trfn_ch.h

   1 /* ligatures with unicode aliases */
   2 static char *ligs_utf8[][2] = {
   3         {"ﬀ", "ff"},
   4         {"ﬁ", "fi"},
   5         {"ﬂ", "fl"},
   6         {"ﬃ", "ffi"},
   7         {"ﬄ", "ffl"},
   8         {"ﬆ", "st"},
   9 };
  10
  11 /* AGL exceptions */
  12 static char *agl_exceptions[][2] = {
  13         {"`", "ga"},    /* grave; using quoteleft for ` */
  14         {"'", "aq"},    /* quotesingle; using quoteright for ' */
  15         {"~", "ti"},    /* asciitilde; using tilde for ~ */
  16         {"^", "ha"},    /* asciicircum; using circumflex for ^ */
  17 };
  18
  19 /* troff aliases */
  20 static char *alts[][8] = {
  21         {"’", "'", "cq"},
  22         {"+", "pl"},
  23         {"-", "hy"},
  24         {"/", "sl"},
  25         {"=", "eq"},
  26         {"\"", "dq"},
  27         {"\\", "bs", "rs"},
  28         {"_", "ru", "ul"},
  29         {"‘", "`", "oq"},
  30         {"|", "or"},
  31         {"¡", "!!", "r!"},
  32         {"¢", "c|", "ct"},
  33         {"£", "L-", "ps"},
  34         {"¤", "xo", "cr"},
  35         {"¥", "Y-", "yn"},
  36         {"¦", "||"},
  37         {"§", "so", "sc"},
  38         {"©", "co"},
  39         {"ª", "a_"},
  40         {"«", "<<", "Fo"},
  41         {"¬", "-,", "no"},
  42         {"®", "ro", "rg"},
  43         {"°", "0^", "de"},
  44         {"±", "+-"},
  45         {"²", "2^"},
  46         {"³", "3^"},
  47         {"¶", "P!", "pg"},
  48         {"·", ".^"},
  49         {"¹", "1^"},
  50         {"º", "o_"},
  51         {"»", ">>", "Fc"},
  52         {"¼", "14"},
  53         {"½", "12"},
  54         {"¾", "34"},
  55         {"¿", "??", "r?"},
  56         {"À", "A`"},
  57         {"Á", "A'"},
  58         {"Â", "A^"},
  59         {"Ã", "A~"},
  60         {"Ä", "A:", "A\""},
  61         {"Å", "A*"},
  62         {"Æ", "AE"},
  63         {"Ç", "C,"},
  64         {"È", "E`"},
  65         {"É", "E'"},
  66         {"Ê", "E^"},
  67         {"Ë", "E:"},
  68         {"Ì", "I`"},
  69         {"Í", "I'"},
  70         {"Î", "I^"},
  71         {"Ï", "I:"},
  72         {"Ð", "D-"},
  73         {"Ñ", "N~"},
  74         {"Ò", "O`"},
  75         {"Ó", "O'"},
  76         {"Ô", "O^"},
  77         {"Õ", "O~"},
  78         {"Ö", "O:"},
  79         {"×", "xx", "mu"},
  80         {"Ø", "O/"},
  81         {"Ù", "U`"},
  82         {"Ú", "U'"},
  83         {"Û", "U^"},
  84         {"Ü", "U:"},
  85         {"Ý", "Y'"},
  86         {"Þ", "TH"},
  87         {"ß", "ss"},
  88         {"à", "a`"},
  89         {"á", "a'"},
  90         {"â", "a^"},
  91         {"ã", "a~"},
  92         {"ä", "a:"},
