Chris@0: <?php
Chris@0: 
Chris@0: /*
Chris@0:  * This file is part of the Symfony package.
Chris@0:  *
Chris@0:  * (c) Fabien Potencier <fabien@symfony.com>
Chris@0:  *
Chris@0:  * For the full copyright and license information, please view the LICENSE
Chris@0:  * file that was distributed with this source code.
Chris@0:  */
Chris@0: 
Chris@0: namespace Symfony\Polyfill\Iconv;
Chris@0: 
Chris@0: /**
Chris@0:  * iconv implementation in pure PHP, UTF-8 centric.
Chris@0:  *
Chris@0:  * Implemented:
Chris@0:  * - iconv              - Convert string to requested character encoding
Chris@0:  * - iconv_mime_decode  - Decodes a MIME header field
Chris@0:  * - iconv_mime_decode_headers - Decodes multiple MIME header fields at once
Chris@0:  * - iconv_get_encoding - Retrieve internal configuration variables of iconv extension
Chris@0:  * - iconv_set_encoding - Set current setting for character encoding conversion
Chris@0:  * - iconv_mime_encode  - Composes a MIME header field
Chris@0:  * - iconv_strlen       - Returns the character count of string
Chris@0:  * - iconv_strpos       - Finds position of first occurrence of a needle within a haystack
Chris@0:  * - iconv_strrpos      - Finds the last occurrence of a needle within a haystack
Chris@0:  * - iconv_substr       - Cut out part of a string
Chris@0:  *
Chris@0:  * Charsets available for conversion are defined by files
Chris@0:  * in the charset/ directory and by Iconv::$alias below.
Chris@0:  * You're welcome to send back any addition you make.
Chris@0:  *
Chris@0:  * @author Nicolas Grekas <p@tchwork.com>
Chris@0:  *
Chris@0:  * @internal
Chris@0:  */
Chris@0: final class Iconv
Chris@0: {
Chris@0:     const ERROR_ILLEGAL_CHARACTER = 'iconv(): Detected an illegal character in input string';
Chris@0:     const ERROR_WRONG_CHARSET = 'iconv(): Wrong charset, conversion from `%s\' to `%s\' is not allowed';
Chris@0: 
Chris@0:     public static $inputEncoding = 'utf-8';
Chris@0:     public static $outputEncoding = 'utf-8';
Chris@0:     public static $internalEncoding = 'utf-8';
Chris@0: 
Chris@0:     private static $alias = array(
Chris@0:         'utf8' => 'utf-8',
Chris@0:         'ascii' => 'us-ascii',
Chris@0:         'tis-620' => 'iso-8859-11',
Chris@0:         'cp1250' => 'windows-1250',
Chris@0:         'cp1251' => 'windows-1251',
Chris@0:         'cp1252' => 'windows-1252',
Chris@0:         'cp1253' => 'windows-1253',
Chris@0:         'cp1254' => 'windows-1254',
Chris@0:         'cp1255' => 'windows-1255',
Chris@0:         'cp1256' => 'windows-1256',
Chris@0:         'cp1257' => 'windows-1257',
Chris@0:         'cp1258' => 'windows-1258',
Chris@0:         'shift-jis' => 'cp932',
Chris@0:         'shift_jis' => 'cp932',
Chris@0:         'latin1' => 'iso-8859-1',
Chris@0:         'latin2' => 'iso-8859-2',
Chris@0:         'latin3' => 'iso-8859-3',
Chris@0:         'latin4' => 'iso-8859-4',
Chris@0:         'latin5' => 'iso-8859-9',
Chris@0:         'latin6' => 'iso-8859-10',
Chris@0:         'latin7' => 'iso-8859-13',
Chris@0:         'latin8' => 'iso-8859-14',
Chris@0:         'latin9' => 'iso-8859-15',
Chris@0:         'latin10' => 'iso-8859-16',
Chris@0:         'iso8859-1' => 'iso-8859-1',
Chris@0:         'iso8859-2' => 'iso-8859-2',
Chris@0:         'iso8859-3' => 'iso-8859-3',
Chris@0:         'iso8859-4' => 'iso-8859-4',
Chris@0:         'iso8859-5' => 'iso-8859-5',
Chris@0:         'iso8859-6' => 'iso-8859-6',
Chris@0:         'iso8859-7' => 'iso-8859-7',
Chris@0:         'iso8859-8' => 'iso-8859-8',
Chris@0:         'iso8859-9' => 'iso-8859-9',
Chris@0:         'iso8859-10' => 'iso-8859-10',
Chris@0:         'iso8859-11' => 'iso-8859-11',
