utf8_encode

(PHP 4, PHP 5, PHP 7)

utf8_encode — 将 ISO-8859-1 编码的字符串转换为 UTF-8 编码

描述

utf8_encode ( string $data ) : string

该函数将 data 字符串转换为 UTF-8 编码，并返回编码后的字符串。UTF-8 是一种用于将宽字符值转换为字节流的 Unicode 的标准机制。UTF-8 对于纯 ASCII 字符来说是透明的，且是自同步的（也就是说这使得程序能够得知字符从字节流的何处开始），并可被普通字符串比较函数用以比较等操作。PHP 可将 UTF-8 编码为多达四个字节的字符，如：

UTF-8 编码
字节（bytes） 位（bits） 表 示
1 7 0bbbbbbb
2 11 110bbbbb 10bbbbbb
3 16 1110bbbb 10bbbbbb 10bbbbbb
4 21 11110bbb 10bbbbbb 10bbbbbb 10bbbbbb

每个 UTF-8 表示一个能被用以储存字符数据的位。

User Contributed Notes

Pini 16-Nov-2015 03:14
My version of utf8_encode_deep,
In case you need one that returns a value without changing the original.

        /**
        * Convert Anything To UTF-8
        * @param mixed $var The variable you want to convert.
        * @param boolean $deep Deep convertion? (*Default: TRUE).
        * @return mixed
        */
        function anything_to_utf8($var,$deep=TRUE){
            if(is_array($var)){
                foreach($var as $key => $value){
                    if($deep){
                        $var[$key] = anything_to_utf8($value,$deep);
                    }elseif(!is_array($value) && !is_object($value) && !mb_detect_encoding($value,'utf-8',true)){
                         $var[$key] = utf8_encode($var);
                    }
                }
                return $var;
            }elseif(is_object($var)){
                foreach($var as $key => $value){
                    if($deep){
                        $var->$key = anything_to_utf8($value,$deep);
                    }elseif(!is_array($value) && !is_object($value) && !mb_detect_encoding($value,'utf-8',true)){
                         $var->$key = utf8_encode($var);
                    }
                }
                return $var;
            }else{
                return (!mb_detect_encoding($var,'utf-8',true))?utf8_encode($var):$var;
            }
        }
a dot rueedlinger at gmail dot com 22-Jun-2013 01:06
If you need a function which converts a string array into a utf8 encoded string array then this function might be useful for you:

<?php
function utf8_string_array_encode(&$array){
    $func = function(&$value,&$key){
        if(is_string($value)){
            $value = utf8_encode($value);
        }
        if(is_string($key)){
            $key = utf8_encode($key);
        }
        if(is_array($value)){
            utf8_string_array_encode($value);
        }
    };
    array_walk($array,$func);
    return $array;
}
?>
Oscar Broman 06-Sep-2012 11:15
Walk through nested arrays/objects and utf8 encode all strings.

<?php
// Usage
class Foo {
    public $somevar = 'whoop whoop';
}

$structure = array(
    'object' => (object) array(
        'entry' => 'hello w?rld',
        'another_array' => array(
            'string',
            1234,
            'another string'
        )
    ),
    'string' => 'foo',
    'foo_object' => new Foo
);

utf8_encode_deep($structure);

// $structure is now utf8 encoded
print_r($structure);

// The function
function utf8_encode_deep(&$input) {
    if (is_string($input)) {
        $input = utf8_encode($input);
    } else if (is_array($input)) {
        foreach ($input as &$value) {
            utf8_encode_deep($value);
        }

        unset($value);
    } else if (is_object($input)) {
        $vars = array_keys(get_object_vars($input));

        foreach ($vars as $var) {
            utf8_encode_deep($input->$var);
        }
    }
}
?>
deceze at gmail dot com 14-Jul-2011 06:51
Please note that utf8_encode only converts a string encoded in ISO-8859-1 to UTF-8. A more appropriate name for it would be "iso88591_to_utf8". If your text is not encoded in  ISO-8859-1, you do not need this function. If your text is already in UTF-8, you do not need this function. In fact, applying this function to text that is not encoded in ISO-8859-1 will most likely simply garble that text.

If you need to convert text from any encoding to any other encoding, look at iconv() instead.
powtac 4t gmx d0t de 11-Feb-2011 01:11
I tried a lot of things, but this seems to be the final fail save method to convert any string to proper UTF-8.