  93         {"å", "a*"},
  94         {"æ", "ae"},
  95         {"ç", "c,"},
  96         {"è", "e`"},
  97         {"é", "e'"},
  98         {"ê", "e^"},
  99         {"ë", "e:"},
 100         {"ì", "i`"},
 101         {"í", "i'"},
 102         {"î", "i^"},
 103         {"ï", "i:"},
 104         {"ð", "d-"},
 105         {"ñ", "n~"},
 106         {"ò", "o`"},
 107         {"ó", "o'"},
 108         {"ô", "o^"},
 109         {"õ", "o~"},
 110         {"ö", "o:"},
 111         {"÷", "di", "-:"},
 112         {"ø", "o/"},
 113         {"ù", "u`"},
 114         {"ú", "u'"},
 115         {"û", "u^"},
 116         {"ü", "u:"},
 117         {"ý", "y'"},
 118         {"þ", "th"},
 119         {"ÿ", "y:"},
 120         {"Č", "C<"},
 121         {"č", "c<"},
 122         {"Ď", "D<"},
 123         {"ď", "d<"},
 124         {"ě", "e<"},
 125         {"ň", "n<"},
 126         {"Ő", "O\""},
 127         {"ő", "o\""},
 128         {"Ř", "R<"},
 129         {"ř", "r<"},
 130         {"Š", "S<"},
 131         {"š", "s<"},
 132         {"Ť", "T<"},
 133         {"ť", "t<"},
 134         {"Ů", "U*"},
 135         {"ů", "u*"},
 136         {"Ű", "U\""},
 137         {"ű", "u\""},
 138         {"Ÿ", "Y:"},
 139         {"Ž", "Z<"},
 140         {"ž", "z<"},
 141         {"ƒ", "fn",},
 142         {"¸", ",,", ",a"},
 143         {"´", "aa", "\\'"},
 144         {"¯", "-a"},
 145         {"¨", "\"\"", ":a"},
 146         {"ga", "\\`"},
 147         {"ˆ", "^", "^a"},
 148         {"ˇ", "va"},
 149         {"˘", "Ua"},
 150         {"˙", ".a"},
 151         {"˚", "oa"},
 152         {"˛", "Ca"},
 153         {"˝", "\"a"},
 154         {"˜", "~"},
 155         {"Α", "*A"},
 156         {"Β", "*B"},
 157         {"Γ", "*G"},
 158         {"Ε", "*E"},
 159         {"Ζ", "*Z"},
 160         {"Η", "*Y"},
 161         {"Θ", "*H"},
 162         {"Ι", "*I"},
 163         {"Κ", "*K"},
 164         {"Λ", "*L"},
 165         {"Μ", "*M"},
 166         {"Ν", "*N"},
 167         {"Ξ", "*C"},
 168         {"Ο", "*O"},
 169         {"Π", "*P"},
 170         {"Ρ", "*R"},
 171         {"Σ", "*S"},
 172         {"Τ", "*T"},
 173         {"Υ", "*U"},
 174         {"Φ", "*F"},
 175         {"Χ", "*X"},
 176         {"Ψ", "*Q"},
 177         {"Ω", "*W", "Ω"}, {"Ω", "Ω", "*W"},
 178         {"α", "*a"},
 179         {"β", "*b"},
 180         {"γ", "*g"},
 181         {"δ", "*d"},
 182         {"ε", "*e"},
 183         {"ζ", "*z"},
 184         {"η", "*y"},
 185         {"θ", "*h"},
 186         {"ι", "*i"},
 187         {"κ", "*k"},
 188         {"λ", "*l"},
 189         {"μ", "*m", "µ", "/u"}, {"µ", "/u", "*m", "μ"},
 190         {"ν", "*n"},