Chris@0:         'iso8859-12' => 'iso-8859-12',
Chris@0:         'iso8859-13' => 'iso-8859-13',
Chris@0:         'iso8859-14' => 'iso-8859-14',
Chris@0:         'iso8859-15' => 'iso-8859-15',
Chris@0:         'iso8859-16' => 'iso-8859-16',
Chris@0:         'iso_8859-1' => 'iso-8859-1',
Chris@0:         'iso_8859-2' => 'iso-8859-2',
Chris@0:         'iso_8859-3' => 'iso-8859-3',
Chris@0:         'iso_8859-4' => 'iso-8859-4',
Chris@0:         'iso_8859-5' => 'iso-8859-5',
Chris@0:         'iso_8859-6' => 'iso-8859-6',
Chris@0:         'iso_8859-7' => 'iso-8859-7',
Chris@0:         'iso_8859-8' => 'iso-8859-8',
Chris@0:         'iso_8859-9' => 'iso-8859-9',
Chris@0:         'iso_8859-10' => 'iso-8859-10',
Chris@0:         'iso_8859-11' => 'iso-8859-11',
Chris@0:         'iso_8859-12' => 'iso-8859-12',
Chris@0:         'iso_8859-13' => 'iso-8859-13',
Chris@0:         'iso_8859-14' => 'iso-8859-14',
Chris@0:         'iso_8859-15' => 'iso-8859-15',
Chris@0:         'iso_8859-16' => 'iso-8859-16',
Chris@0:         'iso88591' => 'iso-8859-1',
Chris@0:         'iso88592' => 'iso-8859-2',
Chris@0:         'iso88593' => 'iso-8859-3',
Chris@0:         'iso88594' => 'iso-8859-4',
Chris@0:         'iso88595' => 'iso-8859-5',
Chris@0:         'iso88596' => 'iso-8859-6',
Chris@0:         'iso88597' => 'iso-8859-7',
Chris@0:         'iso88598' => 'iso-8859-8',
Chris@0:         'iso88599' => 'iso-8859-9',
Chris@0:         'iso885910' => 'iso-8859-10',
Chris@0:         'iso885911' => 'iso-8859-11',
Chris@0:         'iso885912' => 'iso-8859-12',
Chris@0:         'iso885913' => 'iso-8859-13',
Chris@0:         'iso885914' => 'iso-8859-14',
Chris@0:         'iso885915' => 'iso-8859-15',
Chris@0:         'iso885916' => 'iso-8859-16',
Chris@0:     );
Chris@0:     private static $translitMap = array();
Chris@0:     private static $convertMap = array();
Chris@0:     private static $errorHandler;
Chris@0:     private static $lastError;
Chris@0: 
Chris@0:     private static $ulenMask = array("\xC0" => 2, "\xD0" => 2, "\xE0" => 3, "\xF0" => 4);
Chris@0:     private static $isValidUtf8;
Chris@0: 
Chris@0:     public static function iconv($inCharset, $outCharset, $str)
Chris@0:     {
Chris@12:         $str = (string) $str;
Chris@12:         if ('' === $str) {
Chris@0:             return '';
Chris@0:         }
Chris@0: 
Chris@0:         // Prepare for //IGNORE and //TRANSLIT
Chris@0: 
Chris@0:         $translit = $ignore = '';
Chris@0: 
Chris@0:         $outCharset = strtolower($outCharset);
Chris@0:         $inCharset = strtolower($inCharset);
Chris@0: 
Chris@0:         if ('' === $outCharset) {
Chris@0:             $outCharset = 'iso-8859-1';
Chris@0:         }
Chris@0:         if ('' === $inCharset) {
Chris@0:             $inCharset = 'iso-8859-1';
Chris@0:         }
Chris@0: 
Chris@12:         do {
Chris@12:             $loop = false;
Chris@0: 
Chris@12:             if ('//translit' === substr($outCharset, -10)) {
Chris@12:                 $loop = $translit = true;
Chris@12:                 $outCharset = substr($outCharset, 0, -10);
Chris@12:             }
Chris@0: 
Chris@12:             if ('//ignore' === substr($outCharset, -8)) {
Chris@12:                 $loop = $ignore = true;
Chris@12:                 $outCharset = substr($outCharset, 0, -8);
Chris@12:             }
Chris@12:         } while ($loop);
Chris@12: 
Chris@12:         do {
Chris@12:             $loop = false;
Chris@12: 
Chris@12:             if ('//translit' === substr($inCharset, -10)) {
Chris@12:                 $loop = true;
Chris@12:                 $inCharset = substr($inCharset, 0, -10);
Chris@12:             }
Chris@12: 
Chris@12:             if ('//ignore' === substr($inCharset, -8)) {
