unicode.c 53 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272127312741275127612771278127912801281128212831284128512861287128812891290129112921293129412951296129712981299130013011302130313041305130613071308130913101311131213131314131513161317131813191320132113221323132413251326132713281329133013311332133313341335133613371338133913401341134213431344134513461347134813491350135113521353135413551356135713581359136013611362136313641365136613671368136913701371137213731374137513761377137813791380138113821383138413851386138713881389139013911392139313941395139613971398139914001401140214031404140514061407140814091410141114121413
  1. #include <stdio.h>
  2. #include <stdlib.h>
  3. #include <ctype.h>
  4. #include <time.h>
  5. #include <assert.h>
  6. #include "putty.h"
  7. #ifndef MPEXT
  8. #include "terminal.h"
  9. #include "misc.h"
  10. /* Character conversion arrays; they are usually taken from windows,
  11. * the xterm one has the four scanlines that have no unicode 2.0
  12. * equivalents mapped to their unicode 3.0 locations.
  13. */
  14. static const WCHAR unitab_xterm_std[32] = {
  15. 0x2666, 0x2592, 0x2409, 0x240c, 0x240d, 0x240a, 0x00b0, 0x00b1,
  16. 0x2424, 0x240b, 0x2518, 0x2510, 0x250c, 0x2514, 0x253c, 0x23ba,
  17. 0x23bb, 0x2500, 0x23bc, 0x23bd, 0x251c, 0x2524, 0x2534, 0x252c,
  18. 0x2502, 0x2264, 0x2265, 0x03c0, 0x2260, 0x00a3, 0x00b7, 0x0020
  19. };
  20. /*
  21. * If the codepage is non-zero it's a window codepage, zero means use a
  22. * local codepage. The name is always converted to the first of any
  23. * duplicate definitions.
  24. */
  25. /*
  26. * Tables for ISO-8859-{1-10,13-16} derived from those downloaded
  27. * 2001-10-02 from <http://www.unicode.org/Public/MAPPINGS/> -- jtn
  28. * Table for ISO-8859-11 derived from same on 2002-11-18. -- bjh21
  29. */
  30. /* XXX: This could be done algorithmically, but I'm not sure it's
  31. * worth the hassle -- jtn */
  32. /* ISO/IEC 8859-1:1998 (Latin-1, "Western", "West European") */
  33. static const wchar_t iso_8859_1[] = {
  34. 0x00A0, 0x00A1, 0x00A2, 0x00A3, 0x00A4, 0x00A5, 0x00A6, 0x00A7,
  35. 0x00A8, 0x00A9, 0x00AA, 0x00AB, 0x00AC, 0x00AD, 0x00AE, 0x00AF,
  36. 0x00B0, 0x00B1, 0x00B2, 0x00B3, 0x00B4, 0x00B5, 0x00B6, 0x00B7,
  37. 0x00B8, 0x00B9, 0x00BA, 0x00BB, 0x00BC, 0x00BD, 0x00BE, 0x00BF,
  38. 0x00C0, 0x00C1, 0x00C2, 0x00C3, 0x00C4, 0x00C5, 0x00C6, 0x00C7,
  39. 0x00C8, 0x00C9, 0x00CA, 0x00CB, 0x00CC, 0x00CD, 0x00CE, 0x00CF,
  40. 0x00D0, 0x00D1, 0x00D2, 0x00D3, 0x00D4, 0x00D5, 0x00D6, 0x00D7,
  41. 0x00D8, 0x00D9, 0x00DA, 0x00DB, 0x00DC, 0x00DD, 0x00DE, 0x00DF,
  42. 0x00E0, 0x00E1, 0x00E2, 0x00E3, 0x00E4, 0x00E5, 0x00E6, 0x00E7,
  43. 0x00E8, 0x00E9, 0x00EA, 0x00EB, 0x00EC, 0x00ED, 0x00EE, 0x00EF,
  44. 0x00F0, 0x00F1, 0x00F2, 0x00F3, 0x00F4, 0x00F5, 0x00F6, 0x00F7,
  45. 0x00F8, 0x00F9, 0x00FA, 0x00FB, 0x00FC, 0x00FD, 0x00FE, 0x00FF
  46. };
  47. /* ISO/IEC 8859-2:1999 (Latin-2, "Central European", "East European") */
  48. static const wchar_t iso_8859_2[] = {
  49. 0x00A0, 0x0104, 0x02D8, 0x0141, 0x00A4, 0x013D, 0x015A, 0x00A7,
  50. 0x00A8, 0x0160, 0x015E, 0x0164, 0x0179, 0x00AD, 0x017D, 0x017B,
  51. 0x00B0, 0x0105, 0x02DB, 0x0142, 0x00B4, 0x013E, 0x015B, 0x02C7,
  52. 0x00B8, 0x0161, 0x015F, 0x0165, 0x017A, 0x02DD, 0x017E, 0x017C,
  53. 0x0154, 0x00C1, 0x00C2, 0x0102, 0x00C4, 0x0139, 0x0106, 0x00C7,
  54. 0x010C, 0x00C9, 0x0118, 0x00CB, 0x011A, 0x00CD, 0x00CE, 0x010E,
  55. 0x0110, 0x0143, 0x0147, 0x00D3, 0x00D4, 0x0150, 0x00D6, 0x00D7,
  56. 0x0158, 0x016E, 0x00DA, 0x0170, 0x00DC, 0x00DD, 0x0162, 0x00DF,
  57. 0x0155, 0x00E1, 0x00E2, 0x0103, 0x00E4, 0x013A, 0x0107, 0x00E7,
  58. 0x010D, 0x00E9, 0x0119, 0x00EB, 0x011B, 0x00ED, 0x00EE, 0x010F,
  59. 0x0111, 0x0144, 0x0148, 0x00F3, 0x00F4, 0x0151, 0x00F6, 0x00F7,
  60. 0x0159, 0x016F, 0x00FA, 0x0171, 0x00FC, 0x00FD, 0x0163, 0x02D9
  61. };
  62. /* ISO/IEC 8859-3:1999 (Latin-3, "South European", "Maltese & Esperanto") */
  63. static const wchar_t iso_8859_3[] = {
  64. 0x00A0, 0x0126, 0x02D8, 0x00A3, 0x00A4, 0xFFFD, 0x0124, 0x00A7,
  65. 0x00A8, 0x0130, 0x015E, 0x011E, 0x0134, 0x00AD, 0xFFFD, 0x017B,
  66. 0x00B0, 0x0127, 0x00B2, 0x00B3, 0x00B4, 0x00B5, 0x0125, 0x00B7,
  67. 0x00B8, 0x0131, 0x015F, 0x011F, 0x0135, 0x00BD, 0xFFFD, 0x017C,
  68. 0x00C0, 0x00C1, 0x00C2, 0xFFFD, 0x00C4, 0x010A, 0x0108, 0x00C7,
  69. 0x00C8, 0x00C9, 0x00CA, 0x00CB, 0x00CC, 0x00CD, 0x00CE, 0x00CF,
  70. 0xFFFD, 0x00D1, 0x00D2, 0x00D3, 0x00D4, 0x0120, 0x00D6, 0x00D7,
  71. 0x011C, 0x00D9, 0x00DA, 0x00DB, 0x00DC, 0x016C, 0x015C, 0x00DF,
  72. 0x00E0, 0x00E1, 0x00E2, 0xFFFD, 0x00E4, 0x010B, 0x0109, 0x00E7,
  73. 0x00E8, 0x00E9, 0x00EA, 0x00EB, 0x00EC, 0x00ED, 0x00EE, 0x00EF,
  74. 0xFFFD, 0x00F1, 0x00F2, 0x00F3, 0x00F4, 0x0121, 0x00F6, 0x00F7,
  75. 0x011D, 0x00F9, 0x00FA, 0x00FB, 0x00FC, 0x016D, 0x015D, 0x02D9
  76. };
  77. /* ISO/IEC 8859-4:1998 (Latin-4, "North European") */
  78. static const wchar_t iso_8859_4[] = {
  79. 0x00A0, 0x0104, 0x0138, 0x0156, 0x00A4, 0x0128, 0x013B, 0x00A7,
  80. 0x00A8, 0x0160, 0x0112, 0x0122, 0x0166, 0x00AD, 0x017D, 0x00AF,
  81. 0x00B0, 0x0105, 0x02DB, 0x0157, 0x00B4, 0x0129, 0x013C, 0x02C7,
  82. 0x00B8, 0x0161, 0x0113, 0x0123, 0x0167, 0x014A, 0x017E, 0x014B,
  83. 0x0100, 0x00C1, 0x00C2, 0x00C3, 0x00C4, 0x00C5, 0x00C6, 0x012E,
  84. 0x010C, 0x00C9, 0x0118, 0x00CB, 0x0116, 0x00CD, 0x00CE, 0x012A,
  85. 0x0110, 0x0145, 0x014C, 0x0136, 0x00D4, 0x00D5, 0x00D6, 0x00D7,
  86. 0x00D8, 0x0172, 0x00DA, 0x00DB, 0x00DC, 0x0168, 0x016A, 0x00DF,
  87. 0x0101, 0x00E1, 0x00E2, 0x00E3, 0x00E4, 0x00E5, 0x00E6, 0x012F,
  88. 0x010D, 0x00E9, 0x0119, 0x00EB, 0x0117, 0x00ED, 0x00EE, 0x012B,
  89. 0x0111, 0x0146, 0x014D, 0x0137, 0x00F4, 0x00F5, 0x00F6, 0x00F7,
  90. 0x00F8, 0x0173, 0x00FA, 0x00FB, 0x00FC, 0x0169, 0x016B, 0x02D9
  91. };
  92. /* ISO/IEC 8859-5:1999 (Latin/Cyrillic) */
  93. static const wchar_t iso_8859_5[] = {
  94. 0x00A0, 0x0401, 0x0402, 0x0403, 0x0404, 0x0405, 0x0406, 0x0407,
  95. 0x0408, 0x0409, 0x040A, 0x040B, 0x040C, 0x00AD, 0x040E, 0x040F,
  96. 0x0410, 0x0411, 0x0412, 0x0413, 0x0414, 0x0415, 0x0416, 0x0417,
  97. 0x0418, 0x0419, 0x041A, 0x041B, 0x041C, 0x041D, 0x041E, 0x041F,
  98. 0x0420, 0x0421, 0x0422, 0x0423, 0x0424, 0x0425, 0x0426, 0x0427,
  99. 0x0428, 0x0429, 0x042A, 0x042B, 0x042C, 0x042D, 0x042E, 0x042F,
  100. 0x0430, 0x0431, 0x0432, 0x0433, 0x0434, 0x0435, 0x0436, 0x0437,
