feat: add ndarray/base/where

[LoayAhmed304]

Resolves #{{TODO: add issue number}}.

Description

What is the purpose of this pull request?

This pull request:

• Implements the where function for ndarrays, which applies a condition to elements in two input ndarrays and assign results to elements in an output ndarray.

Related Issues

Does this pull request have any related issues?

This pull request has the following related issues:

• #{{TODO: add related issue number}}

Questions

Any questions for reviewers of this pull request?

No.

Other

Any other information relevant to this pull request? This may include screenshots, references, and/or implementation notes.

No.

Checklist

Please ensure the following tasks are completed before submitting this pull request.

• Read, understood, and followed the contributing guidelines.

AI Assistance

When authoring the changes proposed in this PR, did you use any kind of AI assistance?

• Yes
• No

If you answered "yes" above, how did you use AI assistance?

• Code generation (e.g., when writing an implementation or fixing a bug)
• Test/benchmark generation
• Documentation (including examples)
• Research and understanding

Disclosure

If you answered "yes" to using AI assistance, please provide a short disclosure indicating how you used AI assistance. This helps reviewers determine how much scrutiny to apply when reviewing your contribution. Example disclosures: "This PR was written primarily by Claude Code." or "I consulted ChatGPT to understand the codebase, but the proposed changes were fully authored manually by myself.".

{{TODO: add disclosure if applicable}}

---

@stdlib-js/reviewers

[stdlib-bot]

Coverage Report

Package	Statements	Branches	Functions	Lines
ndarray/base/where	$\color{red}7704/14877$ $\color{green}+51.78%$	$\color{red}188/196$ $\color{green}+95.92%$	$\color{red}16/46$ $\color{green}+34.78%$	$\color{red}7704/14877$ $\color{green}+51.78%$

The above coverage report was generated for the changes in this PR.

[headlessNode]

@LoayAhmed304 If someone passes arrays with mismatched shapes but valid dtypes, the function will do unnecessary work (reinterpreting arrays) before eventually throwing.

[LoayAhmed304]

Done. Moved shapes validation to the top.

I believe I took it from assign here, but we're dealing with 4 arrays here not 2, so maybe the unnecessary work will affect performance here.

[LoayAhmed304]

Note that this will cause re-assignment of shapes again after the reinterpretation, since reinterpreting complex ndarrays mutates the shape.

[LoayAhmed304]

Note that tests will have updated cases to cover the varying dtypes of x and y, since the test cases were made while assuming they would have the exact same dtype.

[headlessNode]

Suggested change

	var dc0;
	var dx0;
	var dy0;
	var do0;
	var dc;
	var dx;
	var dy;
	var do;

Applies to below and 1d.js as well.

[LoayAhmed304]

do is a reserved keyword. I'll use dout instead, but will keep it do# for the rest of the kernels.

[headlessNode]

Not sure if this is okay since in the main description 'a condition' and 'two input ndarrays' is used.

[headlessNode]

If you decide to change, it would apply to repl and index.d.ts file as well.

[headlessNode]

Why is this required here? condition becomes 3d while all other are 2d. What happens if the condition is 2d like others? Is the output affected?

[LoayAhmed304]

This is what main.js does when dealing with complex input ndarrays, and actually this is something I wanted to discuss.

// main.js

if ( isComplexArray( x.data ) && isComplexArray( y.data ) && isComplexArray( o.data ) ) {
		x = complex2real( x );
		y = complex2real( y );
		o = complex2real( o );

		c.shape.push( 2 ); // real and imaginary components
		c.strides.push( 0 ); // broadcast
// ....

Currently what I'm doing is something similar to what complex2real does when interpreting the input complex ndarray, which manipulates the shape and stride after reinterpretation. So this basically does the same to the condition array, so each element in the condition ndarray compares both the real and imaginary parts. Hence for complex, the condition ndarray buffer can be half the size of the complex buffer. Like this:

// test.js

var cbuf = new BooleanArray([
		true,
		false,
		false,
		true
	]);
var xbuf = new Complex128Array([
		10.0,
		10.0,
		10.0,
		10.0,
		10.0,
		10.0,
		10.0,
		10.0
	]);
// ....

[LoayAhmed304]

However I believe this will break when having 1 complex input ndarray and 1 real input ndarray, since I was assuming both will have the same dtype.

[headlessNode]

@LoayAhmed304 Would good to check all the files against the above comments, especially the dimensional kernels.