Chris@12:                 $loop = true;
Chris@12:                 $inCharset = substr($inCharset, 0, -8);
Chris@12:             }
Chris@12:         } while ($loop);
Chris@0: 
Chris@17:         if (isset(self::$alias[$inCharset])) {
Chris@17:             $inCharset = self::$alias[$inCharset];
Chris@0:         }
Chris@0:         if (isset(self::$alias[$outCharset])) {
Chris@0:             $outCharset = self::$alias[$outCharset];
Chris@0:         }
Chris@0: 
Chris@0:         // Load charset maps
Chris@0: 
Chris@0:         if (('utf-8' !== $inCharset && !self::loadMap('from.', $inCharset, $inMap))
Chris@0:           || ('utf-8' !== $outCharset && !self::loadMap('to.', $outCharset, $outMap))) {
Chris@0:             trigger_error(sprintf(self::ERROR_WRONG_CHARSET, $inCharset, $outCharset));
Chris@0: 
Chris@0:             return false;
Chris@0:         }
Chris@0: 
Chris@0:         if ('utf-8' !== $inCharset) {
Chris@0:             // Convert input to UTF-8
Chris@0:             $result = '';
Chris@0:             if (self::mapToUtf8($result, $inMap, $str, $ignore)) {
Chris@0:                 $str = $result;
Chris@0:             } else {
Chris@0:                 $str = false;
Chris@0:             }
Chris@0:             self::$isValidUtf8 = true;
Chris@0:         } else {
Chris@0:             self::$isValidUtf8 = preg_match('//u', $str);
Chris@0: 
Chris@0:             if (!self::$isValidUtf8 && !$ignore) {
Chris@0:                 trigger_error(self::ERROR_ILLEGAL_CHARACTER);
Chris@0: 
Chris@0:                 return false;
Chris@0:             }
Chris@0: 
Chris@0:             if ('utf-8' === $outCharset) {
Chris@0:                 // UTF-8 validation
Chris@0:                 $str = self::utf8ToUtf8($str, $ignore);
Chris@0:             }
Chris@0:         }
Chris@0: 
Chris@0:         if ('utf-8' !== $outCharset && false !== $str) {
Chris@0:             // Convert output to UTF-8
Chris@0:             $result = '';
Chris@0:             if (self::mapFromUtf8($result, $outMap, $str, $ignore, $translit)) {
Chris@0:                 return $result;
Chris@0:             }
Chris@0: 
Chris@0:             return false;
Chris@0:         }
Chris@0: 
Chris@0:         return $str;
Chris@0:     }
Chris@0: 
Chris@0:     public static function iconv_mime_decode_headers($str, $mode = 0, $charset = null)
Chris@0:     {
Chris@0:         if (null === $charset) {
Chris@0:             $charset = self::$internalEncoding;
Chris@0:         }
Chris@0: 
Chris@0:         if (false !== strpos($str, "\r")) {
Chris@0:             $str = strtr(str_replace("\r\n", "\n", $str), "\r", "\n");
Chris@0:         }
Chris@0:         $str = explode("\n\n", $str, 2);
Chris@0: 
Chris@0:         $headers = array();
Chris@0: 
Chris@0:         $str = preg_split('/\n(?![ \t])/', $str[0]);
Chris@0:         foreach ($str as $str) {
Chris@0:             $str = self::iconv_mime_decode($str, $mode, $charset);
Chris@0:             if (false === $str) {
Chris@0:                 return false;
Chris@0:             }
Chris@0:             $str = explode(':', $str, 2);
Chris@0: 
Chris@16:             if (2 === \count($str)) {
Chris@0:                 if (isset($headers[$str[0]])) {
Chris@16:                     if (!\is_array($headers[$str[0]])) {
Chris@0:                         $headers[$str[0]] = array($headers[$str[0]]);
Chris@0:                     }
Chris@0:                     $headers[$str[0]][] = ltrim($str[1]);
Chris@0:                 } else {
Chris@0:                     $headers[$str[0]] = ltrim($str[1]);
Chris@0:                 }
Chris@0:             }
Chris@0:         }
Chris@0: 
Chris@0:         return $headers;
Chris@0:     }
Chris@0: 