  101. 0x0438, 0x0439, 0x043A, 0x043B, 0x043C, 0x043D, 0x043E, 0x043F,
  102. 0x0440, 0x0441, 0x0442, 0x0443, 0x0444, 0x0445, 0x0446, 0x0447,
  103. 0x0448, 0x0449, 0x044A, 0x044B, 0x044C, 0x044D, 0x044E, 0x044F,
  104. 0x2116, 0x0451, 0x0452, 0x0453, 0x0454, 0x0455, 0x0456, 0x0457,
  105. 0x0458, 0x0459, 0x045A, 0x045B, 0x045C, 0x00A7, 0x045E, 0x045F
  106. };
  107. /* ISO/IEC 8859-6:1999 (Latin/Arabic) */
  108. static const wchar_t iso_8859_6[] = {
  109. 0x00A0, 0xFFFD, 0xFFFD, 0xFFFD, 0x00A4, 0xFFFD, 0xFFFD, 0xFFFD,
  110. 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0x060C, 0x00AD, 0xFFFD, 0xFFFD,
  111. 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD,
  112. 0xFFFD, 0xFFFD, 0xFFFD, 0x061B, 0xFFFD, 0xFFFD, 0xFFFD, 0x061F,
  113. 0xFFFD, 0x0621, 0x0622, 0x0623, 0x0624, 0x0625, 0x0626, 0x0627,
  114. 0x0628, 0x0629, 0x062A, 0x062B, 0x062C, 0x062D, 0x062E, 0x062F,
  115. 0x0630, 0x0631, 0x0632, 0x0633, 0x0634, 0x0635, 0x0636, 0x0637,
  116. 0x0638, 0x0639, 0x063A, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD,
  117. 0x0640, 0x0641, 0x0642, 0x0643, 0x0644, 0x0645, 0x0646, 0x0647,
  118. 0x0648, 0x0649, 0x064A, 0x064B, 0x064C, 0x064D, 0x064E, 0x064F,
  119. 0x0650, 0x0651, 0x0652, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD,
  120. 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD
  121. };
  122. /* ISO 8859-7:1987 (Latin/Greek) */
  123. static const wchar_t iso_8859_7[] = {
  124. 0x00A0, 0x2018, 0x2019, 0x00A3, 0xFFFD, 0xFFFD, 0x00A6, 0x00A7,
  125. 0x00A8, 0x00A9, 0xFFFD, 0x00AB, 0x00AC, 0x00AD, 0xFFFD, 0x2015,
  126. 0x00B0, 0x00B1, 0x00B2, 0x00B3, 0x0384, 0x0385, 0x0386, 0x00B7,
  127. 0x0388, 0x0389, 0x038A, 0x00BB, 0x038C, 0x00BD, 0x038E, 0x038F,
  128. 0x0390, 0x0391, 0x0392, 0x0393, 0x0394, 0x0395, 0x0396, 0x0397,
  129. 0x0398, 0x0399, 0x039A, 0x039B, 0x039C, 0x039D, 0x039E, 0x039F,
  130. 0x03A0, 0x03A1, 0xFFFD, 0x03A3, 0x03A4, 0x03A5, 0x03A6, 0x03A7,
  131. 0x03A8, 0x03A9, 0x03AA, 0x03AB, 0x03AC, 0x03AD, 0x03AE, 0x03AF,
  132. 0x03B0, 0x03B1, 0x03B2, 0x03B3, 0x03B4, 0x03B5, 0x03B6, 0x03B7,
  133. 0x03B8, 0x03B9, 0x03BA, 0x03BB, 0x03BC, 0x03BD, 0x03BE, 0x03BF,
  134. 0x03C0, 0x03C1, 0x03C2, 0x03C3, 0x03C4, 0x03C5, 0x03C6, 0x03C7,
  135. 0x03C8, 0x03C9, 0x03CA, 0x03CB, 0x03CC, 0x03CD, 0x03CE, 0xFFFD
  136. };
  137. /* ISO/IEC 8859-8:1999 (Latin/Hebrew) */
  138. static const wchar_t iso_8859_8[] = {
  139. 0x00A0, 0xFFFD, 0x00A2, 0x00A3, 0x00A4, 0x00A5, 0x00A6, 0x00A7,
  140. 0x00A8, 0x00A9, 0x00D7, 0x00AB, 0x00AC, 0x00AD, 0x00AE, 0x00AF,
  141. 0x00B0, 0x00B1, 0x00B2, 0x00B3, 0x00B4, 0x00B5, 0x00B6, 0x00B7,
  142. 0x00B8, 0x00B9, 0x00F7, 0x00BB, 0x00BC, 0x00BD, 0x00BE, 0xFFFD,
  143. 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD,
  144. 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD,
  145. 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD,
  146. 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0x2017,
  147. 0x05D0, 0x05D1, 0x05D2, 0x05D3, 0x05D4, 0x05D5, 0x05D6, 0x05D7,
  148. 0x05D8, 0x05D9, 0x05DA, 0x05DB, 0x05DC, 0x05DD, 0x05DE, 0x05DF,
  149. 0x05E0, 0x05E1, 0x05E2, 0x05E3, 0x05E4, 0x05E5, 0x05E6, 0x05E7,
  150. 0x05E8, 0x05E9, 0x05EA, 0xFFFD, 0xFFFD, 0x200E, 0x200F, 0xFFFD
  151. };
  152. /* ISO/IEC 8859-9:1999 (Latin-5, "Turkish") */
  153. static const wchar_t iso_8859_9[] = {
  154. 0x00A0, 0x00A1, 0x00A2, 0x00A3, 0x00A4, 0x00A5, 0x00A6, 0x00A7,
  155. 0x00A8, 0x00A9, 0x00AA, 0x00AB, 0x00AC, 0x00AD, 0x00AE, 0x00AF,
  156. 0x00B0, 0x00B1, 0x00B2, 0x00B3, 0x00B4, 0x00B5, 0x00B6, 0x00B7,
  157. 0x00B8, 0x00B9, 0x00BA, 0x00BB, 0x00BC, 0x00BD, 0x00BE, 0x00BF,
  158. 0x00C0, 0x00C1, 0x00C2, 0x00C3, 0x00C4, 0x00C5, 0x00C6, 0x00C7,
  159. 0x00C8, 0x00C9, 0x00CA, 0x00CB, 0x00CC, 0x00CD, 0x00CE, 0x00CF,
  160. 0x011E, 0x00D1, 0x00D2, 0x00D3, 0x00D4, 0x00D5, 0x00D6, 0x00D7,
  161. 0x00D8, 0x00D9, 0x00DA, 0x00DB, 0x00DC, 0x0130, 0x015E, 0x00DF,
  162. 0x00E0, 0x00E1, 0x00E2, 0x00E3, 0x00E4, 0x00E5, 0x00E6, 0x00E7,
  163. 0x00E8, 0x00E9, 0x00EA, 0x00EB, 0x00EC, 0x00ED, 0x00EE, 0x00EF,
  164. 0x011F, 0x00F1, 0x00F2, 0x00F3, 0x00F4, 0x00F5, 0x00F6, 0x00F7,
  165. 0x00F8, 0x00F9, 0x00FA, 0x00FB, 0x00FC, 0x0131, 0x015F, 0x00FF
  166. };
  167. /* ISO/IEC 8859-10:1998 (Latin-6, "Nordic" [Sami, Inuit, Icelandic]) */
  168. static const wchar_t iso_8859_10[] = {
  169. 0x00A0, 0x0104, 0x0112, 0x0122, 0x012A, 0x0128, 0x0136, 0x00A7,
  170. 0x013B, 0x0110, 0x0160, 0x0166, 0x017D, 0x00AD, 0x016A, 0x014A,
  171. 0x00B0, 0x0105, 0x0113, 0x0123, 0x012B, 0x0129, 0x0137, 0x00B7,
  172. 0x013C, 0x0111, 0x0161, 0x0167, 0x017E, 0x2015, 0x016B, 0x014B,
  173. 0x0100, 0x00C1, 0x00C2, 0x00C3, 0x00C4, 0x00C5, 0x00C6, 0x012E,
  174. 0x010C, 0x00C9, 0x0118, 0x00CB, 0x0116, 0x00CD, 0x00CE, 0x00CF,
  175. 0x00D0, 0x0145, 0x014C, 0x00D3, 0x00D4, 0x00D5, 0x00D6, 0x0168,
  176. 0x00D8, 0x0172, 0x00DA, 0x00DB, 0x00DC, 0x00DD, 0x00DE, 0x00DF,
  177. 0x0101, 0x00E1, 0x00E2, 0x00E3, 0x00E4, 0x00E5, 0x00E6, 0x012F,
  178. 0x010D, 0x00E9, 0x0119, 0x00EB, 0x0117, 0x00ED, 0x00EE, 0x00EF,
  179. 0x00F0, 0x0146, 0x014D, 0x00F3, 0x00F4, 0x00F5, 0x00F6, 0x0169,
  180. 0x00F8, 0x0173, 0x00FA, 0x00FB, 0x00FC, 0x00FD, 0x00FE, 0x0138
  181. };
  182. /* ISO/IEC 8859-11:2001 ("Thai", "TIS620") */
  183. static const wchar_t iso_8859_11[] = {
  184. 0x00A0, 0x0E01, 0x0E02, 0x0E03, 0x0E04, 0x0E05, 0x0E06, 0x0E07,
  185. 0x0E08, 0x0E09, 0x0E0A, 0x0E0B, 0x0E0C, 0x0E0D, 0x0E0E, 0x0E0F,
  186. 0x0E10, 0x0E11, 0x0E12, 0x0E13, 0x0E14, 0x0E15, 0x0E16, 0x0E17,
  187. 0x0E18, 0x0E19, 0x0E1A, 0x0E1B, 0x0E1C, 0x0E1D, 0x0E1E, 0x0E1F,
  188. 0x0E20, 0x0E21, 0x0E22, 0x0E23, 0x0E24, 0x0E25, 0x0E26, 0x0E27,
  189. 0x0E28, 0x0E29, 0x0E2A, 0x0E2B, 0x0E2C, 0x0E2D, 0x0E2E, 0x0E2F,
  190. 0x0E30, 0x0E31, 0x0E32, 0x0E33, 0x0E34, 0x0E35, 0x0E36, 0x0E37,
  191. 0x0E38, 0x0E39, 0x0E3A, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0x0E3F,
  192. 0x0E40, 0x0E41, 0x0E42, 0x0E43, 0x0E44, 0x0E45, 0x0E46, 0x0E47,
  193. 0x0E48, 0x0E49, 0x0E4A, 0x0E4B, 0x0E4C, 0x0E4D, 0x0E4E, 0x0E4F,
  194. 0x0E50, 0x0E51, 0x0E52, 0x0E53, 0x0E54, 0x0E55, 0x0E56, 0x0E57,
  195. 0x0E58, 0x0E59, 0x0E5A, 0x0E5B, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD
  196. };
  197. /* ISO/IEC 8859-13:1998 (Latin-7, "Baltic Rim") */
  198. static const wchar_t iso_8859_13[] = {
  199. 0x00A0, 0x201D, 0x00A2, 0x00A3, 0x00A4, 0x201E, 0x00A6, 0x00A7,
