如何使用AWS Lambda从Firebase获取食谱并对接Amazon Alexa

Hey there! Let me walk you through building your recipe skill that connects Firebase, AWS Lambda, and Amazon Alexa step by step. I’ll break everything down in plain, easy-to-follow steps—no prior experience needed.

First, let’s get clear on how all the pieces fit together:

• Firebase Firestore: Stores all your recipe data (names, ingredients, steps)
• AWS Lambda: Acts as the backend brain—pulls data from Firebase, processes Alexa’s requests, and sends back responses
• Amazon Alexa Skill: The user-facing interface—lets users ask for recipes via voice, and reads back the steps

1. Set up Firebase Project & Firestore
   • Go to the Firebase Console, create a new project, then enable Firestore Database.
   • Create a collection named recipes, and add documents with this structure (adjust to your recipes):

     {
       "name": "番茄炒蛋",
       "ingredients": ["番茄2个", "鸡蛋3个", "盐1勺"],
       "steps": ["1. 鸡蛋打散加盐搅匀", "2. 番茄切块备用", "3. 热锅倒油炒鸡蛋盛出", "4. 炒番茄出汁后加鸡蛋翻炒"]
     }
     

2. Adjust Firestore Security Rules
   • For testing, temporarily set read access to open (we’ll lock this down later):

     rules_version = '2';
     service cloud.firestore {
       match /databases/{database}/documents {
         match /recipes/{document} {
           allow read: if true; // 上线前替换为严格权限
         }
       }
     }
     

3. Download Firebase Service Account Key
   • In Firebase Console → Project Settings → Service Accounts → Generate New Private Key. Save this as firebase-key.json—you’ll need it for Lambda.

1. Create Lambda Function
   • Log into AWS Console, navigate to Lambda, click "Create function". Choose "Author from scratch", name it AlexaRecipeLambda, pick Node.js as the runtime.
2. Package Firebase Dependencies
   • Lambda doesn’t come with Firebase SDK pre-installed, so set up a local Node project:

     npm init -y
     npm install firebase-admin
     

   • Zip up the node_modules folder, firebase-key.json, and your Lambda code (we’ll write that next) into a single zip file, then upload it to your Lambda function’s code tab.
3. Write Lambda Code
   • Here’s the core code to handle Alexa requests and fetch data from Firebase:

     const admin = require('firebase-admin');
     
     // 初始化Firebase Admin
     const serviceAccount = require('./firebase-key.json');
     admin.initializeApp({
       credential: admin.credential.cert(serviceAccount)
     });
     const db = admin.firestore();
     
     // 处理Alexa请求的核心函数
     exports.handler = async (event) => {
       const requestType = event.request.type;
     
       // 用户打开技能时的回复
       if (requestType === 'LaunchRequest') {
         return buildResponse('欢迎使用食谱助手！请告诉我你想找的食谱名称。');
       }
       // 处理用户的意图请求
       else if (requestType === 'IntentRequest') {
         const intentName = event.request.intent.name;
     
         if (intentName === 'SearchRecipeIntent') {
           // 获取用户说的食谱名称
           const recipeName = event.request.intent.slots.RecipeName.value;
     
           // 从Firebase查询食谱
           const recipeRef = db.collection('recipes').where('name', '==', recipeName).limit(1);
           const snapshot = await recipeRef.get();
     
           if (snapshot.empty) {
             return buildResponse(`抱歉，没有找到名为${recipeName}的食谱。`);
           }
     
           const recipe = snapshot.docs[0].data();
           // 把步骤整理成语音友好的文本
           const stepsText = recipe.steps.join('，接下来，');
           return buildResponse(`找到了${recipeName}的食谱，需要的食材有${recipe.ingredients.join('、')}。步骤是：${stepsText}。`);
         }
       }
     
       // 默认回复
       return buildResponse('抱歉，我没听懂，请再说一遍。');
     };
     
     // 辅助函数：构建Alexa的响应格式
     function buildResponse(outputText) {
       return {
         version: '1.0',
         response: {
           outputSpeech: {
             type: 'PlainText',
             text: outputText
           },
           shouldEndSession: false
         }
       };
     }
     

4. Add Alexa Skills Kit Trigger
   • Go to your Lambda function’s "Configuration" → "Triggers" → "Add trigger". Select "Alexa Skills Kit", then enter your Alexa Skill’s ID (you’ll get this when creating the skill next).

1. Set Up Alexa Skill
   • Log into Alexa Developer Console, click "Create Skill". Choose "Custom" as the model, pick your language, and select "Provision your own" for hosting.
2. Build the Interaction Model
   • Go to "Intents" → "Add Intent", create an intent named SearchRecipeIntent.
   • Add a slot named RecipeName, use the pre-built AMAZON.Food type (或创建自定义槽位，添加所有食谱名称以提高识别准确率).
   • Add sample utterances like:
     • "我想找{RecipeName}的食谱"
     • "请告诉我{RecipeName}的做法"
     • "{RecipeName}怎么做"
3. Configure Endpoint
   • Go to "Endpoint" → Select "AWS Lambda ARN", paste the ARN of your Lambda function (found in Lambda’s function details), then save.

• In Alexa Developer Console, go to the "Test" tab and enable testing mode. Try asking: "打开食谱助手" then "番茄炒蛋怎么做".
• If something breaks, check Lambda’s CloudWatch logs (under "Monitor" → "Logs") to debug issues like Firebase permission errors or code logic bugs.

• Lock Down Firebase Permissions: Replace the open allow read: if true; rule with strict access—only allow your Lambda’s IAM role to read the recipes collection.
• Improve Alexa Recognition: Use a custom slot for RecipeName and add all your recipe names to it—this helps Alexa understand exactly what users are asking for.
• Adjust Lambda Settings: If Firebase queries are slow, increase Lambda’s memory to 256MB and timeout to 5 seconds (under "Configuration" → "General configuration").

内容的提问来源于stack exchange，提问作者Alien Bae