Chris@0:     public static function iconv_mime_decode($str, $mode = 0, $charset = null)
Chris@0:     {
Chris@0:         if (null === $charset) {
Chris@0:             $charset = self::$internalEncoding;
Chris@0:         }
Chris@0:         if (ICONV_MIME_DECODE_CONTINUE_ON_ERROR & $mode) {
Chris@0:             $charset .= '//IGNORE';
Chris@0:         }
Chris@0: 
Chris@0:         if (false !== strpos($str, "\r")) {
Chris@0:             $str = strtr(str_replace("\r\n", "\n", $str), "\r", "\n");
Chris@0:         }
Chris@0:         $str = preg_split('/\n(?![ \t])/', rtrim($str), 2);
Chris@0:         $str = preg_replace('/[ \t]*\n[ \t]+/', ' ', rtrim($str[0]));
Chris@0:         $str = preg_split('/=\?([^?]+)\?([bqBQ])\?(.*?)\?=/', $str, -1, PREG_SPLIT_DELIM_CAPTURE);
Chris@0: 
Chris@0:         $result = self::iconv('utf-8', $charset, $str[0]);
Chris@0:         if (false === $result) {
Chris@0:             return false;
Chris@0:         }
Chris@0: 
Chris@0:         $i = 1;
Chris@16:         $len = \count($str);
Chris@0: 
Chris@0:         while ($i < $len) {
Chris@0:             $c = strtolower($str[$i]);
Chris@0:             if ((ICONV_MIME_DECODE_CONTINUE_ON_ERROR & $mode)
Chris@0:               && 'utf-8' !== $c
Chris@0:               && !isset(self::$alias[$c])
Chris@17:               && !self::loadMap('from.', $c, $d)) {
Chris@0:                 $d = false;
Chris@0:             } elseif ('B' === strtoupper($str[$i + 1])) {
Chris@0:                 $d = base64_decode($str[$i + 2]);
Chris@0:             } else {
Chris@0:                 $d = rawurldecode(strtr(str_replace('%', '%25', $str[$i + 2]), '=_', '% '));
Chris@0:             }
Chris@0: 
Chris@0:             if (false !== $d) {
Chris@0:                 if ('' !== $d) {
Chris@0:                     if ('' === $d = self::iconv($c, $charset, $d)) {
Chris@0:                         $str[$i + 3] = substr($str[$i + 3], 1);
Chris@0:                     } else {
Chris@0:                         $result .= $d;
Chris@0:                     }
Chris@0:                 }
Chris@0:                 $d = self::iconv('utf-8', $charset, $str[$i + 3]);
Chris@0:                 if ('' !== trim($d)) {
Chris@0:                     $result .= $d;
Chris@0:                 }
Chris@0:             } elseif (ICONV_MIME_DECODE_CONTINUE_ON_ERROR & $mode) {
Chris@0:                 $result .= "=?{$str[$i]}?{$str[$i + 1]}?{$str[$i + 2]}?={$str[$i + 3]}";
Chris@0:             } else {
Chris@0:                 $result = false;
Chris@0:                 break;
Chris@0:             }
Chris@0: 
Chris@0:             $i += 4;
Chris@0:         }
Chris@0: 
Chris@0:         return $result;
Chris@0:     }
Chris@0: 
Chris@0:     public static function iconv_get_encoding($type = 'all')
Chris@0:     {
Chris@0:         switch ($type) {
Chris@0:             case 'input_encoding': return self::$inputEncoding;
Chris@0:             case 'output_encoding': return self::$outputEncoding;
Chris@0:             case 'internal_encoding': return self::$internalEncoding;
Chris@0:         }
Chris@0: 
Chris@0:         return array(
Chris@0:             'input_encoding' => self::$inputEncoding,
Chris@0:             'output_encoding' => self::$outputEncoding,
Chris@0:             'internal_encoding' => self::$internalEncoding,
Chris@0:         );
Chris@0:     }
Chris@0: 
Chris@0:     public static function iconv_set_encoding($type, $charset)
Chris@0:     {
Chris@0:         switch ($type) {
Chris@0:             case 'input_encoding': self::$inputEncoding = $charset; break;
Chris@0:             case 'output_encoding': self::$outputEncoding = $charset; break;
Chris@0:             case 'internal_encoding': self::$internalEncoding = $charset; break;