  200. 0x00D8, 0x00A9, 0x0156, 0x00AB, 0x00AC, 0x00AD, 0x00AE, 0x00C6,
  201. 0x00B0, 0x00B1, 0x00B2, 0x00B3, 0x201C, 0x00B5, 0x00B6, 0x00B7,
  202. 0x00F8, 0x00B9, 0x0157, 0x00BB, 0x00BC, 0x00BD, 0x00BE, 0x00E6,
  203. 0x0104, 0x012E, 0x0100, 0x0106, 0x00C4, 0x00C5, 0x0118, 0x0112,
  204. 0x010C, 0x00C9, 0x0179, 0x0116, 0x0122, 0x0136, 0x012A, 0x013B,
  205. 0x0160, 0x0143, 0x0145, 0x00D3, 0x014C, 0x00D5, 0x00D6, 0x00D7,
  206. 0x0172, 0x0141, 0x015A, 0x016A, 0x00DC, 0x017B, 0x017D, 0x00DF,
  207. 0x0105, 0x012F, 0x0101, 0x0107, 0x00E4, 0x00E5, 0x0119, 0x0113,
  208. 0x010D, 0x00E9, 0x017A, 0x0117, 0x0123, 0x0137, 0x012B, 0x013C,
  209. 0x0161, 0x0144, 0x0146, 0x00F3, 0x014D, 0x00F5, 0x00F6, 0x00F7,
  210. 0x0173, 0x0142, 0x015B, 0x016B, 0x00FC, 0x017C, 0x017E, 0x2019
  211. };
  212. /* ISO/IEC 8859-14:1998 (Latin-8, "Celtic", "Gaelic/Welsh") */
  213. static const wchar_t iso_8859_14[] = {
  214. 0x00A0, 0x1E02, 0x1E03, 0x00A3, 0x010A, 0x010B, 0x1E0A, 0x00A7,
  215. 0x1E80, 0x00A9, 0x1E82, 0x1E0B, 0x1EF2, 0x00AD, 0x00AE, 0x0178,
  216. 0x1E1E, 0x1E1F, 0x0120, 0x0121, 0x1E40, 0x1E41, 0x00B6, 0x1E56,
  217. 0x1E81, 0x1E57, 0x1E83, 0x1E60, 0x1EF3, 0x1E84, 0x1E85, 0x1E61,
  218. 0x00C0, 0x00C1, 0x00C2, 0x00C3, 0x00C4, 0x00C5, 0x00C6, 0x00C7,
  219. 0x00C8, 0x00C9, 0x00CA, 0x00CB, 0x00CC, 0x00CD, 0x00CE, 0x00CF,
  220. 0x0174, 0x00D1, 0x00D2, 0x00D3, 0x00D4, 0x00D5, 0x00D6, 0x1E6A,
  221. 0x00D8, 0x00D9, 0x00DA, 0x00DB, 0x00DC, 0x00DD, 0x0176, 0x00DF,
  222. 0x00E0, 0x00E1, 0x00E2, 0x00E3, 0x00E4, 0x00E5, 0x00E6, 0x00E7,
  223. 0x00E8, 0x00E9, 0x00EA, 0x00EB, 0x00EC, 0x00ED, 0x00EE, 0x00EF,
  224. 0x0175, 0x00F1, 0x00F2, 0x00F3, 0x00F4, 0x00F5, 0x00F6, 0x1E6B,
  225. 0x00F8, 0x00F9, 0x00FA, 0x00FB, 0x00FC, 0x00FD, 0x0177, 0x00FF
  226. };
  227. /* ISO/IEC 8859-15:1999 (Latin-9 aka -0, "euro") */
  228. static const wchar_t iso_8859_15[] = {
  229. 0x00A0, 0x00A1, 0x00A2, 0x00A3, 0x20AC, 0x00A5, 0x0160, 0x00A7,
  230. 0x0161, 0x00A9, 0x00AA, 0x00AB, 0x00AC, 0x00AD, 0x00AE, 0x00AF,
  231. 0x00B0, 0x00B1, 0x00B2, 0x00B3, 0x017D, 0x00B5, 0x00B6, 0x00B7,
  232. 0x017E, 0x00B9, 0x00BA, 0x00BB, 0x0152, 0x0153, 0x0178, 0x00BF,
  233. 0x00C0, 0x00C1, 0x00C2, 0x00C3, 0x00C4, 0x00C5, 0x00C6, 0x00C7,
  234. 0x00C8, 0x00C9, 0x00CA, 0x00CB, 0x00CC, 0x00CD, 0x00CE, 0x00CF,
  235. 0x00D0, 0x00D1, 0x00D2, 0x00D3, 0x00D4, 0x00D5, 0x00D6, 0x00D7,
  236. 0x00D8, 0x00D9, 0x00DA, 0x00DB, 0x00DC, 0x00DD, 0x00DE, 0x00DF,
  237. 0x00E0, 0x00E1, 0x00E2, 0x00E3, 0x00E4, 0x00E5, 0x00E6, 0x00E7,
  238. 0x00E8, 0x00E9, 0x00EA, 0x00EB, 0x00EC, 0x00ED, 0x00EE, 0x00EF,
  239. 0x00F0, 0x00F1, 0x00F2, 0x00F3, 0x00F4, 0x00F5, 0x00F6, 0x00F7,
  240. 0x00F8, 0x00F9, 0x00FA, 0x00FB, 0x00FC, 0x00FD, 0x00FE, 0x00FF
  241. };
  242. /* ISO/IEC 8859-16:2001 (Latin-10, "Balkan") */
  243. static const wchar_t iso_8859_16[] = {
  244. 0x00A0, 0x0104, 0x0105, 0x0141, 0x20AC, 0x201E, 0x0160, 0x00A7,
  245. 0x0161, 0x00A9, 0x0218, 0x00AB, 0x0179, 0x00AD, 0x017A, 0x017B,
  246. 0x00B0, 0x00B1, 0x010C, 0x0142, 0x017D, 0x201D, 0x00B6, 0x00B7,
  247. 0x017E, 0x010D, 0x0219, 0x00BB, 0x0152, 0x0153, 0x0178, 0x017C,
  248. 0x00C0, 0x00C1, 0x00C2, 0x0102, 0x00C4, 0x0106, 0x00C6, 0x00C7,
  249. 0x00C8, 0x00C9, 0x00CA, 0x00CB, 0x00CC, 0x00CD, 0x00CE, 0x00CF,
  250. 0x0110, 0x0143, 0x00D2, 0x00D3, 0x00D4, 0x0150, 0x00D6, 0x015A,
  251. 0x0170, 0x00D9, 0x00DA, 0x00DB, 0x00DC, 0x0118, 0x021A, 0x00DF,
  252. 0x00E0, 0x00E1, 0x00E2, 0x0103, 0x00E4, 0x0107, 0x00E6, 0x00E7,
  253. 0x00E8, 0x00E9, 0x00EA, 0x00EB, 0x00EC, 0x00ED, 0x00EE, 0x00EF,
  254. 0x0111, 0x0144, 0x00F2, 0x00F3, 0x00F4, 0x0151, 0x00F6, 0x015B,
  255. 0x0171, 0x00F9, 0x00FA, 0x00FB, 0x00FC, 0x0119, 0x021B, 0x00FF
  256. };
  257. static const wchar_t roman8[] = {
  258. 0x00A0, 0x00C0, 0x00C2, 0x00C8, 0x00CA, 0x00CB, 0x00CE, 0x00CF,
  259. 0x00B4, 0x02CB, 0x02C6, 0x00A8, 0x02DC, 0x00D9, 0x00DB, 0x20A4,
  260. 0x00AF, 0x00DD, 0x00FD, 0x00B0, 0x00C7, 0x00E7, 0x00D1, 0x00F1,
  261. 0x00A1, 0x00BF, 0x00A4, 0x00A3, 0x00A5, 0x00A7, 0x0192, 0x00A2,
  262. 0x00E2, 0x00EA, 0x00F4, 0x00FB, 0x00E1, 0x00E9, 0x00F3, 0x00FA,
  263. 0x00E0, 0x00E8, 0x00F2, 0x00F9, 0x00E4, 0x00EB, 0x00F6, 0x00FC,
  264. 0x00C5, 0x00EE, 0x00D8, 0x00C6, 0x00E5, 0x00ED, 0x00F8, 0x00E6,
  265. 0x00C4, 0x00EC, 0x00D6, 0x00DC, 0x00C9, 0x00EF, 0x00DF, 0x00D4,
  266. 0x00C1, 0x00C3, 0x00E3, 0x00D0, 0x00F0, 0x00CD, 0x00CC, 0x00D3,
  267. 0x00D2, 0x00D5, 0x00F5, 0x0160, 0x0161, 0x00DA, 0x0178, 0x00FF,
  268. 0x00DE, 0x00FE, 0x00B7, 0x00B5, 0x00B6, 0x00BE, 0x2014, 0x00BC,
  269. 0x00BD, 0x00AA, 0x00BA, 0x00AB, 0x25A0, 0x00BB, 0x00B1, 0xFFFD
  270. };
  271. static const wchar_t koi8_u[] = {
  272. 0x2500, 0x2502, 0x250C, 0x2510, 0x2514, 0x2518, 0x251C, 0x2524,
  273. 0x252C, 0x2534, 0x253C, 0x2580, 0x2584, 0x2588, 0x258C, 0x2590,
  274. 0x2591, 0x2592, 0x2593, 0x2320, 0x25A0, 0x2022, 0x221A, 0x2248,
  275. 0x2264, 0x2265, 0x00A0, 0x2321, 0x00B0, 0x00B2, 0x00B7, 0x00F7,
  276. 0x2550, 0x2551, 0x2552, 0x0451, 0x0454, 0x2554, 0x0456, 0x0457,
  277. 0x2557, 0x2558, 0x2559, 0x255A, 0x255B, 0x0491, 0x255D, 0x255E,
  278. 0x255F, 0x2560, 0x2561, 0x0401, 0x0404, 0x2563, 0x0406, 0x0407,
  279. 0x2566, 0x2567, 0x2568, 0x2569, 0x256A, 0x0490, 0x256C, 0x00A9,
  280. 0x044E, 0x0430, 0x0431, 0x0446, 0x0434, 0x0435, 0x0444, 0x0433,
  281. 0x0445, 0x0438, 0x0439, 0x043A, 0x043B, 0x043C, 0x043D, 0x043E,
  282. 0x043F, 0x044F, 0x0440, 0x0441, 0x0442, 0x0443, 0x0436, 0x0432,
  283. 0x044C, 0x044B, 0x0437, 0x0448, 0x044D, 0x0449, 0x0447, 0x044A,
  284. 0x042E, 0x0410, 0x0411, 0x0426, 0x0414, 0x0415, 0x0424, 0x0413,
  285. 0x0425, 0x0418, 0x0419, 0x041A, 0x041B, 0x041C, 0x041D, 0x041E,
  286. 0x041F, 0x042F, 0x0420, 0x0421, 0x0422, 0x0423, 0x0416, 0x0412,
  287. 0x042C, 0x042B, 0x0417, 0x0428, 0x042D, 0x0429, 0x0427, 0x042A
  288. };
  289. static const wchar_t vscii[] = {
  290. 0x0000, 0x0001, 0x1EB2, 0x0003, 0x0004, 0x1EB4, 0x1EAA, 0x0007,
  291. 0x0008, 0x0009, 0x000a, 0x000b, 0x000c, 0x000d, 0x000e, 0x000f,
  292. 0x0010, 0x0011, 0x0012, 0x0013, 0x1EF6, 0x0015, 0x0016, 0x0017,
  293. 0x0018, 0x1EF8, 0x001a, 0x001b, 0x001c, 0x001d, 0x1EF4, 0x001f,
  294. 0x0020, 0x0021, 0x0022, 0x0023, 0x0024, 0x0025, 0x0026, 0x0027,
  295. 0x0028, 0x0029, 0x002A, 0x002B, 0x002C, 0x002D, 0x002E, 0x002F,
  296. 0x0030, 0x0031, 0x0032, 0x0033, 0x0034, 0x0035, 0x0036, 0x0037,
  297. 0x0038, 0x0039, 0x003A, 0x003B, 0x003C, 0x003D, 0x003E, 0x003F,