 191         {"ξ", "*c"},
 192         {"ο", "*o"},
 193         {"π", "*p"},
 194         {"ρ", "*r"},
 195         {"ς", "ts"},
 196         {"σ", "*s"},
 197         {"τ", "*t"},
 198         {"υ", "*u"},
 199         {"φ", "*f"},
 200         {"χ", "*x"},
 201         {"ψ", "*q"},
 202         {"ω", "*w"},
 203         {"–", "en", "\\-"},
 204         {"—", "em", "--"},
 205         {"‚", "bq"},
 206         {"“", "``", "lq"},
 207         {"”", "''", "rq"},
 208         {"†", "dg"},
 209         {"‡", "dd"},
 210         {"•", "bu"},
 211         {"…", "el"},
 212         {"‰", "%0"},
 213         {"′", "fm"},
 214         {"‹", "fo"},
 215         {"›", "fc"},
 216         {"⁄", "fr"},
 217         {"ℑ", "If"},
 218         {"ℛ", "ws"},
 219         {"ℜ", "Rf"},
 220         {"ℵ", "af"},
 221         {"←", "<-"},
 222         {"↑", "ua"},
 223         {"→", "->"},
 224         {"↓", "da"},
 225         {"↔", "ab", "<>"},
 226         {"↵", "CR"},
 227         {"∀", "fa"},
 228         {"∂", "pd"},
 229         {"∃", "te"},
 230         {"∅", "es"},
 231         {"∆", "*D", "Δ"}, {"Δ", "∆", "*D"},
 232         {"∇", "gr"},
 233         {"∈", "mo"},
 234         {"∉", "!m"},
 235         {"∋", "st"},
 236         {"∏", "pr"},
 237         {"∑", "su"},
 238         {"−", "mi"},
 239         {"∓", "-+"},
 240         {"∗", "**"},
 241         {"√", "sr"},
 242         {"∝", "pt"},
 243         {"∞", "if"},
 244         {"∠", "an"},
 245         {"∧", "l&"},
 246         {"∨", "l|"},
 247         {"∩", "ca"},
 248         {"∪", "cu"},
 249         {"∫", "is"},
 250         {"∴", "tf"},
 251         {"∼", "ap"},
 252         {"≅", "cg", "=~"},
 253         {"≈", "~~"},
 254         {"≠", "!="},
 255         {"≡", "=="},
 256         {"≤", "<="},
 257         {"≥", ">="},
 258         {"⊂", "sb"},
 259         {"⊃", "sp"},
 260         {"⊄", "!b"},
 261         {"⊆", "ib"},
 262         {"⊇", "ip"},
 263         {"⊕", "O+"},
 264         {"⊗", "Ox"},
 265         {"⊥", "pp"},
 266         {"⋅", "c."},
 267         {"〈", "b<"},
 268         {"〉", "b>"},
 269         {"◊", "lz"},
 270         {"○", "ci"},
 271         {"⟨", "la"},
 272         {"⟩", "ra"},
 273         {"", "co"},
 274         {"", "rg"},
 275         {"", "tm"},
 276         {"", "rn"},
 277         {"", "av"},
 278         {"", "ah"},
 279         {"", "RG"},
 280         {"", "CO"},
 281         {"", "TM"},
 282         {"", "LT"},
 283         {"", "br", "LX"},
 284         {"", "LB"},
 285         {"⎛", "LT"},
 286         {"⎜", "LX"},
 287         {"⎝", "LB"},
 288         {"", "lc"},
 289         {"", "lx"},
 290         {"", "lf"},
