CodePage.php 5.2 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138
  1. <?php
  2. namespace PhpOffice\PhpSpreadsheet\Shared;
  3. use PhpOffice\PhpSpreadsheet\Exception as PhpSpreadsheetException;
  4. class CodePage
  5. {
  6. /**
  7. * Convert Microsoft Code Page Identifier to Code Page Name which iconv
  8. * and mbstring understands.
  9. *
  10. * @param int $codePage Microsoft Code Page Indentifier
  11. *
  12. * @throws PhpSpreadsheetException
  13. *
  14. * @return string Code Page Name
  15. */
  16. public static function numberToName($codePage)
  17. {
  18. switch ($codePage) {
  19. case 367:
  20. return 'ASCII'; // ASCII
  21. case 437:
  22. return 'CP437'; // OEM US
  23. case 720:
  24. throw new PhpSpreadsheetException('Code page 720 not supported.'); // OEM Arabic
  25. case 737:
  26. return 'CP737'; // OEM Greek
  27. case 775:
  28. return 'CP775'; // OEM Baltic
  29. case 850:
  30. return 'CP850'; // OEM Latin I
  31. case 852:
  32. return 'CP852'; // OEM Latin II (Central European)
  33. case 855:
  34. return 'CP855'; // OEM Cyrillic
  35. case 857:
  36. return 'CP857'; // OEM Turkish
  37. case 858:
  38. return 'CP858'; // OEM Multilingual Latin I with Euro
  39. case 860:
  40. return 'CP860'; // OEM Portugese
  41. case 861:
  42. return 'CP861'; // OEM Icelandic
  43. case 862:
  44. return 'CP862'; // OEM Hebrew
  45. case 863:
  46. return 'CP863'; // OEM Canadian (French)
  47. case 864:
  48. return 'CP864'; // OEM Arabic
  49. case 865:
  50. return 'CP865'; // OEM Nordic
  51. case 866:
  52. return 'CP866'; // OEM Cyrillic (Russian)
  53. case 869:
  54. return 'CP869'; // OEM Greek (Modern)
  55. case 874:
  56. return 'CP874'; // ANSI Thai
  57. case 932:
  58. return 'CP932'; // ANSI Japanese Shift-JIS
  59. case 936:
  60. return 'CP936'; // ANSI Chinese Simplified GBK
  61. case 949:
  62. return 'CP949'; // ANSI Korean (Wansung)
  63. case 950:
  64. return 'CP950'; // ANSI Chinese Traditional BIG5
  65. case 1200:
  66. return 'UTF-16LE'; // UTF-16 (BIFF8)
  67. case 1250:
  68. return 'CP1250'; // ANSI Latin II (Central European)
  69. case 1251:
  70. return 'CP1251'; // ANSI Cyrillic
  71. case 0:
  72. // CodePage is not always correctly set when the xls file was saved by Apple's Numbers program
  73. case 1252:
  74. return 'CP1252'; // ANSI Latin I (BIFF4-BIFF7)
  75. case 1253:
  76. return 'CP1253'; // ANSI Greek
  77. case 1254:
  78. return 'CP1254'; // ANSI Turkish
  79. case 1255:
  80. return 'CP1255'; // ANSI Hebrew
  81. case 1256:
  82. return 'CP1256'; // ANSI Arabic
  83. case 1257:
  84. return 'CP1257'; // ANSI Baltic
  85. case 1258:
  86. return 'CP1258'; // ANSI Vietnamese
  87. case 1361:
  88. return 'CP1361'; // ANSI Korean (Johab)
  89. case 10000:
  90. return 'MAC'; // Apple Roman
  91. case 10001:
  92. return 'CP932'; // Macintosh Japanese
  93. case 10002:
  94. return 'CP950'; // Macintosh Chinese Traditional
  95. case 10003:
  96. return 'CP1361'; // Macintosh Korean
  97. case 10004:
  98. return 'MACARABIC'; // Apple Arabic
  99. case 10005:
  100. return 'MACHEBREW'; // Apple Hebrew
  101. case 10006:
  102. return 'MACGREEK'; // Macintosh Greek
  103. case 10007:
  104. return 'MACCYRILLIC'; // Macintosh Cyrillic
  105. case 10008:
  106. return 'CP936'; // Macintosh - Simplified Chinese (GB 2312)
  107. case 10010:
  108. return 'MACROMANIA'; // Macintosh Romania
  109. case 10017:
  110. return 'MACUKRAINE'; // Macintosh Ukraine
  111. case 10021:
  112. return 'MACTHAI'; // Macintosh Thai
  113. case 10029:
  114. return 'MACCENTRALEUROPE'; // Macintosh Central Europe
  115. case 10079:
  116. return 'MACICELAND'; // Macintosh Icelandic
  117. case 10081:
  118. return 'MACTURKISH'; // Macintosh Turkish
  119. case 10082:
  120. return 'MACCROATIAN'; // Macintosh Croatian
  121. case 21010:
  122. return 'UTF-16LE'; // UTF-16 (BIFF8) This isn't correct, but some Excel writer libraries erroneously use Codepage 21010 for UTF-16LE
  123. case 32768:
  124. return 'MAC'; // Apple Roman
  125. case 32769:
  126. throw new PhpSpreadsheetException('Code page 32769 not supported.'); // ANSI Latin I (BIFF2-BIFF3)
  127. case 65000:
  128. return 'UTF-7'; // Unicode (UTF-7)
  129. case 65001:
  130. return 'UTF-8'; // Unicode (UTF-8)
  131. }
  132. throw new PhpSpreadsheetException('Unknown codepage: ' . $codePage);
  133. }
  134. }