacks/create/template?stackName=SwiftChatAPI&templateURL=https://aws-gcr-solutions.s3.amazonaws.com/swift-chat/latest/SwiftChatLambda.template) @@ -59,33 +63,36 @@ this [example](https://github.com/awslabs/aws-lambda-web-adapter/tree/main/examp acknowledge that AWS CloudFormation might create IAM resources" checkbox at the bottom. 4. Click **Next**, In the "Review and create" Review your configuration and click **Submit**. -Wait about 3-5 minutes for the deployment to finish. +Wait about 3-5 minutes for the deployment to finish, and get your API URL: -**For using App Runner**, click the CloudFormation stack and go to Outputs tab, you can find the **API URL** +**For App Runner**, click the CloudFormation stack and go to **Outputs** tab, you can find the **API URL** (which looks like `https://xxx.xxx.awsapprunner.com`) -**For using Lambda**, follow these steps to get your **API URL** by enabling Lambda Function URL manually: +**For Lambda**, follow these steps to get your **API URL**: -1. Click the CloudFormation stack and go to **Resources** tab -2. Expand APIHandler under Physical ID, click the Lambda link to open Lambda page -3. In Lambda page, go to **Configuration** -> **Function URL** to enable Lambda Function URL with **RESPONSE_STREAM** - invoke mode -4. You'll find your Lambda Function URL like `https://xxx.lambda-url.xxx.on.aws` - use this as your **API URL** +1. Find your Lambda function in CloudFormation -> **Resources** -> APIHandler +2. Click to open the Lambda function (`SwiftChatLambda-xxx`) +3. Set up Function URL: + - Go to **Configuration** -> **Function URL**, Click **Create function URL** + - Select: **Auth type**: `NONE` + - Enable: **Invoke mode**: `RESPONSE_STREAM` under Additional settings +4. Click Save to get your API URL (`https://xxx.lambda-url.xxx.on.aws`) +5. Note: We use Bearer Token for API security verification. You can upgrade to `AWS_IAM` Auth type manually. ### Step 3: Download the app and setup with API URL and API Key 1. Download the App - - Android App [Download](https://github.com/aws-samples/swift-chat/releases/download/v1.5.0/SwiftChat.apk) - - macOS App [Download](https://github.com/aws-samples/swift-chat/releases/download/v1.5.0/SwiftChat.dmg) + - Android App click to [Download](https://github.com/aws-samples/swift-chat/releases/download/v1.5.0/SwiftChat.apk) + - macOS App click to [Download](https://github.com/aws-samples/swift-chat/releases/download/v1.5.0/SwiftChat.dmg) - iOS (Currently we do not provide the iOS version, you can build it locally with Xcode) -2. Launch your App, Click the menu button in the top left to open the drawer page, click **Settings** in the bottom of - the drawer. -3. Paste the API URL and API Key then select the Region. -4. You can change the default text and image model, and **make sure you have enabled these models in your AWS console**. +2. Launch your App, click the left menu button to open the drawer page, click **Settings** in the bottom of the drawer. +3. Paste the `API URL` and `API Key` then select the Region. +4. You can change the default text and image model, and **make sure these models are enabled in the corresponding Region + in the AWS Console**. 