feat(spark): implement Spark datetime function last_day

[Standing-Man]

Which issue does this PR close?

• Closes [datafusion-spark] Implement Spark datetime function last_day #16774.
• Part of [EPIC] Complete datafusion-spark Spark Compatible Functions #15914.

Rationale for this change

What changes are included in this PR?

Implement Spark datetime function last_day

Are these changes tested?

I added tests for last_day function.

Are there any user-facing changes?

yes, new function.

[Standing-Man]

Hi @alamb, I’ve added the last_day function. However, running cargo test --test sqllogictests -- spark produces some errors. I’m looking into it, but please let me know if you have any insights.

1. query failed: DataFusion error: Error during planning: Invalid function 'last_day'.
Did you mean 'list_cat'?

[alamb]

Hi @alamb, I’ve added the last_day function. However, running cargo test --test sqllogictests -- spark produces some errors. I’m looking into it, but please let me know if you have any insights.

1. query failed: DataFusion error: Error during planning: Invalid function 'last_day'.
Did you mean 'list_cat'?

I think the issue is that the function is named spark_last_day in the PR (rather than last_day)

    fn name(&self) -> &str {
        "spark_last_day"
    }

[2010YOUY01]

I think you can set the signature to be taking exactly one Date32 type.
After, the planning routine will automatically do all the validations and coercions, then:

• We can assume inside invoke_with_args() the input must have valid type, so those exec_errs can be changed to internal_err just for sanity check
• No need to implement coerce_types()

[2010YOUY01]

Is it select last_day(null)?

perhaps we can add more bad case tests like

select last_day('foo');
select last_day(123);
select last_day();
select last_day(last_day('2016-02-07'::string, 'foo');
select last_day(last_day('2016-02-31'::string);

And ensure it's returning expected errors.