  298. 0x0040, 0x0041, 0x0042, 0x0043, 0x0044, 0x0045, 0x0046, 0x0047,
  299. 0x0048, 0x0049, 0x004A, 0x004B, 0x004C, 0x004D, 0x004E, 0x004F,
  300. 0x0050, 0x0051, 0x0052, 0x0053, 0x0054, 0x0055, 0x0056, 0x0057,
  301. 0x0058, 0x0059, 0x005A, 0x005B, 0x005C, 0x005D, 0x005E, 0x005F,
  302. 0x0060, 0x0061, 0x0062, 0x0063, 0x0064, 0x0065, 0x0066, 0x0067,
  303. 0x0068, 0x0069, 0x006A, 0x006B, 0x006C, 0x006D, 0x006E, 0x006F,
  304. 0x0070, 0x0071, 0x0072, 0x0073, 0x0074, 0x0075, 0x0076, 0x0077,
  305. 0x0078, 0x0079, 0x007A, 0x007B, 0x007C, 0x007D, 0x007E, 0x007f,
  306. 0x1EA0, 0x1EAE, 0x1EB0, 0x1EB6, 0x1EA4, 0x1EA6, 0x1EA8, 0x1EAC,
  307. 0x1EBC, 0x1EB8, 0x1EBE, 0x1EC0, 0x1EC2, 0x1EC4, 0x1EC6, 0x1ED0,
  308. 0x1ED2, 0x1ED4, 0x1ED6, 0x1ED8, 0x1EE2, 0x1EDA, 0x1EDC, 0x1EDE,
  309. 0x1ECA, 0x1ECE, 0x1ECC, 0x1EC8, 0x1EE6, 0x0168, 0x1EE4, 0x1EF2,
  310. 0x00D5, 0x1EAF, 0x1EB1, 0x1EB7, 0x1EA5, 0x1EA7, 0x1EA8, 0x1EAD,
  311. 0x1EBD, 0x1EB9, 0x1EBF, 0x1EC1, 0x1EC3, 0x1EC5, 0x1EC7, 0x1ED1,
  312. 0x1ED3, 0x1ED5, 0x1ED7, 0x1EE0, 0x01A0, 0x1ED9, 0x1EDD, 0x1EDF,
  313. 0x1ECB, 0x1EF0, 0x1EE8, 0x1EEA, 0x1EEC, 0x01A1, 0x1EDB, 0x01AF,
  314. 0x00C0, 0x00C1, 0x00C2, 0x00C3, 0x1EA2, 0x0102, 0x1EB3, 0x1EB5,
  315. 0x00C8, 0x00C9, 0x00CA, 0x1EBA, 0x00CC, 0x00CD, 0x0128, 0x1EF3,
  316. 0x0110, 0x1EE9, 0x00D2, 0x00D3, 0x00D4, 0x1EA1, 0x1EF7, 0x1EEB,
  317. 0x1EED, 0x00D9, 0x00DA, 0x1EF9, 0x1EF5, 0x00DD, 0x1EE1, 0x01B0,
  318. 0x00E0, 0x00E1, 0x00E2, 0x00E3, 0x1EA3, 0x0103, 0x1EEF, 0x1EAB,
  319. 0x00E8, 0x00E9, 0x00EA, 0x1EBB, 0x00EC, 0x00ED, 0x0129, 0x1EC9,
  320. 0x0111, 0x1EF1, 0x00F2, 0x00F3, 0x00F4, 0x00F5, 0x1ECF, 0x1ECD,
  321. 0x1EE5, 0x00F9, 0x00FA, 0x0169, 0x1EE7, 0x00FD, 0x1EE3, 0x1EEE
  322. };
  323. static const wchar_t dec_mcs[] = {
  324. 0x00A0, 0x00A1, 0x00A2, 0x00A3, 0xFFFD, 0x00A5, 0xFFFD, 0x00A7,
  325. 0x00A4, 0x00A9, 0x00AA, 0x00AB, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD,
  326. 0x00B0, 0x00B1, 0x00B2, 0x00B3, 0xFFFD, 0x00B5, 0x00B6, 0x00B7,
  327. 0xFFFD, 0x00B9, 0x00BA, 0x00BB, 0x00BC, 0x00BD, 0xFFFD, 0x00BF,
  328. 0x00C0, 0x00C1, 0x00C2, 0x00C3, 0x00C4, 0x00C5, 0x00C6, 0x00C7,
  329. 0x00C8, 0x00C9, 0x00CA, 0x00CB, 0x00CC, 0x00CD, 0x00CE, 0x00CF,
  330. 0xFFFD, 0x00D1, 0x00D2, 0x00D3, 0x00D4, 0x00D5, 0x00D6, 0x0152,
  331. 0x00D8, 0x00D9, 0x00DA, 0x00DB, 0x00DC, 0x0178, 0xFFFD, 0x00DF,
  332. 0x00E0, 0x00E1, 0x00E2, 0x00E3, 0x00E4, 0x00E5, 0x00E6, 0x00E7,
  333. 0x00E8, 0x00E9, 0x00EA, 0x00EB, 0x00EC, 0x00ED, 0x00EE, 0x00EF,
  334. 0xFFFD, 0x00F1, 0x00F2, 0x00F3, 0x00F4, 0x00F5, 0x00F6, 0x0153,
  335. 0x00F8, 0x00F9, 0x00FA, 0x00FB, 0x00FC, 0x00FF, 0xFFFD, 0xFFFD
  336. };
  337. /* Mazovia (Polish) aka CP620
  338. * from "Mazowia to Unicode table", 04/24/96, Mikolaj Jedrzejak */
  339. static const wchar_t mazovia[] = {
  340. /* Code point 0x9B is "zloty" symbol (z&#0142;), which is not
  341. * widely used and for which there is no Unicode equivalent.
  342. * One reference shows 0xA8 as U+00A7 SECTION SIGN, but we're
  343. * told that's incorrect. */
  344. 0x00C7, 0x00FC, 0x00E9, 0x00E2, 0x00E4, 0x00E0, 0x0105, 0x00E7,
  345. 0x00EA, 0x00EB, 0x00E8, 0x00EF, 0x00EE, 0x0107, 0x00C4, 0x0104,
  346. 0x0118, 0x0119, 0x0142, 0x00F4, 0x00F6, 0x0106, 0x00FB, 0x00F9,
  347. 0x015a, 0x00D6, 0x00DC, 0xFFFD, 0x0141, 0x00A5, 0x015b, 0x0192,
  348. 0x0179, 0x017b, 0x00F3, 0x00d3, 0x0144, 0x0143, 0x017a, 0x017c,
  349. 0x00BF, 0x2310, 0x00AC, 0x00BD, 0x00BC, 0x00A1, 0x00AB, 0x00BB,
  350. 0x2591, 0x2592, 0x2593, 0x2502, 0x2524, 0x2561, 0x2562, 0x2556,
  351. 0x2555, 0x2563, 0x2551, 0x2557, 0x255D, 0x255C, 0x255B, 0x2510,
  352. 0x2514, 0x2534, 0x252C, 0x251C, 0x2500, 0x253C, 0x255E, 0x255F,
  353. 0x255A, 0x2554, 0x2569, 0x2566, 0x2560, 0x2550, 0x256C, 0x2567,
  354. 0x2568, 0x2564, 0x2565, 0x2559, 0x2558, 0x2552, 0x2553, 0x256B,
  355. 0x256A, 0x2518, 0x250C, 0x2588, 0x2584, 0x258C, 0x2590, 0x2580,
  356. 0x03B1, 0x00DF, 0x0393, 0x03C0, 0x03A3, 0x03C3, 0x00B5, 0x03C4,
  357. 0x03A6, 0x0398, 0x03A9, 0x03B4, 0x221E, 0x03C6, 0x03B5, 0x2229,
  358. 0x2261, 0x00B1, 0x2265, 0x2264, 0x2320, 0x2321, 0x00F7, 0x2248,
  359. 0x00B0, 0x2219, 0x00B7, 0x221A, 0x207F, 0x00B2, 0x25A0, 0x00A0
  360. };
  361. struct cp_list_item {
  362. char *name;
  363. int codepage;
  364. int cp_size;
  365. const wchar_t *cp_table;
  366. };
  367. static const struct cp_list_item cp_list[] = {
  368. {"UTF-8", CP_UTF8},
  369. {"ISO-8859-1:1998 (Latin-1, West Europe)", 0, 96, iso_8859_1},
  370. {"ISO-8859-2:1999 (Latin-2, East Europe)", 0, 96, iso_8859_2},
  371. {"ISO-8859-3:1999 (Latin-3, South Europe)", 0, 96, iso_8859_3},
  372. {"ISO-8859-4:1998 (Latin-4, North Europe)", 0, 96, iso_8859_4},
  373. {"ISO-8859-5:1999 (Latin/Cyrillic)", 0, 96, iso_8859_5},
  374. {"ISO-8859-6:1999 (Latin/Arabic)", 0, 96, iso_8859_6},
  375. {"ISO-8859-7:1987 (Latin/Greek)", 0, 96, iso_8859_7},
  376. {"ISO-8859-8:1999 (Latin/Hebrew)", 0, 96, iso_8859_8},
  377. {"ISO-8859-9:1999 (Latin-5, Turkish)", 0, 96, iso_8859_9},
  378. {"ISO-8859-10:1998 (Latin-6, Nordic)", 0, 96, iso_8859_10},
  379. {"ISO-8859-11:2001 (Latin/Thai)", 0, 96, iso_8859_11},
  380. {"ISO-8859-13:1998 (Latin-7, Baltic)", 0, 96, iso_8859_13},
  381. {"ISO-8859-14:1998 (Latin-8, Celtic)", 0, 96, iso_8859_14},
  382. {"ISO-8859-15:1999 (Latin-9, \"euro\")", 0, 96, iso_8859_15},
  383. {"ISO-8859-16:2001 (Latin-10, Balkan)", 0, 96, iso_8859_16},
  384. {"KOI8-U", 0, 128, koi8_u},
  385. {"KOI8-R", 20866},
  386. {"HP-ROMAN8", 0, 96, roman8},
  387. {"VSCII", 0, 256, vscii},
  388. {"DEC-MCS", 0, 96, dec_mcs},
  389. {"Win1250 (Central European)", 1250},
  390. {"Win1251 (Cyrillic)", 1251},
  391. {"Win1252 (Western)", 1252},
  392. {"Win1253 (Greek)", 1253},
  393. {"Win1254 (Turkish)", 1254},
  394. {"Win1255 (Hebrew)", 1255},
  395. {"Win1256 (Arabic)", 1256},
  396. {"Win1257 (Baltic)", 1257},
  397. {"Win1258 (Vietnamese)", 1258},
  398. {"CP437", 437},
  399. {"CP620 (Mazovia)", 0, 128, mazovia},
  400. {"CP819", 28591},
  401. {"CP852", 852},
  402. {"CP878", 20866},
  403. {"Use font encoding", -1},
  404. {0, 0}
  405. };
  406. static void link_font(WCHAR *line_tbl, WCHAR *font_tbl, WCHAR attr);
  407. /*
  408. * We keep a collection of reverse mappings from Unicode back to code pages,
  409. * in the form of array[256] of array[256] of char. These live forever in a
  410. * local tree234, and we just make a new one whenever we find a need.