Chris@0: 
Chris@0:             default: return false;
Chris@0:         }
Chris@0: 
Chris@0:         return true;
Chris@0:     }
Chris@0: 
Chris@0:     public static function iconv_mime_encode($fieldName, $fieldValue, $pref = null)
Chris@0:     {
Chris@16:         if (!\is_array($pref)) {
Chris@0:             $pref = array();
Chris@0:         }
Chris@0: 
Chris@0:         $pref += array(
Chris@0:             'scheme' => 'B',
Chris@0:             'input-charset' => self::$internalEncoding,
Chris@0:             'output-charset' => self::$internalEncoding,
Chris@0:             'line-length' => 76,
Chris@0:             'line-break-chars' => "\r\n",
Chris@0:         );
Chris@0: 
Chris@0:         if (preg_match('/[\x80-\xFF]/', $fieldName)) {
Chris@0:             $fieldName = '';
Chris@0:         }
Chris@0: 
Chris@0:         $scheme = strtoupper(substr($pref['scheme'], 0, 1));
Chris@0:         $in = strtolower($pref['input-charset']);
Chris@0:         $out = strtolower($pref['output-charset']);
Chris@0: 
Chris@0:         if ('utf-8' !== $in && false === $fieldValue = self::iconv($in, 'utf-8', $fieldValue)) {
Chris@0:             return false;
Chris@0:         }
Chris@0: 
Chris@0:         preg_match_all('/./us', $fieldValue, $chars);
Chris@0: 
Chris@0:         $chars = isset($chars[0]) ? $chars[0] : array();
Chris@0: 
Chris@0:         $lineBreak = (int) $pref['line-length'];
Chris@0:         $lineStart = "=?{$pref['output-charset']}?{$scheme}?";
Chris@16:         $lineLength = \strlen($fieldName) + 2 + \strlen($lineStart) + 2;
Chris@16:         $lineOffset = \strlen($lineStart) + 3;
Chris@0:         $lineData = '';
Chris@0: 
Chris@0:         $fieldValue = array();
Chris@0: 
Chris@0:         $Q = 'Q' === $scheme;
Chris@0: 
Chris@0:         foreach ($chars as $c) {
Chris@0:             if ('utf-8' !== $out && false === $c = self::iconv('utf-8', $out, $c)) {
Chris@0:                 return false;
Chris@0:             }
Chris@0: 
Chris@0:             $o = $Q
Chris@0:                 ? $c = preg_replace_callback(
Chris@0:                     '/[=_\?\x00-\x1F\x80-\xFF]/',
Chris@0:                     array(__CLASS__, 'qpByteCallback'),
Chris@0:                     $c
Chris@0:                 )
Chris@0:                 : base64_encode($lineData.$c);
Chris@0: 
Chris@0:             if (isset($o[$lineBreak - $lineLength])) {
Chris@0:                 if (!$Q) {
Chris@0:                     $lineData = base64_encode($lineData);
Chris@0:                 }
Chris@0:                 $fieldValue[] = $lineStart.$lineData.'?=';
Chris@0:                 $lineLength = $lineOffset;
Chris@0:                 $lineData = '';
Chris@0:             }
Chris@0: 
Chris@0:             $lineData .= $c;
Chris@16:             $Q && $lineLength += \strlen($c);
Chris@0:         }
Chris@0: 
Chris@0:         if ('' !== $lineData) {
Chris@0:             if (!$Q) {
Chris@0:                 $lineData = base64_encode($lineData);
Chris@0:             }
Chris@0:             $fieldValue[] = $lineStart.$lineData.'?=';
Chris@0:         }
Chris@0: 
Chris@0:         return $fieldName.': '.implode($pref['line-break-chars'].' ', $fieldValue);
Chris@0:     }
Chris@0: 
Chris@0:     public static function iconv_strlen($s, $encoding = null)
Chris@0:     {
Chris@0:         static $hasXml = null;
Chris@0:         if (null === $hasXml) {
Chris@17:             $hasXml = \extension_loaded('xml');
Chris@0:         }
Chris@0: 
Chris@0:         if ($hasXml) {
Chris@0:             return self::strlen1($s, $encoding);
Chris@0:         }
Chris@0: 
Chris@0:         return self::strlen2($s, $encoding);
Chris@0:     }
Chris@0: 
Chris@0:     public static function strlen1($s, $encoding = null)
