Re: [Tiki-devel] Unicode-aware sort plugin

Volker Wysk <post-hhF2Jplw28UoZk/[email protected]> Thu, 06 Jul 2023 15:08:31 +0200
Newsgroups	gmane.comp.cms.tiki.devel
Message-ID	<[email protected]>
Am Mittwoch, dem 05.07.2023 um 09:45 +0300 schrieb Victor Emanouilov via
TikiWiki-devel:
> Hi Volker,
> 
> Looks good. Just a small note that Collator depends on enchant library 
> and might not be available on all systems. 

That's strange. I don't have the PHP enchant library installed:

desktop ~ $ LANG=C apt list '*php*enchant*' 
Listing... Done
php-enchant/jammy,jammy 2:8.2+93+ubuntu22.04.1+deb.sury.org+2 all
php5.6-enchant/jammy 5.6.40-66+ubuntu22.04.1+deb.sury.org+1 amd64
php7.0-enchant/jammy 7.0.33-66+ubuntu22.04.1+deb.sury.org+1 amd64
php7.1-enchant/jammy 7.1.33-53+ubuntu22.04.1+deb.sury.org+1 amd64
php7.2-enchant/jammy 7.2.34-39+ubuntu22.04.1+deb.sury.org+1 amd64
php7.3-enchant/jammy 7.3.33-11+ubuntu22.04.1+deb.sury.org+1 amd64
php7.4-enchant/jammy 1:7.4.33-6+ubuntu22.04.1+deb.sury.org+1 amd64
php8.0-enchant/jammy 1:8.0.29-1+ubuntu22.04.1+deb.sury.org+1 amd64
php8.1-enchant/jammy 8.1.20-1+ubuntu22.04.1+deb.sury.org+1 amd64
php8.1-enchant/jammy 8.1.8-1+ubuntu22.04.1+deb.sury.org+1 i386
php8.2-enchant/jammy 8.2.7-1+ubuntu22.04.1+deb.sury.org+1 amd64

desktop ~ $ LANG=C apt list '*php*enchant*' --installed
Listing... Done

But the wikiplugin_sort.php version, which I've built, and uses it, DOES
work. Are you sure about the enchant library?

I've looked in the php.net pages, and it says that the Collator class is
part of the intl library, which depends on the C/C++ ICU library (and not
enchant).


> I think you should make the 
> use of Collator sort optional - if the feature is not present, fall back 
> to original sorting methods.


I'm not sure if this is such a great idea. I'd rather check in the Tiki
installer if the library is available and output a warning or error message
if it isn't.

I think this ordering, which is locale aware and at the same time
natural, is needed in quite some places. The natsort and natcasesort
functions are used about a dozen times, each, in Tiki 25 (many in
vendor_bundled). I expect that most of those occurences should made be
Unicode aware, meaning they need to be modified.

We could make mb_natsort() and mb_natcasesort() functions which do just
that, by using the Collator class. Then just "mb_" needs to be added each
time.


Cheers,
Volker 

> On 7/1/23 12:49 PM, Volker Wysk wrote:
> > Am Donnerstag, dem 29.06.2023 um 19:06 +0200 schrieb Volker Wysk:
> > > Hi!
> > > 
> > > The Sort Plugin is kinda broken, because it isn't Unicode aware. That's not
> > > that easy to fix, because the PHP natcasesort() function isn't Unicode
> > > aware. And there is no mb_natcasesort(). It needs to be done differently.
> > > 
> > > I've found on the web, how it is to be done, here:
> > > 
> > > https://stackoverflow.com/questions/58577558/how-to-use-multiple-sort-flags-in-php-array-sorting-using-sort-locale-string
> > > 
> > > I've applied it to the wikiplugin_sort and it works. However, this plugin
> > > needs to use the correct locale for the user. I don't know how to get this
> > > locale. This bit is still missing.
> > > 
> > > I've attached the new wikiplugin_sort.php.
> > > 
> > > Someone with more Tiki knowledge should improve this version, by querying
> > > the locale to be used and pass it to the collator_create function.
> > I've figured it out (hopefully). The new version sorts naturally and in
> > accordance to the user's locale. I've attached it, again.
> > 
> > I'm about to do a merge request next.
> > 
> > Cheers,
> > Volker
> > 
> > 
> > _______________________________________________
> > TikiWiki-devel mailing list
> > [email protected]
> > https://lists.sourceforge.net/lists/listinfo/tikiwiki-devel
> 
> 
> _______________________________________________
> TikiWiki-devel mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/tikiwiki-devel

_______________________________________________
TikiWiki-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/tikiwiki-devel
signature.asc (application/pgp-signature, 833 B)