  411. */
  412. typedef struct reverse_mapping {
  413. int codepage;
  414. char **blocks;
  415. } reverse_mapping;
  416. static tree234 *reverse_mappings = NULL;
  417. static int reverse_mapping_cmp(void *av, void *bv)
  418. {
  419. const reverse_mapping *a = (const reverse_mapping *)av;
  420. const reverse_mapping *b = (const reverse_mapping *)bv;
  421. if (a->codepage < b->codepage)
  422. return -1;
  423. if (a->codepage > b->codepage)
  424. return +1;
  425. return 0;
  426. }
  427. static int reverse_mapping_find(void *av, void *bv)
  428. {
  429. const reverse_mapping *a = (const reverse_mapping *)av;
  430. int b_codepage = *(const int *)bv;
  431. if (a->codepage < b_codepage)
  432. return -1;
  433. if (a->codepage > b_codepage)
  434. return +1;
  435. return 0;
  436. }
  437. static reverse_mapping *get_existing_reverse_mapping(int codepage)
  438. {
  439. if (!reverse_mappings)
  440. return NULL;
  441. return find234(reverse_mappings, &codepage, reverse_mapping_find);
  442. }
  443. static reverse_mapping *make_reverse_mapping_inner(
  444. int codepage, const wchar_t *mapping)
  445. {
  446. if (!reverse_mappings)
  447. reverse_mappings = newtree234(reverse_mapping_cmp);
  448. reverse_mapping *rmap = snew(reverse_mapping);
  449. rmap->blocks = snewn(256, char *);
  450. memset(rmap->blocks, 0, 256 * sizeof(char *));
  451. for (size_t i = 0; i < 256; i++) {
  452. /* These special kinds of value correspond to no Unicode character */
  453. if (DIRECT_CHAR(mapping[i]))
  454. continue;
  455. if (DIRECT_FONT(mapping[i]))
  456. continue;
  457. size_t chr = mapping[i];
  458. size_t block = chr >> 8, index = chr & 0xFF;
  459. if (!rmap->blocks[block]) {
  460. rmap->blocks[block] = snewn(256, char);
  461. memset(rmap->blocks[block], 0, 256);
  462. }
  463. rmap->blocks[block][index] = i;
  464. }
  465. rmap->codepage = codepage;
  466. reverse_mapping *added = add234(reverse_mappings, rmap);
  467. assert(added == rmap); /* we already checked it wasn't already in there */
  468. return added;
  469. }
  470. static void make_reverse_mapping(int codepage, const wchar_t *mapping)
  471. {
  472. if (get_existing_reverse_mapping(codepage))
  473. return; /* we've already got this one */
  474. make_reverse_mapping_inner(codepage, mapping);
  475. }
  476. static reverse_mapping *get_reverse_mapping(int codepage)
  477. {
  478. /*
  479. * Try harder to get a reverse mapping for a codepage we implement
  480. * internally via a translation table, by hastily making it if it doesn't
  481. * already exist.
  482. */
  483. reverse_mapping *rmap = get_existing_reverse_mapping(codepage);
  484. if (rmap)
  485. return rmap;
  486. if (codepage < 65536)
  487. return NULL;
  488. if (codepage >= 65536 + lenof(cp_list))
  489. return NULL;
  490. const struct cp_list_item *cp = &cp_list[codepage - 65536];
  491. if (!cp->cp_table)
  492. return NULL;
  493. wchar_t mapping[256];
  494. get_unitab(codepage, mapping, 0);
  495. return make_reverse_mapping_inner(codepage, mapping);
  496. }
  497. void init_ucs(Conf *conf, struct unicode_data *ucsdata)
  498. {
  499. int i;
  500. bool used_dtf = false;
  501. int vtmode;
  502. /* Decide on the Line and Font codepages */
  503. ucsdata->line_codepage = decode_codepage(conf_get_str(conf,
  504. CONF_line_codepage));
  505. if (ucsdata->font_codepage <= 0) {
  506. ucsdata->font_codepage=0;
  507. ucsdata->dbcs_screenfont=false;
  508. }
  509. vtmode = conf_get_int(conf, CONF_vtmode);
  510. if (vtmode == VT_OEMONLY) {
  511. ucsdata->font_codepage = 437;
  512. ucsdata->dbcs_screenfont = false;
  513. if (ucsdata->line_codepage <= 0)
  514. ucsdata->line_codepage = GetACP();
  515. } else if (ucsdata->line_codepage <= 0)
  516. ucsdata->line_codepage = ucsdata->font_codepage;
  517. /* Collect screen font ucs table */
  518. if (ucsdata->dbcs_screenfont || ucsdata->font_codepage == 0) {
  519. get_unitab(ucsdata->font_codepage, ucsdata->unitab_font, 2);
  520. for (i = 128; i < 256; i++)
  521. ucsdata->unitab_font[i] = (WCHAR) (CSET_ACP + i);
  522. } else {
  523. get_unitab(ucsdata->font_codepage, ucsdata->unitab_font, 1);
  524. /* CP437 fonts are often broken ... */
  525. if (ucsdata->font_codepage == 437)
  526. ucsdata->unitab_font[0] = ucsdata->unitab_font[255] = 0xFFFF;
  527. }
  528. if (vtmode == VT_XWINDOWS)
  529. memcpy(ucsdata->unitab_font + 1, unitab_xterm_std,
  530. sizeof(unitab_xterm_std));
  531. /* Collect OEMCP ucs table */
  532. get_unitab(CP_OEMCP, ucsdata->unitab_oemcp, 1);
  533. /* Collect CP437 ucs table for SCO acs */
  534. if (vtmode == VT_OEMANSI || vtmode == VT_XWINDOWS)
  535. memcpy(ucsdata->unitab_scoacs, ucsdata->unitab_oemcp,
  536. sizeof(ucsdata->unitab_scoacs));
  537. else
  538. get_unitab(437, ucsdata->unitab_scoacs, 1);
  539. /* Collect line set ucs table */
  540. if (ucsdata->line_codepage == ucsdata->font_codepage &&
  541. (ucsdata->dbcs_screenfont ||
  542. vtmode == VT_POORMAN || ucsdata->font_codepage==0)) {
  543. /* For DBCS and POOR fonts force direct to font */
  544. used_dtf = true;
  545. for (i = 0; i < 32; i++)
  546. ucsdata->unitab_line[i] = (WCHAR) i;
  547. for (i = 32; i < 256; i++)
  548. ucsdata->unitab_line[i] = (WCHAR) (CSET_ACP + i);
  549. ucsdata->unitab_line[127] = (WCHAR) 127;
  550. } else {
  551. get_unitab(ucsdata->line_codepage, ucsdata->unitab_line, 0);
  552. }
  553. #if 0
  554. debug("Line cp%d, Font cp%d%s\n", ucsdata->line_codepage,
  555. ucsdata->font_codepage, ucsdata->dbcs_screenfont ? " DBCS" : "");
  556. for (i = 0; i < 256; i += 16) {
  557. for (j = 0; j < 16; j++) {
  558. debug("%04x%s", ucsdata->unitab_line[i + j], j == 15 ? "" : ",");
  559. }
  560. debug("\n");
  561. }
  562. #endif
  563. /* VT100 graphics - NB: Broken for non-ascii CP's */
  564. memcpy(ucsdata->unitab_xterm, ucsdata->unitab_line,
  565. sizeof(ucsdata->unitab_xterm));
  566. memcpy(ucsdata->unitab_xterm + '`', unitab_xterm_std,
  567. sizeof(unitab_xterm_std));
  568. ucsdata->unitab_xterm['_'] = ' ';
  569. if (!used_dtf) {
  570. /* Make sure a reverse mapping exists for this code page. */
  571. make_reverse_mapping(ucsdata->line_codepage, ucsdata->unitab_line);
  572. }
  573. /* Find the line control characters. */
  574. for (i = 0; i < 256; i++)
  575. if (ucsdata->unitab_line[i] < ' '
  576. || (ucsdata->unitab_line[i] >= 0x7F &&
  577. ucsdata->unitab_line[i] < 0xA0))
  578. ucsdata->unitab_ctrl[i] = i;
  579. else
  580. ucsdata->unitab_ctrl[i] = 0xFF;
  581. /* Generate line->screen direct conversion links. */
  582. if (vtmode == VT_OEMANSI || vtmode == VT_XWINDOWS)
  583. link_font(ucsdata->unitab_scoacs, ucsdata->unitab_oemcp, CSET_OEMCP);
  584. link_font(ucsdata->unitab_line, ucsdata->unitab_font, CSET_ACP);
  585. link_font(ucsdata->unitab_scoacs, ucsdata->unitab_font, CSET_ACP);
  586. link_font(ucsdata->unitab_xterm, ucsdata->unitab_font, CSET_ACP);
  587. if (vtmode == VT_OEMANSI || vtmode == VT_XWINDOWS) {
  588. link_font(ucsdata->unitab_line, ucsdata->unitab_oemcp, CSET_OEMCP);
  589. link_font(ucsdata->unitab_xterm, ucsdata->unitab_oemcp, CSET_OEMCP);
  590. }
  591. if (ucsdata->dbcs_screenfont &&