<?php
function _convert($content) {
    if(!mb_check_encoding($content, 'UTF-8')
        OR !($content === mb_convert_encoding(mb_convert_encoding($content, 'UTF-32', 'UTF-8' ), 'UTF-8', 'UTF-32'))) {

        $content = mb_convert_encoding($content, 'UTF-8');

        if (mb_check_encoding($content, 'UTF-8')) {
            // log('Converted to UTF-8');
        } else {
            // log('Could not converted to UTF-8');
        }
    }
    return $content;
}
?>
Yumok 15-Sep-2010 04:26
Avoiding use of preg_match to detect if utf8_encode is needed:

<?php
                $string = $string_input; // avoid being destructive

                $string = preg_replace("#[\x09\x0A\x0D\x20-\x7E]#"        ,"",$string);         // ASCII
                $string = preg_replace("#[\xC2-\xDF][\x80-\xBF]#"            ,"",$string);             // non-overlong 2-byte
                $string = preg_replace("#\xE0[\xA0-\xBF][\x80-\xBF]#"    ,"",$string);     // excluding overlongs
                $string = preg_replace("#[\xE1-\xEC\xEE\xEF][\x80-\xBF]{2}#","",$string);     // straight 3-byte
                $string = preg_replace("#\xED[\x80-\x9F][\x80-\xBF]#"    ,"",$string);     // excluding surrogates
                $string = preg_replace("#\xF0[\x90-\xBF][\x80-\xBF]{2}#","",$string);     // planes 1-3
                $string = preg_replace("#[\xF1-\xF3][\x80-\xBF]{3}#"    ,"",$string);     //  planes 4-15
                $string = preg_replace("#\xF4[\x80-\x8F][\x80-\xBF]{2}#","",$string);     // plane 16

                $rc = ($string == ""?true:false);
?>
rogeriogirodo at gmail dot com 19-Mar-2009 04:24
This function may be useful do encode array keys and values [and checks first to see if it's already in UTF format]:

<?php
public static function to_utf8($in)
{
        if (is_array($in)) {
            foreach ($in as $key => $value) {
                $out[to_utf8($key)] = to_utf8($value);
            }
        } elseif(is_string($in)) {
            if(mb_detect_encoding($in) != "UTF-8")
                return utf8_encode($in);
            else
                return $in;
        } else {
            return $in;
        }
        return $out;
}
?>

Hope this may help.

[NOTE BY danbrown AT php DOT net: Original function written by (cmyk777 AT gmail DOT com) on 28-JAN-09.]
www.tricinty.com 11-Jun-2008 03:43
<?php
    /**
    * Encodes an ISO-8859-1 mixed variable to UTF-8 (PHP 4, PHP 5 compat)
    * @param    mixed    $input An array, associative or simple
    * @param    boolean  $encode_keys optional
    * @return    mixed     ( utf-8 encoded $input)
    */

    function utf8_encode_mix($input, $encode_keys=false)
    {
        if(is_array($input))
        {
            $result = array();
            foreach($input as $k => $v)
            {               
                $key = ($encode_keys)? utf8_encode($k) : $k;
                $result[$key] = utf8_encode_mix( $v, $encode_keys);
            }
        }
        else
        {
            $result = utf8_encode($input);
        }

        return $result;
    }
?>
emze at donazga dot net 17-Dec-2006 09:42
/*
Every function seen so far is incomplete or resource consumpting. Here are two -- integer 2 utf sequence (i3u) and utf sequence to integer (u3i). Below is a code snippet that checks well behavior at the range boundaries.

Someday they might be hardcoded into PHP...
*/

function i3u($i) { // returns UCS-16 or UCS-32 to UTF-8 from an integer
  $i=(int)$i; // integer?
  if ($i<0) return false; // positive?
  if ($i<=0x7f) return chr($i); // range 0
  if (($i & 0x7fffffff) <> $i) return '?'; // 31 bit?
  if ($i<=0x7ff) return chr(0xc0 | ($i >> 6)) . chr(0x80 | ($i & 0x3f));
  if ($i<=0xffff) return chr(0xe0 | ($i >> 12)) . chr(0x80 | ($i >> 6) & 0x3f)
      . chr(0x80  | $i & 0x3f);
  if ($i<=0x1fffff) return chr(0xf0 | ($i >> 18)) . chr(0x80 | ($i >> 12) & 0x3f)
      . chr(0x80 | ($i >> 6) & 0x3f) . chr(0x80  | $i & 0x3f);
  if ($i<=0x3ffffff) return chr(0xf8 | ($i >> 24)) . chr(0x80 | ($i >> 18) & 0x3f)
      . chr(0x80 | ($i >> 12) & 0x3f) . chr(0x80 | ($i >> 6) & 0x3f) . chr(0x80  | $i & 0x3f);
  return chr(0xfc | ($i >> 30)) . chr(0x80 | ($i >> 24) & 0x3f) . chr(0x80 | ($i >> 18) & 0x3f)
      . chr(0x80 | ($i >> 12) & 0x3f) . chr(0x80 | ($i >> 6) & 0x3f) . chr(0x80  | $i & 0x3f);
}

