dmd/entity.c

   1
   2 // Copyright (c) 1999-2006 by Digital Mars
   3 // All Rights Reserved
   4 // written by Walter Bright
   5 // http://www.digitalmars.com
   6 // License for redistribution is by either the Artistic License
   7 // in artistic.txt, or the GNU General Public License in gnu.txt.
   8 // See the included readme.txt for details.
   9
  10
  11 #include <string.h>
  12
  13 /*********************************************
  14  * Convert from named entity to its encoding.
  15  * For reference:
  16  *      http://www.htmlhelp.com/reference/html40/entities/
  17  *      http://www.w3.org/TR/1999/REC-html401-19991224/sgml/entities.html
  18  */
  19
  20 struct NameId
  21 {
  22     char *name;
  23     unsigned short value;
  24 };
  25
  26 #if IN_GCC
  27 static NameId namesA[]={
  28         "Aacgr",        0x0386,
  29         "aacgr",        0x03AC,
  30         "Aacute",       0x00C1,
  31         "aacute",       0x00E1,
  32         "Abreve",       0x0102,
  33         "abreve",       0x0103,
  34         "Acirc",        0x00C2,
  35         "acirc",        0x00E2,
  36         "acute",        0x00B4,
  37         "Acy",          0x0410,
  38         "acy",          0x0430,
  39         "AElig",        0x00C6,
  40         "aelig",        0x00E6,
  41         "Agr",          0x0391,
  42         "agr",          0x03B1,
  43         "Agrave",       0x00C0,
  44         "agrave",       0x00E0,
  45         "aleph",        0x2135,
  46         "alpha",        0x03B1,
  47         "Amacr",        0x0100,
  48         "amacr",        0x0101,
  49         "amalg",        0x2210,
  50         "amp",          0x0026,
  51         "and",          0x2227,
  52         "ang",          0x2220,
  53         "ang90",        0x221F,
  54         "angmsd",       0x2221,
  55         "angsph",       0x2222,
  56         "angst",        0x212B,
  57         "Aogon",        0x0104,
  58         "aogon",        0x0105,
  59         "ap",           0x2248,
  60         "ape",          0x224A,
  61         "apos",         0x0027,
  62         "Aring",        0x00C5,
  63         "aring",        0x00E5,
  64         "ast",          0x002A,
  65         "asymp",        0x224D,
  66         "Atilde",       0x00C3,
  67         "atilde",       0x00E3,
  68         "Auml",         0x00C4,
  69         "auml",         0x00E4,
  70         NULL,           0
  71 };
  72
  73 static NameId namesB[]={
  74         "barwed",       0x22BC,
  75         "Barwed",       0x2306,
  76         "bcong",        0x224C,
  77         "Bcy",          0x0411,
  78         "bcy",          0x0431,
  79         "becaus",       0x2235,
  80         "bepsi",        0x220D,
  81         "bernou",       0x212C,
  82         "beta",         0x03B2,
  83         "beth",         0x2136,
  84         "Bgr",          0x0392,
  85         "bgr",          0x03B2,
  86         "blank",        0x2423,
  87         "blk12",        0x2592,
  88         "blk14",        0x2591,
  89         "blk34",        0x2593,
  90         "block",        0x2588,
  91         "bottom",       0x22A5,
  92         "bowtie",       0x22C8,
  93         "boxdl",        0x2510,
  94         "boxDL",        0x2555,
  95         "boxdL",        0x2556,
  96         "boxDl",        0x2557,
  97         "boxdr",        0x250C,
  98         "boxDR",        0x2552,
  99         "boxDr",        0x2553,
 100         "boxdR",        0x2554,
 101         "boxh",         0x2500,
 102         "boxH",         0x2550,
 103         "boxhd",        0x252C,
 104         "boxhD",        0x2564,
 105         "boxHD",        0x2565,
 106         "boxHd",        0x2566,
 107         "boxhu",        0x2534,
 108         "boxhU",        0x2567,
 109         "boxHU",        0x2568,
 110         "boxHu",        0x2569,
 111         "boxul",        0x2518,
 112         "boxUL",        0x255B,
 113         "boxUl",        0x255C,
 114         "boxuL",        0x255D,
 115         "boxur",        0x2514,
 116         "boxUR",        0x2558,
 117         "boxuR",        0x2559,
 118         "boxUr",        0x255A,
 119         "boxv",         0x2502,
 120         "boxV",         0x2551,
 121         "boxvh",        0x253C,
 122         "boxvH",        0x256A,
 123         "boxVH",        0x256B,
 124         "boxVh",        0x256C,
 125         "boxvl",        0x2524,
 126         "boxvL",        0x2561,
 127         "boxVL",        0x2562,
 128         "boxVl",        0x2563,
 129         "boxvr",        0x251C,
 130         "boxvR",        0x255E,
 131         "boxVR",        0x255F,
 132         "boxVr",        0x2560,
 133         "bprime",       0x2035,
 134         "breve",        0x02D8,
 135         "brvbar",       0x00A6,
 136         "bsim",         0x223D,
 137         "bsime",        0x22CD,
 138         "bsol",         0x005C,
 139         "bull",         0x2022,
 140         "bump",         0x224E,
 141         "bumpe",        0x224F,
 142         NULL,           0
 143 };
 144
 145 static NameId namesC[]={
 146         "Cacute",       0x0106,
 147         "cacute",       0x0107,
 148         "cap",          0x2229,
 149         "Cap",          0x22D2,
 150         "caret",        0x2041,
 151         "caron",        0x02C7,
 152         "Ccaron",       0x010C,
 153         "ccaron",       0x010D,
 154         "Ccedil",       0x00C7,
 155         "ccedil",       0x00E7,
 156         "Ccirc",        0x0108,
 157         "ccirc",        0x0109,
 158         "Cdot",         0x010A,
 159         "cdot",         0x010B,
 160         "cedil",        0x00B8,
 161         "cent",         0x00A2,
 162         "CHcy",         0x0427,
 163         "chcy",         0x0447,
