Configula

Configula is a configuration generation language and processor. It's goal is to make the programmatic definition of declarative configuration easy and intuitive.

Inspiration

Configula is inspired by the JSX language in React that combines Javascript and HTML tags. Configula defines a similar pattern for Python and YAML (for now).

Consider this declaration:

my_object = foo: bar

It's neither Python, nor YAML. It combines the syntax of both. Instead of being a templating language like Jinja or others, or a Domain-Specific Language (DSL) like HCL, ARM templates or jsonnet, it combines the power of a full programming language with the ease of a declarative syntax like YAML.

Programmatic Example

Let's generate a Kubernetes namespace for a collection of users. Each namespace gets the user's name.

Here is what this looks like in Configula:

# Simple example of creating 3 Kubernetes namespaces

# Our users in need of namespaces
users = ['jim', 'sally', 'sue']

# The namespaces objects from YAML
namespaces = map(lambda user: <
        apiVersion: v1
        kind: Namespace
        metadata:
          # Use the !~ tag syntax to inline Python code
          name: !~ user
    >, users)

# Output
render(namespaces)

An alternate version that uses a for loop looks like this:

# Simple example of creating 3 Kubernetes namespaces
users = [ 'bob', 'sue', 'sally']

# This Python object is defined by inline YAML
namespace = <
  apiVersion: v1
  kind: Namespace
  metadata:
    # The !~ syntax enables python substitution, in this case a variable named `userName`
    name: !~ user
>

for user in users:
  namespace.render()
  print("---")

In either case you run this through the configula interpreter:

configula namespaces.py

and the output of this looks like:

apiVersion: v1
kind: Namespace
metadata:
  name: bob
---
apiVersion: v1
kind: Namespace
metadata:
  name: sue
---
apiVersion: v1
kind: Namespace
metadata:
  name: sally
---

Templating Example

The previous example is programming centric. If your configuration is mostly data and not much code, sometimes it makes sense to have a data-centric template. Such templates look more like YAML than Python.

Here is a simple example:

first = "brendan"
last = "burns"

apiVersion: v1
kind: Namespace
metadata:
  name: !~ first + ' ' + last

When you run this with configula namespace.yml it will output:

apiVersion: v1
kind: Namespace
metadata:
  name: brendan burns

Getting Started

The easiest way to get started is with the Docker image:

# Clone the repo
git clone https://github.com/brendandburns/configula
cd configula

# Run the namespaces example
docker run -v $PWD/examples:/examples brendanburns/configula \
  ./configula examples/namespaces.py

See this directory for more examples.

Read the docs

Installing configula

Getting Started

Download Configula for your platform

• darwin
• linux
• windows

Integrating with Kubernetes

Configula can be used as a kubectl plugin.

To install the plugin run this command:

# assumes that configula is in /usr/local/bin/ change paths as needed.
ln -s /usr/local/bin/kubectl-configula /usr/local/bin/configula

Once that's done you can run:

kubectl configula create -f examples/namespaces.py

FAQ

What about other Javascript? Java? Go? .NET? Other Language?

We're open to other programming languages. What other languages would you want to see? Please file issues.

What about JSON? HCL? Other Data Language?

We're open to other data languages too! What other languages would you want to see? Please file issues.