Unicode is_number and is_digit

[StudebakerGuy]

Context

Odin: dev-2026-02:11d2d3727
OS: OpenMandriva Lx 26.90 (Nickel) Cooker, Linux 6.18.5-desktop-1omv2590
CPU: AMD Ryzen Threadripper PRO 5955WX 16-Cores
RAM: 96430 MiB
Backend: LLVM 21.1.8

Expected Behavior

I would have expected is_digit and is_number in the Unicode pacakge to return true for the code point
U+0C68 since it is in category Nd or decimal number.

Current Behavior

Both functions return false. Looking at is_digit, it only returns true for ASCII 0 through 9.

Steps to Reproduce

package main
import "core:fmt"
import "core:unicode"

main :: proc () {
	r: rune = 0x0c68
	fmt.printfln("Is {} a digit {}", r,  unicode.is_digit(r))
	fmt.printfln("Is {} a number {}", r,  unicode.is_number(r))

}

[Kelimion]

I've updated https://github.com/odin-lang/test-assets/tree/master/XML to Unicode version 17.
I've also restructured the generator so it appends to an array any character that has a category attribute, ready for you to work with and extract contiguous ranges of runes from to generate tables from.

What I'd do is take write a core/unicode/generated.odin that contains a slice of contiguous ranges for each of the character classes - you can start with Nd for is_digit.

Then in core/unicode/letter.odin, you'll see that there's already a table lookup for runes in the ASCII range. I'd keep that lookup and the existing tables for those, but then add a binary search on your ranges. You can see how it's done in other procedures in core:unicode, e.g. is_letter.

So then is_digit would become something like:

@(require_results)
is_digit :: proc(r: rune) -> bool {
	if r <= MAX_LATIN1 {
		return '0' <= r && r <= '9'
	}
	p := binary_search(c, nd_ranges[:], len(nd_ranges)/2, 2)
	if p >= 0 && nd_ranges[p] <= c && c <= nd_ranges[p+1] {
		return true
	}
	return false
}

[StudebakerGuy]

This is going to take a bit longer. I'm currently working on a set of tests that use the Unicode Database to test the digit functions along with is_letter, is_graphic, etc.