 164         "check",        0x2713,
 165         "chi",          0x03C7,
 166         "cir",          0x25CB,
 167         "circ",         0x005E,
 168         "cire",         0x2257,
 169         "clubs",        0x2663,
 170         "colon",        0x003A,
 171         "colone",       0x2254,
 172         "comma",        0x002C,
 173         "commat",       0x0040,
 174         "comp",         0x2201,
 175         "compfn",       0x2218,
 176         "cong",         0x2245,
 177         "conint",       0x222E,
 178         "coprod",       0x2210,
 179         "copy",         0x00A9,
 180         "copysr",       0x2117,
 181         "cross",        0x2717,
 182         "cuepr",        0x22DE,
 183         "cuesc",        0x22DF,
 184         "cularr",       0x21B6,
 185         "cup",          0x222A,
 186         "Cup",          0x22D3,
 187         "cupre",        0x227C,
 188         "curarr",       0x21B7,
 189         "curren",       0x00A4,
 190         "cuvee",        0x22CE,
 191         "cuwed",        0x22CF,
 192         NULL,           0
 193 };
 194
 195 static NameId namesD[]={
 196         "dagger",       0x2020,
 197         "Dagger",       0x2021,
 198         "daleth",       0x2138,
 199         "darr",         0x2193,
 200         "dArr",         0x21D3,
 201         "darr2",        0x21CA,
 202         "dash",         0x2010,
 203         "dashv",        0x22A3,
 204         "dblac",        0x02DD,
 205         "Dcaron",       0x010E,
 206         "dcaron",       0x010F,
 207         "Dcy",          0x0414,
 208         "dcy",          0x0434,
 209         "deg",          0x00B0,
 210         "Delta",        0x0394,
 211         "delta",        0x03B4,
 212         "Dgr",          0x0394,
 213         "dgr",          0x03B4,
 214         "dharl",        0x21C3,
 215         "dharr",        0x21C2,
 216         "diam",         0x22C4,
 217         "diams",        0x2666,
 218         "die",          0x00A8,
 219         "divide",       0x00F7,
 220         "divonx",       0x22C7,
 221         "DJcy",         0x0402,
 222         "djcy",         0x0452,
 223         "dlarr",        0x2199,
 224         "dlcorn",       0x231E,
 225         "dlcrop",       0x230D,
 226         "dollar",       0x0024,
 227         "Dot",          0x00A8,
 228         "dot",          0x02D9,
 229         "DotDot",       0x20DC,
 230         "drarr",        0x2198,
 231         "drcorn",       0x231F,
 232         "drcrop",       0x230C,
 233         "DScy",         0x0405,
 234         "dscy",         0x0455,
 235         "Dstrok",       0x0110,
 236         "dstrok",       0x0111,
 237         "dtri",         0x25BF,
 238         "dtrif",        0x25BE,
 239         "DZcy",         0x040F,
 240         "dzcy",         0x045F,
 241         NULL,           0
 242 };
 243
 244 static NameId namesE[]={
 245         "Eacgr",        0x0388,
 246         "eacgr",        0x03AD,
 247         "Eacute",       0x00C9,
 248         "eacute",       0x00E9,
 249         "Ecaron",       0x011A,
 250         "ecaron",       0x011B,
 251         "ecir",         0x2256,
 252         "Ecirc",        0x00CA,
 253         "ecirc",        0x00EA,
 254         "ecolon",       0x2255,
 255         "Ecy",          0x042D,
 256         "ecy",          0x044D,
 257         "Edot",         0x0116,
 258         "edot",         0x0117,
 259         "eDot",         0x2251,
 260         "EEacgr",       0x0389,
 261         "eeacgr",       0x03AE,
 262         "EEgr",         0x0397,
 263         "eegr",         0x03B7,
 264         "efDot",        0x2252,
 265         "Egr",          0x0395,
 266         "egr",          0x03B5,
 267         "Egrave",       0x00C8,
 268         "egrave",       0x00E8,
 269         "egs",          0x22DD,
 270         "ell",          0x2113,
 271         "els",          0x22DC,
 272         "Emacr",        0x0112,
 273         "emacr",        0x0113,
 274         "empty",        0x2205,
 275         "emsp",         0x2003,
 276         "emsp13",       0x2004,
 277         "emsp14",       0x2005,
 278         "ENG",          0x014A,
 279         "eng",          0x014B,
 280         "ensp",         0x2002,
 281         "Eogon",        0x0118,
 282         "eogon",        0x0119,
 283         "epsi",         0x220A,
 284         "epsis",        0x220A,
 285         "epsiv",        0x03B5,
 286         "equals",       0x003D,
 287         "equiv",        0x2261,
 288         "erDot",        0x2253,
 289         "esdot",        0x2250,
 290         "eta",          0x03B7,
 291         "ETH",          0x00D0,
 292         "eth",          0x00F0,
 293         "Euml",         0x00CB,
 294         "euml",         0x00EB,
 295         "excl",         0x0021,
 296         "exist",        0x2203,
 297         NULL,           0
 298 };
 299
 300 static NameId namesF[]={
 301         "Fcy",          0x0424,
 302         "fcy",          0x0444,
 303         "female",       0x2640,
 304         "ffilig",       0xFB03,
 305         "fflig",        0xFB00,
 306         "ffllig",       0xFB04,
 307         "filig",        0xFB01,
 308         "flat",         0x266D,
 309         "fllig",        0xFB02,
 310         "fnof",         0x0192,
 311         "forall",       0x2200,
 312         "fork",         0x22D4,
 313         "frac12",       0x00BD,
 314         "frac13",       0x2153,
 315         "frac14",       0x00BC,
 316         "frac15",       0x2155,
 317         "frac16",       0x2159,
 318         "frac18",       0x215B,
 319         "frac23",       0x2154,
 320         "frac25",       0x2156,
 321         "frac34",       0x00BE,
 322         "frac35",       0x2157,