Chris@0:     {
Chris@0:         if (null === $encoding) {
Chris@0:             $encoding = self::$internalEncoding;
Chris@0:         }
Chris@0:         if (0 !== stripos($encoding, 'utf-8') && false === $s = self::iconv($encoding, 'utf-8', $s)) {
Chris@0:             return false;
Chris@0:         }
Chris@0: 
Chris@16:         return \strlen(utf8_decode($s));
Chris@0:     }
Chris@0: 
Chris@0:     public static function strlen2($s, $encoding = null)
Chris@0:     {
Chris@0:         if (null === $encoding) {
Chris@0:             $encoding = self::$internalEncoding;
Chris@0:         }
Chris@0:         if (0 !== stripos($encoding, 'utf-8') && false === $s = self::iconv($encoding, 'utf-8', $s)) {
Chris@0:             return false;
Chris@0:         }
Chris@0: 
Chris@0:         $ulenMask = self::$ulenMask;
Chris@0: 
Chris@0:         $i = 0;
Chris@0:         $j = 0;
Chris@16:         $len = \strlen($s);
Chris@0: 
Chris@0:         while ($i < $len) {
Chris@0:             $u = $s[$i] & "\xF0";
Chris@0:             $i += isset($ulenMask[$u]) ? $ulenMask[$u] : 1;
Chris@0:             ++$j;
Chris@0:         }
Chris@0: 
Chris@0:         return $j;
Chris@0:     }
Chris@0: 
Chris@0:     public static function iconv_strpos($haystack, $needle, $offset = 0, $encoding = null)
Chris@0:     {
Chris@0:         if (null === $encoding) {
Chris@0:             $encoding = self::$internalEncoding;
Chris@0:         }
Chris@0: 
Chris@0:         if (0 !== stripos($encoding, 'utf-8')) {
Chris@0:             if (false === $haystack = self::iconv($encoding, 'utf-8', $haystack)) {
Chris@0:                 return false;
Chris@0:             }
Chris@0:             if (false === $needle = self::iconv($encoding, 'utf-8', $needle)) {
Chris@0:                 return false;
Chris@0:             }
Chris@0:         }
Chris@0: 
Chris@0:         if ($offset = (int) $offset) {
Chris@0:             $haystack = self::iconv_substr($haystack, $offset, 2147483647, 'utf-8');
Chris@0:         }
Chris@0:         $pos = strpos($haystack, $needle);
Chris@0: 
Chris@0:         return false === $pos ? false : ($offset + ($pos ? self::iconv_strlen(substr($haystack, 0, $pos), 'utf-8') : 0));
Chris@0:     }
Chris@0: 
Chris@0:     public static function iconv_strrpos($haystack, $needle, $encoding = null)
Chris@0:     {
Chris@0:         if (null === $encoding) {
Chris@0:             $encoding = self::$internalEncoding;
Chris@0:         }
Chris@0: 
Chris@0:         if (0 !== stripos($encoding, 'utf-8')) {
Chris@0:             if (false === $haystack = self::iconv($encoding, 'utf-8', $haystack)) {
Chris@0:                 return false;
Chris@0:             }
Chris@0:             if (false === $needle = self::iconv($encoding, 'utf-8', $needle)) {
Chris@0:                 return false;
Chris@0:             }
Chris@0:         }
Chris@0: 
Chris@0:         $pos = isset($needle[0]) ? strrpos($haystack, $needle) : false;
Chris@0: 
Chris@0:         return false === $pos ? false : self::iconv_strlen($pos ? substr($haystack, 0, $pos) : $haystack, 'utf-8');
Chris@0:     }
Chris@0: 
Chris@0:     public static function iconv_substr($s, $start, $length = 2147483647, $encoding = null)
Chris@0:     {
Chris@0:         if (null === $encoding) {
Chris@0:             $encoding = self::$internalEncoding;
Chris@0:         }
Chris@0:         if (0 !== stripos($encoding, 'utf-8')) {
Chris@0:             $encoding = null;
Chris@0:         } elseif (false === $s = self::iconv($encoding, 'utf-8', $s)) {
Chris@0:             return false;
Chris@0:         }
Chris@0: 
Chris@12:         $s = (string) $s;
Chris@0:         $slen = self::iconv_strlen($s, 'utf-8');
Chris@0:         $start = (int) $start;
