mb_convert_encoding breaks well-formed character (PHP 5.5RC3, Mac OSX 10.8).

01_mb_convert_encoding_breaks_well-formed_character.php

<?php
// https://en.wikipedia.org/wiki/UTF-8#Examples
// 2-byte character: U+00A2 (CENT SIGN)

mb_substitute_character(0xFFFD);

$data = [
                          // ill-formed
    "\xC2\xA2"."\xC2\xA2"."\xC2",
                         // ill-formed
    "\xC2\xA2"."\xC2\xA2".     "\xA2",
];

$expected  = [
                         // U+FFFD   
    "\xC2\xA2"."\xC2\xA2"."\xEF\xBF\xBD",
                         // U+FFFD
    "\xC2\xA2"."\xC2\xA2"."\xEF\xBF\xBD",
];
$expected2 = [
     "\xC2\xA2"."\xC2\xA2",
                           // U+FFFD
     "\xC2\xA2"."\xC2\xA2"."\xEF\xBF\xBD",
];

var_dump(
'2-byte character: U+00A2 (CENT SIGN)',
[
    $expected === array_map(function($str) { return UConverter::transcode($str, 'UTF-8', 'UTF-8'); }, $data),
    $expected === array_map(function($str) { return htmlspecialchars($str, ENT_SUBSTITUTE, 'UTF-8'); }, $data),
    $expected2 === array_map(function($str) { return mb_convert_encoding($str, 'UTF-8', 'UTF-8'); }, $data)
],[
    // mb_convert_encoding deletes trailing byte without replacing U+FFFD
    $expected[0] !== $expected2[0],
    $expected[1] === $expected2[1]
]);

02_mb_convert_encoding_breaks_well-formed_character.php

<?php
// https://en.wikipedia.org/wiki/UTF-8#Examples
// 3-byte character: U+20AC (EURO SIGN)

mb_substitute_character(0xFFFD);

$data2 = [
    // ill-formed
    "\xE2\x82"    ."\xE2\x82\xAC"."\xE2\x82\xAC",
    // ill-formed
        "\x82\xAC"."\xE2\x82\xAC"."\xE2\x82\xAC",
                                  // ill-formed
    "\xE2\x82\xAC"."\xE2\x82\xAC"."\xE2\x82",
                                  // ill-formed
    "\xE2\x82\xAC"."\xE2\x82\xAC".    "\x82\xAC",
];

$expected3 = [
    // U+FFFD
    "\xEF\xBF\xBD"."\xE2\x82\xAC"."\xE2\x82\xAC",
    // U+FFFD       U+FFFD
    "\xEF\xBF\xBD"."\xEF\xBF\xBD"."\xE2\x82\xAC"."\xE2\x82\xAC",
                                  // U+FFFD
    "\xE2\x82\xAC"."\xE2\x82\xAC"."\xEF\xBF\xBD",
                                  //  U+FFFD        U+FFFD
    "\xE2\x82\xAC"."\xE2\x82\xAC"."\xEF\xBF\xBD"."\xEF\xBF\xBD"
];

$expected4 = [
    // U+FFFD      U+FFFD         U+FFFD
    "\xEF\xBF\xBD"."\xEF\xBF\xBD"."\xEF\xBF\xBD"."\xE2\x82\xAC",
    // U+FFFD
    "\xEF\xBF\xBD"."\xEF\xBF\xBD"."\xE2\x82\xAC"."\xE2\x82\xAC",
    
    "\xE2\x82\xAC"."\xE2\x82\xAC",
                                   // U+FFFD     U+FFFD
    "\xE2\x82\xAC"."\xE2\x82\xAC"."\xEF\xBF\xBD"."\xEF\xBF\xBD"
];

var_dump(
'3-byte character: U+20AC (EURO SIGN)',
[
    $expected3 === array_map(function($str) { return UConverter::transcode($str, 'UTF-8', 'UTF-8'); }, $data2),
    $expected3 === array_map(function($str) { return htmlspecialchars($str, ENT_SUBSTITUTE, 'UTF-8'); }, $data2),
    $expected4 === array_map(function($str) { return mb_convert_encoding($str, 'UTF-8', 'UTF-8'); }, $data2)
],[
    // mb_convert_encoding breaks well-formed character
    $expected3[0] !== $expected4[0],
    $expected3[1] === $expected4[1],
    // mb_convert_encoding deletes trailing bytes without replacing U+FFFD
    $expected3[2] !== $expected4[2],
    $expected3[3] === $expected4[3]
]);