function u3i($s,$strict=1) { // returns integer on valid UTF-8 seq, NULL on empty, else FALSE
  // NOT strict: takes only DATA bits, present or not; strict: length and bits checking
  if ($s=='') return NULL;
  $l=strlen($s); $o=ord($s{0});
  if ($o <= 0x7f && $l==1) return $o;
  if ($l>6 && $strict) return false;
  if ($strict) for ($i=1;$i<$l;$i++) if (ord($s{$i}) > 0xbf || ord($s{$i})< 0x80) return false;
  if ($o < 0xc2) return false; // no-go even if strict=0
  if ($o <= 0xdf && ($l=2 && $strict)) return (($o & 0x1f) << 6 | (ord($s{1}) & 0x3f));
  if ($o <= 0xef && ($l=3 && $strict)) return (($o & 0x0f) << 12 | (ord($s{1}) & 0x3f) << 6
     |  (ord($s{2}) & 0x3f));
  if ($o <= 0xf7 && ($l=4 && $strict)) return (($o & 0x07) << 18 | (ord($s{1}) & 0x3f) << 12
     | (ord($s{2}) & 0x3f) << 6 |  (ord($s{3}) & 0x3f));
  if ($o <= 0xfb && ($l=5 && $strict)) return (($o & 0x03) << 24 | (ord($s{1}) & 0x3f) << 18
     | (ord($s{2}) & 0x3f) << 12 | (ord($s{3}) & 0x3f) << 6 |  (ord($s{4}) & 0x3f));
  if ($o <= 0xfd && ($l=6 && $strict)) return (($o & 0x01) << 30 | (ord($s{1}) & 0x3f) << 24
     | (ord($s{2}) & 0x3f) << 18 | (ord($s{3}) & 0x3f) << 12
     | (ord($s{4}) & 0x3f) << 6 |  (ord($s{5}) & 0x3f));
  return false;
}

// boundary behavior checking
$do=array(0x7f,0x7ff,0xffff,0x1fffff,0x3ffffff,0x7fffffff);
foreach ($do as $ii) for ($i=$ii;$i<=$ii+1; $i++) {
  $o=i3u($i);
  for ($j=0;$j<strlen($o);$j++) print "O[$j]=" . sprintf('%08b',ord($o{$j})) . ", ";
  print "c=$i, o=[$o].\n";
  print "Back: [$o] => [" . u3i($o) . "]\n";
}
rocketman 16-Mar-2006 04:46
If you are looking for a function to replace special characters with the hex-utf-8 value (e.g. für Webservice-Security/WSS4J compliancy) you might use this:

$textstart = "Gr??e";
$utf8 ='';
$max = strlen($txt);

for ($i = 0; $i < $max; $i++) {

if ($txt{i} == "&"){
$neu = "&x26;";
}
elseif ((ord($txt{$i}) < 32) or (ord($txt{$i}) > 127)){
$neu = urlencode(utf8_encode($txt{$i}));
$neu = preg_replace('#\%(..)\%(..)\%(..)#','&#x\1;&#x\2;&#x\3;',$neu);
$neu = preg_replace('#\%(..)\%(..)#','&#x\1;&#x\2;',$neu);
$neu = preg_replace('#\%(..)#','&#x\1;',$neu);
}
else {
$neu = $txt{$i};
}
       
$utf8 .= $neu;
} // for $i

$textnew = $utf8;

In this example $textnew will be "Gr&#xC3;&#xB6;&#xC3;&#x9F;e"
mailing at jcn50 dot com 20-Jan-2006 10:40
I recommend using this alternative for every language:

$new=mb_convert_encoding($s,"UTF-8","auto");

Don't forget to set all your pages to "utf-8" encoding, otherwise just use HTML entities.

jcn50.
04-Nov-2005 02:34
// Reads a file story.txt ascii (as typed on keyboard)
// converts it to Georgian character using utf8 encoding
// if I am correct(?) just as it should be when typed on Georgian computer
// it outputs it as an html file
//
// http://www.comweb.nl/keys_to_georgian.html
// http://www.comweb.nl/keys_to_georgian.php
// http://www.comweb.nl/story.txt

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">

<HTML>
<HEAD>
<TITLE>keys to unicode code</TITLE>

// this meta tag is needed
<meta http-equiv="Content-Type" content="text/html;CHARSET=gb2312" >

// note the sylfean font seems to be standard installed on Windows XP
// It supports Georgian
 
<style TYPE="text/css">
<!--
body {font-family:sylfaen; }
-->
</style>
</HEAD>

<BODY>

<?
$eng=array(97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,
112,113,114,115,116,117,118,119,120,121,122,87,82,84,83,
67,74,90);
$geo=array(4304,4305,4330,4307,4308,4324,4306,4336,4312,4335,4313,
4314,4315,4316,4317,4318,4325,4320,4321,4322,4323,4309,
4332,4334,4327,4310,4333,4326,4311,4328,4329,4319,4331,
91,93,59,39,44,46,96);