 323         "frac38",       0x215C,
 324         "frac45",       0x2158,
 325         "frac56",       0x215A,
 326         "frac58",       0x215D,
 327         "frac78",       0x215E,
 328         "frown",        0x2322,
 329         NULL,           0
 330 };
 331
 332 static NameId namesG[]={
 333         "gacute",       0x01F5,
 334         "Gamma",        0x0393,
 335         "gamma",        0x03B3,
 336         "gammad",       0x03DC,
 337         "gap",          0x2273,
 338         "Gbreve",       0x011E,
 339         "gbreve",       0x011F,
 340         "Gcedil",       0x0122,
 341         "Gcirc",        0x011C,
 342         "gcirc",        0x011D,
 343         "Gcy",          0x0413,
 344         "gcy",          0x0433,
 345         "Gdot",         0x0120,
 346         "gdot",         0x0121,
 347         "ge",           0x2265,
 348         "gE",           0x2267,
 349         "gel",          0x22DB,
 350         "gEl",          0x22DB,
 351         "ges",          0x2265,
 352         "Gg",           0x22D9,
 353         "Ggr",          0x0393,
 354         "ggr",          0x03B3,
 355         "gimel",        0x2137,
 356         "GJcy",         0x0403,
 357         "gjcy",         0x0453,
 358         "gl",           0x2277,
 359         "gnap",         0xE411,
 360         "gne",          0x2269,
 361         "gnE",          0x2269,
 362         "gnsim",        0x22E7,
 363         "grave",        0x0060,
 364         "gsdot",        0x22D7,
 365         "gsim",         0x2273,
 366         "gt",           0x003E,
 367         "Gt",           0x226B,
 368         "gvnE",         0x2269,
 369         NULL,           0
 370 };
 371
 372 static NameId namesH[]={
 373         "hairsp",       0x200A,
 374         "half",         0x00BD,
 375         "hamilt",       0x210B,
 376         "HARDcy",       0x042A,
 377         "hardcy",       0x044A,
 378         "harr",         0x2194,
 379         "hArr",         0x21D4,
 380         "harrw",        0x21AD,
 381         "Hcirc",        0x0124,
 382         "hcirc",        0x0125,
 383         "hearts",       0x2665,
 384         "hellip",       0x2026,
 385         "horbar",       0x2015,
 386         "Hstrok",       0x0126,
 387         "hstrok",       0x0127,
 388         "hybull",       0x2043,
 389         "hyphen",       0x002D,
 390         NULL,           0
 391 };
 392
 393 static NameId namesI[]={
 394         "Iacgr",        0x038A,
 395         "iacgr",        0x03AF,
 396         "Iacute",       0x00CD,
 397         "iacute",       0x00ED,
 398         "Icirc",        0x00CE,
 399         "icirc",        0x00EE,
 400         "Icy",          0x0418,
 401         "icy",          0x0438,
 402         "idiagr",       0x0390,
 403         "Idigr",        0x03AA,
 404         "idigr",        0x03CA,
 405         "Idot",         0x0130,
 406         "IEcy",         0x0415,
 407         "iecy",         0x0435,
 408         "iexcl",        0x00A1,
 409         "iff",          0x21D4,
 410         "Igr",          0x0399,
 411         "igr",          0x03B9,
 412         "Igrave",       0x00CC,
 413         "igrave",       0x00EC,
 414         "IJlig",        0x0132,
 415         "ijlig",        0x0133,
 416         "Imacr",        0x012A,
 417         "imacr",        0x012B,
 418         "image",        0x2111,
 419         "incare",       0x2105,
 420         "infin",        0x221E,
 421         "inodot",       0x0131,
 422         "int",          0x222B,
 423         "intcal",       0x22BA,
 424         "IOcy",         0x0401,
 425         "iocy",         0x0451,
 426         "Iogon",        0x012E,
 427         "iogon",        0x012F,
 428         "iota",         0x03B9,
 429         "iquest",       0x00BF,
 430         "isin",         0x220A,
 431         "Itilde",       0x0128,
 432         "itilde",       0x0129,
 433         "Iukcy",        0x0406,
 434         "iukcy",        0x0456,
 435         "Iuml",         0x00CF,
 436         "iuml",         0x00EF,
 437         NULL,           0
 438 };
 439
 440 static NameId namesJ[]={
 441         "Jcirc",        0x0134,
 442         "jcirc",        0x0135,
 443         "Jcy",          0x0419,
 444         "jcy",          0x0439,
 445         "Jsercy",       0x0408,
 446         "jsercy",       0x0458,
 447         "Jukcy",        0x0404,
 448         "jukcy",        0x0454,
 449         NULL,           0
 450 };
 451
 452 static NameId namesK[]={
 453         "kappa",        0x03BA,
 454         "kappav",       0x03F0,
 455         "Kcedil",       0x0136,
 456         "kcedil",       0x0137,
 457         "Kcy",          0x041A,
 458         "kcy",          0x043A,
 459         "Kgr",          0x039A,
 460         "kgr",          0x03BA,
 461         "kgreen",       0x0138,
 462         "KHcy",         0x0425,
 463         "khcy",         0x0445,
 464         "KHgr",         0x03A7,
 465         "khgr",         0x03C7,
 466         "KJcy",         0x040C,
 467         "kjcy",         0x045C,
 468         NULL,           0
 469 };
 470
 471 static NameId namesL[]={
 472         "lAarr",        0x21DA,
 473         "Lacute",       0x0139,
 474         "lacute",       0x013A,
 475         "lagran",       0x2112,
 476         "Lambda",       0x039B,
 477         "lambda",       0x03BB,
 478         "lang",         0x3008,
 479         "lap",          0x2272,
 480         "laquo",        0x00AB,
 481         "larr",         0x2190,
 482         "Larr",         0x219E,
 483         "lArr",         0x21D0,
 484         "larr2",        0x21C7,
 485         "larrhk",       0x21A9,
 486         "larrlp",       0x21AB,
 487         "larrtl",       0x21A2,
 488         "Lcaron",       0x013D,