  592. ucsdata->font_codepage != ucsdata->line_codepage) {
  593. /* F***ing Microsoft fonts, Japanese and Korean codepage fonts
  594. * have a currency symbol at 0x5C but their unicode value is
  595. * still given as U+005C not the correct U+00A5. */
  596. ucsdata->unitab_line['\\'] = CSET_OEMCP + '\\';
  597. }
  598. /* Last chance, if !unicode then try poorman links. */
  599. if (vtmode != VT_UNICODE) {
  600. static const char poorman_scoacs[] =
  601. "CueaaaaceeeiiiAAE**ooouuyOUc$YPsaiounNao?++**!<>###||||++||++++++--|-+||++--|-+----++++++++##||#aBTPEsyt******EN=+><++-=... n2* ";
  602. static const char poorman_latin1[] =
  603. " !cL.Y|S\"Ca<--R~o+23'u|.,1o>///?AAAAAAACEEEEIIIIDNOOOOOxOUUUUYPBaaaaaaaceeeeiiiionooooo/ouuuuypy";
  604. static const char poorman_vt100[] = "*#****o~**+++++-----++++|****L.";
  605. for (i = 160; i < 256; i++)
  606. if (!DIRECT_FONT(ucsdata->unitab_line[i]) &&
  607. ucsdata->unitab_line[i] >= 160 &&
  608. ucsdata->unitab_line[i] < 256) {
  609. ucsdata->unitab_line[i] =
  610. (WCHAR) (CSET_ACP +
  611. poorman_latin1[ucsdata->unitab_line[i] - 160]);
  612. }
  613. for (i = 96; i < 127; i++)
  614. if (!DIRECT_FONT(ucsdata->unitab_xterm[i]))
  615. ucsdata->unitab_xterm[i] =
  616. (WCHAR) (CSET_ACP + poorman_vt100[i - 96]);
  617. for(i=128;i<256;i++)
  618. if (!DIRECT_FONT(ucsdata->unitab_scoacs[i]))
  619. ucsdata->unitab_scoacs[i] =
  620. (WCHAR) (CSET_ACP + poorman_scoacs[i - 128]);
  621. }
  622. }
  623. static void link_font(WCHAR *line_tbl, WCHAR *font_tbl, WCHAR attr)
  624. {
  625. int font_index, line_index, i;
  626. for (line_index = 0; line_index < 256; line_index++) {
  627. if (DIRECT_FONT(line_tbl[line_index]))
  628. continue;
  629. for(i = 0; i < 256; i++) {
  630. font_index = ((32 + i) & 0xFF);
  631. if (line_tbl[line_index] == font_tbl[font_index]) {
  632. line_tbl[line_index] = (WCHAR) (attr + font_index);
  633. break;
  634. }
  635. }
  636. }
  637. }
  638. wchar_t xlat_uskbd2cyrllic(int ch)
  639. {
  640. static const wchar_t cyrtab[] = {
  641. 0, 1, 2, 3, 4, 5, 6, 7,
  642. 8, 9, 10, 11, 12, 13, 14, 15,
  643. 16, 17, 18, 19, 20, 21, 22, 23,
  644. 24, 25, 26, 27, 28, 29, 30, 31,
  645. 32, 33, 0x042d, 35, 36, 37, 38, 0x044d,
  646. 40, 41, 42, 0x0406, 0x0431, 0x0454, 0x044e, 0x002e,
  647. 48, 49, 50, 51, 52, 53, 54, 55,
  648. 56, 57, 0x0416, 0x0436, 0x0411, 0x0456, 0x042e, 0x002c,
  649. 64, 0x0424, 0x0418, 0x0421, 0x0412, 0x0423, 0x0410, 0x041f,
  650. 0x0420, 0x0428, 0x041e, 0x041b, 0x0414, 0x042c, 0x0422, 0x0429,
  651. 0x0417, 0x0419, 0x041a, 0x042b, 0x0415, 0x0413, 0x041c, 0x0426,
  652. 0x0427, 0x041d, 0x042f, 0x0445, 0x0457, 0x044a, 94, 0x0404,
  653. 96, 0x0444, 0x0438, 0x0441, 0x0432, 0x0443, 0x0430, 0x043f,
  654. 0x0440, 0x0448, 0x043e, 0x043b, 0x0434, 0x044c, 0x0442, 0x0449,
  655. 0x0437, 0x0439, 0x043a, 0x044b, 0x0435, 0x0433, 0x043c, 0x0446,
  656. 0x0447, 0x043d, 0x044f, 0x0425, 0x0407, 0x042a, 126, 127
  657. };
  658. return cyrtab[ch&0x7F];
  659. }
  660. static int check_compose_internal(int first, int second, int recurse)
  661. {
  662. static const struct {
  663. char first, second;
  664. wchar_t composed;
  665. } composetbl[] = {
  666. {0x2b, 0x2b, 0x0023},
  667. {0x41, 0x41, 0x0040},
  668. {0x28, 0x28, 0x005b},
  669. {0x2f, 0x2f, 0x005c},
  670. {0x29, 0x29, 0x005d},
  671. {0x28, 0x2d, 0x007b},
  672. {0x2d, 0x29, 0x007d},
  673. {0x2f, 0x5e, 0x007c},
  674. {0x21, 0x21, 0x00a1},
  675. {0x43, 0x2f, 0x00a2},
  676. {0x43, 0x7c, 0x00a2},
  677. {0x4c, 0x2d, 0x00a3},
  678. {0x4c, 0x3d, 0x20a4},
  679. {0x58, 0x4f, 0x00a4},
  680. {0x58, 0x30, 0x00a4},
  681. {0x59, 0x2d, 0x00a5},
  682. {0x59, 0x3d, 0x00a5},
  683. {0x7c, 0x7c, 0x00a6},
  684. {0x53, 0x4f, 0x00a7},
  685. {0x53, 0x21, 0x00a7},
  686. {0x53, 0x30, 0x00a7},
  687. {0x22, 0x22, 0x00a8},
  688. {0x43, 0x4f, 0x00a9},
  689. {0x43, 0x30, 0x00a9},
  690. {0x41, 0x5f, 0x00aa},
  691. {0x3c, 0x3c, 0x00ab},
  692. {0x2c, 0x2d, 0x00ac},
  693. {0x2d, 0x2d, 0x00ad},
  694. {0x52, 0x4f, 0x00ae},
  695. {0x2d, 0x5e, 0x00af},
  696. {0x30, 0x5e, 0x00b0},
  697. {0x2b, 0x2d, 0x00b1},
  698. {0x32, 0x5e, 0x00b2},
  699. {0x33, 0x5e, 0x00b3},
  700. {0x27, 0x27, 0x00b4},
  701. {0x2f, 0x55, 0x00b5},
  702. {0x50, 0x21, 0x00b6},
  703. {0x2e, 0x5e, 0x00b7},
  704. {0x2c, 0x2c, 0x00b8},
  705. {0x31, 0x5e, 0x00b9},
  706. {0x4f, 0x5f, 0x00ba},
  707. {0x3e, 0x3e, 0x00bb},
  708. {0x31, 0x34, 0x00bc},
  709. {0x31, 0x32, 0x00bd},
  710. {0x33, 0x34, 0x00be},
  711. {0x3f, 0x3f, 0x00bf},
  712. {0x60, 0x41, 0x00c0},
  713. {0x27, 0x41, 0x00c1},
  714. {0x5e, 0x41, 0x00c2},
  715. {0x7e, 0x41, 0x00c3},
  716. {0x22, 0x41, 0x00c4},
  717. {0x2a, 0x41, 0x00c5},
  718. {0x41, 0x45, 0x00c6},
  719. {0x2c, 0x43, 0x00c7},
  720. {0x60, 0x45, 0x00c8},
  721. {0x27, 0x45, 0x00c9},
  722. {0x5e, 0x45, 0x00ca},
  723. {0x22, 0x45, 0x00cb},
  724. {0x60, 0x49, 0x00cc},
  725. {0x27, 0x49, 0x00cd},
  726. {0x5e, 0x49, 0x00ce},
  727. {0x22, 0x49, 0x00cf},
  728. {0x2d, 0x44, 0x00d0},
  729. {0x7e, 0x4e, 0x00d1},
  730. {0x60, 0x4f, 0x00d2},
  731. {0x27, 0x4f, 0x00d3},
  732. {0x5e, 0x4f, 0x00d4},
  733. {0x7e, 0x4f, 0x00d5},
  734. {0x22, 0x4f, 0x00d6},
  735. {0x58, 0x58, 0x00d7},
  736. {0x2f, 0x4f, 0x00d8},
  737. {0x60, 0x55, 0x00d9},
  738. {0x27, 0x55, 0x00da},
  739. {0x5e, 0x55, 0x00db},
  740. {0x22, 0x55, 0x00dc},
  741. {0x27, 0x59, 0x00dd},
  742. {0x48, 0x54, 0x00de},
  743. {0x73, 0x73, 0x00df},
  744. {0x60, 0x61, 0x00e0},
  745. {0x27, 0x61, 0x00e1},
  746. {0x5e, 0x61, 0x00e2},
  747. {0x7e, 0x61, 0x00e3},
  748. {0x22, 0x61, 0x00e4},
  749. {0x2a, 0x61, 0x00e5},
  750. {0x61, 0x65, 0x00e6},
  751. {0x2c, 0x63, 0x00e7},
  752. {0x60, 0x65, 0x00e8},
  753. {0x27, 0x65, 0x00e9},
  754. {0x5e, 0x65, 0x00ea},
  755. {0x22, 0x65, 0x00eb},
  756. {0x60, 0x69, 0x00ec},
  757. {0x27, 0x69, 0x00ed},
  758. {0x5e, 0x69, 0x00ee},
  759. {0x22, 0x69, 0x00ef},
  760. {0x2d, 0x64, 0x00f0},
  761. {0x7e, 0x6e, 0x00f1},
  762. {0x60, 0x6f, 0x00f2},
  763. {0x27, 0x6f, 0x00f3},
  764. {0x5e, 0x6f, 0x00f4},
  765. {0x7e, 0x6f, 0x00f5},
  766. {0x22, 0x6f, 0x00f6},
  767. {0x3a, 0x2d, 0x00f7},
  768. {0x6f, 0x2f, 0x00f8},
  769. {0x60, 0x75, 0x00f9},
  770. {0x27, 0x75, 0x00fa},
  771. {0x5e, 0x75, 0x00fb},
  772. {0x22, 0x75, 0x00fc},
  773. {0x27, 0x79, 0x00fd},
  774. {0x68, 0x74, 0x00fe},
  775. {0x22, 0x79, 0x00ff},
  776. /* Unicode extras. */
  777. {0x6f, 0x65, 0x0153},
  778. {0x4f, 0x45, 0x0152},
  779. /* Compose pairs from UCS */
  780. {0x41, 0x2D, 0x0100},
  781. {0x61, 0x2D, 0x0101},
  782. {0x43, 0x27, 0x0106},
  783. {0x63, 0x27, 0x0107},
  784. {0x43, 0x5E, 0x0108},
  785. {0x63, 0x5E, 0x0109},
  786. {0x45, 0x2D, 0x0112},
  787. {0x65, 0x2D, 0x0113},
  788. {0x47, 0x5E, 0x011C},
  789. {0x67, 0x5E, 0x011D},
  790. {0x47, 0x2C, 0x0122},
  791. {0x67, 0x2C, 0x0123},
  792. {0x48, 0x5E, 0x0124},
  793. {0x68, 0x5E, 0x0125},
  794. {0x49, 0x7E, 0x0128},
  795. {0x69, 0x7E, 0x0129},
  796. {0x49, 0x2D, 0x012A},
  797. {0x69, 0x2D, 0x012B},
  798. {0x4A, 0x5E, 0x0134},
  799. {0x6A, 0x5E, 0x0135},
  800. {0x4B, 0x2C, 0x0136},
  801. {0x6B, 0x2C, 0x0137},
  802. {0x4C, 0x27, 0x0139},
  803. {0x6C, 0x27, 0x013A},
  804. {0x4C, 0x2C, 0x013B},
  805. {0x6C, 0x2C, 0x013C},
