winucs.c 46 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211
  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. void init_ucs(Conf *conf, struct unicode_data *ucsdata)
  408. {
  409. int i, j;
  410. bool used_dtf = false;
  411. int vtmode;
  412. /* Decide on the Line and Font codepages */
  413. ucsdata->line_codepage = decode_codepage(conf_get_str(conf,
  414. CONF_line_codepage));
  415. if (ucsdata->font_codepage <= 0) {
  416. ucsdata->font_codepage=0;
  417. ucsdata->dbcs_screenfont=false;
  418. }
  419. vtmode = conf_get_int(conf, CONF_vtmode);
  420. if (vtmode == VT_OEMONLY) {
  421. ucsdata->font_codepage = 437;
  422. ucsdata->dbcs_screenfont = false;
  423. if (ucsdata->line_codepage <= 0)
  424. ucsdata->line_codepage = GetACP();
  425. } else if (ucsdata->line_codepage <= 0)
  426. ucsdata->line_codepage = ucsdata->font_codepage;
  427. /* Collect screen font ucs table */
  428. if (ucsdata->dbcs_screenfont || ucsdata->font_codepage == 0) {
  429. get_unitab(ucsdata->font_codepage, ucsdata->unitab_font, 2);
  430. for (i = 128; i < 256; i++)
  431. ucsdata->unitab_font[i] = (WCHAR) (CSET_ACP + i);
  432. } else {
  433. get_unitab(ucsdata->font_codepage, ucsdata->unitab_font, 1);
  434. /* CP437 fonts are often broken ... */
  435. if (ucsdata->font_codepage == 437)
  436. ucsdata->unitab_font[0] = ucsdata->unitab_font[255] = 0xFFFF;
  437. }
  438. if (vtmode == VT_XWINDOWS)
  439. memcpy(ucsdata->unitab_font + 1, unitab_xterm_std,
  440. sizeof(unitab_xterm_std));
  441. /* Collect OEMCP ucs table */
  442. get_unitab(CP_OEMCP, ucsdata->unitab_oemcp, 1);
  443. /* Collect CP437 ucs table for SCO acs */
  444. if (vtmode == VT_OEMANSI || vtmode == VT_XWINDOWS)
  445. memcpy(ucsdata->unitab_scoacs, ucsdata->unitab_oemcp,
  446. sizeof(ucsdata->unitab_scoacs));
  447. else
  448. get_unitab(437, ucsdata->unitab_scoacs, 1);
  449. /* Collect line set ucs table */
  450. if (ucsdata->line_codepage == ucsdata->font_codepage &&
  451. (ucsdata->dbcs_screenfont ||
  452. vtmode == VT_POORMAN || ucsdata->font_codepage==0)) {
  453. /* For DBCS and POOR fonts force direct to font */
  454. used_dtf = true;
  455. for (i = 0; i < 32; i++)
  456. ucsdata->unitab_line[i] = (WCHAR) i;
  457. for (i = 32; i < 256; i++)
  458. ucsdata->unitab_line[i] = (WCHAR) (CSET_ACP + i);
  459. ucsdata->unitab_line[127] = (WCHAR) 127;
  460. } else {
  461. get_unitab(ucsdata->line_codepage, ucsdata->unitab_line, 0);
  462. }
  463. #if 0
  464. debug("Line cp%d, Font cp%d%s\n", ucsdata->line_codepage,
  465. ucsdata->font_codepage, ucsdata->dbcs_screenfont ? " DBCS" : "");
  466. for (i = 0; i < 256; i += 16) {
  467. for (j = 0; j < 16; j++) {
  468. debug("%04x%s", ucsdata->unitab_line[i + j], j == 15 ? "" : ",");
  469. }
  470. debug("\n");
  471. }
  472. #endif
  473. /* VT100 graphics - NB: Broken for non-ascii CP's */
  474. memcpy(ucsdata->unitab_xterm, ucsdata->unitab_line,
  475. sizeof(ucsdata->unitab_xterm));