03_mb_convert_encoding_breaks_well-formed_character.php

<?php
// https://en.wikipedia.org/wiki/UTF-8#Examples
// 4-byte character: U+24B62 (Unicode Han Character)

mb_substitute_character(0xFFFD);

$data3 = [
    // ill-formed
    "\xF0\xA4\xAD"    ."\xF0\xA4\xAD\xA2"."\xF0\xA4\xAD\xA2",
    // ill-formed
        "\xA4\xAD\xA2"."\xF0\xA4\xAD\xA2"."\xF0\xA4\xAD\xA2",
                                          // ill-formed
    "\xF0\xA4\xAD\xA2"."\xF0\xA4\xAD\xA2"."\xF0\xA4\xAD",
                                          // ill-formed
    "\xF0\xA4\xAD\xA2"."\xF0\xA4\xAD\xA2".    "\xA4\xAD\xA2",
];

$expected5 = [
    // U+FFFD
    "\xEF\xBF\xBD"."\xF0\xA4\xAD\xA2"."\xF0\xA4\xAD\xA2",
    // U+FFFD      U+FFFD         U+FFFD
    "\xEF\xBF\xBD"."\xEF\xBF\xBD"."\xEF\xBF\xBD"."\xF0\xA4\xAD\xA2"."\xF0\xA4\xAD\xA2",
                                        // U+FFFD
    "\xF0\xA4\xAD\xA2"."\xF0\xA4\xAD\xA2"."\xEF\xBF\xBD",
                                        // U+FFFD        U+FFFD          U+FFFD
    "\xF0\xA4\xAD\xA2"."\xF0\xA4\xAD\xA2"."\xEF\xBF\xBD"."\xEF\xBF\xBD"."\xEF\xBF\xBD"
];

$expected6 = [
    // U+FFFD      U+FFFD         U+FFFD         U+FFFD
    "\xEF\xBF\xBD"."\xEF\xBF\xBD"."\xEF\xBF\xBD"."\xEF\xBF\xBD"."\xF0\xA4\xAD\xA2",
    // U+FFFD      U+FFFD         U+FFFD
    "\xEF\xBF\xBD"."\xEF\xBF\xBD"."\xEF\xBF\xBD"."\xF0\xA4\xAD\xA2"."\xF0\xA4\xAD\xA2",

    "\xF0\xA4\xAD\xA2"."\xF0\xA4\xAD\xA2",
                                           // U+FFFD      U+FFFD        U+FFFD
    "\xF0\xA4\xAD\xA2"."\xF0\xA4\xAD\xA2"."\xEF\xBF\xBD"."\xEF\xBF\xBD"."\xEF\xBF\xBD"
];

var_dump(
'4-byte character: U+24B62 (Unicode Han Character)',
[
    $expected5 === array_map(function($str) { return UConverter::transcode($str, 'UTF-8', 'UTF-8'); }, $data3),
    $expected5 === array_map(function($str) { return htmlspecialchars($str, ENT_SUBSTITUTE, 'UTF-8'); }, $data3),
    $expected6 === array_map(function($str) { return mb_convert_encoding($str, 'UTF-8', 'UTF-8'); }, $data3)
],[
    // mb_convert_encoding breaks well-formed character
    $expected5[0] !== $expected6[0],
    $expected5[1] === $expected6[1],
    // mb_convert_encoding deletes trailing bytes without replacing U+FFFD
    $expected5[2] !== $expected6[2],
    $expected5[3] === $expected6[3]
]);