$fc=file("story.txt");
foreach($fc as $line)
{
   $spacestart=1;
   for ($i=0; $i<strlen($line); $i+=1)
   {
      $character=ord(substr($line,$i,1));
      $found=0;
      for ($k=0; $k<count($eng); $k+=1)
      {
         if ($eng[$k]==$character)
         {
             print code2utf( $geo[$k] );
             $found=1;
         }
      }
      if ($found==0)
      {
         if ($character==126 || $character==32 || $character==10 || $character==9)
         {
            if ($character==9)  { print '&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;'; }
            if ($character==10) { print "<BR>\n"; }
            if ($character==32)
            {
               if ($spacestart==1) {print '&nbsp;'; } else { print " "; }
            }
            if ($character==126){ print "~";      }
         } else
         {
            print substr($line,$i,1);
         }
      }
      if ($character!=32) { $spacestart=0; }
   }
}

/**
 * Function coverts number of utf char into that character.
 * Function taken from: http://sk2.php.net/manual/en/function.utf8-encode.php#49336
 *
 * @param int $num
 * @return utf8char
*/
function code2utf($num)
{
   if($num<128)return chr($num);
   if($num<2048)return chr(($num>>6)+192).chr(($num&63)+128);
   if($num<65536)return chr(($num>>12)+224).chr((($num>>6)&63)+128).chr(($num&63)+128);
   if($num<2097152)return chr(($num>>18)+240).chr((($num>>12)&63)+128).chr((($num>>6)&63)+128) .chr(($num&63)+128);
   return '';
}
?>

</BODY>
</HTML>
Janci 04-Nov-2005 04:00
I was searching for a function similar to Javascript's unescape(). In most cases it is OK to use url_decode() function but not if you've got UTF characters in the strings. They are converted into %uXXXX entities that url_decode() cannot handle.
I googled the net and found a function which actualy converts these entities into HTML entities (&#xxx;) that your browser can show correctly. If you're OK with that, the function can be found here: http://pure-essence.net/stuff/code/utf8RawUrlDecode.phps

But it was not OK with me because I needed a string in my charset to make some comparations and other stuff. So I have modified the above function and in conjuction with code2utf() function mentioned in some other note here, I have managed to achieve my goal:

<?php
/**
 * Function converts an Javascript escaped string back into a string with specified charset (default is UTF-8).
 * Modified function from http://pure-essence.net/stuff/code/utf8RawUrlDecode.phps
 *
 * @param string $source escaped with Javascript's escape() function
 * @param string $iconv_to destination character set will be used as second paramether in the iconv function. Default is UTF-8.
 * @return string
 */
function unescape($source, $iconv_to = 'UTF-8') {
    $decodedStr = '';
    $pos = 0;
    $len = strlen ($source);
    while ($pos < $len) {
        $charAt = substr ($source, $pos, 1);
        if ($charAt == '%') {
            $pos++;
            $charAt = substr ($source, $pos, 1);
            if ($charAt == 'u') {
                // we got a unicode character
                $pos++;
                $unicodeHexVal = substr ($source, $pos, 4);
                $unicode = hexdec ($unicodeHexVal);
                $decodedStr .= code2utf($unicode);
                $pos += 4;
            }
            else {
                // we have an escaped ascii character
                $hexVal = substr ($source, $pos, 2);
                $decodedStr .= chr (hexdec ($hexVal));
                $pos += 2;
            }
        }
        else {
            $decodedStr .= $charAt;
            $pos++;
        }
    }

    if ($iconv_to != "UTF-8") {
        $decodedStr = iconv("UTF-8", $iconv_to, $decodedStr);
    }
   
    return $decodedStr;
}

/**
 * Function coverts number of utf char into that character.
 * Function taken from: http://sk2.php.net/manual/en/function.utf8-encode.php#49336
 *
 * @param int $num
 * @return utf8char
 */
function code2utf($num){
    if($num<128)return chr($num);
    if($num<2048)return chr(($num>>6)+192).chr(($num&63)+128);
    if($num<65536)return chr(($num>>12)+224).chr((($num>>6)&63)+128).chr(($num&63)+128);
    if($num<2097152)return chr(($num>>18)+240).chr((($num>>12)&63)+128).chr((($num>>6)&63)+128) .chr(($num&63)+128);
    return '';
}
?>
suttichai at ceforce dot com 28-May-2005 12:26
This function I use convert Thai font (iso-8859-11) to UTF-8. For my case, It work properly. Please try to use this function if you have a problem to convert charset iso-8859-11 to UTF-8.

function iso8859_11toUTF8($string) {
 
     if ( ! ereg("[\241-\377]", $string) )
         return $string;
 
