Since the support for the yield keyword in templates, there is a lot more motivation to have complex question answers for dynamic project generation

e.g.

environments_config:
  type: yaml
  multiline: true
  default: |-
    dev:
      db_host: localhost
      db_port: '5432'
  # As data structures get more complex, validation gets harder
  validator: |-
      [% if environments_config is not mapping %]
      Environment config must be a yaml map with keys as the environment name
      [% endif %]
      [% for env, config in environments_config.items() %]
      [% if config is not mapping %]
      Environment '[[ env ]]' value must be a yaml map with keys: [[ required_config | join(',') ]]
      [% endif %]
      [% for var in required_config %]
      [% if var not in config %]
      [[ env ]] must contain [[ var ]] setting
      [% endif %]
      [% if config[var] is not string %]
      [[ var ]] must be a string setting
      [% endif %]
      [% endfor %]
      [% endfor %]

I've been playing around with a few ideas for a better user experience with more complex questions/answers, and I'm looking for some input on how to make these more accessible if there is interest, and/or for people to share other related ideas

Longer validation messages

The single-line limit always felt very constraining when trying to give a helpful error

POC: https://github.com/cbrown1234/example-template/blob/e4e0acb40011faf787877e7a696f32293d622679/extensions/multiline_validation.py

The implementation as a jinja extension to apply a monkeypatch is obviously quite hacky but this could maybe be made into an opt-in flag in the copier.yml via a change to copier itself

(And maybe some work could be done to make the setting more easily accessible in prompt_toolkit)

Pydantic model validation for answers

POC: https://github.com/cbrown1234/example-template/blob/8b6f1c18fb69661278ed4e21454cc4b1a1a67408/extensions/validators.py

I've played around with a few methods of trying to decouple the extension logic from the model definition, not got anything to stick quite yet, but somewhere between the approaches of jinja2_jsonschema and copier_templates_extensions might be able to work. (Updated POC link with separated models definition in the root of template repo). Or it could just be shared as a copier template itself

I see from the discussion here https://github.com/orgs/copier-org/discussions/376, that adding features that increase the surface area for untrusted code to run is preferred to not be included in copier itself. So not expecting this would be something people would want to add to copier, even though pydantic is already a dependency

Related

I also recently found that https://github.com/copier-org/jinja2-jsonschema exists, and can help solve similar problems. This extension could also pair well with the ability to see longer error messages

(Wish I had found it earlier, as it would probably have saved a lot of time reading docs :D)