  476. memcpy(ucsdata->unitab_xterm + '`', unitab_xterm_std,
  477. sizeof(unitab_xterm_std));
  478. ucsdata->unitab_xterm['_'] = ' ';
  479. /* Generate UCS ->line page table. */
  480. if (ucsdata->uni_tbl) {
  481. for (i = 0; i < 256; i++)
  482. if (ucsdata->uni_tbl[i])
  483. sfree(ucsdata->uni_tbl[i]);
  484. sfree(ucsdata->uni_tbl);
  485. ucsdata->uni_tbl = 0;
  486. }
  487. if (!used_dtf) {
  488. for (i = 0; i < 256; i++) {
  489. if (DIRECT_CHAR(ucsdata->unitab_line[i]))
  490. continue;
  491. if (DIRECT_FONT(ucsdata->unitab_line[i]))
  492. continue;
  493. if (!ucsdata->uni_tbl) {
  494. ucsdata->uni_tbl = snewn(256, char *);
  495. memset(ucsdata->uni_tbl, 0, 256 * sizeof(char *));
  496. }
  497. j = ((ucsdata->unitab_line[i] >> 8) & 0xFF);
  498. if (!ucsdata->uni_tbl[j]) {
  499. ucsdata->uni_tbl[j] = snewn(256, char);
  500. memset(ucsdata->uni_tbl[j], 0, 256 * sizeof(char));
  501. }
  502. ucsdata->uni_tbl[j][ucsdata->unitab_line[i] & 0xFF] = i;
  503. }
  504. }
  505. /* Find the line control characters. */
  506. for (i = 0; i < 256; i++)
  507. if (ucsdata->unitab_line[i] < ' '
  508. || (ucsdata->unitab_line[i] >= 0x7F &&
  509. ucsdata->unitab_line[i] < 0xA0))
  510. ucsdata->unitab_ctrl[i] = i;
  511. else
  512. ucsdata->unitab_ctrl[i] = 0xFF;
  513. /* Generate line->screen direct conversion links. */
  514. if (vtmode == VT_OEMANSI || vtmode == VT_XWINDOWS)
  515. link_font(ucsdata->unitab_scoacs, ucsdata->unitab_oemcp, CSET_OEMCP);
  516. link_font(ucsdata->unitab_line, ucsdata->unitab_font, CSET_ACP);
  517. link_font(ucsdata->unitab_scoacs, ucsdata->unitab_font, CSET_ACP);
  518. link_font(ucsdata->unitab_xterm, ucsdata->unitab_font, CSET_ACP);
  519. if (vtmode == VT_OEMANSI || vtmode == VT_XWINDOWS) {
  520. link_font(ucsdata->unitab_line, ucsdata->unitab_oemcp, CSET_OEMCP);
  521. link_font(ucsdata->unitab_xterm, ucsdata->unitab_oemcp, CSET_OEMCP);
  522. }
  523. if (ucsdata->dbcs_screenfont &&
  524. ucsdata->font_codepage != ucsdata->line_codepage) {
  525. /* F***ing Microsoft fonts, Japanese and Korean codepage fonts
  526. * have a currency symbol at 0x5C but their unicode value is
  527. * still given as U+005C not the correct U+00A5. */
  528. ucsdata->unitab_line['\\'] = CSET_OEMCP + '\\';
  529. }
  530. /* Last chance, if !unicode then try poorman links. */
  531. if (vtmode != VT_UNICODE) {
  532. static const char poorman_scoacs[] =
  533. "CueaaaaceeeiiiAAE**ooouuyOUc$YPsaiounNao?++**!<>###||||++||++++++--|-+||++--|-+----++++++++##||#aBTPEsyt******EN=+><++-=... n2* ";
  534. static const char poorman_latin1[] =
  535. " !cL.Y|S\"Ca<--R~o+23'u|.,1o>///?AAAAAAACEEEEIIIIDNOOOOOxOUUUUYPBaaaaaaaceeeeiiiionooooo/ouuuuypy";
  536. static const char poorman_vt100[] = "*#****o~**+++++-----++++|****L.";
  537. for (i = 160; i < 256; i++)
  538. if (!DIRECT_FONT(ucsdata->unitab_line[i]) &&
  539. ucsdata->unitab_line[i] >= 160 &&
  540. ucsdata->unitab_line[i] < 256) {
  541. ucsdata->unitab_line[i] =
