remove autoload section in composer.json

[github/wallabag/wallabag.git] / inc / 3rdparty / htmlpurifier / HTMLPurifier / Encoder.php

<?php\r
\r
/**\r
 * A UTF-8 specific character encoder that handles cleaning and transforming.\r
 * @note All functions in this class should be static.\r
 */\r
class HTMLPurifier_Encoder\r
{\r
\r
    /**\r
     * Constructor throws fatal error if you attempt to instantiate class\r
     */\r
    private function __construct()\r
    {\r
        trigger_error('Cannot instantiate encoder, call methods statically', E_USER_ERROR);\r
    }\r
\r
    /**\r
     * Error-handler that mutes errors, alternative to shut-up operator.\r
     */\r
    public static function muteErrorHandler()\r
    {\r
    }\r
\r
    /**\r
     * iconv wrapper which mutes errors, but doesn't work around bugs.\r
     * @param string $in Input encoding\r
     * @param string $out Output encoding\r
     * @param string $text The text to convert\r
     * @return string\r
     */\r
    public static function unsafeIconv($in, $out, $text)\r
    {\r
        set_error_handler(array('HTMLPurifier_Encoder', 'muteErrorHandler'));\r
        $r = iconv($in, $out, $text);\r
        restore_error_handler();\r
        return $r;\r
    }\r
\r
    /**\r
     * iconv wrapper which mutes errors and works around bugs.\r
     * @param string $in Input encoding\r
     * @param string $out Output encoding\r
     * @param string $text The text to convert\r
     * @param int $max_chunk_size\r
     * @return string\r
     */\r
    public static function iconv($in, $out, $text, $max_chunk_size = 8000)\r
    {\r
        $code = self::testIconvTruncateBug();\r
        if ($code == self::ICONV_OK) {\r
            return self::unsafeIconv($in, $out, $text);\r
        } elseif ($code == self::ICONV_TRUNCATES) {\r
            // we can only work around this if the input character set\r
            // is utf-8\r
            if ($in == 'utf-8') {\r
                if ($max_chunk_size < 4) {\r
                    trigger_error('max_chunk_size is too small', E_USER_WARNING);\r
                    return false;\r
                }\r
                // split into 8000 byte chunks, but be careful to handle\r
                // multibyte boundaries properly\r
                if (($c = strlen($text)) <= $max_chunk_size) {\r
                    return self::unsafeIconv($in, $out, $text);\r
                }\r
                $r = '';\r
                $i = 0;\r
                while (true) {\r
                    if ($i + $max_chunk_size >= $c) {\r
                        $r .= self::unsafeIconv($in, $out, substr($text, $i));\r
                        break;\r
                    }\r
                    // wibble the boundary\r
                    if (0x80 != (0xC0 & ord($text[$i + $max_chunk_size]))) {\r
                        $chunk_size = $max_chunk_size;\r
                    } elseif (0x80 != (0xC0 & ord($text[$i + $max_chunk_size - 1]))) {\r
                        $chunk_size = $max_chunk_size - 1;\r
                    } elseif (0x80 != (0xC0 & ord($text[$i + $max_chunk_size - 2]))) {\r
                        $chunk_size = $max_chunk_size - 2;\r
                    } elseif (0x80 != (0xC0 & ord($text[$i + $max_chunk_size - 3]))) {\r
                        $chunk_size = $max_chunk_size - 3;\r
                    } else {\r
                        return false; // rather confusing UTF-8...\r
                    }\r
                    $chunk = substr($text, $i, $chunk_size); // substr doesn't mind overlong lengths\r
                    $r .= self::unsafeIconv($in, $out, $chunk);\r
                    $i += $chunk_size;\r
                }\r
                return $r;\r
            } else {\r
                return false;\r
            }\r
        } else {\r
            return false;\r
        }\r
    }\r
\r
    /**\r
     * Cleans a UTF-8 string for well-formedness and SGML validity\r
     *\r
     * It will parse according to UTF-8 and return a valid UTF8 string, with\r
     * non-SGML codepoints excluded.\r
     *\r
     * @param string $str The string to clean\r
     * @param bool $force_php\r
     * @return string\r
     *\r
     * @note Just for reference, the non-SGML code points are 0 to 31 and\r
     *       127 to 159, inclusive.  However, we allow code points 9, 10\r
     *       and 13, which are the tab, line feed and carriage return\r
     *       respectively. 128 and above the code points map to multibyte\r
