How to reverse a Unicode string

Question

It was hinted in a comment to an answer to this question that PHP can not reverse Unicode strings.

As for Unicode, it works in PHP because most apps process it as binary. Yes, PHP is 8-bit clean. Try the equivalent of this in PHP: perl -Mutf8 -e 'print scalar reverse("ほげほげ")' You will get garbage, not "げほげほ". – jrockway

And unfortunately it is correct that PHPs unicode support atm is at best "lacking". This will hopefully change drastically with PHP6.

PHPs MultiByte functions does provide the basic functionality you need to deal with unicode, but it is inconsistent and does lack a lot of functions. One of these is a function to reverse a string.

I of course wanted to reverse this text for no other reason then to figure out if it was possible. And I made a function to accomplish this enormous complex task of reversing this Unicode text, so you can relax a bit longer until PHP6.

Test code:

$enc = 'UTF-8';
$text = "ほげほげ";
$defaultEnc = mb_internal_encoding();

echo "Showing results with encoding $defaultEnc.\n\n";

$revNormal = strrev($text);
$revInt = mb_strrev($text);
$revEnc = mb_strrev($text, $enc);

echo "Original text is: $text .\n";
echo "Normal strrev output: " . $revNormal . ".\n";
echo "mb_strrev without encoding output: $revInt.\n";
echo "mb_strrev with encoding $enc output: $revEnc.\n";

if (mb_internal_encoding($enc)) {
    echo "\nSetting internal encoding to $enc from $defaultEnc.\n\n";

    $revNormal = strrev($text);
    $revInt = mb_strrev($text);
    $revEnc = mb_strrev($text, $enc);

    echo "Original text is: $text .\n";
    echo "Normal strrev output: " . $revNormal . ".\n";
    echo "mb_strrev without encoding output: $revInt.\n";
    echo "mb_strrev with encoding $enc output: $revEnc.\n";

} else {
    echo "\nCould not set internal encoding to $enc!\n";
}

Answer 1

Grapheme functions handle UTF-8 string more correctly than mbstring and PCRE functions/ Mbstring and PCRE may break characters. You can see the defference between them by executing the following code.

function str_to_array($string)
{
    $length = grapheme_strlen($string);
    $ret = [];

    for ($i = 0; $i < $length; $i += 1) {

        $ret[] = grapheme_substr($string, $i, 1);
    }

    return $ret;
}

function str_to_array2($string)
{
    $length = mb_strlen($string, "UTF-8");
    $ret = [];

    for ($i = 0; $i < $length; $i += 1) {

    $ret[] = mb_substr($string, $i, 1, "UTF-8");
}

    return $ret;
}

function str_to_array3($string)
{
    return preg_split('//u', $string, -1, PREG_SPLIT_NO_EMPTY);
}

function utf8_strrev($string)
{
    return implode(array_reverse(str_to_array($string)));
}

function utf8_strrev2($string)
{
    return implode(array_reverse(str_to_array2($string)));
}

function utf8_strrev3($string)
{
    return implode(array_reverse(str_to_array3($string)));
}

// http://www.php.net/manual/en/function.grapheme-strlen.php
$string = "a\xCC\x8A"  // 'LATIN SMALL LETTER A WITH RING ABOVE' (U+00E5)
         ."o\xCC\x88"; // 'LATIN SMALL LETTER O WITH DIAERESIS'  (U+00F6)

var_dump(array_map(function($elem) { return strtoupper(bin2hex($elem)); },
[
  'should be' => "o\xCC\x88"."a\xCC\x8A",
  'grapheme' => utf8_strrev($string),
  'mbstring' => utf8_strrev2($string),
  'pcre' => utf8_strrev3($string)
]));

The result is here.

array(4) {
  ["should be"]=>
  string(12) "6FCC8861CC8A"
  ["grapheme"]=>
  string(12) "6FCC8861CC8A"
  ["mbstring"]=>
  string(12) "CC886FCC8A61"
  ["pcre"]=>
  string(12) "CC886FCC8A61"
}

IntlBreakIterator can be used since PHP 5.5 (intl 3.0);

function utf8_strrev($str)
{
    $it = IntlBreakIterator::createCodePointInstance();
    $it->setText($str);

    $ret = '';
    $pos = 0;
    $prev = 0;

    foreach ($it as $pos) {
        $ret = substr($str, $prev, $pos - $prev) . $ret;
        $prev = $pos;
    }

    return $ret;  
}

Answer 2

here's another approach using regex:

function utf8_strrev($str){
 preg_match_all('/./us', $str, $ar);
 return implode(array_reverse($ar[0]));
}

