Load defaults with configvars and keep defined parser arguments

[yt-invers]

I try to load defaults from a yaml file (i know the defaults step), but to keep my cli arguments.
This would overwrite the cli argument play with the one defined inside the configvars file.

context_parser: pypyr.parser.keyvaluepairs

steps:
  - pypyr.steps.configvars
  - name: pypyr.steps.echo
    in:
      echoMe: '{play}'

An unfortunately wrong solution from ChatGPT. Exists something like this in another form?

[yaythomas]

Goodness, I don't even know where ChatGPT would have got that from! I don't recognise the flag name at all! (If someone somewhere wrote some extension like this, feel free to get in touch!)

Yes, you're right @yt-invers, pypyr.steps.configvars will always overwrite on update - so it will overwrite cli arguments.

Configvars does a bit more than just loading from yaml - it interacts with the pypyr config system, checks multiple file locations and also pyproject.toml.

If you're happy just to load from a specific yaml file, you can do this:

context_parser: pypyr.parser.keyvaluepairs
steps:
  - name: pypyr.steps.fetchyaml    
    in:
      fetchYaml:
        path: my-file.yaml
        key: defaults
  - pypyr.steps.default
  - name: pypyr.steps.echo
    in:
      echoMe: '{play}'

In this case your yaml file will look like this (note to vars parent key here, since we're directly loading the yaml, it's not going via the config system):

# ./my-file.yaml
play: set in yaml

The magic here is that the fetchYaml step is pumping the values from yaml into the defaults key, pre-filling it for the default step that comes next. The default step will then only write values from the defaults key to the context root if they don't exist yet.

The net result is that it you will keep your CLI arguments - i.e the CLI argument will always take precedence over the yaml file values.

You can achieve the same sort of idea using configvars, as long as you save you "defaults" under a defaults key in the config file:

# ./pypyr-config.yaml
vars:
  defaults:
    play: set in yaml in config file

and the pipeline:

context_parser: pypyr.parser.keyvaluepairs
steps:
  - pypyr.steps.configvars
  - pypyr.steps.default
  - name: pypyr.steps.echo
    in:
      echoMe: '{play}'

Doing it like this means you could set your play value in any valid pypyr config source, for example pyproject.toml:

# ./pyproject.toml
[tool.pypyr.vars.defaults]
play = "from pyproject.toml"

But you raise a good point, this is a frequent use-case: use cli args as an override for a set of defaults. Although you can achieve this using the code above, to interact with the pypyr config system natively we should probably add a configvarsdefault* step or similar that only sets values if they don't exist already.

*not the world's greatest name 🤔

[yt-invers]

It worked! Thank you very much!
I didn't think of to define a variable in a pre-step and that would be/can be used directly as an in variable!

This knowledge can come very handy (or dangerous if people don't know this). It should be added to the configvars or/and variables documentation :D