Parse json data from token if object is dict-like.

[BrianPugh]

If an object is dict/dataclass-like, attempts to parse input data as a json-string.

@AngellusMortis can you play around with this and see if it satisfies your needs? I need to flesh out the rest of the PR and think a little harder if this is the appropriate mechanism, but it naively seems to work.

Addresses #282.

TODO

• unit tests
• docs

[codecov]

Codecov Report

Attention: Patch coverage is 87.50000% with 2 lines in your changes missing coverage. Please review.

Project coverage is 93.97%. Comparing base (6284a12) to head (4aa51b9).

Files with missing lines	Patch %	Lines
cyclopts/argument.py	87.50%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #285      +/-   ##
==========================================
- Coverage   94.00%   93.97%   -0.04%     
==========================================
  Files          26       26              
  Lines        3020     3035      +15     
  Branches      636      640       +4     
==========================================
+ Hits         2839     2852      +13     
- Misses         88       90       +2     
  Partials       93       93              

Flag	Coverage Δ
unittests	93.97% <87.50%> (-0.04%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[AngellusMortis]

It looks like data is getting populated correctly, but the correct values are not being passed to Pydantic correctly (I am using pydantic). It looks like if it was just a basic dataclass, it would work. (lines at the top are manual prints I added in)

[AngellusMortis]

That commit looks good!

[BrianPugh]

@AngellusMortis I just pushed a fix for that; please give it a try! Pydantic is a more-special-than-usual case within cyclopts; as it's the only type that cyclopts practically disables it's own coercion engine for.

[BrianPugh]

that was a lightning fast reply!

[AngellusMortis]

Watching a show right now. So Github is up and just automatically updating for me. lol.

[BrianPugh]

I'd appreciate it if you'd keep playing with it and let me know if you discover any other issues. Otherwise tonight/this weekend I'll be adding tests/docs for this new feature! Thanks!

[AngellusMortis]

Find an issue: If you have aliases for your pydantic class, it does not accept the fields from the alias value.

i.e. If you have a config that converts camelCase to snake_case:

class BaseK8sModel(BaseModel):
    """Base model for k8s spec."""

    model_config = ConfigDict(
        alias_generator=to_camel,
        populate_by_name=True,
        from_attributes=True,
    )

It cannot load the values from camelCase, only snake_case.

I did confirm that passing both the ENV and one of the nested properties via an option still works.

[BrianPugh]

I'll look into this, thanks!

[AngellusMortis]

@BrianPugh Can you rebase this branch off of main (to get the fix for the nested meta apps)? Thanks!

[BrianPugh]

rebased! I still want to rethink this a little more; because as of right now it sort of bypasses a lot of Cyclopts mechanisms. It probably works well for most general use-cases (once I investigate the pydantic issue you brought up), but would like it to work for even those advanced use-cases.