Creating a Surveys API with the AWS HTTP API and Python

Reading time ~5 minutes

Want to learn more about AWS and Projects? Sign up for my mailing list!

This is project 15 in my Twenty Projects in Twenty Days series! This time, we’re looking at how to create an AWS HTTP API with Python. This will be the Python version of yesterday’s project where we created a Surveys service API to track three entities - customers, customer surveys, and survey responses. Let’s get started!


Make sure that you get all the requirements setup first:

  1. Your AWS Account
  2. The AWS CLI installed and configured
  3. The Serverless Framework
  4. The code

Our Code

After getting all these dependencies, you can look at the project code:

├── backend
│   ├──
│   ├──
│   └──
└── serverless.yml

For everything in the serverless.yml except for the runtime: python3.8 it is the same as yesterday’s project! This is because I’ve kept the same structure between both projects and just kept the JavaScript file names, moved the code to Python, and changed the extension. For a full explanation of the serverless.yml file just take a look at my description of it in yesterday’s post.

The main thing you need to recognize is that you might need to change or remove the first two lines with the org and app value on them depending on if you’re using Serverless Framework Pro or not.

So let’s look at how we handle our entities inside our Python files.

First, we import some dependencies, including boto3 which will be used to interact with AWS services like DynamoDB.

import boto3
import json
import os

Then, we create the DynamoDB client, and import the name of the table from the DYNAMODB_TABLE environment variable we defined in serverless.yml. We use both of these to create a table resource in order to easily interact with the DynamoDB table.

dynamodb = boto3.resource('dynamodb')
table = dynamodb.Table(TABLE_NAME)

Next, we write our first function to create customers. It starts by loading in a JSON payload coming from the API and then extracting the customer_id and profile_data from that JSON object.

def create(event, context):
    body = json.loads(event['body'])
    customer_id = body['customer_id']
    profile_data = body['profile_data']
    # ...

Then, we create the item we’ll be storing in DynamoDB and use the put_item() operation to send it in.

    # ...
    item = {
        'pk': 'CUSTOMER#' + customer_id,
        'sk': 'PROFILE#' + customer_id,
        'profile_data': profile_data

Finally, we return JSON of the item we put into DynamoDB.

    return {
        'statusCode': 200,
        'body': json.dumps(item)

This sort of structure is basically what we’re doing with all the other entities when we create them. The only other difference is sometimes we create a unique UUID for them when we create them (like surveys or responses).

The other way we use to process requests can be seen in the get() function which we use to get a customer’s information out of our table. We start by getting a path parameter from the URL by looking at the pathParameters part of the incoming event:

def get(event, body):
    customer_id = event['pathParameters']['id']
    # ...

Then, we fetch the item with the get_item() operation using the customer_id we pulled from the URL path parameters:

    # ...
    item = table.get_item(
            'pk': 'CUSTOMER#' + customer_id,
            'sk': 'PROFILE#' + customer_id
    # ...

Finally, we return the item as a JSON object back through the API:

    # ... 
    return {
        'statusCode': 200,
        'body': json.dumps(item)

And that’s it! That’s how the majority of these requests work.

Deploying and Testing the Application

We deploy and test this the same way as yesterday’s project. First, we run serverless deploy to get the project deployed.

If you see an error like this: An error occurred: surveysTable

You may have already deployed yesterday’s project and need to either remove it first or to change the name of the DynamoDB table in this project in the serverless.yml file first.

Then, we should see something like this:

Service Information
service: surveys
stage: dev
region: us-east-1
stack: surveys-dev
resources: 54
api keys:
  POST -
  GET -{id}
  POST -
  GET -{id}
  GET -{id}/surveys
  POST -
  GET -{id}
  GET -{id}/responses
  createCustomer: surveys-dev-createCustomer
  getCustomer: surveys-dev-getCustomer
  createSurvey: surveys-dev-createSurvey
  getSurvey: surveys-dev-getSurvey
  getAllCustomerSurveys: surveys-dev-getAllCustomerSurveys
  createResponse: surveys-dev-createResponse
  getResponse: surveys-dev-getResponse
  getAllResponses: surveys-dev-getAllResponses

In order to test the API we can use curl or something like Postman. Make sure to change the API URL from my URL to whatever comes out when you deploy your service. Here’s the command I used:

curl --request POST '' \
--header 'Content-Type: application/json' \
--data-raw '{
	"customer_id": "1",
	"profile_data": {
		"cool": "things"

When you do this, you should see a response that looks like this:

{"pk": "CUSTOMER#1", "sk": "PROFILE#1", "profile_data": {"cool": "things"}}

To make sure it worked, run another curl request to get the data out of the API:

curl --request GET ''

You should see the same response:

{"pk": "CUSTOMER#1", "sk": "PROFILE#1", "profile_data": {"cool": "things"}}%  

Now you can go through and test the other endpoints out if you’d like!

What Next?

With a newly deployed API that you’ve confirmed works properly, you might want to learn more about some of these concepts!

To learn more about DynamoDB you can sign up for a free trial to Pluralsight where I have a course on connecting DynamoDB to your applications.

You can also learn about other ways to design serverless APIs like this in some of my previous posts:

If you want more projects to play with then keep an eye on my Twenty Projects in Twenty Days series as I keep adding more!

Have questions or comments or just want to learn more? Join me (Fernando) in the Moonrise Labs Discord!