Add Polars Builtin Check Tests

[AndriiG13]

Fixes #1421

The structure for most of the tests was taken from the builtin check tests for PySpark

[AndriiG13]

We need to explicitly alias here. If min_value is None, then resulting column name will be literal, which will break Polars lazy computation. Seems to be related to pola-rs/polars#14382

[cosmicBboy]

Would this make sense to do instead:

if min_value is None and max_value is None:
    raise ValueError("Must provide at least on of 'min_value' and 'max_value'")

n_chars = pl.col(data.key).str.n_chars()
if min_value is None:
    expr = n_chars.le(max_value)
elif max_value is None:
    expr = n_chars.ge(min_value)
else:
    expr = n_chars.between(min_value, max_value)

return expr

[AndriiG13]

I decided to remove the str_matches and keep only str_contains check with .str.contains(..., literal=False) since it works for both strings and regex.

Also I see that the str_matches check for Pandas searches for matches from the beginning of a string using the Pandas str.match() method and I don't think there is an equivalent method in Polars. So sticking just with str.contains seemed appropriate

[cosmicBboy]

I think where we can, check backends should implement the built-in check functions, even if there isn't a direct method supported in the underlying framework.

So the implementation would be something like

    if isinstance(pattern, str):
        pattern = f"^{pattern}"
    else:
        pattern = re.compile(f"^{pattern.pattern}")
    return data.dataframe.select(
        pl.col(data.key).str.contains(pattern=pattern)
    )

[cosmicBboy]

This should be consistent with the public Check api: https://github.com/unionai-oss/pandera/blob/main/pandera/api/checks.py#L430

Suggested change

	pattern: re.Pattern,
	pattern: Union[str, re.Pattern],

[cosmicBboy]

I think where we can, check backends should implement the built-in check functions, even if there isn't a direct method supported in the underlying framework.

So the implementation would be something like

    if isinstance(pattern, str):
        pattern = f"^{pattern}"
    else:
        pattern = re.compile(f"^{pattern.pattern}")
    return data.dataframe.select(
        pl.col(data.key).str.contains(pattern=pattern)
    )

[cosmicBboy]

Would this make sense to do instead:

if min_value is None and max_value is None:
    raise ValueError("Must provide at least on of 'min_value' and 'max_value'")

n_chars = pl.col(data.key).str.n_chars()
if min_value is None:
    expr = n_chars.le(max_value)
elif max_value is None:
    expr = n_chars.ge(min_value)
else:
    expr = n_chars.between(min_value, max_value)

return expr

[cosmicBboy]

Thanks for the contribution @AndriiG13! I updated some of the checks and tests, will merge this into polars-dev now

[codecov]

Codecov Report

Attention: Patch coverage is 0% with 14 lines in your changes are missing coverage. Please review.

❗ No coverage uploaded for pull request base (polars-dev@0a24f3e). Click here to learn what that means.

Files	Patch %	Lines
pandera/backends/polars/builtin_checks.py	0.00%	14 Missing ⚠️

Additional details and impacted files

@@              Coverage Diff              @@
##             polars-dev    #1518   +/-   ##
=============================================
  Coverage              ?   83.06%           
=============================================
  Files                 ?      110           
  Lines                 ?     8018           
  Branches              ?        0           
=============================================
  Hits                  ?     6660           
  Misses                ?     1358           
  Partials              ?        0           

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