Testing global exception handling with CliRunner

Hello everyone,

I'm having an issue with testing my click application that has a custom exception handler with click. I would like to use the CliRunner to test custom exception handling with click. Therefore I'm using a setup similar to the one found below.

Execute this script below with the parameter action (e.g. python3 script_name.py action) to reproduce this error.

Also I'd like to ask if this is the correct way to handle global exceptions. For me it does not make sense to handle the error in each subcommand, when there is e.g. a connection error to the server, which will always be handled the same way.

Expected Behavior

The Exception will be handled by the global handler in both cases, leading to a stdout of

Executing manually:
Exception was catched by application
Executing with CliRunner:
Exception was catched by application

Minimal example:

import click
from click.testing import CliRunner

class ExampleError(Exception):
    pass

class GlobalExceptionHandler(click.Group):
    def __call__(self, *args, **kwargs):
        try:
            return self.main(*args, **kwargs)
        except ExampleError:
            print("Exception was catched by application")

@click.group(cls=GlobalExceptionHandler)
def cli():
    pass

@cli.command(short_help="Upload a Dockerfile or Folder to Database")
def action():
    raise ExampleError()

# Manual running
def main_1():
    cli()

# Option 2
def main_2():
    runner = CliRunner()
    runner.invoke(cli, ["action"])

if __name__ == '__main__':
    print("Executing manually:")
    main_1()
    print("Executing with CliRunner:")
    main_2()

Actual Behavior

Tell us what happens instead.

The Exception is handled when manually executing, but not when ran with CliRunner.

Executing manually:
Exception was catched by application
Executing with CliRunner:

Environment

Python version: 3.8
Click version: 7.1.2

pallets / click