GCF-Python

This utility is a simple tool that allows running Google Cloud Functions using the Python runtime via a local Flask app. You can then deploy the function as-is to Google Cloud.

What it's for

It's mostly a prototyping utility used to interact with other APIs or perform light data transformations. Its authentication method should not be used alone if you're running your Cloud Functions in production (read more).

The Functions created with this utility can return a Pandas DataFrame or a boolean, and can, therefore, be used to run simple tests or retrieve and format data for basic use cases.

Don't forget limitations around Cloud Functions.

Prepare your Python environment on Ubuntu

1. Use venv to setup a Python 3.7 environment
2. Install Pandas, Flask
3. flask run

Run a function on your local environment

Before running flask run, make sure you set the relevant environment variables.

Flask

• export FLASK_APP=server.py
• export FLASK_ENV=development

Your function

• export GCFP_TOKEN='securetoken'
  • securetoken is a required value that you will pass in your Cloud Function http calls. You can generate one using openssl rand -base64 256. Note that this value is required even if your Cloud Function requires authentication at the Google Cloud level.

Basic API call

Url encode securetoken and call:

http://127.0.0.1:5000/?gcfp_token={{urlencoded_securetoken}}&param1=value1&param2=value2&script_name=demo_script

Load different scripts

One thing you can do, is re-use the same function to invoke different custom scripts.

1. Create your custom script using demo_script.py as a template
2. Call that script my passing its name in your call's script_name parameter

Deploy a function to gcloud

• Make sure to be in /functions and have gcloud installed when running this command (see gcloud quick start).
• This command is using some defaults, be careful, it has --allow-unauthenticated and you should remove this flag if you want to rely on Google Cloud authentication

gcloud functions deploy {{function_name}} --entry-point main --runtime python37 --trigger-http --allow-unauthenticated --region europe-west2 --set-env-vars=GCFP_TOKEN='{{securetoken}}' --memory=128MB --timeout=120

Errors

Errors will show in your CLI with more details. The API returns generic HTTP codes

• 403 Forbidden: This is likely due to using the wrong gcfp_token, not using a gcfp_token, or having no token environment variable set for that function
• 500 Internal Server Error: This is likely due to an error loading the custom script

Limitations

• Supports only one gcfp token
• Limited use cases