For my last project I needed to convert several CSV files from Windows-1250 to UTF-8, and after several days of searching around I found a function that is partially solved my problem, but it still has not transformed all the characters. So I made ​​this:

function w1250_to_utf8($text) {
    // map based on:
    // http://konfiguracja.c0.pl/iso02vscp1250en.html
    // http://konfiguracja.c0.pl/webpl/index_en.html#examp
    // http://www.htmlentities.com/html/entities/
    $map = array(
        chr(0x8A) => chr(0xA9),
        chr(0x8C) => chr(0xA6),
        chr(0x8D) => chr(0xAB),
        chr(0x8E) => chr(0xAE),
        chr(0x8F) => chr(0xAC),
        chr(0x9C) => chr(0xB6),
        chr(0x9D) => chr(0xBB),
        chr(0xA1) => chr(0xB7),
        chr(0xA5) => chr(0xA1),
        chr(0xBC) => chr(0xA5),
        chr(0x9F) => chr(0xBC),
        chr(0xB9) => chr(0xB1),
        chr(0x9A) => chr(0xB9),
        chr(0xBE) => chr(0xB5),
        chr(0x9E) => chr(0xBE),
        chr(0x80) => '€',
        chr(0x82) => '‚',
        chr(0x84) => '„',
        chr(0x85) => '…',
        chr(0x86) => '†',
        chr(0x87) => '‡',
        chr(0x89) => '‰',
        chr(0x8B) => '‹',
        chr(0x91) => '‘',
        chr(0x92) => '’',
        chr(0x93) => '“',
        chr(0x94) => '”',
        chr(0x95) => '•',
        chr(0x96) => '–',
        chr(0x97) => '—',
        chr(0x99) => '™',
        chr(0x9B) => '’',
        chr(0xA6) => '¦',
        chr(0xA9) => '©',
        chr(0xAB) => '«',
        chr(0xAE) => '®',
        chr(0xB1) => '±',
        chr(0xB5) => 'µ',
        chr(0xB6) => '¶',
        chr(0xB7) => '·',
        chr(0xBB) => '»',
    );
    return html_entity_decode(mb_convert_encoding(strtr($text, $map), 'UTF-8', 'ISO-8859-2'), ENT_QUOTES, 'UTF-8');
}

Hey guys. For everybody who's looking for a function that is converting an iso-string to utf8 or an utf8-string to iso, here's your solution:

public function encodeToUtf8($string) {
     return mb_convert_encoding($string, "UTF-8", mb_detect_encoding($string, "UTF-8, ISO-8859-1, ISO-8859-15", true));
}

public function encodeToIso($string) {
     return mb_convert_encoding($string, "ISO-8859-1", mb_detect_encoding($string, "UTF-8, ISO-8859-1, ISO-8859-15", true));
}

For me these functions are working fine. Give it a try

I've been trying to find the charset of a norwegian (with a lot of ø, æ, å) txt file written on a Mac, i've found it in this way:

<?php
$text = "A strange string to pass, maybe with some ø, æ, å characters.";

foreach(mb_list_encodings() as $chr){
        echo mb_convert_encoding($text, 'UTF-8', $chr)." : ".$chr."<br>";    
 } 
?>

The line that looks good, gives you the encoding it was written in.

Hope can help someone

If you have what looks like ISO-8859-1, but it includes "smart quotes" courtesy of Microsoft software, or people cutting and pasting content from Microsoft software, then what you're actually dealing with is probably Windows-1252. Try this:

<?php
$cleanText = mb_convert_encoding($text, 'UTF-8', 'Windows-1252');
?>

The annoying part is that the auto detection (ie: the mb_detect_encoding function) will often think Windows-1252 is ISO-8859-1. Close, but no cigar. This is critical if you're then trying to do unserialize on the resulting text, because the byte count of the string needs to be perfect.

Text-encoding HTML-ENTITIES will be deprecated as of PHP 8.2.

