MongoDB + CodeWhisperer

Quick demo that shows how using CodeWhisperer and MongoDB can boost developer productivity.

Context: This is an API (built with FastAPI) that provides access to a articles (posts) stored in MongoDB. During the demo we'll augment the API to provide access to the posts via a customizable link (permalink) and to be able to retrieve the total posts by tags. Most of the changes will be writen by CodeWhisperer, showcasing how it can boost developer productivity.

Key Takeaways: How CodeWhisperer, trained with MongoDB dataset, can boost developer productivity in real-world API & Application developement.

Step 0 (Preparation)

a) Make sure python3 is installed. Step 1 (see below) will ensure you have all dependencies installed on a virtual environment.

b) Ensure the AWS Toolkit extension is installed and that you sign-in to use CodeWhisperer. You may have to sign-in using a personal account (as the MongoDB account currently have some permission restrictions). You can follow this guide.

Once the extension is properly installed you should see the following:

1 - AWS buttom visible on the Activity Bar.
2 - Amazon Q buttom visible on the Activity Bar.
3 - In AWS's Toolkit (click on AWS buttom), Auto-Suggestion enabled.

c) Make sure you have a MongoDB Atlas cluster with the sample dataset loaded. This demo uses the sample_training dataset.

d) Copy the .env.example file to .env and replace ATLAS_URI with the proper connection string of your Atlas cluster.

Step 1

Create a virtual environment (an isolated python enrivonment), install dependencies and run the application using the following commands:

python3 -m venv venv            # create python virtual env
source venv/bin/activate        # activate virtual env
pip install -r requirements.txt # install dependencies

uvicorn main:app --reload       # run web server

Now that the application is running with uvicorn, proceed to show how it works.

Step 2

Show how the API works

Access a list of documents http://127.0.0.1:8000/posts/

Access a single document using its ObjectId http://127.0.0.1:8000/posts/50ab0f8bbcf1bfe2536dc3fa

Show structure of a single document, and tell that it would be ideal to be able to get a document by its permalink and not by its ObjectId. The permalink is a customizable string that can reflect the actual article title.

Try to access a document by its permalink http://127.0.0.1:8000/posts/jNsgObovWyKEoXNydtis

ℹ️ The ObjectId and the permalink above already exist in the sample dataset

Step 3

Let's use CodeWhisperer to update the get route to use the permalink instead of the ObjectId

On routes.py delete everything below def find_post(id: id, and alter the route declaration to the following:

@router.get("/{permalink}", response_description="Get a single post by permalink", response_model=Post)
def find_post(permalink: str, request: Request):
  	█

Place your cursor on the position of the block character above ( █) and CodeWhisperer should autocomplete with the proper suggestion. Ideally the suggestion will look like the following:

@router.get("/{permalink}", response_description="Get a single post by permalink", response_model=Post)
def find_post(permalink: str, request: Request):
    if (post := request.app.db["posts"].find_one({"permalink": permalink})) is not None:
        return post
    raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail=f"Post with permalink {permalink} not found")

Save the file.

Step 4

Since uvicorn is running, it should auto-reload the changes automatically.

Try to access a document by its permalink. This time it should work. ✅ http://127.0.0.1:8000/posts/jNsgObovWyKEoXNydtis

Step 5

Now tell the audience about a requirement of displaying the total amount of posts of each tag. Explain how it would require a new route to run a MongoDB aggregation to make it possible.

before @router.get("/{permalink}" add the following code snippet:

⚠️ it's important this is added before the permalink route!

@router.get("/tags", response_description="Get a total of posts of each tag")
def post_totals_per_tag(request: Request):
    """
    Get the total of posts of each tag using MongoDB, return list
    """
    █

Place your cursor on the position of the block character above ( █) and CodeWhisperer should autocomplete with the proper suggestion. Ideally the suggestion will look like the following:

@router.get("/tags", response_description="Get a total of posts of each tag")
def post_totals_per_tag(request: Request):
    """
    Get the total of posts of each tag using MongoDB, return list
    """
    return list(request.app.db["posts"].aggregate([
        {"$unwind": "$tags"},
        {"$group": {"_id": "$tags", "total": {"$sum": 1}}}
        # {"$project": {"_id": 0, "tag": "$_id", "total": 1}} # valid (CodeWhisperer might suggests this)
    ]))

⚠️ it's important the result is converted into a list. AKA: list()

ℹ️ CodeWhisperer might suggest additional lines (often comments showing what the output looks like). You can discard those lines or use Cmd/CTRL+-> to accept just the code portion of its suggestion.

Step 6

1 - Select the aggregation the body of the function that was generated in the previous step. In the example above, this is the portion that should be selected:

    return list(request.app.db["posts"].aggregate([
        {"$unwind": "$tags"},
        {"$group": {"_id": "$tags", "total": {"$sum": 1}}}
        # {"$project": {"_id": 0, "tag": "$_id", "total": 1}} # valid (CodeWhisperer might suggests this)
    ]))

2 - In the Activity Bar, select Amazon Q and ask the following to it:

3 - "Explain selected code"

ℹ️ feel free to click in the "explain selected code" suggestion if it shows up

4 - Quickly read the response and highlight how useful it is for developers to quickly understand portions of the code that they are not familiar with.

Here's how Step 6 should look like:

Step 7

Debrief

Shut down uvicorn with Ctrl+C

Deactivate's python virtual environment by running deactivate on the terminal

deactivate # once done, deactivate venv