src/ctype/towctrans.c

   1 #include <ctype.h>
   2 #include <wctype.h>
   3 #include "libc.h"
   4
   5 #define CASEMAP(u1,u2,l) { (u1), (l)-(u1), (u2)-(u1)+1 }
   6 #define CASELACE(u1,u2) CASEMAP((u1),(u2),(u1)+1)
   7
   8 static const struct {
   9         unsigned short upper;
  10         signed char lower;
  11         unsigned char len;
  12 } casemaps[] = {
  13         CASEMAP(0xc0,0xde,0xe0),
  14
  15         CASELACE(0x0100,0x012e),
  16         CASELACE(0x0132,0x0136),
  17         CASELACE(0x0139,0x0147),
  18         CASELACE(0x014a,0x0176),
  19         CASELACE(0x0179,0x017d),
  20
  21         CASELACE(0x370,0x372),
  22         CASEMAP(0x391,0x3a1,0x3b1),
  23         CASEMAP(0x3a3,0x3ab,0x3c3),
  24         CASEMAP(0x400,0x40f,0x450),
  25         CASEMAP(0x410,0x42f,0x430),
  26
  27         CASELACE(0x460,0x480),
  28         CASELACE(0x48a,0x4be),
  29         CASELACE(0x4c1,0x4cd),
  30         CASELACE(0x4d0,0x50e),
  31
  32         CASELACE(0x514,0x52e),
  33         CASEMAP(0x531,0x556,0x561),
  34
  35         CASELACE(0x01a0,0x01a4),
  36         CASELACE(0x01b3,0x01b5),
  37         CASELACE(0x01cd,0x01db),
  38         CASELACE(0x01de,0x01ee),
  39         CASELACE(0x01f8,0x021e),
  40         CASELACE(0x0222,0x0232),
  41         CASELACE(0x03d8,0x03ee),
  42
  43         CASELACE(0x1e00,0x1e94),
  44         CASELACE(0x1ea0,0x1efe),
  45
  46         CASEMAP(0x1f08,0x1f0f,0x1f00),
  47         CASEMAP(0x1f18,0x1f1d,0x1f10),
  48         CASEMAP(0x1f28,0x1f2f,0x1f20),
  49         CASEMAP(0x1f38,0x1f3f,0x1f30),
  50         CASEMAP(0x1f48,0x1f4d,0x1f40),
  51
  52         CASEMAP(0x1f68,0x1f6f,0x1f60),
  53         CASEMAP(0x1f88,0x1f8f,0x1f80),
  54         CASEMAP(0x1f98,0x1f9f,0x1f90),
  55         CASEMAP(0x1fa8,0x1faf,0x1fa0),
  56         CASEMAP(0x1fb8,0x1fb9,0x1fb0),
  57         CASEMAP(0x1fba,0x1fbb,0x1f70),
  58         CASEMAP(0x1fc8,0x1fcb,0x1f72),
  59         CASEMAP(0x1fd8,0x1fd9,0x1fd0),
  60         CASEMAP(0x1fda,0x1fdb,0x1f76),
  61         CASEMAP(0x1fe8,0x1fe9,0x1fe0),
  62         CASEMAP(0x1fea,0x1feb,0x1f7a),
  63         CASEMAP(0x1ff8,0x1ff9,0x1f78),
  64         CASEMAP(0x1ffa,0x1ffb,0x1f7c),
  65
  66         CASEMAP(0x13f0,0x13f5,0x13f8),
  67         CASELACE(0xa698,0xa69a),
  68         CASELACE(0xa796,0xa79e),
  69
  70         CASELACE(0x246,0x24e),
  71         CASELACE(0x510,0x512),
  72         CASEMAP(0x2160,0x216f,0x2170),
  73         CASEMAP(0x2c00,0x2c2e,0x2c30),
  74         CASELACE(0x2c67,0x2c6b),
  75         CASELACE(0x2c80,0x2ce2),
  76         CASELACE(0x2ceb,0x2ced),
  77
  78         CASELACE(0xa640,0xa66c),
  79         CASELACE(0xa680,0xa696),
  80
  81         CASELACE(0xa722,0xa72e),
  82         CASELACE(0xa732,0xa76e),
  83         CASELACE(0xa779,0xa77b),