 489         "lcaron",       0x013E,
 490         "Lcedil",       0x013B,
 491         "lcedil",       0x013C,
 492         "lceil",        0x2308,
 493         "lcub",         0x007B,
 494         "Lcy",          0x041B,
 495         "lcy",          0x043B,
 496         "ldot",         0x22D6,
 497         "ldquo",        0x201C,
 498         "ldquor",       0x201E,
 499         "le",           0x2264,
 500         "lE",           0x2266,
 501         "leg",          0x22DA,
 502         "lEg",          0x22DA,
 503         "les",          0x2264,
 504         "lfloor",       0x230A,
 505         "lg",           0x2276,
 506         "Lgr",          0x039B,
 507         "lgr",          0x03BB,
 508         "lhard",        0x21BD,
 509         "lharu",        0x21BC,
 510         "lhblk",        0x2584,
 511         "LJcy",         0x0409,
 512         "ljcy",         0x0459,
 513         "Ll",           0x22D8,
 514         "Lmidot",       0x013F,
 515         "lmidot",       0x0140,
 516         "lnap",         0xE2A2,
 517         "lne",          0x2268,
 518         "lnE",          0x2268,
 519         "lnsim",        0x22E6,
 520         "lowast",       0x2217,
 521         "lowbar",       0x005F,
 522         "loz",          0x25CA,
 523         "lozf",         0x2726,
 524         "lpar",         0x0028,
 525         "lrarr2",       0x21C6,
 526         "lrhar2",       0x21CB,
 527         "lsh",          0x21B0,
 528         "lsim",         0x2272,
 529         "lsqb",         0x005B,
 530         "lsquo",        0x2018,
 531         "lsquor",       0x201A,
 532         "Lstrok",       0x0141,
 533         "lstrok",       0x0142,
 534         "lt",           0x003C,
 535         "Lt",           0x226A,
 536         "lthree",       0x22CB,
 537         "ltimes",       0x22C9,
 538         "ltri",         0x25C3,
 539         "ltrie",        0x22B4,
 540         "ltrif",        0x25C2,
 541         "lvnE",         0x2268,
 542         NULL,           0
 543 };
 544
 545 static NameId namesM[]={
 546         "macr",         0x00AF,
 547         "male",         0x2642,
 548         "malt",         0x2720,
 549         "map",          0x21A6,
 550         "marker",       0x25AE,
 551         "Mcy",          0x041C,
 552         "mcy",          0x043C,
 553         "mdash",        0x2014,
 554         "Mgr",          0x039C,
 555         "mgr",          0x03BC,
 556         "micro",        0x00B5,
 557         "mid",          0x2223,
 558         "middot",       0x00B7,
 559         "minus",        0x2212,
 560         "minusb",       0x229F,
 561         "mldr",         0x2026,
 562         "mnplus",       0x2213,
 563         "models",       0x22A7,
 564         "mu",           0x03BC,
 565         "mumap",        0x22B8,
 566         NULL,           0
 567 };
 568
 569 static NameId namesN[]={
 570         "nabla",        0x2207,
 571         "Nacute",       0x0143,
 572         "nacute",       0x0144,
 573         "nap",          0x2249,
 574         "napos",        0x0149,
 575         "natur",        0x266E,
 576 //      "nbsp",         0x00A0,
 577         "nbsp",         32,    // make non-breaking space appear as space
 578         "Ncaron",       0x0147,
 579         "ncaron",       0x0148,
 580         "Ncedil",       0x0145,
 581         "ncedil",       0x0146,
 582         "ncong",        0x2247,
 583         "Ncy",          0x041D,
 584         "ncy",          0x043D,
 585         "ndash",        0x2013,
 586         "ne",           0x2260,
 587         "nearr",        0x2197,
 588         "nequiv",       0x2262,
 589         "nexist",       0x2204,
 590         "nge",          0x2271,
 591         "ngE",          0x2271,
 592         "nges",         0x2271,
 593         "Ngr",          0x039D,
 594         "ngr",          0x03BD,
 595         "ngt",          0x226F,
 596         "nharr",        0x21AE,
 597         "nhArr",        0x21CE,
 598         "ni",           0x220D,
 599         "NJcy",         0x040A,
 600         "njcy",         0x045A,
 601         "nlarr",        0x219A,
 602         "nlArr",        0x21CD,
 603         "nldr",         0x2025,
 604         "nle",          0x2270,
 605         "nlE",          0x2270,
 606         "nles",         0x2270,
 607         "nlt",          0x226E,
 608         "nltri",        0x22EA,
 609         "nltrie",       0x22EC,
 610         "nmid",         0x2224,
 611         "not",          0x00AC,
 612         "notin",        0x2209,
 613         "npar",         0x2226,
 614         "npr",          0x2280,
 615         "npre",         0x22E0,
 616         "nrarr",        0x219B,
 617         "nrArr",        0x21CF,
 618         "nrtri",        0x22EB,
 619         "nrtrie",       0x22ED,
 620         "nsc",          0x2281,
 621         "nsce",         0x22E1,
 622         "nsim",         0x2241,
 623         "nsime",        0x2244,
 624         "nsmid",        0xE2AA,
 625         "nspar",        0x2226,
 626         "nsub",         0x2284,
 627         "nsube",        0x2288,
 628         "nsubE",        0x2288,
 629         "nsup",         0x2285,
 630         "nsupe",        0x2289,
 631         "nsupE",        0x2289,
 632         "Ntilde",       0x00D1,
 633         "ntilde",       0x00F1,
 634         "nu",           0x03BD,
 635         "num",          0x0023,
 636         "numero",       0x2116,
 637         "numsp",        0x2007,
 638         "nvdash",       0x22AC,
 639         "nvDash",       0x22AD,
 640         "nVdash",       0x22AE,
 641         "nVDash",       0x22AF,
 642         "nwarr",        0x2196,
 643         NULL,           0
 644 };
 645
 646 static NameId namesO[]={