     *       UTF-8 representations.\r
     *\r
     * @note Fallback code adapted from utf8ToUnicode by Henri Sivonen and\r
     *       hsivonen@iki.fi at <http://iki.fi/hsivonen/php-utf8/> under the\r
     *       LGPL license.  Notes on what changed are inside, but in general,\r
     *       the original code transformed UTF-8 text into an array of integer\r
     *       Unicode codepoints. Understandably, transforming that back to\r
     *       a string would be somewhat expensive, so the function was modded to\r
     *       directly operate on the string.  However, this discourages code\r
     *       reuse, and the logic enumerated here would be useful for any\r
     *       function that needs to be able to understand UTF-8 characters.\r
     *       As of right now, only smart lossless character encoding converters\r
     *       would need that, and I'm probably not going to implement them.\r
     *       Once again, PHP 6 should solve all our problems.\r
     */\r
    public static function cleanUTF8($str, $force_php = false)\r
    {\r
        // UTF-8 validity is checked since PHP 4.3.5\r
        // This is an optimization: if the string is already valid UTF-8, no\r
        // need to do PHP stuff. 99% of the time, this will be the case.\r
        // The regexp matches the XML char production, as well as well as excluding\r
        // non-SGML codepoints U+007F to U+009F\r
        if (preg_match(\r
            '/^[\x{9}\x{A}\x{D}\x{20}-\x{7E}\x{A0}-\x{D7FF}\x{E000}-\x{FFFD}\x{10000}-\x{10FFFF}]*$/Du',\r
            $str\r
        )) {\r
            return $str;\r
        }\r
\r
        $mState = 0; // cached expected number of octets after the current octet\r
                     // until the beginning of the next UTF8 character sequence\r
        $mUcs4  = 0; // cached Unicode character\r
        $mBytes = 1; // cached expected number of octets in the current sequence\r
\r
        // original code involved an $out that was an array of Unicode\r
        // codepoints.  Instead of having to convert back into UTF-8, we've\r
        // decided to directly append valid UTF-8 characters onto a string\r
        // $out once they're done.  $char accumulates raw bytes, while $mUcs4\r
        // turns into the Unicode code point, so there's some redundancy.\r
\r
        $out = '';\r
        $char = '';\r
\r
        $len = strlen($str);\r
        for ($i = 0; $i < $len; $i++) {\r
            $in = ord($str{$i});\r
            $char .= $str[$i]; // append byte to char\r
            if (0 == $mState) {\r
                // When mState is zero we expect either a US-ASCII character\r
                // or a multi-octet sequence.\r
                if (0 == (0x80 & ($in))) {\r
                    // US-ASCII, pass straight through.\r
                    if (($in <= 31 || $in == 127) &&\r
                        !($in == 9 || $in == 13 || $in == 10) // save \r\t\n\r
                    ) {\r
                        // control characters, remove\r
                    } else {\r
                        $out .= $char;\r
                    }\r
                    // reset\r
                    $char = '';\r
                    $mBytes = 1;\r
                } elseif (0xC0 == (0xE0 & ($in))) {\r
                    // First octet of 2 octet sequence\r
                    $mUcs4 = ($in);\r
                    $mUcs4 = ($mUcs4 & 0x1F) << 6;\r
                    $mState = 1;\r
                    $mBytes = 2;\r
                } elseif (0xE0 == (0xF0 & ($in))) {\r
                    // First octet of 3 octet sequence\r
                    $mUcs4 = ($in);\r
                    $mUcs4 = ($mUcs4 & 0x0F) << 12;\r
                    $mState = 2;\r
                    $mBytes = 3;\r
                } elseif (0xF0 == (0xF8 & ($in))) {\r
                    // First octet of 4 octet sequence\r
                    $mUcs4 = ($in);\r
                    $mUcs4 = ($mUcs4 & 0x07) << 18;\r
                    $mState = 3;\r
                    $mBytes = 4;\r
                } elseif (0xF8 == (0xFC & ($in))) {\r
                    // First octet of 5 octet sequence.\r
                    //\r
                    // This is illegal because the encoded codepoint must be\r
                    // either:\r
                    // (a) not the shortest form or\r
                    // (b) outside the Unicode range of 0-0x10FFFF.\r
                    // Rather than trying to resynchronize, we will carry on\r
