PHP Multi Byte str_replace?

Question

I'm trying to do accented character replacement in PHP but get funky results, my guess being because i'm using a UTF-8 string and str_replace can't properly handle multi-byte strings..

$accents_search     = array('á','à','â','ã','ª','ä','å','Á','À','Â','Ã','Ä','é','è',
'ê','ë','É','È','Ê','Ë','í','ì','î','ï','Í','Ì','Î','Ï','œ','ò','ó','ô','õ','º','ø',
'Ø','Ó','Ò','Ô','Õ','ú','ù','û','Ú','Ù','Û','ç','Ç','Ñ','ñ'); 

$accents_replace    = array('a','a','a','a','a','a','a','A','A','A','A','A','e','e',
'e','e','E','E','E','E','i','i','i','i','I','I','I','I','oe','o','o','o','o','o','o',
'O','O','O','O','O','u','u','u','U','U','U','c','C','N','n'); 

$str = str_replace($accents_search, $accents_replace, $str);

Results I get:

Ørjan Nilsen -> �orjan Nilsen

Expected Result:

Ørjan Nilsen -> Orjan Nilsen

Edit: I've got my internal character handler set to UTF-8 (according to mb_internal_encoding()), also the value of $str is UTF-8, so from what I can tell, all the strings involved are UTF-8. Does str_replace() detect char sets and use them properly?

Answer 1

Looks like the string was not replaced because your input encoding and the file encoding mismatch.

Answer 2

Try this function definition:

if (!function_exists('mb_str_replace')) {
    function mb_str_replace($search, $replace, $subject) {
        if (is_array($subject)) {
            foreach ($subject as $key => $val) {
                $subject[$key] = mb_str_replace((string)$search, $replace, $subject[$key]);
            }
            return $subject;
        }
        $pattern = '/['.preg_quote(implode('', (array)$search), '/').']/u';
        if (is_array($search)) {
            if (is_array($replace)) {
                $len = min(count($search), count($replace));
                $table = array_combine(array_slice($search, 0, $len), array_slice($replace, 0, $len));
                $f = create_function('$match', '$table = '.var_export($table, true).'; return array_key_exists($match[0], $table) ? $table[$match[0]] : $match[0];');
                $subject = preg_replace_callback($pattern, $f, $subject);
                return $subject;
            }
        }
        $subject = preg_replace($pattern, (string)$replace, $subject);
        return $subject;
    }
}

Answer 3

It's possible to remove diacritics using Unicode normalization form D (NFD) and Unicode character properties.

NFD converts something like the "ü" umlaut from "LATIN SMALL LETTER U WITH DIAERESIS" (which is a letter) to "LATIN SMALL LETTER U" (letter) and "COMBINING DIAERESIS" (not a letter).

header('Content-Type: text/plain; charset=utf-8');

$test = implode('', array('á','à','â','ã','ª','ä','å','Á','À','Â','Ã','Ä','é','è',
'ê','ë','É','È','Ê','Ë','í','ì','î','ï','Í','Ì','Î','Ï','œ','ò','ó','ô','õ','º','ø',
'Ø','Ó','Ò','Ô','Õ','ú','ù','û','Ú','Ù','Û','ç','Ç','Ñ','ñ'));

$test = Normalizer::normalize($test, Normalizer::FORM_D);

// Remove everything that's not a "letter" or a space (e.g. diacritics)
// (see http://de2.php.net/manual/en/regexp.reference.unicode.php)
$pattern = '/[^\pL ]/u';

echo preg_replace($pattern, '', $test);

Output:

aaaaªaaAAAAAeeeeEEEEiiiiIIIIœooooºøØOOOOuuuUUUcCNn

The Normalizer class is part of the PECL intl package. (The algorithm itself isn't very complicated but needs to load a lot of character mappings afaik. I wrote a PHP implementation a while ago.)

(I'm adding this two months late because I think it's a nice technique that's not known widely enough.)