 647         "Oacgr",        0x038C,
 648         "oacgr",        0x03CC,
 649         "Oacute",       0x00D3,
 650         "oacute",       0x00F3,
 651         "oast",         0x229B,
 652         "ocir",         0x229A,
 653         "Ocirc",        0x00D4,
 654         "ocirc",        0x00F4,
 655         "Ocy",          0x041E,
 656         "ocy",          0x043E,
 657         "odash",        0x229D,
 658         "Odblac",       0x0150,
 659         "odblac",       0x0151,
 660         "odot",         0x2299,
 661         "OElig",        0x0152,
 662         "oelig",        0x0153,
 663         "ogon",         0x02DB,
 664         "Ogr",          0x039F,
 665         "ogr",          0x03BF,
 666         "Ograve",       0x00D2,
 667         "ograve",       0x00F2,
 668         "OHacgr",       0x038F,
 669         "ohacgr",       0x03CE,
 670         "OHgr",         0x03A9,
 671         "ohgr",         0x03C9,
 672         "ohm",          0x2126,
 673         "olarr",        0x21BA,
 674         "Omacr",        0x014C,
 675         "omacr",        0x014D,
 676         "Omega",        0x03A9,
 677         "omega",        0x03C9,
 678         "ominus",       0x2296,
 679         "oplus",        0x2295,
 680         "or",           0x2228,
 681         "orarr",        0x21BB,
 682         "order",        0x2134,
 683         "ordf",         0x00AA,
 684         "ordm",         0x00BA,
 685         "oS",           0x24C8,
 686         "Oslash",       0x00D8,
 687         "oslash",       0x00F8,
 688         "osol",         0x2298,
 689         "Otilde",       0x00D5,
 690         "otilde",       0x00F5,
 691         "otimes",       0x2297,
 692         "Ouml",         0x00D6,
 693         "ouml",         0x00F6,
 694         NULL,           0
 695 };
 696
 697 static NameId namesP[]={
 698         "par",          0x2225,
 699         "para",         0x00B6,
 700         "part",         0x2202,
 701         "Pcy",          0x041F,
 702         "pcy",          0x043F,
 703         "percnt",       0x0025,
 704         "period",       0x002E,
 705         "permil",       0x2030,
 706         "perp",         0x22A5,
 707         "Pgr",          0x03A0,
 708         "pgr",          0x03C0,
 709         "PHgr",         0x03A6,
 710         "phgr",         0x03C6,
 711         "Phi",          0x03A6,
 712         "phis",         0x03C6,
 713         "phiv",         0x03D5,
 714         "phmmat",       0x2133,
 715         "phone",        0x260E,
 716         "Pi",           0x03A0,
 717         "pi",           0x03C0,
 718         "piv",          0x03D6,
 719         "planck",       0x210F,
 720         "plus",         0x002B,
 721         "plusb",        0x229E,
 722         "plusdo",       0x2214,
 723         "plusmn",       0x00B1,
 724         "pound",        0x00A3,
 725         "pr",           0x227A,
 726         "prap",         0x227E,
 727         "pre",          0x227C,
 728         "prime",        0x2032,
 729         "Prime",        0x2033,
 730         "prnap",        0x22E8,
 731         "prnE",         0xE2B3,
 732         "prnsim",       0x22E8,
 733         "prod",         0x220F,
 734         "prop",         0x221D,
 735         "prsim",        0x227E,
 736         "PSgr",         0x03A8,
 737         "psgr",         0x03C8,
 738         "Psi",          0x03A8,
 739         "psi",          0x03C8,
 740         "puncsp",       0x2008,
 741         NULL,           0
 742 };
 743
 744 static NameId namesQ[]={
 745         "quest",        0x003F,
 746         "quot",         0x0022,
 747         NULL,           0
 748 };
 749
 750 static NameId namesR[]={
 751         "rAarr",        0x21DB,
 752         "Racute",       0x0154,
 753         "racute",       0x0155,
 754         "radic",        0x221A,
 755         "rang",         0x3009,
 756         "raquo",        0x00BB,
 757         "rarr",         0x2192,
 758         "Rarr",         0x21A0,
 759         "rArr",         0x21D2,
 760         "rarr2",        0x21C9,
 761         "rarrhk",       0x21AA,
 762         "rarrlp",       0x21AC,
 763         "rarrtl",       0x21A3,
 764         "rarrw",        0x219D,
 765         "Rcaron",       0x0158,
 766         "rcaron",       0x0159,
 767         "Rcedil",       0x0156,
 768         "rcedil",       0x0157,
 769         "rceil",        0x2309,
 770         "rcub",         0x007D,
 771         "Rcy",          0x0420,
 772         "rcy",          0x0440,
 773         "rdquo",        0x201D,
 774         "rdquor",       0x201C,
 775         "real",         0x211C,
 776         "rect",         0x25AD,
 777         "reg",          0x00AE,
 778         "rfloor",       0x230B,
 779         "Rgr",          0x03A1,
 780         "rgr",          0x03C1,
 781         "rhard",        0x21C1,
 782         "rharu",        0x21C0,
 783         "rho",          0x03C1,
 784         "rhov",         0x03F1,
 785         "ring",         0x02DA,
 786         "rlarr2",       0x21C4,
 787         "rlhar2",       0x21CC,
 788         "rpar",         0x0029,
 789         "rpargt",       0xE291,
 790         "rsh",          0x21B1,
 791         "rsqb",         0x005D,
 792         "rsquo",        0x2019,
 793         "rsquor",       0x2018,
 794         "rthree",       0x22CC,
 795         "rtimes",       0x22CA,
 796         "rtri",         0x25B9,
 797         "rtrie",        0x22B5,
 798         "rtrif",        0x25B8,
 799         "rx",           0x211E,
 800         NULL,           0
 801 };
 802
 803 static NameId namesS[]={
 804         "Sacute",       0x015A,
 805         "sacute",       0x015B,
 806         "samalg",       0x2210,
 807         "sbsol",        0xFE68,
 808         "sc",           0x227B,
