lfm

The AWS Lambda Function Manager

Hi There!

You're probably wondering what lfm is good for. Well, suppose you wake up one day and suddenly realize you need a Lambda function for AES encryption. You could roll your own solution, but ain't nobody got time for that. So you do a quick GitHub search and -- lo and behold -- there's this beauty.

Normally you'd have to clone the repo, install its dependencies, zip it up, and deploy it yourself. But with lfm, you can just do:

$ lfm deploy gh:willyg302/aws-lambda-aes

What?! No Way!

Yup, and it works with a few other things, too:

Source	Command
Git repo	We just covered this, bro
GitHub Gist, 1 file	lfm deploy gist:willyg302/25b8f32e6784aca03a27
GitHub Gist, 2+ files	lfm deploy gist:willyg302/560ab5d328b37d2cd4cc
Local file	lfm deploy awesome-sauce.js
Local directory	lfm deploy my-sweet-function/
Working directory	lfm deploy
S3 bucket	lfm deploy s3:mah-bucket
File on S3	lfm deploy s3:mah-bucket/hello-world.js
Directory on S3	lfm deploy s3:mah-bucket/my-function
Valid URL	lfm deploy http://www.my-site.com/my-function.js

In addition, you can override any of the usual config (see lfm deploy -h for options).

So What's the Secret?

Glad you asked! The magic of lfm is a special file called .lambda.yml at the root of a repo or directory. It looks something like this:

config:
  FunctionName: my-awesome-sauce-function
  Handler: index.handler
  Runtime: nodejs
  Description: This function will blow your mind
  Role: execution_role_arn  # Not recommended, especially in a public repo
install: npm install --production
ignore:
  - test/
  - README.md

config holds all that junk you're used to passing to create-function, so you don't have to type it out any more. Note that any config you provide to lfm deploy will override the values in this file.

install is a single command for installing dependencies before zipping up the function. In this example, we want to install some Node modules so we use npm install --production, but you can just as well shell out to make for more heavy lifting.

ignore is a list of paths that will not be added to the zip file.

NOTE: You don't need a .lambda.yml to be able to use lfm; it's just easier that way. Without the config file, you'll have to provide all the required config as command-line arguments.

But I Only Have One File...

Cool beans. Just stick that config as YAML in your front matter:

/**
 * config:
 *   FunctionName: hello-world
 *   Handler: hello-world.handler
 *   Runtime: nodejs
 *   Description: A simple AWS Lambda function
 */
console.log('Loading function');

exports.handler = function(event, context) {
    console.log('value1 =', event.key1);
    console.log('value2 =', event.key2);
    console.log('value3 =', event.key3);
    context.succeed(event.key1);  // Echo back the first key value
    // context.fail('Something went wrong');
};

Save this in hello-world.js and you're good to go! It's probably worth mentioning that since you have just a single file, you won't be doing any installing or ignoring of files or anything like that.

But Roles Are So Annoying...

Wish you didn't have to type the role every time? It's a bad idea to include the role in your .lambda.yml or front matter for security reasons, but we've got you covered with one extra layer of config:

# Save this in ~/.aws/lambda.yml
roles:
  default: arn:aws:iam::some-role
  bunnies: arn:aws:iam::a-much-fluffier-role

Now you can rest those poor fingertips of yours:

$ lfm deploy awesome-sauce.js  # Will use default role
$ lfm deploy awesome-sauce.js -r bunnies  # You guessed it!
$ lfm deploy awesome-sauce.js --role arn:aws:iam::blahblah  # If you really want to...

Okay, I'm Sold, Gimme!

lfm is still in early development, so don't expect to be finding it on PyPI. You can grab it with a quick pip install git+https://github.com/willyg302/lfm.git@master.

If, like Rust, lfm decides to eat your laundry, please file an issue.

I Want Moar

Don't fret, we hear you! There are currently plans for the following features:

• A Site: All fancy-shmancy and such, so you can search for functions and be all social from the comfort of your browser.
• Webhooks: A bit far-fetched, but wouldn't it be cool to git push and kick off a Lambda deploy?

Testing

Run unit tests with python setup.py test.

Run integration tests with python integ. See integ/__init__.py for more information.