     $iso8859_11 = array(
"\xa1" => "\xe0\xb8\x81",
"\xa2" => "\xe0\xb8\x82",
"\xa3" => "\xe0\xb8\x83",
"\xa4" => "\xe0\xb8\x84",
"\xa5" => "\xe0\xb8\x85",
"\xa6" => "\xe0\xb8\x86",
"\xa7" => "\xe0\xb8\x87",
"\xa8" => "\xe0\xb8\x88",
"\xa9" => "\xe0\xb8\x89",
"\xaa" => "\xe0\xb8\x8a",
"\xab" => "\xe0\xb8\x8b",
"\xac" => "\xe0\xb8\x8c",
"\xad" => "\xe0\xb8\x8d",
"\xae" => "\xe0\xb8\x8e",
"\xaf" => "\xe0\xb8\x8f",
"\xb0" => "\xe0\xb8\x90",
"\xb1" => "\xe0\xb8\x91",
"\xb2" => "\xe0\xb8\x92",
"\xb3" => "\xe0\xb8\x93",
"\xb4" => "\xe0\xb8\x94",
"\xb5" => "\xe0\xb8\x95",
"\xb6" => "\xe0\xb8\x96",
"\xb7" => "\xe0\xb8\x97",
"\xb8" => "\xe0\xb8\x98",
"\xb9" => "\xe0\xb8\x99",
"\xba" => "\xe0\xb8\x9a",
"\xbb" => "\xe0\xb8\x9b",
"\xbc" => "\xe0\xb8\x9c",
"\xbd" => "\xe0\xb8\x9d",
"\xbe" => "\xe0\xb8\x9e",
"\xbf" => "\xe0\xb8\x9f",
"\xc0" => "\xe0\xb8\xa0",
"\xc1" => "\xe0\xb8\xa1",
"\xc2" => "\xe0\xb8\xa2",
"\xc3" => "\xe0\xb8\xa3",
"\xc4" => "\xe0\xb8\xa4",
"\xc5" => "\xe0\xb8\xa5",
"\xc6" => "\xe0\xb8\xa6",
"\xc7" => "\xe0\xb8\xa7",
"\xc8" => "\xe0\xb8\xa8",
"\xc9" => "\xe0\xb8\xa9",
"\xca" => "\xe0\xb8\xaa",
"\xcb" => "\xe0\xb8\xab",
"\xcc" => "\xe0\xb8\xac",
"\xcd" => "\xe0\xb8\xad",
"\xce" => "\xe0\xb8\xae",
"\xcf" => "\xe0\xb8\xaf",
"\xd0" => "\xe0\xb8\xb0",
"\xd1" => "\xe0\xb8\xb1",
"\xd2" => "\xe0\xb8\xb2",
"\xd3" => "\xe0\xb8\xb3",
"\xd4" => "\xe0\xb8\xb4",
"\xd5" => "\xe0\xb8\xb5",
"\xd6" => "\xe0\xb8\xb6",
"\xd7" => "\xe0\xb8\xb7",
"\xd8" => "\xe0\xb8\xb8",
"\xd9" => "\xe0\xb8\xb9",
"\xda" => "\xe0\xb8\xba",
"\xdf" => "\xe0\xb8\xbf",
"\xe0" => "\xe0\xb9\x80",
"\xe1" => "\xe0\xb9\x81",
"\xe2" => "\xe0\xb9\x82",
"\xe3" => "\xe0\xb9\x83",
"\xe4" => "\xe0\xb9\x84",
"\xe5" => "\xe0\xb9\x85",
"\xe6" => "\xe0\xb9\x86",
"\xe7" => "\xe0\xb9\x87",
"\xe8" => "\xe0\xb9\x88",
"\xe9" => "\xe0\xb9\x89",
"\xea" => "\xe0\xb9\x8a",
"\xeb" => "\xe0\xb9\x8b",
"\xec" => "\xe0\xb9\x8c",
"\xed" => "\xe0\xb9\x8d",
"\xee" => "\xe0\xb9\x8e",
"\xef" => "\xe0\xb9\x8f",
"\xf0" => "\xe0\xb9\x90",
"\xf1" => "\xe0\xb9\x91",
"\xf2" => "\xe0\xb9\x92",
"\xf3" => "\xe0\xb9\x93",
"\xf4" => "\xe0\xb9\x94",
"\xf5" => "\xe0\xb9\x95",
"\xf6" => "\xe0\xb9\x96",
"\xf7" => "\xe0\xb9\x97",
"\xf8" => "\xe0\xb9\x98",
"\xf9" => "\xe0\xb9\x99",
"\xfa" => "\xe0\xb9\x9a",
"\xfb" => "\xe0\xb9\x9b"
 );
 
     $string=strtr($string,$iso8859_11);
     return $string;
 }

Suttichai Mesaard-www.ceforce.com
bisqwit at iki dot fi 20-May-2005 01:15
For reference, it may be insightful to point out that:
  utf8_encode($s)
is actually identical to:
  recode_string('latin1..utf8', $s)
and:
  iconv('iso-8859-1', 'utf-8', $s)
That is, utf8_encode is a specialized case of character set conversions.

