HTTP 请求正文未通过 AWS API Gateway 访问 AWS lambda 函数

Posted 2023-03-28

【中文标题】HTTP 请求正文未通过 AWS API Gateway 访问 AWS lambda 函数

【英文标题】：HTTP request body not getting to AWS lambda function via AWS API Gateway

【发布时间】：2017-04-24 20:59:12

【问题描述】：

我有一个用 Scala 编写的非常基本的 lambda 函数部署到 AWS Lambda。当我通过 AWS Lambda 控制台对其进行测试时，该函数运行良好。

这是为调试目的添加了一些额外日志记录的函数。

package com.spacecorpshandbook.ostium.lambda.handler

import java.util

import com.google.gson.Gson
import temp.ApiGatewayProxyResponse, Appointment, CancelResponse

/**
  * Amazon Lambda handler adapter for the Cancellation application
  */
class CancellationHandler 

  def cancelAppointment(appointment: Appointment): ApiGatewayProxyResponse = 

    System.out.println("++++ appointmentId is: " + appointment.getAppointmentId)

    val apiGatewayProxyResponse = new ApiGatewayProxyResponse
    val cancelResponse = new CancelResponse

    cancelResponse.setMessage("Cancelled appointment with id " + appointment.getAppointmentId)

    val gson: Gson = new Gson

    apiGatewayProxyResponse.setBody(gson.toJson(cancelResponse))

    apiGatewayProxyResponse.setStatusCode("200")

    val headerValues = new util.HashMap[String, String]

    headerValues put("Content-Type", "application/json")

    apiGatewayProxyResponse.setHeaders(headerValues)

    System.out.println("+++++ message before returning: " + apiGatewayProxyResponse.getBody)

    apiGatewayProxyResponse
  

我担心作为 Scala bean 的 POJO 输入/输出可能会导致问题，所以我暂时实施 Java 版本只是为了排除这种情况。

AWS API 网关上的集成请求默认设置为启用 Lambda 代理集成的 Resources ANY。请注意，在此配置中，当我从 AWS API Gateway 控制台进行测试时，数据会被转换并进入，但不会一直到达 lambda 函数

Execution log for request test-request
Fri Dec 09 11:14:40 UTC 2016 : Starting execution for request: test-invoke-request
Fri Dec 09 11:14:40 UTC 2016 : HTTP Method: PUT, Resource Path: /cancel-appointment
Fri Dec 09 11:14:40 UTC 2016 : Method request path: 
Fri Dec 09 11:14:40 UTC 2016 : Method request query string: 
Fri Dec 09 11:14:40 UTC 2016 : Method request headers: Content-Type= application/json
Fri Dec 09 11:14:40 UTC 2016 : Method request body before transformations: 
    "applicationId": "asdfsfa"