                    // until the end of the sequence and let the later error\r
                    // handling code catch it.\r
                    $mUcs4 = ($in);\r
                    $mUcs4 = ($mUcs4 & 0x03) << 24;\r
                    $mState = 4;\r
                    $mBytes = 5;\r
                } elseif (0xFC == (0xFE & ($in))) {\r
                    // First octet of 6 octet sequence, see comments for 5\r
                    // octet sequence.\r
                    $mUcs4 = ($in);\r
                    $mUcs4 = ($mUcs4 & 1) << 30;\r
                    $mState = 5;\r
                    $mBytes = 6;\r
                } else {\r
                    // Current octet is neither in the US-ASCII range nor a\r
                    // legal first octet of a multi-octet sequence.\r
                    $mState = 0;\r
                    $mUcs4  = 0;\r
                    $mBytes = 1;\r
                    $char = '';\r
                }\r
            } else {\r
                // When mState is non-zero, we expect a continuation of the\r
                // multi-octet sequence\r
                if (0x80 == (0xC0 & ($in))) {\r
                    // Legal continuation.\r
                    $shift = ($mState - 1) * 6;\r
                    $tmp = $in;\r
                    $tmp = ($tmp & 0x0000003F) << $shift;\r
                    $mUcs4 |= $tmp;\r
\r
                    if (0 == --$mState) {\r
                        // End of the multi-octet sequence. mUcs4 now contains\r
                        // the final Unicode codepoint to be output\r
\r
                        // Check for illegal sequences and codepoints.\r
\r
                        // From Unicode 3.1, non-shortest form is illegal\r
                        if (((2 == $mBytes) && ($mUcs4 < 0x0080)) ||\r
                            ((3 == $mBytes) && ($mUcs4 < 0x0800)) ||\r
                            ((4 == $mBytes) && ($mUcs4 < 0x10000)) ||\r
                            (4 < $mBytes) ||\r
                            // From Unicode 3.2, surrogate characters = illegal\r
                            (($mUcs4 & 0xFFFFF800) == 0xD800) ||\r
                            // Codepoints outside the Unicode range are illegal\r
                            ($mUcs4 > 0x10FFFF)\r
                        ) {\r
\r
                        } elseif (0xFEFF != $mUcs4 && // omit BOM\r
                            // check for valid Char unicode codepoints\r
                            (\r
                                0x9 == $mUcs4 ||\r
                                0xA == $mUcs4 ||\r
                                0xD == $mUcs4 ||\r
                                (0x20 <= $mUcs4 && 0x7E >= $mUcs4) ||\r
                                // 7F-9F is not strictly prohibited by XML,\r
                                // but it is non-SGML, and thus we don't allow it\r
                                (0xA0 <= $mUcs4 && 0xD7FF >= $mUcs4) ||\r
                                (0x10000 <= $mUcs4 && 0x10FFFF >= $mUcs4)\r
                            )\r
                        ) {\r
                            $out .= $char;\r
                        }\r
                        // initialize UTF8 cache (reset)\r
                        $mState = 0;\r
                        $mUcs4  = 0;\r
                        $mBytes = 1;\r
                        $char = '';\r
                    }\r
                } else {\r
                    // ((0xC0 & (*in) != 0x80) && (mState != 0))\r
                    // Incomplete multi-octet sequence.\r
                    // used to result in complete fail, but we'll reset\r
                    $mState = 0;\r
                    $mUcs4  = 0;\r
                    $mBytes = 1;\r
                    $char ='';\r
                }\r
            }\r
        }\r
        return $out;\r
    }\r
\r
    /**\r
     * Translates a Unicode codepoint into its corresponding UTF-8 character.\r
     * @note Based on Feyd's function at\r
     *       <http://forums.devnetwork.net/viewtopic.php?p=191404#191404>,\r
     *       which is in public domain.\r
     * @note While we're going to do code point parsing anyway, a good\r
     *       optimization would be to refuse to translate code points that\r
     *       are non-SGML characters.  However, this could lead to duplication.\r
     * @note This is very similar to the unichr function in\r