If your string to be converted to utf-8 is something other than iso-8859-1 (such as iso-8859-2 (Polish/Croatian)), you should use recode_string() or iconv() instead rather than trying to devise complex str_replace statements.
JF Sebastian 09-Apr-2005 03:54
The following Perl regular expression tests if a string is well-formed Unicode UTF-8 (Broken up after each | since long lines are not permitted here. Please join as a single line, no spaces, before use.):

^([\x00-\x7f]|
[\xc2-\xdf][\x80-\xbf]|
\xe0[\xa0-\xbf][\x80-\xbf]|
[\xe1-\xec][\x80-\xbf]{2}|
\xed[\x80-\x9f][\x80-\xbf]|
[\xee-\xef][\x80-\xbf]{2}|
f0[\x90-\xbf][\x80-\xbf]{2}|
[\xf1-\xf3][\x80-\xbf]{3}|
\xf4[\x80-\x8f][\x80-\xbf]{2})*$

NOTE: This strictly follows the Unicode standard 4.0, as described in chapter 3.9, table 3-6, "Well-formed UTF-8 byte sequences" ( http://www.unicode.org/versions/Unicode4.0.0/ch03.pdf#G31703 ).

ISO-10646, a super-set of Unicode, uses UTF-8 (there called "UCS", see http://www.unicode.org/faq/utf_bom.html#1 ) in a relaxed variant that supports a 31-bit space encoded into up to six bytes instead of Unicode's 21 bits in up to four bytes. To check for ISO-10646 UTF-8, use the following Perl regular expression (again, broken up, see above):

^([\x00-\x7f]|
[\xc0-\xdf][\x80-\xbf]|
[\xe0-\xef][\x80-\xbf]{2}|
[\xf0-\xf7][\x80-\xbf]{3}|
[\xf8-\xfb][\x80-\xbf]{4}|
[\xfc-\xfd][\x80-\xbf]{5})*$

The following function may be used with above expressions for a quick UTF-8 test, e.g. to distinguish ISO-8859-1-data from UTF-8-data if submitted from a <form accept-charset="utf-8,iso-8859-1" method=..>.

function is_utf8($string) {
   return (preg_match('/[insert regular expression here]/', $string) === 1);
}
hrpeters (at) gmx (dot) net 13-Dec-2004 10:46
// Validate Unicode UTF-8 Version 4
// This function takes as reference the table 3.6 found at http://www.unicode.org/versions/Unicode4.0.0/ch03.pdf
// It also flags overlong bytes as error

function is_validUTF8($str)
{
    // values of -1 represent disalloweded values for the first bytes in current UTF-8
    static $trailing_bytes = array (
        0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
        0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
        0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
        0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
        -1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1, -1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,
        -1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1, -1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,
        -1,-1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
        2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, 3,3,3,3,3,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1
    );

    $ups = unpack('C*', $str);
    if (!($aCnt = count($ups))) return true; // Empty string *is* valid UTF-8
    for ($i = 1; $i <= $aCnt;)
    {
        if (!($tbytes = $trailing_bytes[($b1 = $ups[$i++])])) continue;
        if ($tbytes == -1) return false;
       
        $first = true;
        while ($tbytes > 0 && $i <= $aCnt)
        {
            $cbyte = $ups[$i++];
            if (($cbyte & 0xC0) != 0x80) return false;
           
            if ($first)
            {
                switch ($b1)
                {
                    case 0xE0:
                        if ($cbyte < 0xA0) return false;
                        break;
                    case 0xED:
                        if ($cbyte > 0x9F) return false;
                        break;
                    case 0xF0:
                        if ($cbyte < 0x90) return false;
                        break;
                    case 0xF4:
                        if ($cbyte > 0x8F) return false;
                        break;
                    default:
                        break;
                }
                $first = false;
            }
            $tbytes--;
        }
        if ($tbytes) return false; // incomplete sequence at EOS
    }       
    return true;
}
Mark AT modernbill DOT com 09-Nov-2004 11:56
If you haven't guessed already: If the UTF-8 character has no representation in the ISO-8859-1 codepage, a ? will be returned. You might want to wrap a function around this to make sure you aren't saving a bunch of ???? into your database.
Aidan Kehoe <php-manual at parhasard dot net> 30-Aug-2004 07:05
Here's some code that addresses the issue that Steven describes in the previous comment;

<?php

/* This structure encodes the difference between ISO-8859-1 and Windows-1252,
   as a map from the UTF-8 encoding of some ISO-8859-1 control characters to
   the UTF-8 encoding of the non-control characters that Windows-1252 places
   at the equivalent code points. */

