This is a GroupMe birthday bot that runs on two AWS Lambda functions, bd_checker and bd_responder, along with an associated DynamoDB instance for storing the birthdays. The name "Bee-Dee" is a playful take on the phrase "b-day".
bd_checker is a simple function that's connected to an EventBridge Scheduler to run every morning; it queries the DynamoDB instance for any birthdays that day, and sends a congratulatory message to the GroupMe group for each one.
bd_responder is slightly more complicated, as it receives message callbacks from the GroupMe group, and has to validate and parse those messages to check if it needs to respond. It begins by checking the user agent of the incoming event to check that it's from GroupMe's BotNotifier agent (to filter out anyone just trawling AWS function URLs), and then checks that the sender of the message is a user (to prevent the bot from responding to itself or any other bots, mostly). Finally, it parses the message for the starting command syntax (!bd, case insensitive), and then for either of its command phrase (hello/hi and me, case insensitive), to which it responds appropriately. If the starting command syntax is found but not a valid command phrase, it responds with an error message.
shared is the final crate in this workspace, and it's a library crate that contains logic that is shared between the two functions: querying the DynamoDB instance, sending GroupMe messages, etc.
I made an earlier version of this bot in Python that ran on a Raspberry Pi, but I decided to use the project as a good starting point for getting familiar with AWS, and since I prefer Rust to Python, I rewrote it in Rust for the fun of it.
As for why I'm open-sourcing my work, the AWS SDKs aren't very well documented at the time of writing this, so I had to figure out a lot by trial and error. And, when you have to re-build and re-upload for every single trial, it takes a long time to figure out little issues. I hope that this repository can be a learning resource to help others looking to tackle similar projects.
-
A 2024 edition of the Rust programming language, which is versions
1.85.0and above. You can download Rust from the Rust website. -
Cargo Lambda, a CLI tool developed by Amazon that integrates with the Rust package manager (
cargo) to build and deploy Lambda functions. You can find download instructions on the Cargo Lambda website.
-
Clone this repository:
git clone https://github.com/chris-t-jansen/bee-dee-lambda.git
-
Navigate to the repository folder:
cd bee-dee-lambda -
Add
secrets.rsto/shared/src/. This file contains the necessary secrets to run the bot, such as the ARN for the DynamoDB instance. Secrets should be declared as public&strconstants (e.g.pub const MY_SECRET: &str = r"my_very_special_secret"). -
Run the build command to build the functions:
cargo lambda build --release
Important
If you're deploying to a Lambda function that is setup to use the arm64 architecture instead of the x86_64 architecture, you'll need to append the --arm64 flag to the above build command.
That should build both binary functions (bd_checker and bd_responder) to their own folders in the /target/lambda/ directory. Both built binaries should be named bootstrap, which they need to be for the Lambda runtime to detect and run them properly.
-
AWS Lambda functions with IAM permissions to read from DynamoDB (for querying birthdays) and write logs to CloudWatch (for logging information/errors).
- The responder function will need a public function URL to use as the callback URL for the GroupMe bot settings (see below).
- The checker function will need to be connected to an EventBridge Scheduler to run it on a schedule.
-
A DynamoDB instance containing the birthdays. The items should all have the following attributes:
user_id: aNumberof the user's GroupMe user ID. You can find this by sending an HTTPGETrequest as detailed in this GroupMe API documentation.fullname: aStringof the user's name as the bot will print it in birthday messages (e.g. for "John D. Smith", it'll print "It's John D. Smith's birthday today!").month_num: aNumberof the person's birth month (e.g.4for April). Should be within the range of1 - 12.day_num: aNumberof the person's calendar birth date (e.g.20if they were born on the 20th of the month). Should be within the range of1 - 31.
The
user_idfield should be the partition key with no sort key, and you'll need to configure a global secondary index (GSI) namedmonth-day-indexwithmonth_numas the partition key andday_numas the sort key so that the bot can query birthdays based on the current date. -
A GroupMe bot, which you can create on this page of the GroupMe Developers website. You'll need the bot ID as one of the secrets in your
secrets.rsfile (see above), and you'll need to set the callback URL to the public function URL of the responder function (see above).
Note
Theoretically, there's a way to deploy the Lambda functions by running cargo lambda deploy (see here). However, I don't have the AWS CLI installed, so I do it the manual way of uploading .zip files.
-
After building the binaries (see above), compress the binaries to individual zip files (the parentheses create a subshell, so no need to
cd -after each command):(cd target/lambda/bd_checker && zip bootstrap.zip bootstrap) (cd target/lambda/bd_responder && zip bootstrap.zip bootstrap)
-
Modify the permissions of the zip files so that they're executable by the Lambda runtime on AWS:
chmod a+x target/lambda/bd_checker/bootstrap.zip chmod a+x target/lambda/bd_responder/bootstrap.zip
-
Navigate to the page for your Lambda function in the AWS Management Console.
-
Under the
Codetab in theCode Sourcebox, click onUpload fromand choose.zip file. -
Select the
bootstrap.zipfile in the build directory for that function (e.g./target/lambda/bd_responder/bootstrap.zipfor the responder Lambda function). -
Upload the zip file and wait for the function to finish updating.
-
Repeat steps 3 - 6 for the other Lambda function.
You should now have both Lambda functions uploaded and automatically running smoothly. If things don't appear to be working, check the latest log streams in CloudWatch. If you can't figure it out, feel free to open an issue.
Licensed under either of
- Apache License, Version 2.0, (LICENSE-APACHE or http://www.apache.org/licenses/LICENSE-2.0)
- MIT license (LICENSE-MIT or http://opensource.org/licenses/MIT)
at your option.