Lambda 函数中的过滤条件

Question

我想使用 AWS CDK 在我的 lambda 上启用 DynamoDB 流，我可以这样做，但我也想在 lambda 上启用过滤条件

但我收到此错误：

过滤器模式定义无效。（服务：AWSLambda；状态代码：400；错误代码：InvalidParameterValueException

这是我从 DynamoDB 流中收到的事件：

{
    "input": {
        "Records": [
            {
                "eventID": "e92e0072a661a06df0e62e411f",
                "eventName": "INSERT",
                "eventVersion": "1.1",
                "eventSource": "aws:dynamodb",
                "awsRegion": "<region>",
                "dynamodb": {
                    "ApproximateCreationDateTime": 1639500357,
                    "Keys": {
                        "service": {
                            "S": "service"
                        },
                        "key": {
                            "S": "key"
                        }
                    },
                    "NewImage": {
                        "service": {
                            "S": "service"
                        },
                        "channel": {
                            "S": "email"
                        },
                        "key": {
                            "S": "key"
                        }
                    },
                    "SequenceNumber": "711500000000015864417",
                    "SizeBytes": 168,
                    "StreamViewType": "NEW_IMAGE"
                },
                "eventSourceARN": "arn:aws:dynamodb:<region>:<account>:table/table-name/stream/2021-12-14T13:00:29.888"
            }
        ]
    },
    "env": {
        "lambdaContext": {
            "callbackWaitsForEmptyEventLoop": true,
            "functionVersion": "$LATEST",
            "functionName": "functionName",
            "memoryLimitInMB": "128",
            "logGroupName": "/aws/lambda/functionName",
            "logStreamName": "2021/12/14/[$LATEST]028531c7b489b8ec69bace700acc0",
            "invokedFunctionArn": "arn:aws:lambda:<region>:<account>:function:functionName",
            "awsRequestId": "c72e80252-4722-b9f0-a03b7f8b820e"
        },
        "region": "<region-name>"
    }
}

事件源映射代码为：

const mapping = new lambda.CfnEventSourceMapping(this, 'event', {
  functionName: "functionName,
  batchSize: 1,
  bisectBatchOnFunctionError: true,
  startingPosition: lambda.StartingPosition.TRIM_HORIZON,
  eventSourceArn: <stream-arn>,
  filterCriteria: filter,
});

我想在此处获取事件名称INSERT和通道电子邮件。过滤条件的值应该是多少？它不适合我

Answer 1

<编辑> v2.42.0 中添加的 CDK 过滤器助手

不再需要原来的解决方法。CDK 现在具有适用于 Lambda、Kinesis 和 SQS 的事件源过滤器。将过滤器传递给 L2EventSourceMapping构造：

const source: EventSourceMapping = new lambda.EventSourceMapping(this, "EventSourceMapping",{
    target: func,
    eventSourceArn: table.tableStreamArn,
    startingPosition: lambda.StartingPosition.TRIM_HORIZON,
    filters: [
      lambda.FilterCriteria.filter({
        eventName: lambda.FilterRule.isEqual("INSERT"),
        dynamodb: {  NewImage: { channel: { S: lambda.FilterRule.isEqual("email") } },},
      }),
    ],
  }
);

</编辑>

以下是Pattern带有channelof的新记录的 DynamoDB 流过滤器语法email：

`{ \"eventName\": [\"INSERT\"], \"dynamodb\": { \"NewImage\": {\"channel\": { \"S\" : [\"email\"]}} } }`

换句话说，这Pattern是一个带有转义引号的字符串化 JSON过滤规则。该模式应用于每个流记录。

这是完整的 CDK 语法。代码以通常的 L2 开头EventSourceMapping。然后，它使用逃生舱口语法来设置FilterCriteria底层L1 CfnEventSourceMapping：

// start with the L2 type - Note: the OP code starts with a L1 `CfnEventSourceMapping`
const source: EventSourceMapping = new lambda.EventSourceMapping(this, 'EventSourceMapping', {
  target: func,
  eventSourceArn: table.tableStreamArn,
  startingPosition: lambda.StartingPosition.TRIM_HORIZON,
});

// escape hatch - get a L1 reference
const cfnSouce = source.node.defaultChild as lambda.CfnEventSourceMapping;

cfnSouce.addPropertyOverride('FilterCriteria', {
  Filters: [
    {
      Pattern: `{ \"eventName\": [\"INSERT\"], \"dynamodb\": { \"NewImage\": {\"channel\": { \"S\" : [\"email\"]}} } }`,
    },
  ],
});