     *       maintenance/generate-entity-file.php (although this is superior,\r
     *       due to its sanity checks).\r
     */\r
\r
    // +----------+----------+----------+----------+\r
    // | 33222222 | 22221111 | 111111   |          |\r
    // | 10987654 | 32109876 | 54321098 | 76543210 | bit\r
    // +----------+----------+----------+----------+\r
    // |          |          |          | 0xxxxxxx | 1 byte 0x00000000..0x0000007F\r
    // |          |          | 110yyyyy | 10xxxxxx | 2 byte 0x00000080..0x000007FF\r
    // |          | 1110zzzz | 10yyyyyy | 10xxxxxx | 3 byte 0x00000800..0x0000FFFF\r
    // | 11110www | 10wwzzzz | 10yyyyyy | 10xxxxxx | 4 byte 0x00010000..0x0010FFFF\r
    // +----------+----------+----------+----------+\r
    // | 00000000 | 00011111 | 11111111 | 11111111 | Theoretical upper limit of legal scalars: 2097151 (0x001FFFFF)\r
    // | 00000000 | 00010000 | 11111111 | 11111111 | Defined upper limit of legal scalar codes\r
    // +----------+----------+----------+----------+\r
\r
    public static function unichr($code)\r
    {\r
        if ($code > 1114111 or $code < 0 or\r
          ($code >= 55296 and $code <= 57343) ) {\r
            // bits are set outside the "valid" range as defined\r
            // by UNICODE 4.1.0\r
            return '';\r
        }\r
\r
        $x = $y = $z = $w = 0;\r
        if ($code < 128) {\r
            // regular ASCII character\r
            $x = $code;\r
        } else {\r
            // set up bits for UTF-8\r
            $x = ($code & 63) | 128;\r
            if ($code < 2048) {\r
                $y = (($code & 2047) >> 6) | 192;\r
            } else {\r
                $y = (($code & 4032) >> 6) | 128;\r
                if ($code < 65536) {\r
                    $z = (($code >> 12) & 15) | 224;\r
                } else {\r
                    $z = (($code >> 12) & 63) | 128;\r
                    $w = (($code >> 18) & 7)  | 240;\r
                }\r
            }\r
        }\r
        // set up the actual character\r
        $ret = '';\r
        if ($w) {\r
            $ret .= chr($w);\r
        }\r
        if ($z) {\r
            $ret .= chr($z);\r
        }\r
        if ($y) {\r
            $ret .= chr($y);\r
        }\r
        $ret .= chr($x);\r
\r
        return $ret;\r
    }\r
\r
    /**\r
     * @return bool\r
     */\r
    public static function iconvAvailable()\r
    {\r
        static $iconv = null;\r
        if ($iconv === null) {\r
            $iconv = function_exists('iconv') && self::testIconvTruncateBug() != self::ICONV_UNUSABLE;\r
        }\r
        return $iconv;\r
    }\r
\r
    /**\r
     * Convert a string to UTF-8 based on configuration.\r
     * @param string $str The string to convert\r
     * @param HTMLPurifier_Config $config\r
     * @param HTMLPurifier_Context $context\r
     * @return string\r
     */\r
    public static function convertToUTF8($str, $config, $context)\r
    {\r
        $encoding = $config->get('Core.Encoding');\r
        if ($encoding === 'utf-8') {\r
            return $str;\r
        }\r
        static $iconv = null;\r
        if ($iconv === null) {\r
            $iconv = self::iconvAvailable();\r
        }\r
        if ($iconv && !$config->get('Test.ForceNoIconv')) {\r
            // unaffected by bugs, since UTF-8 support all characters\r
            $str = self::unsafeIconv($encoding, 'utf-8//IGNORE', $str);\r
            if ($str === false) {\r
                // $encoding is not a valid encoding\r
                trigger_error('Invalid encoding ' . $encoding, E_USER_ERROR);\r
                return '';\r
            }\r
            // If the string is bjorked by Shift_JIS or a similar encoding\r
            // that doesn't support all of ASCII, convert the naughty\r
            // characters to their true byte-wise ASCII/UTF-8 equivalents.\r
            $str = strtr($str, self::testEncodingSupportsASCII($encoding));\r
            return $str;\r
        } elseif ($encoding === 'iso-8859-1') {\r
            $str = utf8_encode($str);\r
            return $str;\r
        }\r
        $bug = HTMLPurifier_Encoder::testIconvTruncateBug();\r
        if ($bug == self::ICONV_OK) {\r
            trigger_error('Encoding not supported, please install iconv', E_USER_ERROR);\r
        } else {\r
            trigger_error(\r
                'You have a buggy version of iconv, see https://bugs.php.net/bug.php?id=48147 ' .\r
                'and http://sourceware.org/bugzilla/show_bug.cgi?id=13541',\r
