Leveraging Amazon Bedrock to enhance customer service with AI-powered Automated Email Response > Setup Lambda Functions > Create Extract Email Function

Create Extract Email Function

This is the first function, with missions:

Extract email content from MIME type to JSON data
Send that data to SQS.

Why i need this function?

Because many reasons:

I want to save money when calling Amazon Bedrock and lambda function in Generate email function (If a email with wrong data, we can parse them to our formatted data first before calling AI agent, reduce the tokens send to LLM model)
Easier for debugging
In real life, maybe some hater will attack this endpoint with spamming, so we can put a process function here.

Let’s create the lambda function

In the AWS Console, search “lambda” and choose Lambda

Choose Create function

In the AWS Console, choose Author from scratch, Function name is ExtractEmailLambda, because i our code is use nodejs so we choose Nodejs 22.x. Then we choose existing role ExtractEmailFunctionRole then click Create function

Next, we need to upload our source code to this function, you can download it index.zip or you can build them with my github repository here. After got that file, you need to upload it to lambda Code > Upload from > .zip file

Click Save to upload it

Upload successful
Next, we need to config environment variable for our function to interact with other service

We have 2 variables

Key	Value	Description
`QUEUE_NAME`	`sqs-generate-email-content-queue`	This is the queue name we will send message (email content) to generate response
`BUCKET_NAME`	`ai-powered-email-auto-replies`	This bucket for getting email content, because SES not emit email content, it just emit the messageId (the key of the email object in our bucket)

Click Save

Update environment variables successfully

Details about our function

Configuration	Value	Description
function_name	`ExtractEmailLambda`	The unique name identifier for the Lambda function
role	`ExtractEmailFunctionRole`	IAM role that defines the permissions and access policies for the function
runtime	nodejs20.x	The execution environment and version for the Lambda function (Node.js 20)
handler	index.handler	The entry point for the function
memory_size	128	Amount of memory allocated to the function in MB (128MB)
timeout	30	Maximum execution time allowed for the function in seconds

Received event from SES and get email content from S3

// Helper function to extract email data from SES event
const extractEmailData = async (event: SESEvent) => {
  const sesRecord = event.Records[0].ses;
  const messageId = sesRecord.mail.messageId;
  const emailData = {
    sender: sesRecord.mail.source,
    key: messageId,
  };

  // Get email content from S3
  const getObjectCommand = new GetObjectCommand({
    Bucket: env.bucketName,
    Key: `received-email/${messageId}`
  });

  const response = await s3Client.send(getObjectCommand);
  const emailContent = await response.Body?.transformToString();

  // Parsing MIME type to JSON object
  const { text, subject } = await simpleParser(emailContent);
  return {
    ...emailData,
    subject,
    content: text || "No email data"
  };
};

Sending parsed content message to SQS

const body = await extractEmailData(event);
const command = new GetQueueUrlCommand({ // GetQueueUrlRequest
    QueueName: env.queueName, // required
});
const queueUrlResponse = await sqsClient.send(command);
const params = {
    QueueUrl: queueUrlResponse.QueueUrl,
    MessageBody: JSON.stringify(body),
};
await sqsClient.send(new SendMessageCommand(params));

Full source code

import { Handler, SESEvent } from 'aws-lambda';
import { GetQueueUrlCommand, SQSClient, SendMessageCommand } from "@aws-sdk/client-sqs";
import { GetObjectCommand, S3Client } from "@aws-sdk/client-s3";
import { Logger } from '@aws-lambda-powertools/logger';
import * as mailparser from 'mailparser';

require('dotenv').config();

const simpleParser = mailparser.simpleParser;
const logger = new Logger();

const env = {
  queueName: process.env.QUEUE_NAME || "",
  bucketName: process.env.BUCKET_NAME || ""
}

const sqsClient = new SQSClient();
const s3Client = new S3Client();

// Helper function to extract email data from SES event
const extractEmailData = async (event: SESEvent) => {
  const sesRecord = event.Records[0].ses;
  const messageId = sesRecord.mail.messageId;
  const emailData = {
    sender: sesRecord.mail.source,
    key: messageId,
  };
  // logger.info("Email Data: ", JSON.stringify(emailData, null, 2));

  // Get email content from S3
  const getObjectCommand = new GetObjectCommand({
    Bucket: env.bucketName,
    Key: `received-email/${messageId}`
  });

  const response = await s3Client.send(getObjectCommand);
  const emailContent = await response.Body?.transformToString();
  const { text, subject } = await simpleParser(emailContent);
  return {
    ...emailData,
    subject,
    content: text || "No email data"
  };
};

export const handler: Handler = async (event: SESEvent): Promise<void> => {
  try {
    const body = await extractEmailData(event);
    // logger.info({
    //     message: "Extracted email data",
    //     body: body
    // });
    const command = new GetQueueUrlCommand({ // GetQueueUrlRequest
      QueueName: env.queueName, // required
    });
    const queueUrlResponse = await sqsClient.send(command);
    const params = {
      QueueUrl: queueUrlResponse.QueueUrl,
      MessageBody: JSON.stringify(body),
    };
    await sqsClient.send(new SendMessageCommand(params));
  }
  catch (error) {
    logger.error(error as string);
  }
};

Create SES rule for trigger extract email function

Go to SES dashboard and scroll down at the left panel Configuration > Email receiving > Create rule set

Enter rule set name TriggerEmailIncoming and choose Create rule set

Then choose “Create rule”

We enter any name ExtractAndSaveToS3 and choose Next

At this step, as you can see, you can add many Recipient conditions look like the Guidelines, for simple i just want to trigger Lambda function and save email to S3. Then choose Next

Choose Add new action > Invoke AWS lambda function

Then we choose our function ExtractEmailLambda and add another action to save email to S3

Next we need to choose existing S3 bucket and folder received-email. We also provide them a policy to allow SES put object to S3, choose Create IAM Role