find-font-iconv.c 4.2 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128
  1. #include <iconv.h>
  2. #include <errno.h>
  3. #include "find-font.h"
  4. struct mac_font_mapping {
  5. unsigned short encoding_id;
  6. unsigned short language_id;
  7. const char *code_page;
  8. };
  9. #define TT_MAC_LANGID_ANY 0xFFFF
  10. static const struct mac_font_mapping mac_codes[] = {
  11. {TT_MAC_ID_ROMAN, TT_MAC_LANGID_ENGLISH, "macintosh"},
  12. {TT_MAC_ID_ROMAN, TT_MAC_LANGID_ICELANDIC, "x-mac-icelandic"},
  13. {TT_MAC_ID_ROMAN, TT_MAC_LANGID_TURKISH, "x-mac-ce"},
  14. {TT_MAC_ID_ROMAN, TT_MAC_LANGID_POLISH, "x-mac-ce"},
  15. {TT_MAC_ID_ROMAN, TT_MAC_LANGID_ROMANIAN, "x-mac-romanian"},
  16. {TT_MAC_ID_ROMAN, TT_MAC_LANGID_CZECH, "x-mac-ce"},
  17. {TT_MAC_ID_ROMAN, TT_MAC_LANGID_SLOVAK, "x-mac-ce"},
  18. {TT_MAC_ID_ROMAN, TT_MAC_LANGID_ANY, "macintosh"},
  19. {TT_MAC_ID_JAPANESE, TT_MAC_LANGID_JAPANESE, "Shift_JIS"},
  20. {TT_MAC_ID_JAPANESE, TT_MAC_LANGID_ANY, "Shift_JIS"},
  21. {TT_MAC_ID_KOREAN, TT_MAC_LANGID_KOREAN, "EUC-KR"},
  22. {TT_MAC_ID_KOREAN, TT_MAC_LANGID_ANY, "EUC-KR"},
  23. {TT_MAC_ID_ARABIC, TT_MAC_LANGID_ARABIC, "x-mac-arabic"},
  24. {TT_MAC_ID_ARABIC, TT_MAC_LANGID_URDU, "x-mac-farsi"},
  25. {TT_MAC_ID_ARABIC, TT_MAC_LANGID_FARSI, "x-mac-farsi"},
  26. {TT_MAC_ID_ARABIC, TT_MAC_LANGID_ANY, "x-mac-arabic"},
  27. {TT_MAC_ID_HEBREW, TT_MAC_LANGID_HEBREW, "x-mac-hebrew"},
  28. {TT_MAC_ID_HEBREW, TT_MAC_LANGID_ANY, "x-mac-hebrew"},
  29. {TT_MAC_ID_GREEK, TT_MAC_LANGID_ANY, "x-mac-greek"},
  30. {TT_MAC_ID_RUSSIAN, TT_MAC_LANGID_ANY, "x-mac-cyrillic"},
  31. {TT_MAC_ID_DEVANAGARI, TT_MAC_LANGID_ANY, "x-mac-devanagari"},
  32. {TT_MAC_ID_GURMUKHI, TT_MAC_LANGID_ANY, "x-mac-gurmukhi"},
  33. {TT_MAC_ID_GUJARATI, TT_MAC_LANGID_ANY, "x-mac-gujarati"},
  34. {TT_MAC_ID_TRADITIONAL_CHINESE, TT_MAC_LANGID_CHINESE_SIMPLIFIED, "Big5"},
  35. {TT_MAC_ID_TRADITIONAL_CHINESE, TT_MAC_LANGID_ANY, "Big5"},
  36. {TT_MAC_ID_SIMPLIFIED_CHINESE, TT_MAC_LANGID_CHINESE_SIMPLIFIED, "GB2312"},
  37. {TT_MAC_ID_SIMPLIFIED_CHINESE, TT_MAC_LANGID_ANY, "GB2312"}};
  38. const char *iso_codes[] = {"us-ascii", NULL, "iso-8859-1"};
  39. const char *ms_codes[] = {"UTF-16BE", "UTF-16BE", "Shift_JIS", NULL, "Big5", NULL, NULL, NULL, NULL, NULL, "UTF-16BE"};
  40. static const size_t mac_code_count = sizeof(mac_codes) / sizeof(mac_codes[0]);
  41. static const size_t iso_code_count = sizeof(iso_codes) / sizeof(iso_codes[0]);
  42. static const size_t ms_code_count = sizeof(ms_codes) / sizeof(ms_codes[0]);
  43. static const char *get_mac_code(uint16_t encoding_id, uint16_t language_id)
  44. {
  45. for (size_t i = 0; i < mac_code_count; i++) {
  46. const struct mac_font_mapping *mac_code = &mac_codes[i];
  47. if (mac_code->encoding_id == encoding_id && mac_code->language_id == language_id)
  48. return mac_code->code_page;
  49. }
  50. return NULL;
  51. }
  52. static const char *get_code_page_for_font(uint16_t platform_id, uint16_t encoding_id, uint16_t language_id)
  53. {
  54. const char *ret;
  55. switch (platform_id) {
  56. case TT_PLATFORM_APPLE_UNICODE:
  57. return "UTF-16BE";
  58. case TT_PLATFORM_MACINTOSH:
  59. ret = get_mac_code(encoding_id, language_id);
  60. if (!ret)
  61. ret = get_mac_code(encoding_id, TT_MAC_LANGID_ANY);
  62. return ret;
  63. case TT_PLATFORM_ISO:
  64. if (encoding_id < iso_code_count)
  65. return iso_codes[encoding_id];
  66. break;
  67. case TT_PLATFORM_MICROSOFT:
  68. if (encoding_id < ms_code_count)
  69. return ms_codes[encoding_id];
  70. break;
  71. }
  72. return NULL;
  73. }
  74. char *sfnt_name_to_utf8(FT_SfntName *sfnt_name)
  75. {
  76. const char *charset =
  77. get_code_page_for_font(sfnt_name->platform_id, sfnt_name->encoding_id, sfnt_name->language_id);
  78. char utf8[256];
  79. char *conv_in, *conv_out;
  80. size_t in_len, out_len;
  81. if (!charset) {
  82. blog(LOG_DEBUG,
  83. "invalid character set found, "
  84. "platform_id: %d, encoding_id: %d, "
  85. "language_id: %d",
  86. sfnt_name->platform_id, sfnt_name->encoding_id, sfnt_name->language_id);
  87. return NULL;
  88. }
  89. iconv_t ic = iconv_open("UTF-8", charset);
  90. if (ic == (iconv_t)-1) {
  91. blog(LOG_DEBUG,
  92. "couldn't intialize font code page "
  93. "conversion: '%s' to 'utf-8': errno = %d",
  94. charset, (int)errno);
  95. return NULL;
  96. }
  97. conv_in = (char *)sfnt_name->string;
  98. conv_out = utf8;
  99. in_len = sfnt_name->string_len;
  100. out_len = 256;
  101. size_t n = iconv(ic, &conv_in, &in_len, &conv_out, &out_len);
  102. if (n == (size_t)-1) {
  103. blog(LOG_WARNING, "couldn't convert font name text: errno = %d", (int)errno);
  104. iconv_close(ic);
  105. return NULL;
  106. }
  107. iconv_close(ic);
  108. *conv_out = 0;
  109. return bstrdup(utf8);
  110. }