                E_USER_ERROR\r
            );\r
        }\r
    }\r
\r
    /**\r
     * Converts a string from UTF-8 based on configuration.\r
     * @param string $str The string to convert\r
     * @param HTMLPurifier_Config $config\r
     * @param HTMLPurifier_Context $context\r
     * @return string\r
     * @note Currently, this is a lossy conversion, with unexpressable\r
     *       characters being omitted.\r
     */\r
    public static function convertFromUTF8($str, $config, $context)\r
    {\r
        $encoding = $config->get('Core.Encoding');\r
        if ($escape = $config->get('Core.EscapeNonASCIICharacters')) {\r
            $str = self::convertToASCIIDumbLossless($str);\r
        }\r
        if ($encoding === 'utf-8') {\r
            return $str;\r
        }\r
        static $iconv = null;\r
        if ($iconv === null) {\r
            $iconv = self::iconvAvailable();\r
        }\r
        if ($iconv && !$config->get('Test.ForceNoIconv')) {\r
            // Undo our previous fix in convertToUTF8, otherwise iconv will barf\r
            $ascii_fix = self::testEncodingSupportsASCII($encoding);\r
            if (!$escape && !empty($ascii_fix)) {\r
                $clear_fix = array();\r
                foreach ($ascii_fix as $utf8 => $native) {\r
                    $clear_fix[$utf8] = '';\r
                }\r
                $str = strtr($str, $clear_fix);\r
            }\r
            $str = strtr($str, array_flip($ascii_fix));\r
            // Normal stuff\r
            $str = self::iconv('utf-8', $encoding . '//IGNORE', $str);\r
            return $str;\r
        } elseif ($encoding === 'iso-8859-1') {\r
            $str = utf8_decode($str);\r
            return $str;\r
        }\r
        trigger_error('Encoding not supported', E_USER_ERROR);\r
        // You might be tempted to assume that the ASCII representation\r
        // might be OK, however, this is *not* universally true over all\r
        // encodings.  So we take the conservative route here, rather\r
        // than forcibly turn on %Core.EscapeNonASCIICharacters\r
    }\r
\r
    /**\r
     * Lossless (character-wise) conversion of HTML to ASCII\r
     * @param string $str UTF-8 string to be converted to ASCII\r
     * @return string ASCII encoded string with non-ASCII character entity-ized\r
     * @warning Adapted from MediaWiki, claiming fair use: this is a common\r
     *       algorithm. If you disagree with this license fudgery,\r
     *       implement it yourself.\r
     * @note Uses decimal numeric entities since they are best supported.\r
     * @note This is a DUMB function: it has no concept of keeping\r
     *       character entities that the projected character encoding\r
     *       can allow. We could possibly implement a smart version\r
     *       but that would require it to also know which Unicode\r
     *       codepoints the charset supported (not an easy task).\r
     * @note Sort of with cleanUTF8() but it assumes that $str is\r
     *       well-formed UTF-8\r
     */\r
    public static function convertToASCIIDumbLossless($str)\r
    {\r
        $bytesleft = 0;\r
        $result = '';\r
        $working = 0;\r
        $len = strlen($str);\r
        for ($i = 0; $i < $len; $i++) {\r
            $bytevalue = ord($str[$i]);\r
            if ($bytevalue <= 0x7F) { //0xxx xxxx\r
                $result .= chr($bytevalue);\r
                $bytesleft = 0;\r
            } elseif ($bytevalue <= 0xBF) { //10xx xxxx\r
                $working = $working << 6;\r
                $working += ($bytevalue & 0x3F);\r
                $bytesleft--;\r
                if ($bytesleft <= 0) {\r
                    $result .= "&#" . $working . ";";\r
                }\r
            } elseif ($bytevalue <= 0xDF) { //110x xxxx\r
                $working = $bytevalue & 0x1F;\r
                $bytesleft = 1;\r
            } elseif ($bytevalue <= 0xEF) { //1110 xxxx\r
                $working = $bytevalue & 0x0F;\r
                $bytesleft = 2;\r
            } else { //1111 0xxx\r
                $working = $bytevalue & 0x07;\r
                $bytesleft = 3;\r
            }\r
        }\r
        return $result;\r
    }\r
\r
    /** No bugs detected in iconv. */\r
    const ICONV_OK = 0;\r
\r
    /** Iconv truncates output if converting from UTF-8 to another\r
     *  character set with //IGNORE, and a non-encodable character is found */\r
    const ICONV_TRUNCATES = 1;\r