To convert all non-ASCII characters into entities (to produce pure 7-bit HTML output), I was using:

<?php
echo mb_convert_encoding( htmlspecialchars( $text, ENT_QUOTES, 'UTF-8' ), 'HTML-ENTITIES', 'UTF-8' );
?>

I can get the identical result with:

<?php
echo mb_encode_numericentity( htmlentities( $text, ENT_QUOTES, 'UTF-8' ), [0x80, 0x10FFFF, 0, ~0], 'UTF-8' );
?>

The output contains well-known named entities for some often used characters and numeric entities for the rest.

aaron, to discard unsupported characters instead of printing a ?, you might as well simply set the configuration directive:

mbstring.substitute_character = "none"

in your php.ini. Be sure to include the quotes around none. Or at run-time with

<?php
ini_set('mbstring.substitute_character', "none");
?>

many people below talk about using 
<?php
    mb_convert_encode($s,'HTML-ENTITIES','UTF-8');
?>
to convert non-ascii code into html-readable stuff.  Due to my webserver being out of my control, I was unable to set the database character set, and whenever PHP made a copy of my $s variable that it had pulled out of the database, it would convert it to nasty latin1 automatically and not leave it in it's beautiful UTF-8 glory.

So [insert korean characters here] turned into ?????.

I found myself needing to pass by reference (which of course is deprecated/nonexistent in recent versions of PHP)
so instead of
<?php
    mb_convert_encode(&$s,'HTML-ENTITIES','UTF-8');
?>
which worked perfectly until I upgraded, so I had to use
<?php
    call_user_func_array('mb_convert_encoding', array(&$s,'HTML-ENTITIES','UTF-8'));
?>

Hope it helps someone else out

My solution below was slightly incorrect, so here is the correct version (I posted at the end of a long day, never a good idea!)

Again, this is a quick and dirty solution to stop mb_convert_encoding from filling your string with question marks whenever it encounters an illegal character for the target encoding. 

<?php
function convert_to ( $source, $target_encoding )
    {
    // detect the character encoding of the incoming file
    $encoding = mb_detect_encoding( $source, "auto" );
       
    // escape all of the question marks so we can remove artifacts from
    // the unicode conversion process
    $target = str_replace( "?", "[question_mark]", $source );
       
    // convert the string to the target encoding
    $target = mb_convert_encoding( $target, $target_encoding, $encoding);
       
    // remove any question marks that have been introduced because of illegal characters
    $target = str_replace( "?", "", $target );
       
    // replace the token string "[question_mark]" with the symbol "?"
    $target = str_replace( "[question_mark]", "?", $target );
   
    return $target;
    }
?>

Hope this helps someone! (Admins should feel free to delete my previous, incorrect, post for clarity)
-A

If you are trying to generate a CSV (with extended chars) to be opened at Exel for Mac, the only that worked for me was:
<?php mb_convert_encoding( $CSV, 'Windows-1252', 'UTF-8'); ?>

I also tried this:

<?php
//Separado OK, chars MAL
iconv('MACINTOSH', 'UTF8', $CSV);
//Separado MAL, chars OK
chr(255).chr(254).mb_convert_encoding( $CSV, 'UCS-2LE', 'UTF-8');
?>

But the first one didn't show extended chars correctly, and the second one, did't separe fields correctly

To add to the Flash conversion comment below, here's how I convert back from what I've stored in a database after converting from Flash HTML text field output, in order to load it back into a Flash HTML text field:

function htmltoflash($htmlstr)
{
  return str_replace("<br />","\n",
    str_replace("<","&lt;",
      str_replace(">","&gt;",
        mb_convert_encoding(html_entity_decode($htmlstr),
        "UTF-8","ISO-8859-1"))));
}

Note that `mb_convert_encoding($val, 'HTML-ENTITIES')` does not escape '\'', '"', '<', '>', or '&'.

It appears that when dealing with an unknown "from encoding" the function will both throw an E_WARNING and proceed to convert the string from ISO-8859-1 to the "to encoding".

