Autocode Setup |
Node |
Web |
Python (alpha) |
Ruby (alpha)

Introduction

Autocode is a fastest and easy way to build web
services and APIs that respond to external SaaS events. The Autocode ecosystem
treats external SaaS APIs as single-line function calls with the use of the
lib package in NPM. The Autocode CLI allows
you to interact seamlessly with the following components of Autocode:

1. Executing APIs on the Autocode standard library
2. Uploading new APIs / web services to Autocode’s hosting platform

Autocode is based on Function as a Service (“serverless”) architecture,
initially popularized by AWS Lambda. You can use Autocode to build modular, scalable APIs
for yourself and other developers in minutes without having to manage servers,
gateways, domains, write documentation, or build SDKs. Your development workflow
has never been easier – focus on writing code you love, let Autocode handle
everything else.
Autocode uses an open specification called
FunctionScript for function definitions and
execution. If you run into concerns or questions as you’re building from this
guide, please reference the FunctionScript repository. 🙂
You can view services published by our large and growing developer community
on the Autocode standard library page.

Table of contents

1. Getting started
2. Creating your first service
3. Connecting service endpoints
4. Accessing your APIs from other applications
5. Accessing your APIs over HTTP
6. Version control and package management
7. Logging
8. Additional functionality
9. Acknowledgements
10. Contact

Getting started

To get started with Autocode, first make sure you have Node 8.x or later installed,
available from the official Node.js website. Next install
the Autocode CLI tools with:

$ npm install lib.cli -g

And you’re now ready to start building!

Creating your first service

The first thing you’ll want to do is create a workspace. Create a new directory
you intend to build your services in and initialize the workspace.

$ mkdir autocode-workspace
$ cd autocode-workspace
$ lib init

You’ll be asked for an e-mail address to log in to the Autocode registry.
If you don’t yet have an account, you can create one by going to https://autocode.com/.
Note that you can skip account creation with lib init --no-login.
You’ll be unable to use the registry, but it’s useful for creating workspaces
when you don’t have internet access.
Next, create your service:

$ lib create <service>

You’ll be asked for a default function name, which is the entry point
into your service (useful if you only want a single entry point). This will automatically
generate a service project scaffold in autocode-workspace/<username>/<service>.
Once created, enter the service directory:

$ cd your_username/your_service

In this directory, you’ll see something like:

- functions/
  - __main__.js
- package.json
- env.json
- WELCOME.md
- README.md

At this point, there’s a “hello world” function that’s been automatically
created (__main__.js). Autocode comes paired with a simple lib command for
testing your functions locally and running them in the cloud.
To test your function:

$ lib .
"hello world"

If we examine the functions/__main__.js file, we see the following:

/**
* A basic Hello World function
* @param {string} name Who you're saying hello to
* @returns {string}
*/
module.exports = async (name = 'world', context) => {
  return `hello ${name}`;
};

We can pass parameters to it using the CLI by specifying named parameters:

$ lib . --name "dolores abernathy"
"hello dolores abernathy"

Note that context is a magic parameter (automatically populated with
execution details, when provided) as is callback (terminates execution),
so these don’t need to be documented and can not be specified as
parameters when executing the function.

Pushing to the cloud

To push your function to a development environment in the cloud…

$ lib up dev
$ lib your_username.your_service[@dev]
"hello world"

And to release it (when you’re ready!)

$ lib release
$ lib your_username.your_service
"hello world"

You can check out your service on the web, and use it in applications using our
functions gateway, api.stdlib.com.

https://your_username.api.stdlib.com/your_service/

That’s it! You haven’t written a line of code yet, and you have mastery over
building a service, testing it in a development (staging) environment online,
and releasing it for private (or public) consumption.
Note: By default, APIs that you publish with lib release will have a
documentation page in the Autocode public registry. You can keep your page private,
as well as restrict execution access or add collaborators to your API,
by modifying your API’s permissions. For more information, see this docs page.
Another Note: Staging environments (like the one created with lib up dev)
are mutable and can be replaced indefinitely. Releases (lib release) are
immutable and can never be overwritten. However, any service can be torn down
with lib down <environment> or lib down -r <version> (but releases
can’t be replaced once removed, to prevent mistakes and / or bad actors).

