Functions Framework for Python

An open source FaaS (Function as a service) framework for writing portable Python functions.

The Functions Framework lets you write lightweight functions that run in many different environments, including:

• OpenFunction
• Knative-based environments
• Dapr-based environments
• Google Cloud Functions
• Your local development machine
• Cloud Run and Cloud Run for Anthos

The framework allows you to go from:

def hello(request):
    return "Hello world!"

To:

curl http://my-url
# Output: Hello world!

All without needing to worry about writing an HTTP server or complicated request handling logic.

Features

• Spin up a local development server for quick testing
• Invoke a function in response to a request
• Automatically unmarshal events conforming to the CloudEvents spec
• Portable between serverless platforms

Installation

Install the Functions Framework via pip:

pip install functions-framework

Or, for deployment, add the Functions Framework to your requirements.txt file:

functions-framework==3.*

Quickstarts

Quickstart: HTTP Function (Hello World)

Create an main.py file with the following contents:

import functions_framework

@functions_framework.http
def hello(request):
    return "Hello world!"

Your function is passed a single parameter, (request), which is a Flask Request object.

Run the following command:

functions-framework --target hello --debug
 * Serving Flask app "hello" (lazy loading)
 * Environment: production
   WARNING: This is a development server. Do not use it in a production deployment.
   Use a production WSGI server instead.
 * Debug mode: on
 * Running on http://0.0.0.0:8080/ (Press CTRL+C to quit)

(You can also use functions-framework-python if you have multiple language frameworks installed).

Open http://localhost:8080/ in your browser and see Hello world!.

Or send requests to this function using curl from another terminal window:

curl localhost:8080
# Output: Hello world!

Quickstart: CloudEvent Function

Create an main.py file with the following contents:

import functions_framework

@functions_framework.cloud_event
def hello_cloud_event(cloud_event):
   print(f"Received event with ID: {cloud_event['id']} and data {cloud_event.data}")

Your function is passed a single CloudEvent parameter.

Run the following command to run hello_cloud_event target locally:

functions-framework --target=hello_cloud_event

In a different terminal, curl the Functions Framework server:

curl -X POST localhost:8080 \
   -H "Content-Type: application/cloudevents+json" \
   -d '{
	"specversion" : "1.0",
	"type" : "example.com.cloud.event",
	"source" : "https://example.com/cloudevents/pull",
	"subject" : "123",
	"id" : "A234-1234-1234",
	"time" : "2018-04-05T17:31:00Z",
	"data" : "hello world"
}'

Output from the terminal running functions-framework:

Received event with ID: A234-1234-1234 and data hello world

More info on sending CloudEvents payloads, see examples/cloud_run_cloud_events instruction.

Quickstart: Error handling

The framework includes an error handler that is similar to the flask.Flask.errorhandler function, which allows you to handle specific error types with a decorator:

import functions_framework


@functions_framework.errorhandler(ZeroDivisionError)
def handle_zero_division(e):
    return "I'm a teapot", 418


def function(request):
    1 / 0
    return "Success", 200

This function will catch the ZeroDivisionError and return a different response instead.

Quickstart: Pub/Sub emulator

1. Create a main.py file with the following contents:

   def hello(event, context):
        print("Received", context.event_id)

2. Start the Functions Framework on port 8080:

   functions-framework --target=hello --signature-type=event --debug --port=8080

3. In a second terminal, start the Pub/Sub emulator on port 8085.

   export PUBSUB_PROJECT_ID=my-project
   gcloud beta emulators pubsub start \
       --project=$PUBSUB_PROJECT_ID \
       --host-port=localhost:8085

   You should see the following after the Pub/Sub emulator has started successfully:

   [pubsub] INFO: Server started, listening on 8085
   

4. In a third terminal, create a Pub/Sub topic and attach a push subscription to the topic, using http://localhost:8080 as its push endpoint. Publish some messages to the topic. Observe your function getting triggered by the Pub/Sub messages.