When you need to convert from HTML-ENTITIES, but your UTF-8 string is partially broken (not all chars in UTF-8) - in this case passing string to mb_convert_encoding($string, 'UTF-8', 'HTML-ENTITIES'); - corrupts chars in string even more. In this case you need to replace html entities gradually to preserve character good encoding. I wrote such closure for this job :
<?php
$decode_entities = function($string) {
        preg_match_all("/&#?\w+;/", $string, $entities, PREG_SET_ORDER);
        $entities = array_unique(array_column($entities, 0));
        foreach ($entities as $entity) {
            $decoded = mb_convert_encoding($entity, 'UTF-8', 'HTML-ENTITIES');
            $string = str_replace($entity, $decoded, $string);
        }
        return $string;
    };
?>

Clean a string for use as filename by simply replacing all unwanted characters with underscore (ASCII converts to 7bit). It removes slightly more chars than necessary. Hope its useful. 

$fileName = 'Test:!"$%&/()=ÖÄÜöäü<<';
echo strtr(mb_convert_encoding($fileName,'ASCII'), 
    ' ,;:?*#!§$%&/(){}<>=`´|\\\'"', 
    '____________________________');

/**
 * Convert Windows-1250 to UTF-8
 * Based on https://www.php.net/manual/en/function.mb-convert-encoding.php#112547
 */
class TextConverter
{
    private const ENCODING_TO = 'UTF-8';
    private const ENCODING_FROM = 'ISO-8859-2';

    private array $mapChrChr = [
        0x8A => 0xA9,
        0x8C => 0xA6,
        0x8D => 0xAB,
        0x8E => 0xAE,
        0x8F => 0xAC,
        0x9C => 0xB6,
        0x9D => 0xBB,
        0xA1 => 0xB7,
        0xA5 => 0xA1,
        0xBC => 0xA5,
        0x9F => 0xBC,
        0xB9 => 0xB1,
        0x9A => 0xB9,
        0xBE => 0xB5,
        0x9E => 0xBE
    ];

    private array $mapChrString = [
        0x80 => '€',
        0x82 => '‚',
        0x84 => '„',
        0x85 => '…',
        0x86 => '†',
        0x87 => '‡',
        0x89 => '‰',
        0x8B => '‹',
        0x91 => '‘',
        0x92 => '’',
        0x93 => '“',
        0x94 => '”',
        0x95 => '•',
        0x96 => '–',
        0x97 => '—',
        0x99 => '™',
        0x9B => '’',
        0xA6 => '¦',
        0xA9 => '©',
        0xAB => '«',
        0xAE => '®',
        0xB1 => '±',
        0xB5 => 'µ',
        0xB6 => '¶',
        0xB7 => '·',
        0xBB => '»'
    ];

    /**
     * @param $text
     * @return string
     */
    public function execute($text): string
    {
        $map = $this->prepareMap();

        return html_entity_decode(
            mb_convert_encoding(strtr($text, $map), self::ENCODING_TO, self::ENCODING_FROM),
            ENT_QUOTES,
            self::ENCODING_TO
        );
    }

    /**
     * @return array
     */
    private function prepareMap(): array
    {
        $maps[] = $this->arrayMapAssoc(function ($k, $v) {
            return [chr($k), chr($v)];
        }, $this->mapChrChr);

        $maps[] = $this->arrayMapAssoc(function ($k, $v) {
            return [chr($k), $v];
        }, $this->mapChrString);

        return array_merge([], ...$maps);
    }

    /**
     * @param callable $function
     * @param array $array
     * @return array
     */
    private function arrayMapAssoc(callable $function, array $array): array
    {
        return array_column(
            array_map(
                $function,
                array_keys($array),
                $array
            ),
            1,
            0
        );
    }
}

instead of ini_set(), you can try this

mb_substitute_character("none");

Another sample of recoding without MultiByte enabling.
(Russian koi->win, if input in win-encoding already, function recode() returns unchanged string)