Chris@0: 
Chris@0:         if (0 > $start) {
Chris@0:             $start += $slen;
Chris@0:         }
Chris@0:         if (0 > $start) {
Chris@0:             return false;
Chris@0:         }
Chris@0:         if ($start >= $slen) {
Chris@0:             return false;
Chris@0:         }
Chris@0: 
Chris@0:         $rx = $slen - $start;
Chris@0: 
Chris@0:         if (0 > $length) {
Chris@0:             $length += $rx;
Chris@0:         }
Chris@0:         if (0 === $length) {
Chris@0:             return '';
Chris@0:         }
Chris@0:         if (0 > $length) {
Chris@0:             return false;
Chris@0:         }
Chris@0: 
Chris@0:         if ($length > $rx) {
Chris@0:             $length = $rx;
Chris@0:         }
Chris@0: 
Chris@0:         $rx = '/^'.($start ? self::pregOffset($start) : '').'('.self::pregOffset($length).')/u';
Chris@0: 
Chris@0:         $s = preg_match($rx, $s, $s) ? $s[1] : '';
Chris@0: 
Chris@0:         if (null === $encoding) {
Chris@0:             return $s;
Chris@0:         }
Chris@0: 
Chris@0:         return self::iconv('utf-8', $encoding, $s);
Chris@0:     }
Chris@0: 
Chris@0:     private static function loadMap($type, $charset, &$map)
Chris@0:     {
Chris@0:         if (!isset(self::$convertMap[$type.$charset])) {
Chris@0:             if (false === $map = self::getData($type.$charset)) {
Chris@0:                 if ('to.' === $type && self::loadMap('from.', $charset, $map)) {
Chris@0:                     $map = array_flip($map);
Chris@0:                 } else {
Chris@0:                     return false;
Chris@0:                 }
Chris@0:             }
Chris@0: 
Chris@0:             self::$convertMap[$type.$charset] = $map;
Chris@0:         } else {
Chris@0:             $map = self::$convertMap[$type.$charset];
Chris@0:         }
Chris@0: 
Chris@0:         return true;
Chris@0:     }
Chris@0: 
Chris@0:     private static function utf8ToUtf8($str, $ignore)
Chris@0:     {
Chris@0:         $ulenMask = self::$ulenMask;
Chris@0:         $valid = self::$isValidUtf8;
Chris@0: 
Chris@0:         $u = $str;
Chris@0:         $i = $j = 0;
Chris@16:         $len = \strlen($str);
Chris@0: 
Chris@0:         while ($i < $len) {
Chris@0:             if ($str[$i] < "\x80") {
Chris@0:                 $u[$j++] = $str[$i++];
Chris@0:             } else {
Chris@0:                 $ulen = $str[$i] & "\xF0";
Chris@0:                 $ulen = isset($ulenMask[$ulen]) ? $ulenMask[$ulen] : 1;
Chris@0:                 $uchr = substr($str, $i, $ulen);
Chris@0: 
Chris@0:                 if (1 === $ulen || !($valid || preg_match('/^.$/us', $uchr))) {
Chris@0:                     if ($ignore) {
Chris@0:                         ++$i;
Chris@0:                         continue;
Chris@0:                     }
Chris@0: 
Chris@0:                     trigger_error(self::ERROR_ILLEGAL_CHARACTER);
Chris@0: 
Chris@0:                     return false;
Chris@0:                 } else {
Chris@0:                     $i += $ulen;
Chris@0:                 }
Chris@0: 
Chris@0:                 $u[$j++] = $uchr[0];
Chris@0: 
Chris@0:                 isset($uchr[1]) && 0 !== ($u[$j++] = $uchr[1])
Chris@0:                     && isset($uchr[2]) && 0 !== ($u[$j++] = $uchr[2])
Chris@0:                     && isset($uchr[3]) && 0 !== ($u[$j++] = $uchr[3]);
Chris@0:             }
Chris@0:         }
Chris@0: 
Chris@0:         return substr($u, 0, $j);
Chris@0:     }
Chris@0: 
Chris@12:     private static function mapToUtf8(&$result, array $map, $str, $ignore)
Chris@0:     {
Chris@16:         $len = \strlen($str);
Chris@0:         for ($i = 0; $i < $len; ++$i) {
Chris@0:             if (isset($str[$i + 1], $map[$str[$i].$str[$i + 1]])) {