  806. {0x4E, 0x27, 0x0143},
  807. {0x6E, 0x27, 0x0144},
  808. {0x4E, 0x2C, 0x0145},
  809. {0x6E, 0x2C, 0x0146},
  810. {0x4F, 0x2D, 0x014C},
  811. {0x6F, 0x2D, 0x014D},
  812. {0x52, 0x27, 0x0154},
  813. {0x72, 0x27, 0x0155},
  814. {0x52, 0x2C, 0x0156},
  815. {0x72, 0x2C, 0x0157},
  816. {0x53, 0x27, 0x015A},
  817. {0x73, 0x27, 0x015B},
  818. {0x53, 0x5E, 0x015C},
  819. {0x73, 0x5E, 0x015D},
  820. {0x53, 0x2C, 0x015E},
  821. {0x73, 0x2C, 0x015F},
  822. {0x54, 0x2C, 0x0162},
  823. {0x74, 0x2C, 0x0163},
  824. {0x55, 0x7E, 0x0168},
  825. {0x75, 0x7E, 0x0169},
  826. {0x55, 0x2D, 0x016A},
  827. {0x75, 0x2D, 0x016B},
  828. {0x55, 0x2A, 0x016E},
  829. {0x75, 0x2A, 0x016F},
  830. {0x57, 0x5E, 0x0174},
  831. {0x77, 0x5E, 0x0175},
  832. {0x59, 0x5E, 0x0176},
  833. {0x79, 0x5E, 0x0177},
  834. {0x59, 0x22, 0x0178},
  835. {0x5A, 0x27, 0x0179},
  836. {0x7A, 0x27, 0x017A},
  837. {0x47, 0x27, 0x01F4},
  838. {0x67, 0x27, 0x01F5},
  839. {0x4E, 0x60, 0x01F8},
  840. {0x6E, 0x60, 0x01F9},
  841. {0x45, 0x2C, 0x0228},
  842. {0x65, 0x2C, 0x0229},
  843. {0x59, 0x2D, 0x0232},
  844. {0x79, 0x2D, 0x0233},
  845. {0x44, 0x2C, 0x1E10},
  846. {0x64, 0x2C, 0x1E11},
  847. {0x47, 0x2D, 0x1E20},
  848. {0x67, 0x2D, 0x1E21},
  849. {0x48, 0x22, 0x1E26},
  850. {0x68, 0x22, 0x1E27},
  851. {0x48, 0x2C, 0x1E28},
  852. {0x68, 0x2C, 0x1E29},
  853. {0x4B, 0x27, 0x1E30},
  854. {0x6B, 0x27, 0x1E31},
  855. {0x4D, 0x27, 0x1E3E},
  856. {0x6D, 0x27, 0x1E3F},
  857. {0x50, 0x27, 0x1E54},
  858. {0x70, 0x27, 0x1E55},
  859. {0x56, 0x7E, 0x1E7C},
  860. {0x76, 0x7E, 0x1E7D},
  861. {0x57, 0x60, 0x1E80},
  862. {0x77, 0x60, 0x1E81},
  863. {0x57, 0x27, 0x1E82},
  864. {0x77, 0x27, 0x1E83},
  865. {0x57, 0x22, 0x1E84},
  866. {0x77, 0x22, 0x1E85},
  867. {0x58, 0x22, 0x1E8C},
  868. {0x78, 0x22, 0x1E8D},
  869. {0x5A, 0x5E, 0x1E90},
  870. {0x7A, 0x5E, 0x1E91},
  871. {0x74, 0x22, 0x1E97},
  872. {0x77, 0x2A, 0x1E98},
  873. {0x79, 0x2A, 0x1E99},
  874. {0x45, 0x7E, 0x1EBC},
  875. {0x65, 0x7E, 0x1EBD},
  876. {0x59, 0x60, 0x1EF2},
  877. {0x79, 0x60, 0x1EF3},
  878. {0x59, 0x7E, 0x1EF8},
  879. {0x79, 0x7E, 0x1EF9},
  880. /* Compatible/possibles from UCS */
  881. {0x49, 0x4A, 0x0132},
  882. {0x69, 0x6A, 0x0133},
  883. {0x4C, 0x4A, 0x01C7},
  884. {0x4C, 0x6A, 0x01C8},
  885. {0x6C, 0x6A, 0x01C9},
  886. {0x4E, 0x4A, 0x01CA},
  887. {0x4E, 0x6A, 0x01CB},
  888. {0x6E, 0x6A, 0x01CC},
  889. {0x44, 0x5A, 0x01F1},
  890. {0x44, 0x7A, 0x01F2},
  891. {0x64, 0x7A, 0x01F3},
  892. {0x2E, 0x2E, 0x2025},
  893. {0x21, 0x21, 0x203C},
  894. {0x3F, 0x21, 0x2048},
  895. {0x21, 0x3F, 0x2049},
  896. {0x52, 0x73, 0x20A8},
  897. {0x4E, 0x6F, 0x2116},
  898. {0x53, 0x4D, 0x2120},
  899. {0x54, 0x4D, 0x2122},
  900. {0x49, 0x49, 0x2161},
  901. {0x49, 0x56, 0x2163},
  902. {0x56, 0x49, 0x2165},
  903. {0x49, 0x58, 0x2168},
  904. {0x58, 0x49, 0x216A},
  905. {0x69, 0x69, 0x2171},
  906. {0x69, 0x76, 0x2173},
  907. {0x76, 0x69, 0x2175},
  908. {0x69, 0x78, 0x2178},
  909. {0x78, 0x69, 0x217A},
  910. {0x31, 0x30, 0x2469},
  911. {0x31, 0x31, 0x246A},
  912. {0x31, 0x32, 0x246B},
  913. {0x31, 0x33, 0x246C},
  914. {0x31, 0x34, 0x246D},
  915. {0x31, 0x35, 0x246E},
  916. {0x31, 0x36, 0x246F},
  917. {0x31, 0x37, 0x2470},
  918. {0x31, 0x38, 0x2471},
  919. {0x31, 0x39, 0x2472},
  920. {0x32, 0x30, 0x2473},
  921. {0x31, 0x2E, 0x2488},
  922. {0x32, 0x2E, 0x2489},
  923. {0x33, 0x2E, 0x248A},
  924. {0x34, 0x2E, 0x248B},
  925. {0x35, 0x2E, 0x248C},
  926. {0x36, 0x2E, 0x248D},
  927. {0x37, 0x2E, 0x248E},
  928. {0x38, 0x2E, 0x248F},
  929. {0x39, 0x2E, 0x2490},
  930. {0x64, 0x61, 0x3372},
  931. {0x41, 0x55, 0x3373},
  932. {0x6F, 0x56, 0x3375},
  933. {0x70, 0x63, 0x3376},
  934. {0x70, 0x41, 0x3380},
  935. {0x6E, 0x41, 0x3381},
  936. {0x6D, 0x41, 0x3383},
  937. {0x6B, 0x41, 0x3384},
  938. {0x4B, 0x42, 0x3385},
  939. {0x4D, 0x42, 0x3386},
  940. {0x47, 0x42, 0x3387},
  941. {0x70, 0x46, 0x338A},
  942. {0x6E, 0x46, 0x338B},
  943. {0x6D, 0x67, 0x338E},
  944. {0x6B, 0x67, 0x338F},
  945. {0x48, 0x7A, 0x3390},
  946. {0x66, 0x6D, 0x3399},
  947. {0x6E, 0x6D, 0x339A},
  948. {0x6D, 0x6D, 0x339C},
  949. {0x63, 0x6D, 0x339D},
  950. {0x6B, 0x6D, 0x339E},
  951. {0x50, 0x61, 0x33A9},
  952. {0x70, 0x73, 0x33B0},
  953. {0x6E, 0x73, 0x33B1},
  954. {0x6D, 0x73, 0x33B3},
  955. {0x70, 0x56, 0x33B4},
  956. {0x6E, 0x56, 0x33B5},
  957. {0x6D, 0x56, 0x33B7},
  958. {0x6B, 0x56, 0x33B8},
  959. {0x4D, 0x56, 0x33B9},
  960. {0x70, 0x57, 0x33BA},
  961. {0x6E, 0x57, 0x33BB},
  962. {0x6D, 0x57, 0x33BD},
  963. {0x6B, 0x57, 0x33BE},
  964. {0x4D, 0x57, 0x33BF},
  965. {0x42, 0x71, 0x33C3},
  966. {0x63, 0x63, 0x33C4},
  967. {0x63, 0x64, 0x33C5},
  968. {0x64, 0x42, 0x33C8},
  969. {0x47, 0x79, 0x33C9},
  970. {0x68, 0x61, 0x33CA},
  971. {0x48, 0x50, 0x33CB},
  972. {0x69, 0x6E, 0x33CC},
  973. {0x4B, 0x4B, 0x33CD},
  974. {0x4B, 0x4D, 0x33CE},
  975. {0x6B, 0x74, 0x33CF},
  976. {0x6C, 0x6D, 0x33D0},
  977. {0x6C, 0x6E, 0x33D1},
  978. {0x6C, 0x78, 0x33D3},
  979. {0x6D, 0x62, 0x33D4},
  980. {0x50, 0x48, 0x33D7},
  981. {0x50, 0x52, 0x33DA},
  982. {0x73, 0x72, 0x33DB},
  983. {0x53, 0x76, 0x33DC},
  984. {0x57, 0x62, 0x33DD},
  985. {0x66, 0x66, 0xFB00},
  986. {0x66, 0x69, 0xFB01},
  987. {0x66, 0x6C, 0xFB02},
  988. {0x73, 0x74, 0xFB06},
  989. {0, 0, 0}
  990. }, *c;
  991. int nc = -1;
  992. for (c = composetbl; c->first; c++) {
  993. if (c->first == first && c->second == second)
  994. return c->composed;
  995. }
  996. if (recurse == 0) {
  997. nc = check_compose_internal(second, first, 1);
  998. if (nc == -1)
  999. nc = check_compose_internal(toupper((unsigned char)first),
  1000. toupper((unsigned char)second), 1);
  1001. if (nc == -1)
  1002. nc = check_compose_internal(toupper((unsigned char)second),
  1003. toupper((unsigned char)first), 1);
  1004. }
  1005. return nc;
  1006. }
  1007. int check_compose(int first, int second)
  1008. {
  1009. return check_compose_internal(first, second, 0);
  1010. }
  1011. int decode_codepage(const char *cp_name)
  1012. {
  1013. const char *s, *d;
  1014. const struct cp_list_item *cpi;
  1015. int codepage = -1;
  1016. CPINFO cpinfo;
  1017. if (!cp_name || !*cp_name)
  1018. return CP_UTF8; /* default */
  1019. for (cpi = cp_list; cpi->name; cpi++) {
  1020. s = cp_name;
  1021. d = cpi->name;
  1022. for (;;) {
  1023. while (*s && !isalnum((unsigned char)*s) && *s != ':')
  1024. s++;
  1025. while (*d && !isalnum((unsigned char)*d) && *d != ':')
  1026. d++;
  1027. if (*s == 0) {
  1028. codepage = cpi->codepage;
  1029. if (codepage == CP_UTF8)
  1030. goto break_break;
  1031. if (codepage == -1)
  1032. return codepage;
  1033. if (codepage == 0) {
  1034. codepage = 65536 + (cpi - cp_list);
  1035. goto break_break;
  1036. }
  1037. if (GetCPInfo(codepage, &cpinfo) != 0)
  1038. goto break_break;
  1039. }
  1040. if (tolower((unsigned char)*s++) != tolower((unsigned char)*d++))
  1041. break;
  1042. }
  1043. }
  1044. d = cp_name;
  1045. if (tolower((unsigned char)d[0]) == 'c' &&
  1046. tolower((unsigned char)d[1]) == 'p')
  1047. d += 2;
  1048. if (tolower((unsigned char)d[0]) == 'i' &&
  1049. tolower((unsigned char)d[1]) == 'b' &&