   export PUBSUB_PROJECT_ID=my-project
   export TOPIC_ID=my-topic
   export PUSH_SUBSCRIPTION_ID=my-subscription
   $(gcloud beta emulators pubsub env-init)
   
   git clone https://github.com/googleapis/python-pubsub.git
   cd python-pubsub/samples/snippets/
   pip install -r requirements.txt
   
   python publisher.py $PUBSUB_PROJECT_ID create $TOPIC_ID
   python subscriber.py $PUBSUB_PROJECT_ID create-push $TOPIC_ID $PUSH_SUBSCRIPTION_ID http://localhost:8080
   python publisher.py $PUBSUB_PROJECT_ID publish $TOPIC_ID

   You should see the following after the commands have run successfully:

   Created topic: projects/my-project/topics/my-topic
   
   topic: "projects/my-project/topics/my-topic"
   push_config {
     push_endpoint: "http://localhost:8080"
   }
   ack_deadline_seconds: 10
   message_retention_duration {
     seconds: 604800
   }
   .
   Endpoint for subscription is: http://localhost:8080
   
   1
   2
   3
   4
   5
   6
   7
   8
   9
   Published messages to projects/my-project/topics/my-topic.
   

   And in the terminal where the Functions Framework is running:

    * Serving Flask app "hello" (lazy loading)
    * Environment: production
      WARNING: This is a development server. Do not use it in a production deployment.
      Use a production WSGI server instead.
    * Debug mode: on
    * Running on http://0.0.0.0:8080/ (Press CTRL+C to quit)
    * Restarting with fsevents reloader
    * Debugger is active!
    * Debugger PIN: 911-794-046
   Received 1
   127.0.0.1 - - [11/Aug/2021 14:42:22] "POST / HTTP/1.1" 200 -
   Received 2
   127.0.0.1 - - [11/Aug/2021 14:42:22] "POST / HTTP/1.1" 200 -
   Received 5
   127.0.0.1 - - [11/Aug/2021 14:42:22] "POST / HTTP/1.1" 200 -
   Received 6
   127.0.0.1 - - [11/Aug/2021 14:42:22] "POST / HTTP/1.1" 200 -
   Received 7
   127.0.0.1 - - [11/Aug/2021 14:42:22] "POST / HTTP/1.1" 200 -
   Received 8
   127.0.0.1 - - [11/Aug/2021 14:42:22] "POST / HTTP/1.1" 200 -
   Received 9
   127.0.0.1 - - [11/Aug/2021 14:42:39] "POST / HTTP/1.1" 200 -
   Received 3
   127.0.0.1 - - [11/Aug/2021 14:42:39] "POST / HTTP/1.1" 200 -
   Received 4
   127.0.0.1 - - [11/Aug/2021 14:42:39] "POST / HTTP/1.1" 200 -
   

For more details on extracting data from a Pub/Sub event, see https://cloud.google.com/functions/docs/tutorials/pubsub#functions_helloworld_pubsub_tutorial-python

Quickstart: Build a Deployable Container

1. Install Docker and the pack tool.

2. Build a container from your function using the Functions buildpacks:

    pack build \
        --builder gcr.io/buildpacks/builder:v1 \
        --env GOOGLE_FUNCTION_SIGNATURE_TYPE=http \
        --env GOOGLE_FUNCTION_TARGET=hello \
        my-first-function
   

3. Start the built container:

    docker run --rm -p 8080:8080 my-first-function
    # Output: Serving function...
   

4. Send requests to this function using curl from another terminal window:

    curl localhost:8080
    # Output: Hello World!
   

Run your function on serverless platforms

Container environments based on Knative

The Functions Framework is designed to be compatible with Knative environments. Build and deploy your container to a Knative environment.

OpenFunction

Besides Knative function support, one notable feature of OpenFunction is embracing Dapr system, so far Dapr pub/sub and bindings have been support.