$cp1252_map = array(
    "\xc2\x80" => "\xe2\x82\xac", /* EURO SIGN */
    "\xc2\x82" => "\xe2\x80\x9a", /* SINGLE LOW-9 QUOTATION MARK */
    "\xc2\x83" => "\xc6\x92",     /* LATIN SMALL LETTER F WITH HOOK */
    "\xc2\x84" => "\xe2\x80\x9e", /* DOUBLE LOW-9 QUOTATION MARK */
    "\xc2\x85" => "\xe2\x80\xa6", /* HORIZONTAL ELLIPSIS */
    "\xc2\x86" => "\xe2\x80\xa0", /* DAGGER */
    "\xc2\x87" => "\xe2\x80\xa1", /* DOUBLE DAGGER */
    "\xc2\x88" => "\xcb\x86",     /* MODIFIER LETTER CIRCUMFLEX ACCENT */
    "\xc2\x89" => "\xe2\x80\xb0", /* PER MILLE SIGN */
    "\xc2\x8a" => "\xc5\xa0",     /* LATIN CAPITAL LETTER S WITH CARON */
    "\xc2\x8b" => "\xe2\x80\xb9", /* SINGLE LEFT-POINTING ANGLE QUOTATION */
    "\xc2\x8c" => "\xc5\x92",     /* LATIN CAPITAL LIGATURE OE */
    "\xc2\x8e" => "\xc5\xbd",     /* LATIN CAPITAL LETTER Z WITH CARON */
    "\xc2\x91" => "\xe2\x80\x98", /* LEFT SINGLE QUOTATION MARK */
    "\xc2\x92" => "\xe2\x80\x99", /* RIGHT SINGLE QUOTATION MARK */
    "\xc2\x93" => "\xe2\x80\x9c", /* LEFT DOUBLE QUOTATION MARK */
    "\xc2\x94" => "\xe2\x80\x9d", /* RIGHT DOUBLE QUOTATION MARK */
    "\xc2\x95" => "\xe2\x80\xa2", /* BULLET */
    "\xc2\x96" => "\xe2\x80\x93", /* EN DASH */
    "\xc2\x97" => "\xe2\x80\x94", /* EM DASH */

    "\xc2\x98" => "\xcb\x9c",     /* SMALL TILDE */
    "\xc2\x99" => "\xe2\x84\xa2", /* TRADE MARK SIGN */
    "\xc2\x9a" => "\xc5\xa1",     /* LATIN SMALL LETTER S WITH CARON */
    "\xc2\x9b" => "\xe2\x80\xba", /* SINGLE RIGHT-POINTING ANGLE QUOTATION*/
    "\xc2\x9c" => "\xc5\x93",     /* LATIN SMALL LIGATURE OE */
    "\xc2\x9e" => "\xc5\xbe",     /* LATIN SMALL LETTER Z WITH CARON */
    "\xc2\x9f" => "\xc5\xb8"      /* LATIN CAPITAL LETTER Y WITH DIAERESIS*/
);

function cp1252_to_utf8($str) {
        global $cp1252_map;
        return  strtr(utf8_encode($str), $cp1252_map);
}

?>
Net Raven 24-Jun-2004 12:58
I often need to convert multi language text sent to me for use in websites and other apps into UTF8 encoded so I can insert it into source code and databases.

I knocked up a small web page with its charset set to UTF8 then set it up so I can paste from the original doc (eg word or excel) and have the page return the UTF8 encoded version.

Of course the browser will convert the unicode to UTF8 for you as part of the submit (I use IE5 or better for this) then all you have to do in the PHP is encode the UTF8 so the browser will show it in its raw form.

Its a bit bulky but I just convert ALL character to html numbered entities (brute force and ignorance does it again.)

I've used this to encode everything from Hebrew to Japanese without problems

<?
header("Content-Type: text/plain; CHARSET=gb2312");
$code = (get_magic_quotes_gpc())?stripslashes($GLOBALS[code]):$GLOBALS[code];
?>
<html>
<head>
    <title>UTF8 ENCODER PAGE</title>
    <meta http-equiv="content-type" content="text/html; CHARSET=gb2312">
</head>
<body>
<form method=post action="?seed=<?=time()?>">
    Original Unicode<br />
    <textarea name="code" cols="80" rows="10"><?=$code?></textarea><br />
    Encoded UTF8<br />
    <textarea name="encd" cols="80" rows="10"><?
        for ($i = 0; $i < strlen($code); $i++) {
            echo '&#'.ord(substr($code,$i,1));
        }
    ?></textarea><br />
    <input type="submit" value="encode">
</form>
</body>
</html>
Karen 01-Oct-2003 12:33
Re the previous post about converting GB2312 code to Unicode code which displayed the following function:

