Sort special characters

[laportjo]

Hi Daniel,

Is there a way to sort correctly special characters in a list via the helper reference() ? For example, I have this list (excerpt) :

• Abbaye
• Bâtiment d'exposition
• Palais
• Édifice
• Église

But I expect this order :

• Abbaye
• Bâtiment d'exposition
• Édifice
• Église
• Palais

Thanks ! Your plugin is very wonderful.

[Daniel-KM]

You have to change the line https://github.com/Daniel-KM/Reference/blob/master/views/helpers/Reference.php#L287 with " utf8_roman_ci" (or add an option to manage it cleanly).

[laportjo]

It doesn't seem to work. I get the same list order.

[Daniel-KM]

So you can sort the resulting array via php just after that.

[silviaegt]

Hi @Daniel-KM, thank you so much for your tool. We find it very useful!
However we encountered the same problem as @laportjo (cf. http://sandbox.colmex.mx/~silvia/omeka25/references/creator#number)
We also changed the Reference.php line you mentioned (we tried utf8_roman_ci; utf8_general_ci; and utf8mb4_unicode_ci [following this post]), but it didn't have any effect.

Also, we don't quite understand what you mean by: "or add an option to manage it cleanly" nor with " you can sort the resulting array via php just after that"

Thanks in advance for any help you could provide us :)

[Daniel-KM]

The file to edit is the view one : https://github.com/Daniel-KM/Omeka-plugin-Reference/blob/master/views/public/common/reference-list.php. It's hard to resolve, because it's a mix of locale format of the server, of the database and of php.
As always with cms, copy this file in your theme (in common/reference-list.php), in order to keep the plugin clean and your site maintainable.

[zerocrates]

MySQL should be handling this sorting fine, whether under the default Omeka collation of utf8_unicode_ci or with utf8_general_ci.

I think the issue here is actually happening because of the PHP-side "re-sorting" that happens when the strip option is enabled. You can turn off strip and thereby rely only on the MySQL-side sorting, or you can replace the uksort in the Reference helper with something like this: (you'll need the PHP intl extension, though)

$collator = new Collator('root');
uksort($references, array($collator, 'compare'));

But, this would also require a change in the logic for generating headings: it will currently "see" the change from A to Á (and back again to A if the accented word appears in the middle of the list) and output extra headings.

[Daniel-KM]

Thanks, I'll give it a try.

[rodyoukai]

Hi all,

I solve the problem.

First, Comment this line:

https://github.com/Daniel-KM/Omeka-plugin-Reference/blob/master/views/helpers/Reference.php#L128

Because natcasesort does not support diacritics.

Second add this function on top of https://github.com/Daniel-KM/Omeka-plugin-Reference/blob/master/views/public/common/reference-list.php

<?php
function normaliza ($cadena){
    $originales = 'ÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖØÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõöøùúûýýþÿŔŕ';
    $modificadas = strtoupper('aaaaaaaceeeeiiiidnoooooouuuuy').'bsaaaaaaaceeeeiiiidnoooooouuuyybyRr';
    $cadena = utf8_decode($cadena);
    $cadena = strtr($cadena, utf8_decode($originales), $modificadas);
    return utf8_encode($cadena);
}

Third apply this function in reference-list.php in lines 22 & 60, as follows:

$first_char = normaliza(function_exists('mb_substr') ? mb_substr($reference, 0, 1) : substr($reference, 0, 1));

Fourth step... enjoy!!!

[DBinaghi]

Sorry, @rodyoukai, but I've tried your method and don't see any change.
Could you please provide a working example, like a live repository?
thanks

[DBinaghi]

$collator = new Collator('root');
uksort($references, array($collator, 'compare'));

This solution works very well for me, thanks. I've also changed the headings functions, like this:

// Get the list of headers.
$collator = new Collator('root');
uksort($references, array($collator, 'compare'));
$alphabet = array('A','B','C','Ĉ','D','E','È','F','G','Ĝ','H','Ĥ','I','J','Ĵ','K','L','M','N','O','P','Q','R','S','Ŝ','T','U','Ŭ','V','X','Y','W','Z');
$letters = array('number' => false) + array_fill_keys($alphabet, false);
foreach ($references as $reference => $referenceData):
	$first_char = mb_substr($reference, 0, 1, 'UTF-8');
	if (strlen($first_char) == 0 || preg_match('/\W|\d/u', $first_char)):
		$letters['number'] = true;
	else:
		$first_char = mb_strtoupper($first_char, 'UTF-8');;
		$letters[$first_char] = true;
	endif;
endforeach;
$pagination_list = '<ul class="pagination_list">';
foreach ($letters as $letter => $isSet):
	$letterDisplay = ($letter == 'number' ? '#0-9' : $letter);
	if ($isSet):
		$pagination_list .= sprintf('<li class="pagination_range"><a href="#%s">%s</a></li>', $letter, $letterDisplay);
	else:
		$pagination_list .= sprintf('<li class="pagination_range"><span>%s</span></li>', $letterDisplay);
	endif;
endforeach;

Similarly, then, for the references list.
Hope this helps.

[DBinaghi]

$alphabet = array('A','B','C','Ĉ','D','E','È','F','G','Ĝ','H','Ĥ','I','J','Ĵ','K','L','M','N','O','P','Q','R','S','Ŝ','T','U','Ŭ','V','X','Y','W','Z');
$letters = array('number' => false) + array_fill_keys($alphabet, false);

I've amended my fork, with a config option to add a custom alphabet to be used for skip links headers:

$alphabet = (get_option('reference_list_alphabet') != '' ? explode(' ', get_option('reference_list_alphabet')) : array_fill_keys(range('A', 'Z'), false)); 
$letters = array('number' => false) + array_fill_keys($alphabet, false);