GH-48478: [Ruby] Fix Ruby list inference for nested non-negative integer arrays

[hypsakata]

Rationale for this change

When building an Arrow::Table from a Ruby Hash passed to Arrow::Table.new, nested Integer arrays are incorrectly inferred as string (utf8) if all values are non-negative. This behavior is unexpected; nested integer arrays should be consistently represented as a list type (e.g., list<item: uint*> or list<item: int*>) rather than falling back to UTF-8 strings.

What changes are included in this PR?

This PR modifies the logic in detect_builder_info(), specifically the when ::Array block, to correctly identify nested non-negative integer arrays as list arrays.

The change ensures that if sub_builder_info contains a valid :builder, it will be used even if sub_builder_info does not yet indicate that the type has been "detected."

Are these changes tested?

Yes. (ruby ruby/red-arrow/test/run-test.rb)

Are there any user-facing changes?

Yes.

GitHub Issue: Closes #48478

• GitHub Issue: [Ruby] Arrow::Table.new infers nested integer arrays as utf8 when all values are non-negative #48478

[github-actions]

⚠️ GitHub issue #48478 has been automatically assigned in GitHub to PR creator.

[kou]

Could you add tests for newly supported cases?

[hypsakata]

I have added test cases for nested array type inference in test/unit/apache-arrow/test-table.rb.
Since I couldn't find existing tests specifically covering this input pattern (Hash with nested arrays), I added them to the .new test block.

[kou]

How about using

arrow/ruby/red-arrow/test/test-array-builder.rb

Lines 133 to 148 in 649a372

	test("list<boolean>s") do
	assert_build(Arrow::ArrayBuilder,
	[
	[nil, true, false],
	nil,
	[false],
	])
	end
	test("list<string>s") do
	assert_build(Arrow::ArrayBuilder,
	[
	["Hello", "World"],
	["Apache Arrow"],
	])
	end

instead of this file?

[kou]

Is this !! for converting nil to false?

If so, we don't need it because nil is also a false value.

[hypsakata]

Thanks for the review!

I've removed the redundant boolean conversion.
Also, I've moved the tests to test/test-array-builder.rb. I should have checked the appropriate location more carefully, so thanks for pointing it out.

[kou]

+1

[kou]

Could you also check values like

arrow/ruby/red-arrow/test/test-array-builder.rb

Lines 97 to 108 in 1fac131

	assert_equal({
	data_type: data_type,
	values: [
	BigDecimal("10.1"),
	BigDecimal("1.11"),
	BigDecimal("1"),
	],
	},
	{
	data_type: array.value_data_type,
	values: array.to_a,
	})

?

[hypsakata]

Thanks, I've updated the tests.

[conbench-apache-arrow]

After merging your PR, Conbench analyzed the 3 benchmarking runs that have been run so far on merge-commit d89c14b.

There were no benchmark performance regressions. 🎉

The full Conbench report has more details. It also includes information about 1 possible false positive for unstable benchmarks that are known to sometimes produce them.