  542. (WCHAR) (CSET_ACP +
  543. poorman_latin1[ucsdata->unitab_line[i] - 160]);
  544. }
  545. for (i = 96; i < 127; i++)
  546. if (!DIRECT_FONT(ucsdata->unitab_xterm[i]))
  547. ucsdata->unitab_xterm[i] =
  548. (WCHAR) (CSET_ACP + poorman_vt100[i - 96]);
  549. for(i=128;i<256;i++)
  550. if (!DIRECT_FONT(ucsdata->unitab_scoacs[i]))
  551. ucsdata->unitab_scoacs[i] =
  552. (WCHAR) (CSET_ACP + poorman_scoacs[i - 128]);
  553. }
  554. }
  555. static void link_font(WCHAR * line_tbl, WCHAR * font_tbl, WCHAR attr)
  556. {
  557. int font_index, line_index, i;
  558. for (line_index = 0; line_index < 256; line_index++) {
  559. if (DIRECT_FONT(line_tbl[line_index]))
  560. continue;
  561. for(i = 0; i < 256; i++) {
  562. font_index = ((32 + i) & 0xFF);
  563. if (line_tbl[line_index] == font_tbl[font_index]) {
  564. line_tbl[line_index] = (WCHAR) (attr + font_index);
  565. break;
  566. }
  567. }
  568. }
  569. }
  570. wchar_t xlat_uskbd2cyrllic(int ch)
  571. {
  572. static const wchar_t cyrtab[] = {
  573. 0, 1, 2, 3, 4, 5, 6, 7,
  574. 8, 9, 10, 11, 12, 13, 14, 15,
  575. 16, 17, 18, 19, 20, 21, 22, 23,
  576. 24, 25, 26, 27, 28, 29, 30, 31,
  577. 32, 33, 0x042d, 35, 36, 37, 38, 0x044d,
  578. 40, 41, 42, 0x0406, 0x0431, 0x0454, 0x044e, 0x002e,
  579. 48, 49, 50, 51, 52, 53, 54, 55,
  580. 56, 57, 0x0416, 0x0436, 0x0411, 0x0456, 0x042e, 0x002c,
  581. 64, 0x0424, 0x0418, 0x0421, 0x0412, 0x0423, 0x0410, 0x041f,
  582. 0x0420, 0x0428, 0x041e, 0x041b, 0x0414, 0x042c, 0x0422, 0x0429,
  583. 0x0417, 0x0419, 0x041a, 0x042b, 0x0415, 0x0413, 0x041c, 0x0426,
  584. 0x0427, 0x041d, 0x042f, 0x0445, 0x0457, 0x044a, 94, 0x0404,
  585. 96, 0x0444, 0x0438, 0x0441, 0x0432, 0x0443, 0x0430, 0x043f,
  586. 0x0440, 0x0448, 0x043e, 0x043b, 0x0434, 0x044c, 0x0442, 0x0449,
  587. 0x0437, 0x0439, 0x043a, 0x044b, 0x0435, 0x0433, 0x043c, 0x0446,
  588. 0x0447, 0x043d, 0x044f, 0x0425, 0x0407, 0x042a, 126, 127
  589. };
  590. return cyrtab[ch&0x7F];
  591. }
  592. int check_compose_internal(int first, int second, int recurse)
  593. {
  594. static const struct {
  595. char first, second;
  596. wchar_t composed;
  597. } composetbl[] = {
  598. {0x2b, 0x2b, 0x0023},
  599. {0x41, 0x41, 0x0040},
  600. {0x28, 0x28, 0x005b},
  601. {0x2f, 0x2f, 0x005c},
  602. {0x29, 0x29, 0x005d},
  603. {0x28, 0x2d, 0x007b},
  604. {0x2d, 0x29, 0x007d},
  605. {0x2f, 0x5e, 0x007c},
  606. {0x21, 0x21, 0x00a1},
  607. {0x43, 0x2f, 0x00a2},
  608. {0x43, 0x7c, 0x00a2},
  609. {0x4c, 0x2d, 0x00a3},
  610. {0x4c, 0x3d, 0x20a4},
  611. {0x58, 0x4f, 0x00a4},
  612. {0x58, 0x30, 0x00a4},
  613. {0x59, 0x2d, 0x00a5},
  614. {0x59, 0x3d, 0x00a5},
  615. {0x7c, 0x7c, 0x00a6},
  616. {0x53, 0x4f, 0x00a7},
  617. {0x53, 0x21, 0x00a7},
  618. {0x53, 0x30, 0x00a7},
  619. {0x22, 0x22, 0x00a8},
  620. {0x43, 0x4f, 0x00a9},
  621. {0x43, 0x30, 0x00a9},
  622. {0x41, 0x5f, 0x00aa},
  623. {0x3c, 0x3c, 0x00ab},
  624. {0x2c, 0x2d, 0x00ac},
  625. {0x2d, 0x2d, 0x00ad},
  626. {0x52, 0x4f, 0x00ae},