\r
    /** Iconv does not support //IGNORE, making it unusable for\r
     *  transcoding purposes */\r
    const ICONV_UNUSABLE = 2;\r
\r
    /**\r
     * glibc iconv has a known bug where it doesn't handle the magic\r
     * //IGNORE stanza correctly.  In particular, rather than ignore\r
     * characters, it will return an EILSEQ after consuming some number\r
     * of characters, and expect you to restart iconv as if it were\r
     * an E2BIG.  Old versions of PHP did not respect the errno, and\r
     * returned the fragment, so as a result you would see iconv\r
     * mysteriously truncating output. We can work around this by\r
     * manually chopping our input into segments of about 8000\r
     * characters, as long as PHP ignores the error code.  If PHP starts\r
     * paying attention to the error code, iconv becomes unusable.\r
     *\r
     * @return int Error code indicating severity of bug.\r
     */\r
    public static function testIconvTruncateBug()\r
    {\r
        static $code = null;\r
        if ($code === null) {\r
            // better not use iconv, otherwise infinite loop!\r
            $r = self::unsafeIconv('utf-8', 'ascii//IGNORE', "\xCE\xB1" . str_repeat('a', 9000));\r
            if ($r === false) {\r
                $code = self::ICONV_UNUSABLE;\r
            } elseif (($c = strlen($r)) < 9000) {\r
                $code = self::ICONV_TRUNCATES;\r
            } elseif ($c > 9000) {\r
                trigger_error(\r
                    'Your copy of iconv is extremely buggy. Please notify HTML Purifier maintainers: ' .\r
                    'include your iconv version as per phpversion()',\r
                    E_USER_ERROR\r
                );\r
            } else {\r
                $code = self::ICONV_OK;\r
            }\r
        }\r
        return $code;\r
    }\r
\r
    /**\r
     * This expensive function tests whether or not a given character\r
     * encoding supports ASCII. 7/8-bit encodings like Shift_JIS will\r
     * fail this test, and require special processing. Variable width\r
     * encodings shouldn't ever fail.\r
     *\r
     * @param string $encoding Encoding name to test, as per iconv format\r
     * @param bool $bypass Whether or not to bypass the precompiled arrays.\r
     * @return Array of UTF-8 characters to their corresponding ASCII,\r
     *      which can be used to "undo" any overzealous iconv action.\r
     */\r
    public static function testEncodingSupportsASCII($encoding, $bypass = false)\r
    {\r
        // All calls to iconv here are unsafe, proof by case analysis:\r
        // If ICONV_OK, no difference.\r
        // If ICONV_TRUNCATE, all calls involve one character inputs,\r
        // so bug is not triggered.\r
        // If ICONV_UNUSABLE, this call is irrelevant\r
        static $encodings = array();\r
        if (!$bypass) {\r
            if (isset($encodings[$encoding])) {\r
                return $encodings[$encoding];\r
            }\r
            $lenc = strtolower($encoding);\r
            switch ($lenc) {\r
                case 'shift_jis':\r
                    return array("\xC2\xA5" => '\\', "\xE2\x80\xBE" => '~');\r
                case 'johab':\r
                    return array("\xE2\x82\xA9" => '\\');\r
            }\r
            if (strpos($lenc, 'iso-8859-') === 0) {\r
                return array();\r
            }\r
        }\r
        $ret = array();\r
        if (self::unsafeIconv('UTF-8', $encoding, 'a') === false) {\r
            return false;\r
        }\r
        for ($i = 0x20; $i <= 0x7E; $i++) { // all printable ASCII chars\r
            $c = chr($i); // UTF-8 char\r
            $r = self::unsafeIconv('UTF-8', "$encoding//IGNORE", $c); // initial conversion\r
            if ($r === '' ||\r
                // This line is needed for iconv implementations that do not\r
                // omit characters that do not exist in the target character set\r
                ($r === $c && self::unsafeIconv($encoding, 'UTF-8//IGNORE', $r) !== $c)\r
            ) {\r
                // Reverse engineer: what's the UTF-8 equiv of this byte\r
                // sequence? This assumes that there's no variable width\r
                // encoding that doesn't support ASCII.\r
                $ret[self::unsafeIconv($encoding, 'UTF-8//IGNORE', $c)] = $c;\r
            }\r
        }\r
        $encodings[$encoding] = $ret;\r
        return $ret;\r
    }\r
}\r
\r
// vim: et sw=4 sts=4\r