 291         {"⎡", "lc"},
 292         {"⎢", "lx"},
 293         {"⎣", "lf"},
 294         {"", "lt"},
 295         {"", "lk"},
 296         {"", "lb"},
 297         {"", "bv", "|"},
 298         {"⎧", "lt"},
 299         {"⎨", "lk"},
 300         {"⎩", "lb"},
 301         {"⎪", "bv"},
 302         {"", "RT"},
 303         {"", "RX"},
 304         {"", "RB"},
 305         {"⎞", "RT"},
 306         {"⎟", "RX"},
 307         {"⎠", "RB"},
 308         {"", "rc"},
 309         {"", "rx"},
 310         {"", "rf"},
 311         {"⎤", "rc"},
 312         {"⎥", "rx"},
 313         {"⎦", "rf"},
 314         {"", "rt"},
 315         {"", "rk"},
 316         {"", "rb"},
 317         {"⎫", "rt"},
 318         {"⎬", "rk"},
 319         {"⎭", "rb"},
 320         {"ﬀ", "ff"},
 321         {"ﬁ", "fi"},
 322         {"ﬂ", "fl"},
 323         {"ﬃ", "ffi", "Fi"},
 324         {"ﬄ", "ffl", "Fl"},
 325         {"ﬆ", "st"},
 326         {"ﺀ", "ء"},
 327         {"ﺁ", "آ"},
 328         {"ﺃ", "أ"},
 329         {"ﺅ", "ؤ"},
 330         {"ﺇ", "إ"},
 331         {"ﺉ", "ئ"},
 332         {"ﺍ", "ا"},
 333         {"ﺏ", "ب"},
 334         {"ﺓ", "ة"},
 335         {"ﺕ", "ت"},
 336         {"ﺙ", "ث"},
 337         {"ﺝ", "ج"},
 338         {"ﺡ", "ح"},
 339         {"ﺥ", "خ"},
 340         {"ﺩ", "د"},
 341         {"ﺫ", "ذ"},
 342         {"ﺭ", "ر"},
 343         {"ﺯ", "ز"},
 344         {"ﺱ", "س"},
 345         {"ﺵ", "ش"},
 346         {"ﺹ", "ص"},
 347         {"ﺽ", "ض"},
 348         {"ﻁ", "ط"},
 349         {"ﻅ", "ظ"},
 350         {"ﻉ", "ع"},
 351         {"ﻍ", "غ"},
 352         {"ـ", "ـ"},
 353         {"ﻑ", "ف"},
 354         {"ﻕ", "ق"},
 355         {"ﻙ", "ك"},
 356         {"ﻝ", "ل"},
 357         {"ﻡ", "م"},
 358         {"ﻥ", "ن"},
 359         {"ﻩ", "ه"},
 360         {"ﻭ", "و"},
 361         {"ﻯ", "ى"},
 362         {"ﻱ", "ي"},
 363         {"ﭖ", "پ"},
 364         {"ﭺ", "چ"},
 365         {"ﮊ", "ژ"},
 366         {"ﮎ", "ک"},
 367         {"ﮒ", "گ"},
 368         {"ﯼ", "ی"},
 369 };
 370
 371 /* different shapes of arabic and farsi characters */
 372 static struct achar {
 373         char *name;
 374         unsigned c;
 375         unsigned s;
 376         unsigned i;
 377         unsigned m;
 378         unsigned f;
 379 } achars[] = {
 380         {"hamza", 0x0621, 0xfe80},
 381         {"alefwithmaddaabove", 0x0622, 0xfe81, 0, 0, 0xfe82},
 382         {"alefwithhamzaabove", 0x0623, 0xfe83, 0, 0, 0xfe84},
 383         {"wawwithhamzaabove", 0x0624, 0xfe85, 0, 0, 0xfe86},
 384         {"alefwithhamzabelow", 0x0625, 0xfe87, 0, 0, 0xfe88},
 385         {"yehwithhamzaabove", 0x0626, 0xfe89, 0xfe8b, 0xfe8c, 0xfe8a},
