string-collate-lessp in elisp

string-collate-lessp in elisp

[Madhu]

I can sort lines with "accents" by calling STRING-COLLATE-LESSP with an
optional "en_US.UTF-8" LOCALE parameter. But would it be possible to
implement the string comparison function in emacs directly using the
unicode data that emacs already has?

If someone has a pointer to the collation rules that have to be
implemented and maybe prior work, I'd appreciate it.  -- Thanks, Madhu

Re: string-collate-lessp in elisp

[Eli Zaretskii]

> From: Madhu <enometh@meer.net>
> Date: Sun, 21 Jul 2024 11:52:53 +0530
> 
> 
> I can sort lines with "accents" by calling STRING-COLLATE-LESSP with an
> optional "en_US.UTF-8" LOCALE parameter. But would it be possible to
> implement the string comparison function in emacs directly using the
> unicode data that emacs already has?

Emacs doesn't import the Unicode collation data, only the codepoints.
string-lessp and friends are based on that, and they compare strings
by codepoints only.  compare-strings does the same, but it also uses
the case-conversion tables (which Emacs does have, partly from
Unicode, partly from its own code).

The collation data is very large, and in addition depends (in minor,
but significant, ways) on the language and country.

> If someone has a pointer to the collation rules that have to be
> implemented and maybe prior work, I'd appreciate it.  -- Thanks, Madhu

The rules are described in this Unicode Technical Standard (UTS#10):

  https://www.unicode.org/reports/tr10/