5. Click the top right finish icon to save your configuration and start your chat. -Congratulations! Your SwiftChat App is ready to use 🎉 +Congratulations 🎉 Your SwiftChat App is ready to use! ### Supported Region @@ -101,7 +108,30 @@ Congratulations! Your SwiftChat App is ready to use 🎉 - Europe (Paris): eu-west-3 - South America (São Paulo): sa-east-1 -## Why Swift? +## What Makes SwiftChat Really "Swift"? + +🚀 **Fast Launch Speed** + +- Thanks to the **AOT** (Ahead of Time) compilation of RN Hermes engine +- Added **lazy loading** of complex components +- App can start and be ready for input **within 1 second** + +🌐 **Fast Request Speed** + +- Speed up end-to-end API requests through **image compression** +- Deploying APIs in the **same region** as Bedrock provides lower latency +- Minimal response payload with **zero parsing** needed for direct display + +📱 **Fast Render Speed** + +- Using `useMemo` and custom caching to creates secondary cache for session content +- Reduce unnecessary re-renders and speed up streaming messages display +- All UI components are rendered as **native components** + +📦 **Fast Storage Speed** + +- By using **react-native-mmkv** Messages can be read, stored, and updated **10x faster** than AsyncStorage +- Optimized session content and session list storage structure to accelerates history list display ## App Privacy & Security diff --git a/README_CN.md b/README_CN.md new file mode 100644 index 0000000..65f256e --- /dev/null +++ b/README_CN.md @@ -0,0 +1,167 @@ +[English](/README.md) + +# SwiftChat - 跨平台 AI 聊天应用 + +SwiftChat是一个使用 React Native 构建并由 Amazon Bedrock 提供支持的快速响应式AI聊天应用。它采用极简设计理念和强大的隐私保护，提供实时流式对话和AI图像生成功能，支持 +Android、iOS 和 macOS 等多个平台。 + +![](images/promo.png) + +**主要特点:** + +- 与 AI 进行实时流式聊天 +- AI 图像生成 +- 跨平台支持（Android、iOS、macOS） +- 针对 iPad 和 Android 平板电脑优化 +- 快速启动和响应性能 +- 支持多种 AI 模型及切换 +- 使用 React Native 构建 +- 由 Amazon Bedrock 提供支持 + +## 架构 + +![](/images/architecture.png) + +默认情况下，我们使用 AWS App Runner（通常用于托管 Python FastAPI 服务器），提供高性能、可扩展性和低延迟。 + +另外，我们提供使用 AWS Lambda 的 Function URL 替代 App Runner +的选项，以获得更具成本效益的解决方案，如 [示例](https://github.com/awslabs/aws-lambda-web-adapter/tree/main/examples/fastapi-response-streaming) +所示。 + +## 入门指南 + +### 第1步: 设置API Key + +1. 右键点击 [Parameter Store](https://console.aws.amazon.com/systems-manager/parameters/) 在新窗口中打开AWS控制台。 +2. 检查您是否在 [支持的区域](#支持的区域)，然后点击"创建参数"按钮。 +3. 名称:为您的参数输入描述性名称(例如"SwiftChatAPIKey"，这是您将在 [第2步](#第2步-部署堆栈并获取api-url) + 中填写的`ApiKeyParam`)。 +4. 层级:选择 **标准**。 +5. 类型:选择 **SecureString**。 +6. 值:任何不含空格的字符串（这是您需要在 [第3步](#第3步-下载应用并设置-api-url-和-api-key) 中配置 App 的 API Key）。 +7. 点击"创建参数"。 + +### 第2步: 部署堆栈并获取API URL + +1. 点击以下按钮在与刚才创建的 API Key 相同的区域启动 CloudFormation 堆栈。 + - App Runner + + [![启动堆栈](images/launch-stack.png)](https://console.aws.amazon.com/cloudformation/home#/stacks/create/template?stackName=SwiftChatAPI&templateURL=https://aws-gcr-solutions.s3.amazonaws.com/swift-chat/latest/SwiftChatAppRunner.template) + + - Lambda (需要手动设置无认证的 Function URL) + + [![启动堆栈](images/launch-stack.png)](https://console.aws.amazon.com/cloudformation/home#/stacks/create/template?stackName=SwiftChatAPI&templateURL=https://aws-gcr-solutions.s3.amazonaws.com/swift-chat/latest/SwiftChatLambda.template) + +2. 