<?php
  // 0 - win
  // 1 - koi
  function detect_encoding($str) {
    $win = 0;
    $koi = 0;

    for($i=0; $i<strlen($str); $i++) {
      if( ord($str[$i]) >224 && ord($str[$i]) < 255) $win++;
      if( ord($str[$i]) >192 && ord($str[$i]) < 223) $koi++;
    }

    if( $win < $koi ) {
      return 1;
    } else return 0;

  }

  // recodes koi to win
  function koi_to_win($string) {

    $kw = array(128, 129, 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183,  184, 185, 186, 187, 188, 189, 190, 191, 254, 224, 225, 246, 228, 229, 244, 227, 245, 232, 233, 234, 235, 236, 237, 238, 239, 255, 240, 241, 242, 243, 230, 226, 252, 251, 231, 248, 253, 249, 247, 250, 222, 192, 193, 214, 196, 197, 212, 195, 213, 200, 201, 202, 203, 204, 205, 206, 207, 223, 208, 209, 210, 211, 198, 194, 220, 219, 199, 216, 221, 217, 215, 218);
    $wk = array(128, 129, 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183,  184, 185, 186, 187, 188, 189, 190, 191, 225, 226, 247, 231, 228, 229, 246, 250, 233, 234, 235, 236, 237, 238, 239, 240, 242,  243, 244, 245, 230, 232, 227, 254, 251, 253, 255, 249, 248, 252, 224, 241, 193, 194, 215, 199, 196, 197, 214, 218, 201, 202, 203, 204, 205, 206, 207, 208, 210, 211, 212, 213, 198, 200, 195, 222, 219, 221, 223, 217, 216, 220, 192, 209);

    $end = strlen($string);
    $pos = 0;
    do {
      $c = ord($string[$pos]);
      if ($c>128) {
        $string[$pos] = chr($kw[$c-128]);
      }

    } while (++$pos < $end);

    return $string;
  }

  function recode($str) {

    $enc = detect_encoding($str);
    if ($enc==1) {
      $str = koi_to_win($str);
    }

    return $str;
  }
?>

// convert UTF8 to DOS = CP850 
//
// $utf8_text=UTF8-Formatted text;
// $dos=CP850-Formatted text;

// have fun

$dos = mb_convert_encoding($utf8_text, "CP850", mb_detect_encoding($utf8_text, "UTF-8, CP850, ISO-8859-15", true));

Why did you use the php html encode functions? mbstring has it's own Encoding which is (as far as I tested it) much more usefull:

HTML-ENTITIES

Example:

$text = mb_convert_encoding($text, 'HTML-ENTITIES', "UTF-8");

If you are attempting to convert "UTF-8" text to "ISO-8859-1" and the result is always returning in "ASCII", place the following line of code before the mb_convert_encoding:

mb_detect_order(array('UTF-8', 'ISO-8859-1'));

It is necessary to force a specific search order for the conversion to work

For those wanting to convert from $set to MacRoman, use iconv():

<?php

$string = iconv('UTF-8', 'macintosh', $string);

?>

('macintosh' is the IANA name for the MacRoman character set.)

As an alternative to Johannes's suggestion for converting strings from other character sets to a 7bit representation while not just deleting latin diacritics, you might try this:

<?php
$text = iconv($from_enc, 'US-ASCII//TRANSLIT', $text);
?>

The only disadvantage is that it does not convert "ä" to "ae", but it handles punctuation and other special characters better.
-- 
David

When converting Japanese strings to ISO-2022-JP or JIS on PHP >= 5.2.1, you can use "ISO-2022-JP-MS" instead of them.
Kishu-Izon (platform dependent) characters are converted correctly with the encoding, as same as with eucJP-win or with SJIS-win.

be careful when converting from iso-8859-1 to utf-8.

even if you explicitly specify the character encoding of a page as iso-8859-1(via headers and strict xml defs), windows 2000 will ignore that and interpret it as whatever character set it has natively installed. 

for example, i wrote char #128 into a page, with char encoding iso-8859-1, and it displayed in internet explorer (& mozilla) as a euro symbol.

it should have displayed a box, denoting that char #128 is undefined in iso-8859-1. The problem was it was displaying in "Windows: western europe" (my native character set).