Fri Dec 09 11:14:40 UTC 2016 : Endpoint request headers: x-amzn-lambda-integration-tag=test-request, Authorization=****************************************************************************************************************************************************************************************************************************************************************************************************************************************5c044d, X-Amz-Date=20161209T111440Z, x-amzn-apigateway-api-id=l5tcmj0vlk,  Accept=application/json, User-Agent=AmazonAPIGateway_l5tcmj0vlk, Host=lambda.us-east-1.amazonaws.com, X-Amz-Content-Sha256=857a062940a7fbb8134bad1c007e9975a10bd8323c39f6040e797a98e87ea1f6, X-Amzn-Trace-Id=Root=1-584a9220-9cd537954952cca7daee32bf, Content-Type=application/json
Fri Dec 09 11:14:40 UTC 2016 : Endpoint request body after transformations: "resource":"/cancel-appointment","path":"/cancel-appointment","httpMethod":"PUT","headers":"Content-Type":" application/json","queryStringParameters":null,"pathParameters":null,"stageVariables":null,"requestContext":"accountId":"456204981758","resourceId":"xznq3u","stage":"test-invoke-stage","requestId":"test-invoke-request","identity":"cognitoIdentityPoolId":null,"accountId":"456204981758","cognitoIdentityId":null,"caller":"456204981758","apiKey":"test-invoke-api-key","sourceIp":"test-invoke-source-ip","accessKey":"ASIAJ5D7KU524H7CTTTQ","cognitoAuthenticationType":null,"cognitoAuthenticationProvider":null,"userArn":"arn:aws:iam::456204981758:root","userAgent":"Apache-HttpClient/4.5.x (Java/1.8.0_102)","user":"456204981758","resourcePath":"/cancel-appointment","httpMethod":"PUT","apiId":"l5tcmj0vlk","body":"\n    \"applicationId\": \"asdfsfa\"\n","isBase64Encoded":false
Fri Dec 09 11:14:40 UTC 2016 : Endpoint response body before transformations: "statusCode":"200","headers":"Content-Type":"application/json","body":"\"message\":\"Cancelled appointment with id null\""
Fri Dec 09 11:14:40 UTC 2016 : Endpoint response headers: x-amzn-Remapped-Content-Length=0, x-amzn-RequestId=adcadf25-be00-11e6-8855-75e96d772946, Connection=keep-alive, Content-Length=128, Date=Fri, 09 Dec 2016 11:14:39 GMT, Content-Type=application/json
Fri Dec 09 11:14:40 UTC 2016 : Method response body after transformations: "message":"Cancelled appointment with id null"
Fri Dec 09 11:14:40 UTC 2016 : Method response headers: Content-Type=application/json, X-Amzn-Trace-Id=Root=1-584a9220-9cd537954952cca7daee32bf
Fri Dec 09 11:14:40 UTC 2016 : Successfully completed execution
Fri Dec 09 11:14:40 UTC 2016 : Method completed with status: 200

如果我添加一个特定的方法，比如 POST 并且 not 将其设置为 Lambda 代理集成，我确实看到提供请求正文数据使其进入 lambda 函数，这是正确的- 序列化到我的 POJO 并返回

Execution log for request test-request
Fri Dec 09 11:22:02 UTC 2016 : Starting execution for request: test-invoke-request
Fri Dec 09 11:22:02 UTC 2016 : HTTP Method: POST, Resource Path: /cancel-appointment
Fri Dec 09 11:22:02 UTC 2016 : Method request path: 
Fri Dec 09 11:22:02 UTC 2016 : Method request query string: 
Fri Dec 09 11:22:02 UTC 2016 : Method request headers: 
Fri Dec 09 11:22:02 UTC 2016 : Method request body before transformations: 
    "appointmentId" : "sfssdf"

Fri Dec 09 11:22:02 UTC 2016 : Endpoint request headers: x-amzn-lambda-integration-tag=test-request, Authorization=****************************************************************************************************************************************************************************************************************************************************************************************************************************************a8dc41, X-Amz-Date=20161209T112202Z, x-amzn-apigateway-api-id=l5tcmj0vlk, Accept=application/json, User-Agent=AmazonAPIGateway_l5tcmj0vlk, Host=lambda.us-east-1.amazonaws.com, X-Amz-Content-Sha256=875dad4d4e05f8c12a7ca8aeaf69046d4153fc7f910e1eff1959cb011e8313a0, X-Amzn-Trace-Id=Root=1-584a93da-f841704d9feb371b31e41cb9, Content-Type=application/json
Fri Dec 09 11:22:02 UTC 2016 : Endpoint request body after transformations: 
    "appointmentId" : "sfssdf"