 809         "scap",         0x227F,
 810         "Scaron",       0x0160,
 811         "scaron",       0x0161,
 812         "sccue",        0x227D,
 813         "sce",          0x227D,
 814         "Scedil",       0x015E,
 815         "scedil",       0x015F,
 816         "Scirc",        0x015C,
 817         "scirc",        0x015D,
 818         "scnap",        0x22E9,
 819         "scnE",         0xE2B5,
 820         "scnsim",       0x22E9,
 821         "scsim",        0x227F,
 822         "Scy",          0x0421,
 823         "scy",          0x0441,
 824         "sdot",         0x22C5,
 825         "sdotb",        0x22A1,
 826         "sect",         0x00A7,
 827         "semi",         0x003B,
 828         "setmn",        0x2216,
 829         "sext",         0x2736,
 830         "sfgr",         0x03C2,
 831         "sfrown",       0x2322,
 832         "Sgr",          0x03A3,
 833         "sgr",          0x03C3,
 834         "sharp",        0x266F,
 835         "SHCHcy",       0x0429,
 836         "shchcy",       0x0449,
 837         "SHcy",         0x0428,
 838         "shcy",         0x0448,
 839         "shy",          0x00AD,
 840         "Sigma",        0x03A3,
 841         "sigma",        0x03C3,
 842         "sigmav",       0x03C2,
 843         "sim",          0x223C,
 844         "sime",         0x2243,
 845         "smid",         0xE301,
 846         "smile",        0x2323,
 847         "SOFTcy",       0x042C,
 848         "softcy",       0x044C,
 849         "sol",          0x002F,
 850         "spades",       0x2660,
 851         "spar",         0x2225,
 852         "sqcap",        0x2293,
 853         "sqcup",        0x2294,
 854         "sqsub",        0x228F,
 855         "sqsube",       0x2291,
 856         "sqsup",        0x2290,
 857         "sqsupe",       0x2292,
 858         "squ",          0x25A1,
 859         "square",       0x25A1,
 860         "squf",         0x25AA,
 861         "ssetmn",       0x2216,
 862         "ssmile",       0x2323,
 863         "sstarf",       0x22C6,
 864         "star",         0x22C6,
 865         "starf",        0x2605,
 866         "sub",          0x2282,
 867         "Sub",          0x22D0,
 868         "sube",         0x2286,
 869         "subE",         0x2286,
 870         "subne",        0x228A,
 871         "subnE",        0x228A,
 872         "sum",          0x2211,
 873         "sung",         0x2669,
 874         "sup",          0x2283,
 875         "Sup",          0x22D1,
 876         "sup1",         0x00B9,
 877         "sup2",         0x00B2,
 878         "sup3",         0x00B3,
 879         "supe",         0x2287,
 880         "supE",         0x2287,
 881         "supne",        0x228B,
 882         "supnE",        0x228B,
 883         "szlig",        0x00DF,
 884         NULL,           0
 885 };
 886
 887 static NameId namesT[]={
 888         "target",       0x2316,
 889         "tau",          0x03C4,
 890         "Tcaron",       0x0164,
 891         "tcaron",       0x0165,
 892         "Tcedil",       0x0162,
 893         "tcedil",       0x0163,
 894         "Tcy",          0x0422,
 895         "tcy",          0x0442,
 896         "tdot",         0x20DB,
 897         "telrec",       0x2315,
 898         "Tgr",          0x03A4,
 899         "tgr",          0x03C4,
 900         "there4",       0x2234,
 901         "Theta",        0x0398,
 902         "thetas",       0x03B8,
 903         "thetav",       0x03D1,
 904         "THgr",         0x0398,
 905         "thgr",         0x03B8,
 906         "thinsp",       0x2009,
 907         "thkap",        0x2248,
 908         "thksim",       0x223C,
 909         "THORN",        0x00DE,
 910         "thorn",        0x00FE,
 911         "tilde",        0x02DC,
 912         "times",        0x00D7,
 913         "timesb",       0x22A0,
 914         "top",          0x22A4,
 915         "tprime",       0x2034,
 916         "trade",        0x2122,
 917         "trie",         0x225C,
 918         "TScy",         0x0426,
 919         "tscy",         0x0446,
 920         "TSHcy",        0x040B,
 921         "tshcy",        0x045B,
 922         "Tstrok",       0x0166,
 923         "tstrok",       0x0167,
 924         "twixt",        0x226C,
 925         NULL,           0
 926 };
 927
 928 static NameId namesU[]={
 929         "Uacgr",        0x038E,
 930         "uacgr",        0x03CD,
 931         "Uacute",       0x00DA,
 932         "uacute",       0x00FA,
 933         "uarr",         0x2191,
 934         "uArr",         0x21D1,
 935         "uarr2",        0x21C8,
 936         "Ubrcy",        0x040E,
 937         "ubrcy",        0x045E,
 938         "Ubreve",       0x016C,
 939         "ubreve",       0x016D,
 940         "Ucirc",        0x00DB,
 941         "ucirc",        0x00FB,
 942         "Ucy",          0x0423,
 943         "ucy",          0x0443,
 944         "Udblac",       0x0170,
 945         "udblac",       0x0171,
 946         "udiagr",       0x03B0,
 947         "Udigr",        0x03AB,
 948         "udigr",        0x03CB,
 949         "Ugr",          0x03A5,
 950         "ugr",          0x03C5,
 951         "Ugrave",       0x00D9,
 952         "ugrave",       0x00F9,
 953         "uharl",        0x21BF,
 954         "uharr",        0x21BE,
 955         "uhblk",        0x2580,
 956         "ulcorn",       0x231C,
 957         "ulcrop",       0x230F,
 958         "Umacr",        0x016A,
 959         "umacr",        0x016B,
 960         "uml",          0x00A8,
 961         "Uogon",        0x0172,
 962         "uogon",        0x0173,
 963         "uplus",        0x228E,
 964         "upsi",         0x03C5,
 965         "Upsi",         0x03D2,
 966         "urcorn",       0x231D,