this led to confusion when i tried to convert this euro to UTF-8 via mb_convert_encoding()  

IE displays UTF-8 correctly- and because PHP correctly converted #128 into a box in UTF-8, IE would show a box.

so all i saw was mb_convert_encoding() converting a euro symbol into a box. It took me a long time to figure out what was going on.

rodrigo at bb2 dot co dot jp wrote that inconv works better than mb_convert_encoding, I find that when converting from uft8 to shift_jis 
$conv_str = mb_convert_encoding($str,$toCS,$fromCS); 
works while
$conv_str = iconv($fromCS,$toCS.'//IGNORE',$str); 
removes tildes from $str.

For those who can´t use mb_convert_encoding() to convert from one charset to another as a metter of lower version of php, try iconv().

I had this problem converting to japanese charset:

$txt=mb_convert_encoding($txt,'SJIS',$this->encode);

And I could fix it by using this:

$txt = iconv('UTF-8', 'SJIS', $txt);

Maybe it´s helpfull for someone else! ;)

If you want to convert japanese to ISO-2022-JP it is highly recommended to use ISO-2022-JP-MS as the target encoding instead. This includes the extended character set and avoids ? in the text. For example the often used "1 in a circle" ① will be correctly converted then.

Here's a tip for anyone using Flash and PHP for storing HTML output submitted from a Flash text field in a database or whatever.

Flash submits its HTML special characters in UTF-8, so you can use the following function to convert those into HTML entity characters:

function utf8html($utf8str)
{
  return htmlentities(mb_convert_encoding($utf8str,"ISO-8859-1","UTF-8"));
}

I\'d like to share some code to convert latin diacritics to their
traditional 7bit representation, like, for example,

- à,ç,é,î,... to a,c,e,i,...
- ß to ss
- ä,Ä,... to ae,Ae,...
- ë,... to e,...

(mb_convert \"7bit\" would simply delete any offending characters). 

I might have missed on your country\'s typographic 
conventions--correct me then. 
<?php
/**
 * @args string $text line of encoded text
 *       string $from_enc (encoding type of $text, e.g. UTF-8, ISO-8859-1)
 *
 * @returns 7bit representation
 */
function to7bit($text,$from_enc) {
    $text = mb_convert_encoding($text,\'HTML-ENTITIES\',$from_enc);
    $text = preg_replace(
        array(\'/ß/\',\'/&(..)lig;/\',
             \'/&([aouAOU])uml;/\',\'/&(.)[^;]*;/\'),
        array(\'ss\',\"$1\",\"$1\".\'e\',\"$1\"),
        $text);
    return $text;
}   
?>

Enjoy :-)
Johannes

==
[EDIT BY danbrown AT php DOT net: Author provided the following update on 27-FEB-2012.]
==

An addendum to my "to7bit" function referenced below in the notes. 
The function is supposed to solve the problem that some languages require a different 7bit rendering of special (umlauted) characters for sorting or other applications. For example, the German ß ligature is usually written "ss" in 7bit context. Dutch ÿ is typically rendered "ij" (not "y"). 

The original function works well with word (alphabet) character entities and I've seen it used in many places. But non-word entities cause funny results:
E.g., "©" is rendered as "c", "­" as "s" and "&rquo;" as "r". 
The following version fixes this by converting non-alphanumeric characters (also chains thereof) to '_'.

<?php
/**
 * @args string $text line of encoded text
 *       string $from_enc (encoding type of $text, e.g. UTF-8, ISO-8859-1)
 *
 * @returns 7bit representation
 */
function to7bit($text,$from_enc) {
    $text = preg_replace(/W+/,'_',$text);
    $text = mb_convert_encoding($text,'HTML-ENTITIES',$from_enc);
    $text = preg_replace(
        array('/ß/','/&(..)lig;/',
             '/&([aouAOU])uml;/','/ÿ/','/&(.)[^;]*;/'),
        array('ss',"$1","$1".'e','ij',"$1"),
        $text);
    return $text;
}  
?>

Enjoy again,
Johannes