Coming from other languages: this made my head explode. Well done!

I suspect a programming bug in the implementation of the tests.

I think "the randomized process generating the input data" is being invoked twice, yielding different results for user code versus test code, which has an exceedingly low probability of being identical. Thus, correct implementations don't match the expected output.

😂

I feel silly. Just noticed the instruction near the start to only count days in 2023.

I'm pretty inexperienced with SQL, so this could totally be a newbie mistake, but I think the provided tests are malfunctioning?

[1, '2022-12-31'],  # Last Saturday of 2022
[1, '2023-01-01'],  # First Sunday of 2023, still ISO week 52 of 2022
[1, '2023-10-07'],  # A Saturday in late 2023
[1, '2023-10-08'],  # A Sunday in late 2023

But the expected in the test comes back as

Just eyeballing the test data, shouldn't that be 2 weekends, 4 days?