 967         "urcrop",       0x230E,
 968         "Uring",        0x016E,
 969         "uring",        0x016F,
 970         "Utilde",       0x0168,
 971         "utilde",       0x0169,
 972         "utri",         0x25B5,
 973         "utrif",        0x25B4,
 974         "Uuml",         0x00DC,
 975         "uuml",         0x00FC,
 976         NULL,           0
 977 };
 978
 979 static NameId namesV[]={
 980         "varr",         0x2195,
 981         "vArr",         0x21D5,
 982         "Vcy",          0x0412,
 983         "vcy",          0x0432,
 984         "vdash",        0x22A2,
 985         "vDash",        0x22A8,
 986         "Vdash",        0x22A9,
 987         "veebar",       0x22BB,
 988         "vellip",       0x22EE,
 989         "verbar",       0x007C,
 990         "Verbar",       0x2016,
 991         "vltri",        0x22B2,
 992         "vprime",       0x2032,
 993         "vprop",        0x221D,
 994         "vrtri",        0x22B3,
 995         "vsubne",       0x228A,
 996         "vsubnE",       0xE2B8,
 997         "vsupne",       0x228B,
 998         "vsupnE",       0x228B,
 999         "Vvdash",       0x22AA,
1000         NULL,           0
1001 };
1002
1003 static NameId namesW[]={
1004         "Wcirc",        0x0174,
1005         "wcirc",        0x0175,
1006         "wedgeq",       0x2259,
1007         "weierp",       0x2118,
1008         "wreath",       0x2240,
1009         NULL,           0
1010 };
1011
1012 static NameId namesX[]={
1013         "xcirc",        0x25CB,
1014         "xdtri",        0x25BD,
1015         "Xgr",          0x039E,
1016         "xgr",          0x03BE,
1017         "xharr",        0x2194,
1018         "xhArr",        0x2194,
1019         "Xi",           0x039E,
1020         "xi",           0x03BE,
1021         "xlArr",        0x21D0,
1022         "xrArr",        0x21D2,
1023         "xutri",        0x25B3,
1024         NULL,           0
1025 };
1026
1027 static NameId namesY[]={
1028         "Yacute",       0x00DD,
1029         "yacute",       0x00FD,
1030         "YAcy",         0x042F,
1031         "yacy",         0x044F,
1032         "Ycirc",        0x0176,
1033         "ycirc",        0x0177,
1034         "Ycy",          0x042B,
1035         "ycy",          0x044B,
1036         "yen",          0x00A5,
1037         "YIcy",         0x0407,
1038         "yicy",         0x0457,
1039         "YUcy",         0x042E,
1040         "yucy",         0x044E,
1041         "yuml",         0x00FF,
1042         "Yuml",         0x0178,
1043         NULL,           0
1044 };
1045
1046 static NameId namesZ[]={
1047         "Zacute",       0x0179,
1048         "zacute",       0x017A,
1049         "Zcaron",       0x017D,
1050         "zcaron",       0x017E,
1051         "Zcy",          0x0417,
1052         "zcy",          0x0437,
1053         "Zdot",         0x017B,
1054         "zdot",         0x017C,
1055         "zeta",         0x03B6,
1056         "Zgr",          0x0396,
1057         "zgr",          0x03B6,
1058         "ZHcy",         0x0416,
1059         "zhcy",         0x0436,
1060         NULL, 0
1061 };
1062
1063 // @todo@ order namesTable and names? by frequency
1064 static NameId* namesTable[] = {
1065         namesA, namesB, namesC, namesD, namesE, namesF, namesG, namesH, namesI,
1066         namesJ, namesK, namesL, namesM, namesN, namesO, namesP, namesQ, namesR,
1067         namesS, namesT, namesU, namesV, namesW, namesX, namesY, namesZ, NULL
1068 };
1069
1070 int HtmlNamedEntity(unsigned char *p, int length)
1071 {
1072     int tableIndex = tolower(*p) - 'a';
1073     if (tableIndex >= 0 && tableIndex < 26) {
1074         NameId* names = namesTable[tableIndex];
1075         int i;
1076
1077         for (i = 0; names[i].name; i++){
1078                 if (strncmp(names[i].name, (char *)p, length) == 0){
1079                         return names[i].value;
1080                 }
1081         }
1082     }
1083     error("unrecognized character entity \"%.*s\"", length, p);
1084     return -1;
1085 }
1086
1087 #else //TODO: Merge Walter's list with Thomas'
1088
1089 static NameId names[] =
1090 {
1091     // Entities
1092     "quot",     34,
1093     "amp",      38,
1094     "lt",       60,
1095     "gt",       62,
1096
1097     "OElig",    338,
1098     "oelig",    339,
1099     "Scaron",   352,
1100     "scaron",   353,
1101     "Yuml",     376,
1102     "circ",     710,
1103     "tilde",    732,
1104     "ensp",     8194,
1105     "emsp",     8195,
1106     "thinsp",   8201,
1107     "zwnj",     8204,
1108     "zwj",      8205,
1109     "lrm",      8206,
1110     "rlm",      8207,
1111     "ndash",    8211,
1112     "mdash",    8212,
1113     "lsquo",    8216,
1114     "rsquo",    8217,
1115     "sbquo",    8218,
1116     "ldquo",    8220,
1117     "rdquo",    8221,
1118     "bdquo",    8222,
1119     "dagger",   8224,
1120     "Dagger",   8225,
1121     "permil",   8240,
1122     "lsaquo",   8249,
1123     "rsaquo",   8250,
1124     "euro",     8364,
1125
1126     // Latin-1 (ISO-8859-1) Entities
1127     "nbsp",     160,
1128     "iexcl",    161,
1129     "cent",     162,
1130     "pound",    163,
1131     "curren",   164,
1132     "yen",      165,
1133     "brvbar",   166,
1134     "sect",     167,
1135     "uml",      168,
1136     "copy",     169,
1137     "ordf",     170,
1138     "laquo",    171,
1139     "not",      172,
1140     "shy",      173,
1141     "reg",      174,
1142     "macr",     175,
1143     "deg",      176,
1144     "plusmn",   177,
1145     "sup2",     178,
1146     "sup3",     179,
1147     "acute",    180,
1148     "micro",    181,
1149     "para",     182,
1150     "middot",   183,
