Skip to content

Python functions

Python function example on K8s

As an example, let's write a Python function which reads a name from an incoming payload and returns a "Hello" message.

Writing a function requires two steps:

  • Writing a function manifest
  • Applying the manifest to your Kubernetes cluster

The Function object spec requires a minimal amount of configuration:

  • The runtime, here we choose python
  • Whether the function is publicly accessible or not using the public keyword.
  • The entrypoint, which specifies the name of the function
  • The code, written in-line with the function manifest

Save the YAML manifest below in a file called function.yaml

apiVersion: extensions.triggermesh.io/v1alpha1
kind: Function
metadata:
  name: python-function-hello
spec:
  runtime: python
  public: true
  entrypoint: endpoint
  code: |
    def endpoint(event, context):
      return "Hello " + event['name']

You can then create the function with:

kubectl apply -f function.yaml

You can find the public endpoint of your function and test it:

$ kubectl get function
NAME                    ADDRESS                                                          READY   REASON
python-function-hello   https://python-function-hello-mvf2bk.sebgoa.dev.triggermesh.io   True

$ curl -ks -d '{"name":"seb"}' https://python-function-hello-mvf2bk.sebgoa.dev.triggermesh.io |jq
{
  "id": "62402f5a-0a82-48e8-8e67-db68d57efdf9",
  "type": "io.triggermesh.function.python",
  "source": "source.py",
  "specversion": "1.0",
  "time": "2021-10-11T16:26:49Z",
  "datacontenttype": "text/plain",
  "data": "Hello seb"
}

Note

The returned event adheres to the CloudEvent specification.

Python random even/odd events example

apiVersion: extensions.triggermesh.io/v1alpha1
kind: Function
metadata:
  name: inline-python-function
spec:
  runtime: python
  responseIsEvent: true
  adapterOverrides:
    public: true
  ceOverrides:
    extensions:
      type: io.triggermesh.python.sample
  entrypoint: endpoint
  code: |
    from random import randrange
    def endpoint(event, context):
      val = randrange(10)
      if (val % 2) == 0:
        result = {
          "type" : "io.triggermesh.klr.even",
          "datacontenttype" : "application/json",
          "data" : {
            "value" : val
          }
        }
      else:
        result = {
          "type" : "io.triggermesh.klr.odd",
          "datacontenttype" : "application/json",
          "data" : {
            "value" : val
          }
        }
      return result