<?
// Program by sadly (www.phpx.com)

function gb2unicode($gb)
{
   if(!trim($gb))
    return $gb;
   $filename="gb2312.txt";
   $tmp=file($filename);
   $codetable=array();
   while(list($key,$value)=each($tmp))
    $codetable[hexdec(substr($value,0,6))]=substr($value,9,4);
   $utf="";
   while($gb)
    {
      if (ord(substr($gb,0,1))>127)
     {
        $this=substr($gb,0,2);
        $gb=substr($gb,2,strlen($gb));
        $utf.="&#x".$codetable[hexdec(bin2hex($this))-0x8080].";";
      }
     else
     {
      $gb=substr($gb,1,strlen($gb));
      $utf.=substr($gb,0,1);
     }
     }
  return $utf;
}
?>

I found that a small change was needed in the code to properly handle latin characters embedded in the middle of gb2312 text, as when the text includes a URL or email address. Just reverse the two lines in the part of the statement above that handles ord vals !>127.

Change:

$gb=substr($gb,1,strlen($gb));
$utf.=substr($gb,0,1);

to:

$utf.=substr($gb,0,1);
$gb=substr($gb,1,strlen($gb));

In the original function, the first latin chacter was dropped and it was not converting the first non-latin character after the latin text (everything was shifted one character too far to the right). Reversing those two lines makes it work correctly in every example I have tried.

Also, the source of the gb2312.txt file needed for this to work has changed. You can find it a couple places:

http://tcl.apache.org/sources/tcl/tools/encoding/gb2312.txt
ftp://ftp.unicode.org/Public/MAPPINGS/OBSOLETE/EASTASIA/GB/GB2312.TXT
ronen at greyzone dot com 07-Mar-2002 12:01
The following function will utf-8 encode unicode entities &#nnn(nn); with n={0..9}

/**
* takes a string of unicode entities and converts it to a utf-8 encoded string
* each unicode entitiy has the form &#nnn(nn); n={0..9} and can be displayed by utf-8 supporting
* browsers.  Ascii will not be modified.
* @param $source string of unicode entities [STRING]
* @return a utf-8 encoded string [STRING]
* @access public
*/
function utf8Encode ($source) {
    $utf8Str = '';
    $entityArray = explode ("&#", $source);
    $size = count ($entityArray);
    for ($i = 0; $i < $size; $i++) {
        $subStr = $entityArray[$i];
        $nonEntity = strstr ($subStr, ';');
        if ($nonEntity !== false) {
            $unicode = intval (substr ($subStr, 0, (strpos ($subStr, ';') + 1)));
            // determine how many chars are needed to reprsent this unicode char
            if ($unicode < 128) {
                $utf8Substring = chr ($unicode);
            }
            else if ($unicode >= 128 && $unicode < 2048) {
                $binVal = str_pad (decbin ($unicode), 11, "0", STR_PAD_LEFT);
                $binPart1 = substr ($binVal, 0, 5);
                $binPart2 = substr ($binVal, 5);
           
                $char1 = chr (192 + bindec ($binPart1));
                $char2 = chr (128 + bindec ($binPart2));
                $utf8Substring = $char1 . $char2;
            }
            else if ($unicode >= 2048 && $unicode < 65536) {
                $binVal = str_pad (decbin ($unicode), 16, "0", STR_PAD_LEFT);
                $binPart1 = substr ($binVal, 0, 4);
                $binPart2 = substr ($binVal, 4, 6);
                $binPart3 = substr ($binVal, 10);
           
                $char1 = chr (224 + bindec ($binPart1));
                $char2 = chr (128 + bindec ($binPart2));
                $char3 = chr (128 + bindec ($binPart3));
                $utf8Substring = $char1 . $char2 . $char3;
            }
            else {
                $binVal = str_pad (decbin ($unicode), 21, "0", STR_PAD_LEFT);
                $binPart1 = substr ($binVal, 0, 3);
                $binPart2 = substr ($binVal, 3, 6);
                $binPart3 = substr ($binVal, 9, 6);
                $binPart4 = substr ($binVal, 15);
       
                $char1 = chr (240 + bindec ($binPart1));
                $char2 = chr (128 + bindec ($binPart2));
                $char3 = chr (128 + bindec ($binPart3));
                $char4 = chr (128 + bindec ($binPart4));
                $utf8Substring = $char1 . $char2 . $char3 . $char4;
            }
           
            if (strlen ($nonEntity) > 1)
                $nonEntity = substr ($nonEntity, 1); // chop the first char (';')
            else
                $nonEntity = '';

            $utf8Str .= $utf8Substring . $nonEntity;
        }
        else {
            $utf8Str .= $subStr;
        }
    }

    return $utf8Str;
}
       
Ronen.