Fri Dec 09 11:22:02 UTC 2016 : Endpoint response body before transformations: "statusCode":"200","headers":"Content-Type":"application/json","body":"\"message\":\"Cancelled appointment with id sfssdf\""
Fri Dec 09 11:22:02 UTC 2016 : Endpoint response headers: x-amzn-Remapped-Content-Length=0, x-amzn-RequestId=b4f5efce-be01-11e6-91c3-5b1e06f831e2, Connection=keep-alive, Content-Length=130, Date=Fri, 09 Dec 2016 11:22:02 GMT, Content-Type=application/json
Fri Dec 09 11:22:02 UTC 2016 : Method response body after transformations: "statusCode":"200","headers":"Content-Type":"application/json","body":"\"message\":\"Cancelled appointment with id sfssdf\""
Fri Dec 09 11:22:02 UTC 2016 : Method response headers: X-Amzn-Trace-Id=Root=1-584a93da-f841704d9feb371b31e41cb9, Content-Type=application/json
Fri Dec 09 11:22:02 UTC 2016 : Successfully completed execution
Fri Dec 09 11:22:02 UTC 2016 : Method completed with status: 200

所以现在一切看起来都很好，但是当我使用 HTTP 方法 POST 对来自 PostMan 的 AWS API URL 进行实际测试时，我得到了带有 null 作为约会 ID 的响应，我可以在 CloudWatch 日志中看到约会 ID 做了未在输入 Appointment 对象上设置。

我觉得我在这里缺少一些基本的东西。任何帮助将不胜感激。

源码可以找到here

更新

通过将 lambda 处理程序函数切换为使用 Stream 而不是尝试将 JSON 序列化/反序列化为 POJO 来解决此问题。使用 API Gateway Lambda 代理时，处理程序的输入是一个复杂的 JSON 结构，我不想尝试将其复制为 Java/Scala 类。将输入作为流处理，将其解析为 JsonObject，然后使用 Gson 或等效库将消息正文转换为我的 POJO 会更容易。下面的示例处理程序，您还可以看到更大的示例here

class CancellationHandler 

  def cancelAppointment(request: InputStream, response: OutputStream, context: Context): Unit = 

    val logger = context.getLogger
    val parser: JsonParser = new JsonParser
    var inputObj: JsonObject = null
    val gson: Gson = new Gson

    try 

      inputObj = parser.parse(IOUtils.toString(request, "UTF-8")).getAsJsonObject

     catch 

      case e: IOException =>

        logger.log("Error while reading request\n" + e.getMessage)
        throw new RuntimeException(e.getMessage)

    

    val body: String = inputObj.get("body").getAsString
    val appointment: Appointment = gson.fromJson(body, classOf[Appointment])

    val apiGatewayProxyResponse = new ApiGatewayProxyResponse
    val cancelResponse = new CancelResponse

    cancelResponse.setMessage("Cancelled appointment with id " + appointment.getAppointmentId)

    apiGatewayProxyResponse.setBody(gson.toJson(cancelResponse))

    apiGatewayProxyResponse.setStatusCode("200")

    val headerValues = new util.HashMap[String, String]

    headerValues put("Content-Type", "application/json")

    apiGatewayProxyResponse.setHeaders(headerValues)

    val output: String = gson.toJson(apiGatewayProxyResponse)

    IOUtils.write(output, response, "UTF-8")
  

【参考方案1】：

Lambda 代理的输入形状将不同于常规非代理 Lambda 集成的形状。当然，这对您的用例很重要，因为您使用的是 Java/Scala，您必须在其中显式构建输入 POJO。

代理输入如下所示：

  "resource": "\/pets",
  "path": "\/pets",
  "httpMethod": "POST",
  "headers": null,
  "queryStringParameters": null,
  "pathParameters": null,
  "stageVariables": null,
  "requestContext": 
    ...
    "stage": "test-invoke-stage",
    "requestId": "test-invoke-request",
    "identity": 
      ...
    ,
    "resourcePath": "\/pets",
    "httpMethod": "POST"
  ,
  "body": "\n    \"foo\":\"bar\"\n", <---- here's what you're looking for
  "isBase64Encoded": false

文档：http://docs.aws.amazon.com/apigateway/latest/developerguide/api-gateway-create-api-as-simple-proxy-for-lambda.html

【讨论】：

谢谢。我现在正在玩这个，我想我需要将我的处理程序转换为使用流，以便我可以更轻松地将输入作为 JSON 处理。不过，这无疑将我推向了正确的方向。当我确定一个可行的解决方案时，我会更新这篇文章。