  627. {0x2d, 0x5e, 0x00af},
  628. {0x30, 0x5e, 0x00b0},
  629. {0x2b, 0x2d, 0x00b1},
  630. {0x32, 0x5e, 0x00b2},
  631. {0x33, 0x5e, 0x00b3},
  632. {0x27, 0x27, 0x00b4},
  633. {0x2f, 0x55, 0x00b5},
  634. {0x50, 0x21, 0x00b6},
  635. {0x2e, 0x5e, 0x00b7},
  636. {0x2c, 0x2c, 0x00b8},
  637. {0x31, 0x5e, 0x00b9},
  638. {0x4f, 0x5f, 0x00ba},
  639. {0x3e, 0x3e, 0x00bb},
  640. {0x31, 0x34, 0x00bc},
  641. {0x31, 0x32, 0x00bd},
  642. {0x33, 0x34, 0x00be},
  643. {0x3f, 0x3f, 0x00bf},
  644. {0x60, 0x41, 0x00c0},
  645. {0x27, 0x41, 0x00c1},
  646. {0x5e, 0x41, 0x00c2},
  647. {0x7e, 0x41, 0x00c3},
  648. {0x22, 0x41, 0x00c4},
  649. {0x2a, 0x41, 0x00c5},
  650. {0x41, 0x45, 0x00c6},
  651. {0x2c, 0x43, 0x00c7},
  652. {0x60, 0x45, 0x00c8},
  653. {0x27, 0x45, 0x00c9},
  654. {0x5e, 0x45, 0x00ca},
  655. {0x22, 0x45, 0x00cb},
  656. {0x60, 0x49, 0x00cc},
  657. {0x27, 0x49, 0x00cd},
  658. {0x5e, 0x49, 0x00ce},
  659. {0x22, 0x49, 0x00cf},
  660. {0x2d, 0x44, 0x00d0},
  661. {0x7e, 0x4e, 0x00d1},
  662. {0x60, 0x4f, 0x00d2},
  663. {0x27, 0x4f, 0x00d3},
  664. {0x5e, 0x4f, 0x00d4},
  665. {0x7e, 0x4f, 0x00d5},
  666. {0x22, 0x4f, 0x00d6},
  667. {0x58, 0x58, 0x00d7},
  668. {0x2f, 0x4f, 0x00d8},
  669. {0x60, 0x55, 0x00d9},
  670. {0x27, 0x55, 0x00da},
  671. {0x5e, 0x55, 0x00db},
  672. {0x22, 0x55, 0x00dc},
  673. {0x27, 0x59, 0x00dd},
  674. {0x48, 0x54, 0x00de},
  675. {0x73, 0x73, 0x00df},
  676. {0x60, 0x61, 0x00e0},
  677. {0x27, 0x61, 0x00e1},
  678. {0x5e, 0x61, 0x00e2},
  679. {0x7e, 0x61, 0x00e3},
  680. {0x22, 0x61, 0x00e4},
  681. {0x2a, 0x61, 0x00e5},
  682. {0x61, 0x65, 0x00e6},
  683. {0x2c, 0x63, 0x00e7},
  684. {0x60, 0x65, 0x00e8},
  685. {0x27, 0x65, 0x00e9},
  686. {0x5e, 0x65, 0x00ea},
  687. {0x22, 0x65, 0x00eb},
  688. {0x60, 0x69, 0x00ec},
  689. {0x27, 0x69, 0x00ed},
  690. {0x5e, 0x69, 0x00ee},
  691. {0x22, 0x69, 0x00ef},
  692. {0x2d, 0x64, 0x00f0},
  693. {0x7e, 0x6e, 0x00f1},
  694. {0x60, 0x6f, 0x00f2},
  695. {0x27, 0x6f, 0x00f3},
  696. {0x5e, 0x6f, 0x00f4},
  697. {0x7e, 0x6f, 0x00f5},
  698. {0x22, 0x6f, 0x00f6},
  699. {0x3a, 0x2d, 0x00f7},
  700. {0x6f, 0x2f, 0x00f8},
  701. {0x60, 0x75, 0x00f9},
  702. {0x27, 0x75, 0x00fa},
  703. {0x5e, 0x75, 0x00fb},
  704. {0x22, 0x75, 0x00fc},
  705. {0x27, 0x79, 0x00fd},
  706. {0x68, 0x74, 0x00fe},
  707. {0x22, 0x79, 0x00ff},
  708. /* Unicode extras. */
  709. {0x6f, 0x65, 0x0153},
  710. {0x4f, 0x45, 0x0152},
  711. /* Compose pairs from UCS */
  712. {0x41, 0x2D, 0x0100},
  713. {0x61, 0x2D, 0x0101},
  714. {0x43, 0x27, 0x0106},
  715. {0x63, 0x27, 0x0107},
  716. {0x43, 0x5E, 0x0108},
  717. {0x63, 0x5E, 0x0109},
  718. {0x45, 0x2D, 0x0112},
  719. {0x65, 0x2D, 0x0113},
  720. {0x47, 0x5E, 0x011C},
  721. {0x67, 0x5E, 0x011D},
  722. {0x47, 0x2C, 0x0122},
  723. {0x67, 0x2C, 0x0123},
  724. {0x48, 0x5E, 0x0124},
  725. {0x68, 0x5E, 0x0125},
  726. {0x49, 0x7E, 0x0128},
  727. {0x69, 0x7E, 0x0129},
  728. {0x49, 0x2D, 0x012A},
  729. {0x69, 0x2D, 0x012B},
