Simplify kedro new flow

[merelcht]

Description

#1361

Development notes

• Removed the prompts for repo_name and python_package. These variables will be generated automatically from the project_name provided by the user.
• Added validation for the project_name input and changed it to only accept alphanumeric symbols, hyphens and underscores and check it's at least 2 characters long. This is to prevent the user from adding in special characters that would then also be used for the repo_name and python_package. See comments on this below.
• Users can still provide all variables in a configuration file if they want custom repository and python package names.
• Added tests to check that users can provide custom prompts as well and fixed some bugs that at first failed those tests.

The new flow looks like:

Left to do:
Need to update the starters as well and remove the repo_name and python_package prompts there.

Checklist

• Read the contributing guidelines
• Opened this PR as a 'Draft Pull Request' if it is work-in-progress
• Updated the documentation to reflect the code changes
• Added a description of this change in the RELEASE.md file
• Added tests to cover my changes

[datajoely]

If at all useful, this is how twine resolves a similar problem https://github.com/pypa/twine/blob/8f5e5d6d42d582ef3ea6ef07da277e0cabd22fd2/twine/package.py#L51

[noklam]

One minor comment, when user input an invalid prompt. It will throw an error ValueError: new project!!!! at the bottom, but a more informative message is printed above the stack trace. Maybe they can all be combined in the ValueError? I don't know if we did this before for the coloring, but with the current Rich coloring I think we don't need both.

'new project!!!!' is an invalid value.
It must contain only alphanumeric symbols and/or underscores and/or hyphens.

    def validate(self, user_input: str) -> None:
        """Validate a given prompt value against the regex validator"""
        if self.regexp and not re.match(self.regexp, user_input):
            click.secho(f"'{user_input}' is an invalid value.", fg="red", err=True)
            click.secho(self.error_message, fg="red", err=True)
            raise ValueError(user_input)

If you scroll up, you see this.

At the bottom, it shows this

[noklam]

Nice work! This makes it easier for the beginners.

[merelcht]

One minor comment, when user input an invalid prompt. It will throw an error ValueError: new project!!!! at the bottom, but a more informative message is printed above the stack trace. Maybe they can all be combined in the ValueError? I don't know if we did this before for the coloring, but with the current Rich coloring I think we don't need both.

'new project!!!!' is an invalid value.
It must contain only alphanumeric symbols and/or underscores and/or hyphens.

    def validate(self, user_input: str) -> None:
        """Validate a given prompt value against the regex validator"""
        if self.regexp and not re.match(self.regexp, user_input):
            click.secho(f"'{user_input}' is an invalid value.", fg="red", err=True)
            click.secho(self.error_message, fg="red", err=True)
            raise ValueError(user_input)

If you scroll up, you see this.

At the bottom, it shows this

Ah good catch! The stacktrace doesn't really add anything and makes it quite a bit harder to find our own message. I'll change it.

[SajidAlamQB]

This looks great, so much more seamless! 🎉

[AhdraMeraliQB]

Great work on this! 🌟

[antonymilne]

Looks good! Just one small question.

Also I think it might be nice just to tidy cookiecutter.json up a bit. It is a constant source of minor irritation to me that it's not quite consistent 😅 Here's what we currently have:

    "repo_name": "{{ cookiecutter.project_name.replace(' ', '-').lower().strip('-') }}",
    "python_package": "{{ cookiecutter.project_name.replace(' ', '_').replace('-', '_').lower() }}",

and what I would suggest:

    "repo_name": "{{ cookiecutter.project_name.strip().replace(' ', '-').replace('_', '-').lower() }}",
    "python_package": "{{ cookiecutter.project_name.strip().replace(' ', '_').replace('-', '_').lower() }}",

[antonymilne]

Why would extra_context not exist here, and what happens in that case? 🤔 project_name would just be an empty string which doesn't seem right.

Why can't we just do project_name = extra_context["project_name"] and python_package = extra_context["python_package"]?

[merelcht]

Fair point! I guess I was just being overly cautious but it might be better if this throws an error if there's no extra_context because then something would be really wrong.

[merelcht]

We can't do python_package = extra_context["python_package"] because the extra_context will only contain the values that have come in through prompts + kedro_version because it's set explicitly or from config.

[merelcht]

Oohh and actually I forgot about this, but we added in backward compatibility for when there's no prompts.yml file. Then you can still run kedro new and it will create a new project with "New Kedro Project" as project name which is the default in cookiecutter.json. So project_name = extra_context["project_name"] would break that and we should instead do extra_context.get("project_name", "New Kedro Project")

[merelcht]

See: https://github.com/kedro-org/kedro/blob/main/tests/framework/cli/test_starters.py#L216-L220

[antonymilne]

Thanks for checking this! I think I understand. We should definitely see if we can get rid of the prompts.yml file at some point though, it is pretty confusing 😬