 386         {"alef", 0x0627, 0xfe8d, 0, 0, 0xfe8e},
 387         {"arabicalef", 0x0627},
 388         {"beh", 0x0628, 0xfe8f, 0xfe91, 0xfe92, 0xfe90},
 389         {"tehmarbuta", 0x0629, 0xfe93, 0, 0, 0xfe94},
 390         {"teh", 0x062a, 0xfe95, 0xfe97, 0xfe98, 0xfe96},
 391         {"theh", 0x062b, 0xfe99, 0xfe9b, 0xfe9c, 0xfe9a},
 392         {"jeem", 0x062c, 0xfe9d, 0xfe9f, 0xfea0, 0xfe9e},
 393         {"hah", 0x062d, 0xfea1, 0xfea3, 0xfea4, 0xfea2},
 394         {"khah", 0x062e, 0xfea5, 0xfea7, 0xfea8, 0xfea6},
 395         {"dal", 0x062f, 0xfea9, 0, 0, 0xfeaa},
 396         {"thal", 0x0630, 0xfeab, 0, 0, 0xfeac},
 397         {"reh", 0x0631, 0xfead, 0, 0, 0xfeae},
 398         {"zain", 0x0632, 0xfeaf, 0, 0, 0xfeb0},
 399         {"seen", 0x0633, 0xfeb1, 0xfeb3, 0xfeb4, 0xfeb2},
 400         {"sheen", 0x0634, 0xfeb5, 0xfeb7, 0xfeb8, 0xfeb6},
 401         {"sad", 0x0635, 0xfeb9, 0xfebb, 0xfebc, 0xfeba},
 402         {"dad", 0x0636, 0xfebd, 0xfebf, 0xfec0, 0xfebe},
 403         {"tah", 0x0637, 0xfec1, 0xfec3, 0xfec4, 0xfec2},
 404         {"zah", 0x0638, 0xfec5, 0xfec7, 0xfec8, 0xfec6},
 405         {"ain", 0x0639, 0xfec9, 0xfecb, 0xfecc, 0xfeca},
 406         {"ghain", 0x063a, 0xfecd, 0xfecf, 0xfed0, 0xfece},
 407         {"tatweel", 0x0640},
 408         {"feh", 0x0641, 0xfed1, 0xfed3, 0xfed4, 0xfed2},
 409         {"qaf", 0x0642, 0xfed5, 0xfed7, 0xfed8, 0xfed6},
 410         {"kaf", 0x0643, 0xfed9, 0xfedb, 0xfedc, 0xfeda},
 411         {"lam", 0x0644, 0xfedd, 0xfedf, 0xfee0, 0xfede},
 412         {"meem", 0x0645, 0xfee1, 0xfee3, 0xfee4, 0xfee2},
 413         {"noon", 0x0646, 0xfee5, 0xfee7, 0xfee8, 0xfee6},
 414         {"heh", 0x0647, 0xfee9, 0xfeeb, 0xfeec, 0xfeea},
 415         {"waw", 0x0648, 0xfeed, 0, 0, 0xfeee},
 416         {"alefmaksura", 0x0649, 0xfeef, 0, 0, 0xfef0},
 417         {"yeh", 0x064a, 0xfef1, 0xfef3, 0xfef4, 0xfef2},
 418         {"fathatan", 0x064b, 0xfe70},
 419         {"dammatan", 0x064c, 0xfe72},
 420         {"kasratan", 0x064d, 0xfe74},
 421         {"fatha", 0x064e, 0xfe76, 0, 0xfe77, 0},
 422         {"damma", 0x064f, 0xfe78, 0, 0xfe79, 0},
 423         {"kasra", 0x0650, 0xfe7a, 0, 0xfe7b, 0},
 424         {"shadda", 0x0651, 0xfe7c, 0, 0xfe7c, 0},
 425         {"sukun", 0x0652, 0xfe7e, 0, 0xfe7f, 0},
 426         {"peh", 0x067e, 0xfb56, 0xfb58, 0xfb59, 0xfb57},
 427         {"tcheh", 0x0686, 0xfb7a, 0xfb7c, 0xfb7d, 0xfb7b},
 428         {"jeh", 0x0698, 0xfb8a, 0, 0, 0xfb8b},
 429         {"keheh", 0x06a9, 0xfb8e, 0xfb90, 0xfb91, 0xfb8f},
 430         {"gaf", 0x06af, 0xfb92, 0xfb94, 0xfb95, 0xfb93},
 431         {"farsiyeh", 0x06cc, 0xfbfc, 0xfbfe, 0xfbff, 0xfbfd},