  730. {0x4A, 0x5E, 0x0134},
  731. {0x6A, 0x5E, 0x0135},
  732. {0x4B, 0x2C, 0x0136},
  733. {0x6B, 0x2C, 0x0137},
  734. {0x4C, 0x27, 0x0139},
  735. {0x6C, 0x27, 0x013A},
  736. {0x4C, 0x2C, 0x013B},
  737. {0x6C, 0x2C, 0x013C},
  738. {0x4E, 0x27, 0x0143},
  739. {0x6E, 0x27, 0x0144},
  740. {0x4E, 0x2C, 0x0145},
  741. {0x6E, 0x2C, 0x0146},
  742. {0x4F, 0x2D, 0x014C},
  743. {0x6F, 0x2D, 0x014D},
  744. {0x52, 0x27, 0x0154},
  745. {0x72, 0x27, 0x0155},
  746. {0x52, 0x2C, 0x0156},
  747. {0x72, 0x2C, 0x0157},
  748. {0x53, 0x27, 0x015A},
  749. {0x73, 0x27, 0x015B},
  750. {0x53, 0x5E, 0x015C},
  751. {0x73, 0x5E, 0x015D},
  752. {0x53, 0x2C, 0x015E},
  753. {0x73, 0x2C, 0x015F},
  754. {0x54, 0x2C, 0x0162},
  755. {0x74, 0x2C, 0x0163},
  756. {0x55, 0x7E, 0x0168},
  757. {0x75, 0x7E, 0x0169},
  758. {0x55, 0x2D, 0x016A},
  759. {0x75, 0x2D, 0x016B},
  760. {0x55, 0x2A, 0x016E},
  761. {0x75, 0x2A, 0x016F},
  762. {0x57, 0x5E, 0x0174},
  763. {0x77, 0x5E, 0x0175},
  764. {0x59, 0x5E, 0x0176},
  765. {0x79, 0x5E, 0x0177},
  766. {0x59, 0x22, 0x0178},
  767. {0x5A, 0x27, 0x0179},
  768. {0x7A, 0x27, 0x017A},
  769. {0x47, 0x27, 0x01F4},
  770. {0x67, 0x27, 0x01F5},
  771. {0x4E, 0x60, 0x01F8},
  772. {0x6E, 0x60, 0x01F9},
  773. {0x45, 0x2C, 0x0228},
  774. {0x65, 0x2C, 0x0229},
  775. {0x59, 0x2D, 0x0232},
  776. {0x79, 0x2D, 0x0233},
  777. {0x44, 0x2C, 0x1E10},
  778. {0x64, 0x2C, 0x1E11},
  779. {0x47, 0x2D, 0x1E20},
  780. {0x67, 0x2D, 0x1E21},
  781. {0x48, 0x22, 0x1E26},
  782. {0x68, 0x22, 0x1E27},
  783. {0x48, 0x2C, 0x1E28},
  784. {0x68, 0x2C, 0x1E29},
  785. {0x4B, 0x27, 0x1E30},
  786. {0x6B, 0x27, 0x1E31},
  787. {0x4D, 0x27, 0x1E3E},
  788. {0x6D, 0x27, 0x1E3F},
  789. {0x50, 0x27, 0x1E54},
  790. {0x70, 0x27, 0x1E55},
  791. {0x56, 0x7E, 0x1E7C},
  792. {0x76, 0x7E, 0x1E7D},
  793. {0x57, 0x60, 0x1E80},
  794. {0x77, 0x60, 0x1E81},
  795. {0x57, 0x27, 0x1E82},
  796. {0x77, 0x27, 0x1E83},
  797. {0x57, 0x22, 0x1E84},
  798. {0x77, 0x22, 0x1E85},
  799. {0x58, 0x22, 0x1E8C},
  800. {0x78, 0x22, 0x1E8D},
  801. {0x5A, 0x5E, 0x1E90},
  802. {0x7A, 0x5E, 0x1E91},
  803. {0x74, 0x22, 0x1E97},
  804. {0x77, 0x2A, 0x1E98},
  805. {0x79, 0x2A, 0x1E99},
  806. {0x45, 0x7E, 0x1EBC},
  807. {0x65, 0x7E, 0x1EBD},
  808. {0x59, 0x60, 0x1EF2},
  809. {0x79, 0x60, 0x1EF3},
  810. {0x59, 0x7E, 0x1EF8},
  811. {0x79, 0x7E, 0x1EF9},
  812. /* Compatible/possibles from UCS */
  813. {0x49, 0x4A, 0x0132},
  814. {0x69, 0x6A, 0x0133},
  815. {0x4C, 0x4A, 0x01C7},
  816. {0x4C, 0x6A, 0x01C8},
  817. {0x6C, 0x6A, 0x01C9},
  818. {0x4E, 0x4A, 0x01CA},
  819. {0x4E, 0x6A, 0x01CB},
  820. {0x6E, 0x6A, 0x01CC},
  821. {0x44, 0x5A, 0x01F1},
  822. {0x44, 0x7A, 0x01F2},
  823. {0x64, 0x7A, 0x01F3},
  824. {0x2E, 0x2E, 0x2025},
  825. {0x21, 0x21, 0x203C},
  826. {0x3F, 0x21, 0x2048},
  827. {0x21, 0x3F, 0x2049},
  828. {0x52, 0x73, 0x20A8},
  829. {0x4E, 0x6F, 0x2116},
  830. {0x53, 0x4D, 0x2120},
  831. {0x54, 0x4D, 0x2122},
  832. {0x49, 0x49, 0x2161},