Connecting service endpoints

You’ll notice that you can create more than one function per service. While
you can structure your project however you’d like internally, it should also
be noted that these functions have zero-latency access to each other. You
can access them internally with the lib package on NPM,
which behaves similarly to the lib command for testing. Use:

$ npm install lib --save

In your main service directory to add it, and use it like so:

functions/add.js

module.exports = async (a = 0, b = 0) => {
  return a + b;
};

functions/add_double.js

const lib = require('lib');

module.exports = async (a = 0, b = 0, context) => {
  let result = await lib[`${context.service.identifier}.add`]({a: a, b: b});
  return result * 2;
};

In this case, calling lib .add --a 1 --b 2 will return 3 and lib .add_double --a 1 --b 2
will return 6. The context magic parameter is used for its
context.service.identifier property, which will return the string "your_username.your_service[@local]"
in the case of local execution, "your_username.your_service[@ENV]" when deployed to an
environment or release (where ENV is your environment name or semver).

Accessing your APIs from other applications

As mentioned in the previous section, you can use the NPM lib package that’s
available on GitHub and NPM to access your
APIs from legacy Node.js applications and even the web browser. We’ll
have more SDKs coming out in the following months.
An existing app would call a function (username.bestTrekChar with version 0.2.1):

const lib = require('lib');

let result;

try {
  result = await lib.username.bestTrekChar['@0.2.1']({name: 'spock'});
} catch (err) {
  // handle error
}

// do something with result

Which would speak to your API…

module.exports = async (name = 'kirk') => {

  if (name === 'kirk') {
    return 'why, thank, you, too, kind';
  } else if (name === 'spock') {
    return 'i think this feeling is called "pleased"';
  } else {
    throw new Error('Only kirk and spock supported.');
  }

};

Accessing your APIs over HTTP

We definitely recommend using the lib library on NPM
to make API calls as specified above, but you can also make HTTPS
requests directly to the Autocode gateway. HTTP query parameters are mapped
automatically to parameters by name.

https://username.api.stdlib.com/liveService@1.12.2/?name=BATMAN

Maps directly to:

/**
* Hello World
* @param {string} name
* @returns {string}
*/
module.exports = async (name = 'world') => {
  // returns "HELLO BATMAN" from above HTTP query
  return `Hello ${name}`;
};

Version control and package management

A quick note on version control – Autocode is not a replacement for normal
git-based workflows, it is a supplement focused around service creation and
execution.
You have unlimited access to any release (that hasn’t been torn down)
with lib download <serviceIdentifier> to download and unpack the
tarball to a working directory.
Tarballs (and package contents) are closed-source.
Nobody but you (and potentially your teammates) has access to these. It’s up to
you whether or not you share the guts of your service with others on GitHub or NPM.
As mentioned above: releases are immutable and can not be overwritten (but can
be removed, just not replaced afterwards) and development / staging environments
are mutable, you can overwrite them as much as you’d like.

Logging

Logging for services is enabled by default. When running a service locally with
lib . or lib .functionname, all logs will be output in your console. The very
last output (normally a JSON-compatible string) is the return value of the function.
To view remote logs (in dev or release environments), use the following syntax:

:: Lists all logs for the service
$ lib logs username.servicename

:: Lists main service endpoint logs for "dev" environment
$ lib logs username.servicename[@dev]

:: Lists service endpoint named "test" logs for "dev" environment
$ lib logs username.servicename[@dev].test

:: Lists all logs for "dev" environment
$ lib logs username.servicename[@dev]*
$ lib logs username.servicename[@dev].*

The default log type is stdout, though you can specify stderr with
lib logs username.servicename -t stderr.
Limit the number of lines to show with the -l argument (or --lines).

Additional functionality

Autocode comes packed with a bunch of other goodies – as we roll out updates to
the platform the serverless builds we’re using may change. You can update
your service to our latest build using lib rebuild. If for any reason your
service goes down and is unrecoverable, you can fix it with this command.
To see a full list of commands available for the CLI tools, type:

$ lib help

We’ve conveniently copy-and-pasted the output here for you to peruse;

*
	-b                   Execute as a Background Function
	-d                   Specify debug mode (prints Gateway logs locally, response logs remotely)
	-i                   Specify information mode (prints tar...