1151     "cedil",    184,
1152     "sup1",     185,
1153     "ordm",     186,
1154     "raquo",    187,
1155     "frac14",   188,
1156     "frac12",   189,
1157     "frac34",   190,
1158     "iquest",   191,
1159     "Agrave",   192,
1160     "Aacute",   193,
1161     "Acirc",    194,
1162     "Atilde",   195,
1163     "Auml",     196,
1164     "Aring",    197,
1165     "AElig",    198,
1166     "Ccedil",   199,
1167     "Egrave",   200,
1168     "Eacute",   201,
1169     "Ecirc",    202,
1170     "Euml",     203,
1171     "Igrave",   204,
1172     "Iacute",   205,
1173     "Icirc",    206,
1174     "Iuml",     207,
1175     "ETH",      208,
1176     "Ntilde",   209,
1177     "Ograve",   210,
1178     "Oacute",   211,
1179     "Ocirc",    212,
1180     "Otilde",   213,
1181     "Ouml",     214,
1182     "times",    215,
1183     "Oslash",   216,
1184     "Ugrave",   217,
1185     "Uacute",   218,
1186     "Ucirc",    219,
1187     "Uuml",     220,
1188     "Yacute",   221,
1189     "THORN",    222,
1190     "szlig",    223,
1191     "agrave",   224,
1192     "aacute",   225,
1193     "acirc",    226,
1194     "atilde",   227,
1195     "auml",     228,
1196     "aring",    229,
1197     "aelig",    230,
1198     "ccedil",   231,
1199     "egrave",   232,
1200     "eacute",   233,
1201     "ecirc",    234,
1202     "euml",     235,
1203     "igrave",   236,
1204     "iacute",   237,
1205     "icirc",    238,
1206     "iuml",     239,
1207     "eth",      240,
1208     "ntilde",   241,
1209     "ograve",   242,
1210     "oacute",   243,
1211     "ocirc",    244,
1212     "otilde",   245,
1213     "ouml",     246,
1214     "divide",   247,
1215     "oslash",   248,
1216     "ugrave",   249,
1217     "uacute",   250,
1218     "ucirc",    251,
1219     "uuml",     252,
1220     "yacute",   253,
1221     "thorn",    254,
1222     "yuml",     255,
1223
1224         // Symbols and Greek letter entities
1225     "fnof",     402,
1226     "Alpha",    913,
1227     "Beta",     914,
1228     "Gamma",    915,
1229     "Delta",    916,
1230     "Epsilon",  917,
1231     "Zeta",     918,
1232     "Eta",      919,
1233     "Theta",    920,
1234     "Iota",     921,
1235     "Kappa",    922,
1236     "Lambda",   923,
1237     "Mu",       924,
1238     "Nu",       925,
1239     "Xi",       926,
1240     "Omicron",  927,
1241     "Pi",       928,
1242     "Rho",      929,
1243     "Sigma",    931,
1244     "Tau",      932,
1245     "Upsilon",  933,
1246     "Phi",      934,
1247     "Chi",      935,
1248     "Psi",      936,
1249     "Omega",    937,
1250     "alpha",    945,
1251     "beta",     946,
1252     "gamma",    947,
1253     "delta",    948,
1254     "epsilon",  949,
1255     "zeta",     950,
1256     "eta",      951,
1257     "theta",    952,
1258     "iota",     953,
1259     "kappa",    954,
1260     "lambda",   955,
1261     "mu",       956,
1262     "nu",       957,
1263     "xi",       958,
1264     "omicron",  959,
1265     "pi",       960,
1266     "rho",      961,
1267     "sigmaf",   962,
1268     "sigma",    963,
1269     "tau",      964,
1270     "upsilon",  965,
1271     "phi",      966,
1272     "chi",      967,
1273     "psi",      968,
1274     "omega",    969,
1275     "thetasym", 977,
1276     "upsih",    978,
1277     "piv",      982,
1278     "bull",     8226,
1279     "hellip",   8230,
1280     "prime",    8242,
1281     "Prime",    8243,
1282     "oline",    8254,
1283     "frasl",    8260,
1284     "weierp",   8472,
1285     "image",    8465,
1286     "real",     8476,
1287     "trade",    8482,
1288     "alefsym",  8501,
1289     "larr",     8592,
1290     "uarr",     8593,
1291     "rarr",     8594,
1292     "darr",     8595,
1293     "harr",     8596,
1294     "crarr",    8629,
1295     "lArr",     8656,
1296     "uArr",     8657,
1297     "rArr",     8658,
1298     "dArr",     8659,
1299     "hArr",     8660,
1300     "forall",   8704,
1301     "part",     8706,
1302     "exist",    8707,
1303     "empty",    8709,
1304     "nabla",    8711,
1305     "isin",     8712,
1306     "notin",    8713,
1307     "ni",       8715,
1308     "prod",     8719,
1309     "sum",      8721,
1310     "minus",    8722,
1311     "lowast",   8727,
1312     "radic",    8730,
1313     "prop",     8733,
1314     "infin",    8734,
1315     "ang",      8736,
1316     "and",      8743,
1317     "or",       8744,
1318     "cap",      8745,
1319     "cup",      8746,
1320     "int",      8747,
1321     "there4",   8756,
1322     "sim",      8764,
1323     "cong",     8773,
1324     "asymp",    8776,
1325     "ne",       8800,
1326     "equiv",    8801,
1327     "le",       8804,
1328     "ge",       8805,
1329     "sub",      8834,
1330     "sup",      8835,
1331     "nsub",     8836,
1332     "sube",     8838,
1333     "supe",     8839,
1334     "oplus",    8853,
1335     "otimes",   8855,
1336     "perp",     8869,
1337     "sdot",     8901,
1338     "lceil",    8968,
1339     "rceil",    8969,
1340     "lfloor",   8970,
1341     "rfloor",   8971,
1342     "lang",     9001,
1343     "rang",     9002,
1344     "loz",      9674,
1345     "spades",   9824,
1346     "clubs",    9827,
1347     "hearts",   9829,
1348     "diams",    9830,
1349 };
1350
1351 int HtmlNamedEntity(unsigned char *p, int length)
1352 {
1353     int i;
1354
1355     // BUG: this is a dumb, slow linear search
1356     for (i = 0; i < sizeof(names) / sizeof(names[0]); i++)
1357     {
1358         // Entries are case sensitive
1359         if (memcmp(names[i].name, (char *)p, length) == 0 &&
1360             !names[i].name[length])
1361             return names[i].value;
1362     }
1363     return -1;
1364 }
1365
1366 #endif