  833. {0x49, 0x56, 0x2163},
  834. {0x56, 0x49, 0x2165},
  835. {0x49, 0x58, 0x2168},
  836. {0x58, 0x49, 0x216A},
  837. {0x69, 0x69, 0x2171},
  838. {0x69, 0x76, 0x2173},
  839. {0x76, 0x69, 0x2175},
  840. {0x69, 0x78, 0x2178},
  841. {0x78, 0x69, 0x217A},
  842. {0x31, 0x30, 0x2469},
  843. {0x31, 0x31, 0x246A},
  844. {0x31, 0x32, 0x246B},
  845. {0x31, 0x33, 0x246C},
  846. {0x31, 0x34, 0x246D},
  847. {0x31, 0x35, 0x246E},
  848. {0x31, 0x36, 0x246F},
  849. {0x31, 0x37, 0x2470},
  850. {0x31, 0x38, 0x2471},
  851. {0x31, 0x39, 0x2472},
  852. {0x32, 0x30, 0x2473},
  853. {0x31, 0x2E, 0x2488},
  854. {0x32, 0x2E, 0x2489},
  855. {0x33, 0x2E, 0x248A},
  856. {0x34, 0x2E, 0x248B},
  857. {0x35, 0x2E, 0x248C},
  858. {0x36, 0x2E, 0x248D},
  859. {0x37, 0x2E, 0x248E},
  860. {0x38, 0x2E, 0x248F},
  861. {0x39, 0x2E, 0x2490},
  862. {0x64, 0x61, 0x3372},
  863. {0x41, 0x55, 0x3373},
  864. {0x6F, 0x56, 0x3375},
  865. {0x70, 0x63, 0x3376},
  866. {0x70, 0x41, 0x3380},
  867. {0x6E, 0x41, 0x3381},
  868. {0x6D, 0x41, 0x3383},
  869. {0x6B, 0x41, 0x3384},
  870. {0x4B, 0x42, 0x3385},
  871. {0x4D, 0x42, 0x3386},
  872. {0x47, 0x42, 0x3387},
  873. {0x70, 0x46, 0x338A},
  874. {0x6E, 0x46, 0x338B},
  875. {0x6D, 0x67, 0x338E},
  876. {0x6B, 0x67, 0x338F},
  877. {0x48, 0x7A, 0x3390},
  878. {0x66, 0x6D, 0x3399},
  879. {0x6E, 0x6D, 0x339A},
  880. {0x6D, 0x6D, 0x339C},
  881. {0x63, 0x6D, 0x339D},
  882. {0x6B, 0x6D, 0x339E},
  883. {0x50, 0x61, 0x33A9},
  884. {0x70, 0x73, 0x33B0},
  885. {0x6E, 0x73, 0x33B1},
  886. {0x6D, 0x73, 0x33B3},
  887. {0x70, 0x56, 0x33B4},
  888. {0x6E, 0x56, 0x33B5},
  889. {0x6D, 0x56, 0x33B7},
  890. {0x6B, 0x56, 0x33B8},
  891. {0x4D, 0x56, 0x33B9},
  892. {0x70, 0x57, 0x33BA},
  893. {0x6E, 0x57, 0x33BB},
  894. {0x6D, 0x57, 0x33BD},
  895. {0x6B, 0x57, 0x33BE},
  896. {0x4D, 0x57, 0x33BF},
  897. {0x42, 0x71, 0x33C3},
  898. {0x63, 0x63, 0x33C4},
  899. {0x63, 0x64, 0x33C5},
  900. {0x64, 0x42, 0x33C8},
  901. {0x47, 0x79, 0x33C9},
  902. {0x68, 0x61, 0x33CA},
  903. {0x48, 0x50, 0x33CB},
  904. {0x69, 0x6E, 0x33CC},
  905. {0x4B, 0x4B, 0x33CD},
  906. {0x4B, 0x4D, 0x33CE},
  907. {0x6B, 0x74, 0x33CF},
  908. {0x6C, 0x6D, 0x33D0},
  909. {0x6C, 0x6E, 0x33D1},
  910. {0x6C, 0x78, 0x33D3},
  911. {0x6D, 0x62, 0x33D4},
  912. {0x50, 0x48, 0x33D7},
  913. {0x50, 0x52, 0x33DA},
  914. {0x73, 0x72, 0x33DB},
  915. {0x53, 0x76, 0x33DC},
  916. {0x57, 0x62, 0x33DD},
  917. {0x66, 0x66, 0xFB00},
  918. {0x66, 0x69, 0xFB01},
  919. {0x66, 0x6C, 0xFB02},
  920. {0x73, 0x74, 0xFB06},
  921. {0, 0, 0}
  922. }, *c;
  923. int nc = -1;
  924. for (c = composetbl; c->first; c++) {
  925. if (c->first == first && c->second == second)
  926. return c->composed;
  927. }
  928. if (recurse == 0) {
  929. nc = check_compose_internal(second, first, 1);
  930. if (nc == -1)
  931. nc = check_compose_internal(toupper(first), toupper(second), 1);
  932. if (nc == -1)
  933. nc = check_compose_internal(toupper(second), toupper(first), 1);