Answer 3

Here's another way. This seems to work without having to specify an output encoding (tested with a couple of different mb_internal_encodings):

function mb_strrev($text)
{
    return join('', array_reverse(
        preg_split('~~u', $text, -1, PREG_SPLIT_NO_EMPTY)
    ));
}

Answer 4

The answer

function mb_strrev($text, $encoding = null)
{
    $funcParams = array($text);
    if ($encoding !== null)
        $funcParams[] = $encoding;
    $length = call_user_func_array('mb_strlen', $funcParams);

    $output = '';
    $funcParams = array($text, $length, 1);
    if ($encoding !== null)
        $funcParams[] = $encoding;
    while ($funcParams[1]--) {
         $output .= call_user_func_array('mb_substr', $funcParams);
    }
    return $output;
}

Answer 5

Another method:

function mb_strrev($str, $enc = null) {
    if(is_null($enc)) $enc = mb_internal_encoding();
    $str = mb_convert_encoding($str, 'UTF-16BE', $enc);
    return mb_convert_encoding(strrev($str), $enc, 'UTF-16LE');
}

Answer 6

It is easy utf8_strrev( $str ). See the relevant source code of my Library that I copied below:

function utf8_strrev( $str )
{
    return implode( array_reverse( utf8_split( $str ) ) );
}

function utf8_split( $str , $split_length = 1 )
{
    $str    = ( string ) $str;

    $ret    = array( );

    if( pcre_utf8_support( ) )
    {
        $str    = utf8_clean( $str );

        $ret    = preg_split('/(?<!^)(?!$)/u', $str );

        // \X is buggy in many recent versions of PHP
        //preg_match_all( '/\X/u' , $str , $ret );
        //$ret  = $ret[0];
    }
    else
    {
        //Fallback

        $len    = strlen( $str );

        for( $i = 0 ; $i < $len ; $i++ )
        {
            if( ( $str[$i] & "\x80" ) === "\x00" )
            {
                $ret[]  = $str[$i];
            }
            else if( ( ( $str[$i] & "\xE0" ) === "\xC0" ) && ( isset( $str[$i+1] ) ) )
            {
                if( ( $str[$i+1] & "\xC0" ) === "\x80" )
                {
                    $ret[]  = $str[$i] . $str[$i+1];

                    $i++;
                }
            }
            else if( ( ( $str[$i] & "\xF0" ) === "\xE0" ) && ( isset( $str[$i+2] ) ) )
            {
                if( ( ( $str[$i+1] & "\xC0" ) === "\x80" ) && ( ( $str[$i+2] & "\xC0" ) === "\x80" ) )
                {
                    $ret[]  = $str[$i] . $str[$i+1] . $str[$i+2];

                    $i  = $i + 2;
                }
            }
            else if( ( ( $str[$i] & "\xF8" ) === "\xF0" ) && ( isset( $str[$i+3] ) ) )
            {
                if( ( ( $str[$i+1] & "\xC0" ) === "\x80" ) && ( ( $str[$i+2] & "\xC0" ) === "\x80" ) && ( ( $str[$i+3] & "\xC0" ) === "\x80" ) )
                {
                    $ret[]  = $str[$i] . $str[$i+1] . $str[$i+2] . $str[$i+3];

                    $i  = $i + 3;
                }
            }
        }
    }


    if( $split_length > 1 )
    {
        $ret = array_chunk( $ret , $split_length );

        $ret    = array_map( 'implode' , $ret );
    }

    if( $ret[0] === '' )
    {
        return array( );
    }

    return $ret;
}


function utf8_clean( $str , $remove_bom = false )
{
    $regx = '/([\x00-\x7F]|[\xC0-\xDF][\x80-\xBF]|[\xE0-\xEF][\x80-\xBF]{2}|[\xF0-\xF7][\x80-\xBF]{3})|./s';

    $str    = preg_replace( $regx , '$1' , $str );

    if( $remove_bom )
    {
        $str    = utf8_str_replace( utf8_bom( ) , '' , $str );
    }

    return $str;
}


function utf8_str_replace( $search , $replace , $subject , &$count = 0 )
{
    return str_replace( $search , $replace , $subject , $count );
}


function utf8_bom( )
{
    return "\xef\xbb\xbf";

}


function pcre_utf8_support( )
{
    static $support;

    if( !isset( $support ) )
    {
        $support = @preg_match( '//u', '' );
        //Cached the response
    }

    return $support;
}