PyFunc

This package provides a mechanism to write Function-as-a-Service style code in Python for handling HTTP events, including CloudEvents delivered via HTTP.

This framework is primarily intended to work with Knative, but also works to provide a generic server for handling CloudEvents over HTTP (e.g. from Kubernetes or on a local machine).

The framework uses reflection to find a suitable function to wrap; it should not be necessary to import any of the following modules in your own code unless you want (e.g. for type definitions):

• pyfunc (this module; on PyPi as pyfunc)
• flask
• cloudevents

Instead, simply ensure that you have a single non-_ prefixed function which uses some combination of the following:

• HTTP request arguments (named req, request, body, headers or of the flask.Request type)
• CloudEvent arguments (named event, payload, data, attributes or of the cloudevents.sdk.event.v1.Event type)

Usage:

import logging
from typing import Any

counter = 0

def handler(data: Any, attributes: dict, req: Any):
    global counter
    counter = counter + 1

    logging.info(f"Got data: {data}")
    logging.info(f"From {req.origin}, my {counter}th request!")

    attributes["type"] = "com.example.reply"
    attributes["datacontenttype"] = "text/plain"

    return attributes, "It's a demo"

Usage

To check the current working directory for a module called handler and function handler:

python -m pyfunc

To check the current working directory for a module called myhandler and function func:

PYTHON_HANDLER=myhandler.func python -m pyfunc

To check a specific directory ./myfuncs/http/ for a module called handler and function handler:

python -m pyfunc ./myfuncs/http/

To check a specific directory ./myfuncs/http/ for a module called myhandler and function func:

PYTHON_HANDLER=myhandler.func python -m pyfunc ./myfuncs/http/