  934. }
  935. return nc;
  936. }
  937. int check_compose(int first, int second)
  938. {
  939. return check_compose_internal(first, second, 0);
  940. }
  941. int decode_codepage(char *cp_name)
  942. {
  943. char *s, *d;
  944. const struct cp_list_item *cpi;
  945. int codepage = -1;
  946. CPINFO cpinfo;
  947. if (!cp_name || !*cp_name)
  948. return CP_UTF8; /* default */
  949. for (cpi = cp_list; cpi->name; cpi++) {
  950. s = cp_name;
  951. d = cpi->name;
  952. for (;;) {
  953. while (*s && !isalnum(*s) && *s != ':')
  954. s++;
  955. while (*d && !isalnum(*d) && *d != ':')
  956. d++;
  957. if (*s == 0) {
  958. codepage = cpi->codepage;
  959. if (codepage == CP_UTF8)
  960. goto break_break;
  961. if (codepage == -1)
  962. return codepage;
  963. if (codepage == 0) {
  964. codepage = 65536 + (cpi - cp_list);
  965. goto break_break;
  966. }
  967. if (GetCPInfo(codepage, &cpinfo) != 0)
  968. goto break_break;
  969. }
  970. if (tolower((unsigned char)*s++) != tolower((unsigned char)*d++))
  971. break;
  972. }
  973. }
  974. d = cp_name;
  975. if (tolower((unsigned char)d[0]) == 'c' &&
  976. tolower((unsigned char)d[1]) == 'p')
  977. d += 2;
  978. if (tolower((unsigned char)d[0]) == 'i' &&
  979. tolower((unsigned char)d[1]) == 'b' &&
  980. tolower((unsigned char)d[2]) == 'm')
  981. d += 3;
  982. for (s = d; *s >= '0' && *s <= '9'; s++);
  983. if (*s == 0 && s != d)
  984. codepage = atoi(d); /* CP999 or IBM999 */
  985. if (codepage == CP_ACP)
  986. codepage = GetACP();
  987. if (codepage == CP_OEMCP)
  988. codepage = GetOEMCP();
  989. if (codepage > 65535)
  990. codepage = -2;
  991. break_break:;
  992. if (codepage != -1) {
  993. if (codepage != CP_UTF8 && codepage < 65536) {
  994. if (GetCPInfo(codepage, &cpinfo) == 0) {
  995. codepage = -2;
  996. } else if (cpinfo.MaxCharSize > 1)
  997. codepage = -3;
  998. }
  999. }
  1000. if (codepage == -1 && *cp_name)
  1001. codepage = -2;
  1002. return codepage;
  1003. }
  1004. const char *cp_name(int codepage)
  1005. {
  1006. const struct cp_list_item *cpi, *cpno;
  1007. static char buf[32];
  1008. if (codepage == -1) {
  1009. sprintf(buf, "Use font encoding");
  1010. return buf;
  1011. }
  1012. if (codepage > 0 && codepage < 65536)
  1013. sprintf(buf, "CP%03d", codepage);
  1014. else
  1015. *buf = 0;
  1016. if (codepage >= 65536) {
  1017. cpno = 0;
  1018. for (cpi = cp_list; cpi->name; cpi++)
  1019. if (cpi == cp_list + (codepage - 65536)) {
  1020. cpno = cpi;
  1021. break;
  1022. }
  1023. if (cpno)
  1024. for (cpi = cp_list; cpi->name; cpi++) {
  1025. if (cpno->cp_table == cpi->cp_table)
  1026. return cpi->name;
  1027. }
  1028. } else {
  1029. for (cpi = cp_list; cpi->name; cpi++) {
  1030. if (codepage == cpi->codepage)
  1031. return cpi->name;
  1032. }
  1033. }
  1034. return buf;
  1035. }
  1036. /*
  1037. * Return the nth code page in the list, for use in the GUI
  1038. * configurer.