点击**下一步**，在"指定堆栈详细信息"页面中提供以下信息: + - 使用存储 API Key 的参数名填写`ApiKeyParam`(例如"SwiftChatAPIKey")。 + - 对于App Runner，根据您的需求选择`InstanceTypeParam`。 +3. 点击**下一步**，保持"配置堆栈选项"页面默认设置，阅读功能并勾选底部的"我确认，AWS CloudFormation可能会创建 IAM 资源"复选框。 +4. 点击**下一步**，在"审核并创建"中检查配置并点击**提交**。 + +等待约3-5分钟部署完成，然后获取 API URL: + +**对于使用 App Runner**，点击CloudFormation堆栈并转到**输出**选项卡，您可以找到**API URL**( +类似`https://xxx.xxx.awsapprunner.com`) + +**对于使用 Lambda**，按照以下步骤获取**API URL**: + +1. 在 CloudFormation -> **资源** -> APIHandler中找到您的Lambda函数 +2. 点击打开 Lambda 函数(`SwiftChatLambda-xxx`) +3. 设置 Function URL: + - 转到 **配置** -> **函数 URL**，点击**创建函数 URL** + - 选择: **授权类型**: `NONE` + - 启用: **调用模式**: 在其他设置下选择`RESPONSE_STREAM` +4. 点击保存获取API URL(`https://xxx.lambda-url.xxx.on.aws`) +5. 注意: 我们默认使用 Bearer Token 进行 API 安全验证。您可以手动升级到`AWS_IAM`认证类型。 + +### 第3步: 下载应用并设置 API URL 和 API Key + +1. 下载应用 + - Android 应用点击 [下载](https://github.com/aws-samples/swift-chat/releases/download/v1.5.0/SwiftChat.apk) + - macOS 应用点击 [下载](https://github.com/aws-samples/swift-chat/releases/download/v1.5.0/SwiftChat.dmg) + - iOS (目前不提供 iOS 版本，您可以使用 Xcode 在本地构建) + +2. 启动应用，点击左侧菜单按钮打开抽屉页面，点击抽屉底部的**Settings**。 +3. 粘贴`API URL`和`API Key`然后选择 Region。 +4. 您可以更改默认的文本和图像模型，并**确保已在 AWS 控制台的对应 Region 中启用了这些模型**。 +5. 点击右上角对勾图标保存配置并开始聊天。 + +恭喜🎉 您的 SwiftChat 应用已准备就绪! + +### 支持的区域 + +- 美国东部(弗吉尼亚北部):us-east-1 +- 美国西部(俄勒冈):us-west-2 +- 亚太地区(孟买):ap-south-1 +- 亚太地区(新加坡):ap-southeast-1 +- 亚太地区(悉尼):ap-southeast-2 +- 亚太地区(东京):ap-northeast-1 +- 加拿大(中部):ca-central-1 +- 欧洲(法兰克福):eu-central-1 +- 欧洲(伦敦):eu-west-2 +- 欧洲(巴黎):eu-west-3 +- 南美洲(圣保罗):sa-east-1 + +## 是什么让SwiftChat如此"迅速"? + +🚀 **快速启动速度** + +- 得益于RN Hermes 引擎的 **AOT**（提前编译） +- 添加复杂组件的 **延迟加载** +- 应用可以在 **1秒内** 启动并准备好接收输入 + +🌐 **快速请求速度** + +- 通过**图像压缩**加速端到端 API 请求 +- 在与 Bedrock **相同区域**部署 API 以提供更低延迟 +- 最小响应有效负载，**零解析**直接显示 + +📱 **快速渲染速度** + +- 使用`useMemo`和自定义缓存为会话内容创建二级缓存 +- 减少不必要的重新渲染并加快流式消息显示 +- 所有 UI 组件都渲染为**原生组件** + +📦 **快速存储速度** + +- 通过使用 **react-native-mmkv** 消息可以比 AsyncStorage **快10倍**读取、存储和更新 +- 优化会话内容和会话列表存储结构以加快历史记录列表显示 + +## 应用隐私和安全 + +- 加密 API Key 的存储 +- 最小权限要求 +- 仅本地数据存储 +- 无用户行为跟踪 +- 无数据收集 +- 隐私优先策略 + +## 构建和开发 + +首先，克隆此仓库并运行`npm i`下载依赖项。 + +### 构建 Android + +```bash +npm run start && npm run android +``` + +## 构建 iOS + +```bash +npm run start && npm run ios +``` + +### 构建 macOS + +1. 在 `/src/App.tsx` 中将 `isMac` 修改为 `true` 并执行 `npm run start`。 +2. 双击 `ios/SwiftChat.xcworkspace` 在 Xcode 中打开项目。 +3. 将构建目标更改为 `My Mac (Mac Catalyst)` 然后点击 ▶ 运行按钮。 + +## 安全 + +更多信息请参见 [CONTRIBUTING](CONTRIBUTING.md#security-issue-notifications)。 + +## 许可证 + +该库使用 MIT-0 许可证。详见 [LICENSE](/LICENSE) 文件。 diff --git a/src/storage/StorageUtils.ts b/src/storage/StorageUtils.ts index 614cab0..35ccf1d 100644 --- a/src/storage/StorageUtils.ts +++ b/src/storage/StorageUtils.ts @@ -180,8 +180,8 @@ export function getTextModel(): Model { export function getDefaultTextModels() { return [ { - modelName: 'Claude 3.5 Sonnet v2', - modelId: 'anthropic.claude-3-5-sonnet-20241022-v2:0', + modelName: 'Claude 3.5 Sonnet', + modelId: 'anthropic.claude-3-5-sonnet-20240620-v1:0', }, ] as Model[]; }