Chris@0:                 $result .= $map[$str[$i].$str[++$i]];
Chris@0:             } elseif (isset($map[$str[$i]])) {
Chris@0:                 $result .= $map[$str[$i]];
Chris@0:             } elseif (!$ignore) {
Chris@0:                 trigger_error(self::ERROR_ILLEGAL_CHARACTER);
Chris@0: 
Chris@0:                 return false;
Chris@0:             }
Chris@0:         }
Chris@0: 
Chris@0:         return true;
Chris@0:     }
Chris@0: 
Chris@12:     private static function mapFromUtf8(&$result, array $map, $str, $ignore, $translit)
Chris@0:     {
Chris@0:         $ulenMask = self::$ulenMask;
Chris@0:         $valid = self::$isValidUtf8;
Chris@0: 
Chris@0:         if ($translit && !self::$translitMap) {
Chris@0:             self::$translitMap = self::getData('translit');
Chris@0:         }
Chris@0: 
Chris@0:         $i = 0;
Chris@16:         $len = \strlen($str);
Chris@0: 
Chris@0:         while ($i < $len) {
Chris@0:             if ($str[$i] < "\x80") {
Chris@0:                 $uchr = $str[$i++];
Chris@0:             } else {
Chris@0:                 $ulen = $str[$i] & "\xF0";
Chris@0:                 $ulen = isset($ulenMask[$ulen]) ? $ulenMask[$ulen] : 1;
Chris@0:                 $uchr = substr($str, $i, $ulen);
Chris@0: 
Chris@0:                 if ($ignore && (1 === $ulen || !($valid || preg_match('/^.$/us', $uchr)))) {
Chris@0:                     ++$i;
Chris@0:                     continue;
Chris@0:                 } else {
Chris@0:                     $i += $ulen;
Chris@0:                 }
Chris@0:             }
Chris@0: 
Chris@0:             if (isset($map[$uchr])) {
Chris@0:                 $result .= $map[$uchr];
Chris@0:             } elseif ($translit) {
Chris@0:                 if (isset(self::$translitMap[$uchr])) {
Chris@0:                     $uchr = self::$translitMap[$uchr];
Chris@0:                 } elseif ($uchr >= "\xC3\x80") {
Chris@0:                     $uchr = \Normalizer::normalize($uchr, \Normalizer::NFD);
Chris@0: 
Chris@0:                     if ($uchr[0] < "\x80") {
Chris@0:                         $uchr = $uchr[0];
Chris@0:                     } elseif ($ignore) {
Chris@0:                         continue;
Chris@0:                     } else {
Chris@0:                         return false;
Chris@0:                     }
Chris@17:                 } elseif ($ignore) {
Chris@17:                     continue;
Chris@17:                 } else {
Chris@17:                     return false;
Chris@0:                 }
Chris@0: 
Chris@0:                 $str = $uchr.substr($str, $i);
Chris@16:                 $len = \strlen($str);
Chris@0:                 $i = 0;
Chris@0:             } elseif (!$ignore) {
Chris@0:                 return false;
Chris@0:             }
Chris@0:         }
Chris@0: 
Chris@0:         return true;
Chris@0:     }
Chris@0: 
Chris@12:     private static function qpByteCallback(array $m)
Chris@0:     {
Chris@16:         return '='.strtoupper(dechex(\ord($m[0])));
Chris@0:     }
Chris@0: 
Chris@0:     private static function pregOffset($offset)
Chris@0:     {
Chris@0:         $rx = array();
Chris@0:         $offset = (int) $offset;
Chris@0: 
Chris@0:         while ($offset > 65535) {
Chris@0:             $rx[] = '.{65535}';
Chris@0:             $offset -= 65535;
Chris@0:         }
Chris@0: 
Chris@0:         return implode('', $rx).'.{'.$offset.'}';
Chris@0:     }
Chris@0: 
Chris@0:     private static function getData($file)
Chris@0:     {
Chris@0:         if (file_exists($file = __DIR__.'/Resources/charset/'.$file.'.php')) {
Chris@0:             return require $file;
Chris@0:         }
Chris@0: 
Chris@0:         return false;
Chris@0:     }
Chris@0: }