  1039. */
  1040. const char *cp_enumerate(int index)
  1041. {
  1042. if (index < 0 || index >= lenof(cp_list))
  1043. return NULL;
  1044. return cp_list[index].name;
  1045. }
  1046. void get_unitab(int codepage, wchar_t * unitab, int ftype)
  1047. {
  1048. char tbuf[4];
  1049. int i, max = 256, flg = MB_ERR_INVALID_CHARS;
  1050. if (ftype)
  1051. flg |= MB_USEGLYPHCHARS;
  1052. if (ftype == 2)
  1053. max = 128;
  1054. if (codepage == CP_UTF8) {
  1055. for (i = 0; i < max; i++)
  1056. unitab[i] = i;
  1057. return;
  1058. }
  1059. if (codepage == CP_ACP)
  1060. codepage = GetACP();
  1061. else if (codepage == CP_OEMCP)
  1062. codepage = GetOEMCP();
  1063. if (codepage > 0 && codepage < 65536) {
  1064. for (i = 0; i < max; i++) {
  1065. tbuf[0] = i;
  1066. if (mb_to_wc(codepage, flg, tbuf, 1, unitab + i, 1)
  1067. != 1)
  1068. unitab[i] = 0xFFFD;
  1069. }
  1070. } else {
  1071. int j = 256 - cp_list[codepage & 0xFFFF].cp_size;
  1072. for (i = 0; i < max; i++)
  1073. unitab[i] = i;
  1074. for (i = j; i < max; i++)
  1075. unitab[i] = cp_list[codepage & 0xFFFF].cp_table[i - j];
  1076. }
  1077. }
  1078. int wc_to_mb(int codepage, int flags, const wchar_t *wcstr, int wclen,
  1079. char *mbstr, int mblen, const char *defchr,
  1080. struct unicode_data *ucsdata)
  1081. {
  1082. char *p;
  1083. int i;
  1084. if (ucsdata && codepage == ucsdata->line_codepage && ucsdata->uni_tbl) {
  1085. /* Do this by array lookup if we can. */
  1086. if (wclen < 0) {
  1087. for (wclen = 0; wcstr[wclen++] ;); /* will include the NUL */
  1088. }
  1089. for (p = mbstr, i = 0; i < wclen; i++) {
  1090. wchar_t ch = wcstr[i];
  1091. int by;
  1092. char *p1;
  1093. if (ucsdata->uni_tbl && (p1 = ucsdata->uni_tbl[(ch >> 8) & 0xFF])
  1094. && (by = p1[ch & 0xFF]))
  1095. *p++ = by;
  1096. else if (ch < 0x80)
  1097. *p++ = (char) ch;
  1098. else if (defchr) {
  1099. int j;
  1100. for (j = 0; defchr[j]; j++)
  1101. *p++ = defchr[j];
  1102. }
  1103. #if 1
  1104. else
  1105. *p++ = '.';
  1106. #endif
  1107. assert(p - mbstr < mblen);
  1108. }
  1109. return p - mbstr;
  1110. } else {
  1111. int defused;
  1112. return WideCharToMultiByte(codepage, flags, wcstr, wclen,
  1113. mbstr, mblen, defchr, &defused);
  1114. }
  1115. }
  1116. #endif
  1117. int mb_to_wc(int codepage, int flags, const char *mbstr, int mblen,
  1118. wchar_t *wcstr, int wclen)
  1119. {
  1120. return MultiByteToWideChar(codepage, flags, mbstr, mblen, wcstr, wclen);
  1121. }
  1122. bool is_dbcs_leadbyte(int codepage, char byte)
  1123. {
  1124. return IsDBCSLeadByteEx(codepage, byte);
  1125. }