Dapr bindings allows you to trigger your applications or services with events coming in from external systems, or interface with external systems. OpenFunction 0.6.0 release adds Dapr output bindings to its synchronous functions which enables HTTP triggers for asynchronous functions. For example, synchronous functions backed by the Knative runtime can now interact with middlewares defined by Dapr output binding or pub/sub, and an asynchronous function will be triggered by the events sent from the synchronous function.

Asynchronous function introduces Dapr pub/sub to provide a platform-agnostic API to send and receive messages. A typical use case is that you can leverage synchronous functions to receive an event in plain JSON or Cloud Events format, and then send the received event to a Dapr output binding or pub/sub component, most likely a message queue (e.g. Kafka, NATS Streaming, GCP PubSub, MQTT). Finally, the asynchronous function could be triggered from the message queue.

More details would be brought up to you in some quickstart samples, stay tuned.

Google Cloud Functions

This Functions Framework is based on the Python Runtime on Google Cloud Functions.

On Cloud Functions, using the Functions Framework is not necessary: you don't need to add it to your requirements.txt file.

After you've written your function, you can simply deploy it from your local machine using the gcloud command-line tool. Check out the Cloud Functions quickstart.

Cloud Run/Cloud Run on GKE

Once you've written your function and added the Functions Framework to your requirements.txt file, all that's left is to create a container image. Check out the Cloud Run quickstart for Python to create a container image and deploy it to Cloud Run. You'll write a Dockerfile when you build your container. This Dockerfile allows you to specify exactly what goes into your container (including custom binaries, a specific operating system, and more). Here is an example Dockerfile that calls Functions Framework.

If you want even more control over the environment, you can deploy your container image to Cloud Run on GKE. With Cloud Run on GKE, you can run your function on a GKE cluster, which gives you additional control over the environment (including use of GPU-based instances, longer timeouts and more).

Container environments based on Knative

Cloud Run and Cloud Run on GKE both implement the Knative Serving API. The Functions Framework is designed to be compatible with Knative environments. Just build and deploy your container to a Knative environment.

Configure the Functions Framework

You can configure the Functions Framework using command-line flags or environment variables. If you specify both, the environment variable will be ignored.

Command-line flag	Environment variable	Description
--host	HOST	The host on which the Functions Framework listens for requests. Default: 0.0.0.0
--port	PORT	The port on which the Functions Framework listens for requests. Default: 8080
--target	FUNCTION_TARGET	The name of the exported function to be invoked in response to requests. Default: function
--signature-type	FUNCTION_SIGNATURE_TYPE	The signature used when writing your function. Controls unmarshalling rules and determines which arguments are used to invoke your function. Default: http; accepted values: http, event or cloudevent
--source	FUNCTION_SOURCE	The path to the file containing your function. Default: main.py (in the current working directory)
--debug	DEBUG	A flag that allows to run functions-framework to run in debug mode, including live reloading. Default: False
--dry-run	DRY_RUN	A flag that allows for testing the function build from the configuration without creating a server. Default: False

Enable Google Cloud Function Events

The Functions Framework can unmarshall incoming Google Cloud Functions event payloads to event and context objects. These will be passed as arguments to your function when it receives a request. Note that your function must use the event-style function signature:

def hello(event, context):
    print(event)
    print(context)

To enable automatic unmarshalling, set the function signature type to event using the --signature-type command-line flag or the FUNCTION_SIGNATURE_TYPE environment variable. By default, the HTTP signature will be used and automatic event unmarshalling will be disabled.

For more details on this signature type, see the Google Cloud Functions documentation on background functions.

See the running example.

Advanced Examples

More advanced guides can be found in the examples/ directory. You can also find examples on using the CloudEvent Python SDK here.

Contributing

Contributions to this library are welcome and encouraged. See CONTRIBUTING for more information on how to get started.