  1050. tolower((unsigned char)d[2]) == 'm')
  1051. d += 3;
  1052. for (s = d; *s >= '0' && *s <= '9'; s++);
  1053. if (*s == 0 && s != d)
  1054. codepage = atoi(d); /* CP999 or IBM999 */
  1055. if (codepage == CP_ACP)
  1056. codepage = GetACP();
  1057. if (codepage == CP_OEMCP)
  1058. codepage = GetOEMCP();
  1059. if (codepage > 65535)
  1060. codepage = -2;
  1061. break_break:;
  1062. if (codepage != -1) {
  1063. if (codepage != CP_UTF8 && codepage < 65536) {
  1064. if (GetCPInfo(codepage, &cpinfo) == 0) {
  1065. codepage = -2;
  1066. } else if (cpinfo.MaxCharSize > 1)
  1067. codepage = -3;
  1068. }
  1069. }
  1070. if (codepage == -1 && *cp_name)
  1071. codepage = -2;
  1072. return codepage;
  1073. }
  1074. const char *cp_name(int codepage)
  1075. {
  1076. const struct cp_list_item *cpi, *cpno;
  1077. static char buf[32];
  1078. if (codepage == -1) {
  1079. sprintf(buf, "Use font encoding");
  1080. return buf;
  1081. }
  1082. if (codepage > 0 && codepage < 65536)
  1083. sprintf(buf, "CP%03d", codepage);
  1084. else
  1085. *buf = 0;
  1086. if (codepage >= 65536) {
  1087. cpno = 0;
  1088. for (cpi = cp_list; cpi->name; cpi++)
  1089. if (cpi == cp_list + (codepage - 65536)) {
  1090. cpno = cpi;
  1091. break;
  1092. }
  1093. if (cpno)
  1094. for (cpi = cp_list; cpi->name; cpi++) {
  1095. if (cpno->cp_table == cpi->cp_table)
  1096. return cpi->name;
  1097. }
  1098. } else {
  1099. for (cpi = cp_list; cpi->name; cpi++) {
  1100. if (codepage == cpi->codepage)
  1101. return cpi->name;
  1102. }
  1103. }
  1104. return buf;
  1105. }
  1106. /*
  1107. * Return the nth code page in the list, for use in the GUI
  1108. * configurer.
  1109. */
  1110. const char *cp_enumerate(int index)
  1111. {
  1112. if (index < 0 || index >= lenof(cp_list))
  1113. return NULL;
  1114. return cp_list[index].name;
  1115. }
  1116. void get_unitab(int codepage, wchar_t *unitab, int ftype)
  1117. {
  1118. char tbuf[4];
  1119. int i, max = 256, flg = MB_ERR_INVALID_CHARS;
  1120. if (ftype)
  1121. flg |= MB_USEGLYPHCHARS;
  1122. if (ftype == 2)
  1123. max = 128;
  1124. if (codepage == CP_UTF8) {
  1125. for (i = 0; i < max; i++)
  1126. unitab[i] = i;
  1127. return;
  1128. }
  1129. if (codepage == CP_ACP)
  1130. codepage = GetACP();
  1131. else if (codepage == CP_OEMCP)
  1132. codepage = GetOEMCP();
  1133. if (codepage > 0 && codepage < 65536) {
  1134. for (i = 0; i < max; i++) {
  1135. tbuf[0] = i;
  1136. if (mb_to_wc(codepage, flg, tbuf, 1, unitab + i, 1)
  1137. != 1)
  1138. unitab[i] = 0xFFFD;
  1139. }
  1140. } else {
  1141. int j = 256 - cp_list[codepage & 0xFFFF].cp_size;
  1142. for (i = 0; i < max; i++)
  1143. unitab[i] = i;
  1144. for (i = j; i < max; i++)
  1145. unitab[i] = cp_list[codepage & 0xFFFF].cp_table[i - j];
  1146. }
  1147. }
  1148. int wc_to_mb(int codepage, int flags, const wchar_t *wcstr, int wclen,
  1149. char *mbstr, int mblen, const char *defchr)
  1150. {
  1151. reverse_mapping *rmap = get_reverse_mapping(codepage);
  1152. if (rmap) {
  1153. /* Do this by array lookup if we can. */
  1154. if (wclen < 0) {
  1155. for (wclen = 0; wcstr[wclen++] ;); /* will include the NUL */
  1156. }
  1157. char *p;
  1158. int i;
  1159. for (p = mbstr, i = 0; i < wclen; i++) {
  1160. wchar_t ch = wcstr[i];
  1161. int by;
  1162. const char *p1;
  1163. #define WRITECH(chr) do \
  1164. { \
  1165. assert(p - mbstr < mblen); \
  1166. *p++ = (char)(chr); \
  1167. } while (0)
  1168. if ((p1 = rmap->blocks[(ch >> 8) & 0xFF]) != NULL &&
  1169. (by = p1[ch & 0xFF]) != '\0')
  1170. WRITECH(by);
  1171. else if (ch < 0x80)
  1172. WRITECH(ch);
  1173. else if (defchr)
  1174. for (const char *q = defchr; *q; q++)
  1175. WRITECH(*q);
  1176. #if 1
  1177. else
  1178. WRITECH('.');
  1179. #endif
  1180. #undef WRITECH
  1181. }
  1182. return p - mbstr;
  1183. } else {
  1184. int defused, ret;
  1185. ret = WideCharToMultiByte(codepage, flags, wcstr, wclen,
  1186. mbstr, mblen, defchr, &defused);
  1187. if (ret)
  1188. return ret;
  1189. #ifdef LEGACY_WINDOWS
  1190. /*
  1191. * Fallback for legacy platforms too old to support UTF-8: if
  1192. * the codepage is UTF-8, we can do the translation ourselves.
  1193. */
  1194. if (codepage == CP_UTF8 && mblen > 0 && wclen > 0) {
  1195. size_t remaining = mblen;
  1196. char *p = mbstr;
  1197. while (wclen > 0) {
  1198. unsigned long wc = (wclen--, *wcstr++);
  1199. if (wclen > 0 && IS_SURROGATE_PAIR(wc, *wcstr)) {
  1200. wc = FROM_SURROGATES(wc, *wcstr);
  1201. wclen--, wcstr++;
  1202. }
  1203. char utfbuf[6];
  1204. size_t utflen = encode_utf8(utfbuf, wc);
  1205. if (utflen <= remaining) {
  1206. memcpy(p, utfbuf, utflen);
  1207. p += utflen;
  1208. remaining -= utflen;
  1209. } else {
  1210. return p - mbstr;
  1211. }
  1212. }
  1213. return p - mbstr;
  1214. }
  1215. #endif
  1216. /* No other fallbacks are available */
  1217. return 0;
  1218. }
  1219. }
  1220. #endif
  1221. int mb_to_wc(int codepage, int flags, const char *mbstr, int mblen,
  1222. wchar_t *wcstr, int wclen)
  1223. {
  1224. #ifdef WINSCP
  1225. pinitassert(codepage == DEFAULT_CODEPAGE);
  1226. #else
  1227. if (codepage >= 65536) {
  1228. /* Character set not known to Windows, so we'll have to
  1229. * translate it ourself */
  1230. size_t index = codepage - 65536;
  1231. if (index >= lenof(cp_list))
  1232. return 0;
  1233. const struct cp_list_item *cp = &cp_list[index];
  1234. if (!cp->cp_table)
  1235. return 0;
  1236. size_t remaining = wclen;
  1237. wchar_t *p = wcstr;
  1238. unsigned tablebase = 256 - cp->cp_size;
  1239. while (mblen > 0) {
  1240. mblen--;
  1241. unsigned c = 0xFF & *mbstr++;
  1242. wchar_t wc = (c < tablebase ? c : cp->cp_table[c - tablebase]);
  1243. if (remaining > 0) {
  1244. remaining--;
  1245. *p++ = wc;
  1246. } else {
  1247. return p - wcstr;
  1248. }
  1249. }
  1250. return p - wcstr;
  1251. }
  1252. #endif
  1253. int ret = MultiByteToWideChar(codepage, flags, mbstr, mblen, wcstr, wclen);
  1254. if (ret)
  1255. return ret;
  1256. #ifdef LEGACY_WINDOWS
  1257. /*
  1258. * Fallback for legacy platforms too old to support UTF-8: if the
  1259. * codepage is UTF-8, we can do the translation ourselves.
  1260. */
  1261. if (codepage == CP_UTF8 && mblen > 0 && wclen > 0) {
  1262. size_t remaining = wclen;
  1263. wchar_t *p = wcstr;
  1264. while (mblen > 0) {
  1265. char utfbuf[7];
  1266. int thissize = mblen < 6 ? mblen : 6;
  1267. memcpy(utfbuf, mbstr, thissize);
  1268. utfbuf[thissize] = '\0';
  1269. const char *utfptr = utfbuf;
  1270. wchar_t wcbuf[2];
  1271. size_t nwc = decode_utf8_to_wchar(&utfptr, wcbuf);
  1272. for (size_t i = 0; i < nwc; i++) {
  1273. if (remaining > 0) {
  1274. remaining--;
  1275. *p++ = wcbuf[i];
  1276. } else {
  1277. return p - wcstr;
  1278. }
  1279. }
  1280. mbstr += (utfptr - utfbuf);
  1281. mblen -= (utfptr - utfbuf);
  1282. }
  1283. return p - wcstr;
  1284. }
  1285. #endif
  1286. /* No other fallbacks are available */
  1287. return 0;
  1288. }
  1289. bool is_dbcs_leadbyte(int codepage, char byte)
  1290. {
  1291. return IsDBCSLeadByteEx(codepage, byte);
  1292. }