 432         {"lamwithalef", 0xfefb, 0xfefb, 0, 0, 0xfefc},
 433         {"arabiccomma", 0x060c},
 434         {"arabicsemicolon", 0x061b},
 435         {"arabicquestionmark", 0x061f},
 436         {"arabicindicdigitzero", 0x0660},
 437         {"arabicindicdigitone", 0x0661},
 438         {"arabicindicdigittwo", 0x0662},
 439         {"arabicindicdigitthree", 0x0663},
 440         {"arabicindicdigitfour", 0x0664},
 441         {"arabicindicdigitfive", 0x0665},
 442         {"arabicindicdigitsix", 0x0666},
 443         {"arabicindicdigitseven", 0x0667},
 444         {"arabicindicdigiteight", 0x0668},
 445         {"arabicindicdigitnine", 0x0669},
 446         {"arabicpercentsign", 0x066a},
 447         {"extendedarabicindicdigitzero", 0x06f0},
 448         {"extendedarabicindicdigitone", 0x06f1},
 449         {"extendedarabicindicdigittwo", 0x06f2},
 450         {"extendedarabicindicdigitthree", 0x06f3},
 451         {"extendedarabicindicdigitfour", 0x06f4},
 452         {"extendedarabicindicdigitfive", 0x06f5},
 453         {"extendedarabicindicdigitsix", 0x06f6},
 454         {"extendedarabicindicdigitseven", 0x06f7},
 455         {"extendedarabicindicdigiteight", 0x06f8},
 456         {"extendedarabicindicdigitnine", 0x06f9},
 457         {"zeronojoin", 0x200c},
 458         {"zerojoin", 0x200d},
 459 };
 460
 461 static int ctype_ascii[128] = {
 462         ['!'] = 2, ['"'] = 2, ['#'] = 2, ['$'] = 2, ['%'] = 2,
 463         ['&'] = 2, ['\''] = 2, ['('] = 3, [')'] = 3, ['*'] = 2,
 464         ['+'] = 0, [','] = 1, ['-'] = 0, ['.'] = 0, ['/'] = 2,
 465         ['0'] = 2, ['1'] = 2, ['2'] = 2, ['3'] = 2, ['4'] = 2,
 466         ['5'] = 2, ['6'] = 2, ['7'] = 2, ['8'] = 2, ['9'] = 2,
 467         [':'] = 0, [';'] = 1, ['<'] = 0, ['='] = 0, ['>'] = 0,
 468         ['?'] = 2, ['@'] = 3, ['A'] = 2, ['B'] = 2, ['C'] = 2,
 469         ['D'] = 2, ['E'] = 2, ['F'] = 2, ['G'] = 2, ['H'] = 2,
 470         ['I'] = 2, ['J'] = 3, ['K'] = 2, ['L'] = 2, ['M'] = 2,
 471         ['N'] = 2, ['O'] = 2, ['P'] = 2, ['Q'] = 3, ['R'] = 2,
 472         ['S'] = 2, ['T'] = 2, ['U'] = 2, ['V'] = 2, ['W'] = 2,
 473         ['X'] = 2, ['Y'] = 2, ['Z'] = 2, ['['] = 3, ['\\'] = 2,
 474         [']'] = 3, ['^'] = 2, ['_'] = 1, ['a'] = 0, ['b'] = 2,
 475         ['c'] = 0, ['d'] = 2, ['e'] = 0, ['f'] = 2, ['g'] = 1,
 476         ['h'] = 2, ['i'] = 2, ['j'] = 3, ['k'] = 2, ['l'] = 2,
 477         ['m'] = 0, ['n'] = 0, ['o'] = 0, ['p'] = 1, ['q'] = 1,
 478         ['r'] = 0, ['s'] = 0, ['t'] = 2, ['u'] = 0, ['v'] = 0,
 479         ['w'] = 0, ['x'] = 0, ['y'] = 1, ['z'] = 0, ['{'] = 3,
 480         ['|'] = 3, ['}'] = 3, ['~'] = 0,
 481 };