| 1 |
<?php |
| 2 |
|
| 3 |
namespace PhpOffice\PhpSpreadsheet\Shared; |
| 4 |
|
| 5 |
use PhpOffice\PhpSpreadsheet\Exception as PhpSpreadsheetException; |
| 6 |
|
| 7 |
class CodePage |
| 8 |
{ |
| 9 |
/** |
| 10 |
* Convert Microsoft Code Page Identifier to Code Page Name which iconv |
| 11 |
* and mbstring understands. |
| 12 |
* |
| 13 |
* @param int $codePage Microsoft Code Page Indentifier |
| 14 |
* |
| 15 |
* @throws PhpSpreadsheetException |
| 16 |
* |
| 17 |
* @return string Code Page Name |
| 18 |
*/ |
| 19 |
public static function numberToName($codePage) |
| 20 |
{ |
| 21 |
switch ($codePage) { |
| 22 |
case 367: |
| 23 |
return 'ASCII'; // ASCII |
| 24 |
case 437: |
| 25 |
return 'CP437'; // OEM US |
| 26 |
case 720: |
| 27 |
throw new PhpSpreadsheetException('Code page 720 not supported.'); // OEM Arabic |
| 28 |
case 737: |
| 29 |
return 'CP737'; // OEM Greek |
| 30 |
case 775: |
| 31 |
return 'CP775'; // OEM Baltic |
| 32 |
case 850: |
| 33 |
return 'CP850'; // OEM Latin I |
| 34 |
case 852: |
| 35 |
return 'CP852'; // OEM Latin II (Central European) |
| 36 |
case 855: |
| 37 |
return 'CP855'; // OEM Cyrillic |
| 38 |
case 857: |
| 39 |
return 'CP857'; // OEM Turkish |
| 40 |
case 858: |
| 41 |
return 'CP858'; // OEM Multilingual Latin I with Euro |
| 42 |
case 860: |
| 43 |
return 'CP860'; // OEM Portugese |
| 44 |
case 861: |
| 45 |
return 'CP861'; // OEM Icelandic |
| 46 |
case 862: |
| 47 |
return 'CP862'; // OEM Hebrew |
| 48 |
case 863: |
| 49 |
return 'CP863'; // OEM Canadian (French) |
| 50 |
case 864: |
| 51 |
return 'CP864'; // OEM Arabic |
| 52 |
case 865: |
| 53 |
return 'CP865'; // OEM Nordic |
| 54 |
case 866: |
| 55 |
return 'CP866'; // OEM Cyrillic (Russian) |
| 56 |
case 869: |
| 57 |
return 'CP869'; // OEM Greek (Modern) |
| 58 |
case 874: |
| 59 |
return 'CP874'; // ANSI Thai |
| 60 |
case 932: |
| 61 |
return 'CP932'; // ANSI Japanese Shift-JIS |
| 62 |
case 936: |
| 63 |
return 'CP936'; // ANSI Chinese Simplified GBK |
| 64 |
case 949: |
| 65 |
return 'CP949'; // ANSI Korean (Wansung) |
| 66 |
case 950: |
| 67 |
return 'CP950'; // ANSI Chinese Traditional BIG5 |
| 68 |
case 1200: |
| 69 |
return 'UTF-16LE'; // UTF-16 (BIFF8) |
| 70 |
case 1250: |
| 71 |
return 'CP1250'; // ANSI Latin II (Central European) |
| 72 |
case 1251: |
| 73 |
return 'CP1251'; // ANSI Cyrillic |
| 74 |
case 0: |
| 75 |
// CodePage is not always correctly set when the xls file was saved by Apple's Numbers program |
| 76 |
case 1252: |
| 77 |
return 'CP1252'; // ANSI Latin I (BIFF4-BIFF7) |
| 78 |
case 1253: |
| 79 |
return 'CP1253'; // ANSI Greek |
| 80 |
case 1254: |
| 81 |
return 'CP1254'; // ANSI Turkish |
| 82 |
case 1255: |
| 83 |
return 'CP1255'; // ANSI Hebrew |
| 84 |
case 1256: |
| 85 |
return 'CP1256'; // ANSI Arabic |
| 86 |
case 1257: |
| 87 |
return 'CP1257'; // ANSI Baltic |
| 88 |
case 1258: |
| 89 |
return 'CP1258'; // ANSI Vietnamese |
| 90 |
case 1361: |
| 91 |
return 'CP1361'; // ANSI Korean (Johab) |
| 92 |
case 10000: |
| 93 |
return 'MAC'; // Apple Roman |
| 94 |
case 10001: |
| 95 |
return 'CP932'; // Macintosh Japanese |
| 96 |
case 10002: |
| 97 |
return 'CP950'; // Macintosh Chinese Traditional |
| 98 |
case 10003: |
| 99 |
return 'CP1361'; // Macintosh Korean |
| 100 |
case 10004: |
| 101 |
return 'MACARABIC'; // Apple Arabic |
| 102 |
case 10005: |
| 103 |
return 'MACHEBREW'; // Apple Hebrew |
| 104 |
case 10006: |
| 105 |
return 'MACGREEK'; // Macintosh Greek |
| 106 |
case 10007: |
| 107 |
return 'MACCYRILLIC'; // Macintosh Cyrillic |
| 108 |
case 10008: |
| 109 |
return 'CP936'; // Macintosh - Simplified Chinese (GB 2312) |
| 110 |
case 10010: |
| 111 |
return 'MACROMANIA'; // Macintosh Romania |
| 112 |
case 10017: |
| 113 |
return 'MACUKRAINE'; // Macintosh Ukraine |
| 114 |
case 10021: |
| 115 |
return 'MACTHAI'; // Macintosh Thai |
| 116 |
case 10029: |
| 117 |
return 'MACCENTRALEUROPE'; // Macintosh Central Europe |
| 118 |
case 10079: |
| 119 |
return 'MACICELAND'; // Macintosh Icelandic |
| 120 |
case 10081: |
| 121 |
return 'MACTURKISH'; // Macintosh Turkish |
| 122 |
case 10082: |
| 123 |
return 'MACCROATIAN'; // Macintosh Croatian |
| 124 |
case 21010: |
| 125 |
return 'UTF-16LE'; // UTF-16 (BIFF8) This isn't correct, but some Excel writer libraries erroneously use Codepage 21010 for UTF-16LE |
| 126 |
case 32768: |
| 127 |
return 'MAC'; // Apple Roman |
| 128 |
case 32769: |
| 129 |
throw new PhpSpreadsheetException('Code page 32769 not supported.'); // ANSI Latin I (BIFF2-BIFF3) |
| 130 |
case 65000: |
| 131 |
return 'UTF-7'; // Unicode (UTF-7) |
| 132 |
case 65001: |
| 133 |
return 'UTF-8'; // Unicode (UTF-8) |
| 134 |
} |
| 135 |
|
| 136 |
throw new PhpSpreadsheetException('Unknown codepage: ' . $codePage); |
| 137 |
} |
| 138 |
} |
| 139 |
|