  84         CASELACE(0xa77e,0xa786),
  85
  86         CASELACE(0xa790,0xa792),
  87         CASELACE(0xa7a0,0xa7a8),
  88
  89         CASELACE(0xa7b4,0xa7b6),
  90
  91         CASEMAP(0xff21,0xff3a,0xff41),
  92         { 0,0,0 }
  93 };
  94
  95 static const unsigned short pairs[][2] = {
  96         { 'I',    0x0131 },
  97         { 'S',    0x017f },
  98         { 0x0130, 'i'    },
  99         { 0x0178, 0x00ff },
 100         { 0x0181, 0x0253 },
 101         { 0x0182, 0x0183 },
 102         { 0x0184, 0x0185 },
 103         { 0x0186, 0x0254 },
 104         { 0x0187, 0x0188 },
 105         { 0x0189, 0x0256 },
 106         { 0x018a, 0x0257 },
 107         { 0x018b, 0x018c },
 108         { 0x018e, 0x01dd },
 109         { 0x018f, 0x0259 },
 110         { 0x0190, 0x025b },
 111         { 0x0191, 0x0192 },
 112         { 0x0193, 0x0260 },
 113         { 0x0194, 0x0263 },
 114         { 0x0196, 0x0269 },
 115         { 0x0197, 0x0268 },
 116         { 0x0198, 0x0199 },
 117         { 0x019c, 0x026f },
 118         { 0x019d, 0x0272 },
 119         { 0x019f, 0x0275 },
 120         { 0x01a6, 0x0280 },
 121         { 0x01a7, 0x01a8 },
 122         { 0x01a9, 0x0283 },
 123         { 0x01ac, 0x01ad },
 124         { 0x01ae, 0x0288 },
 125         { 0x01af, 0x01b0 },
 126         { 0x01b1, 0x028a },
 127         { 0x01b2, 0x028b },
 128         { 0x01b7, 0x0292 },
 129         { 0x01b8, 0x01b9 },
 130         { 0x01bc, 0x01bd },
 131         { 0x01c4, 0x01c6 },
 132         { 0x01c4, 0x01c5 },
 133         { 0x01c5, 0x01c6 },
 134         { 0x01c7, 0x01c9 },
 135         { 0x01c7, 0x01c8 },
 136         { 0x01c8, 0x01c9 },
 137         { 0x01ca, 0x01cc },
 138         { 0x01ca, 0x01cb },
 139         { 0x01cb, 0x01cc },
 140         { 0x01f1, 0x01f3 },
 141         { 0x01f1, 0x01f2 },
 142         { 0x01f2, 0x01f3 },
 143         { 0x01f4, 0x01f5 },
 144         { 0x01f6, 0x0195 },
 145         { 0x01f7, 0x01bf },
 146         { 0x0220, 0x019e },
 147         { 0x0386, 0x03ac },
 148         { 0x0388, 0x03ad },
 149         { 0x0389, 0x03ae },
 150         { 0x038a, 0x03af },
 151         { 0x038c, 0x03cc },
 152         { 0x038e, 0x03cd },
 153         { 0x038f, 0x03ce },
 154         { 0x0399, 0x0345 },
 155         { 0x0399, 0x1fbe },
 156         { 0x03a3, 0x03c2 },
 157         { 0x03f7, 0x03f8 },
 158         { 0x03fa, 0x03fb },
 159         { 0x1e60, 0x1e9b },
 160         { 0x1e9e, 0xdf },
 161
 162         { 0x1f59, 0x1f51 },
 163         { 0x1f5b, 0x1f53 },
 164         { 0x1f5d, 0x1f55 },
 165         { 0x1f5f, 0x1f57 },
 166         { 0x1fbc, 0x1fb3 },
 167         { 0x1fcc, 0x1fc3 },
 168         { 0x1fec, 0x1fe5 },
 169         { 0x1ffc, 0x1ff3 },
 170
 171         { 0x23a, 0x2c65 },
 172         { 0x23b, 0x23c },
 173         { 0x23d, 0x19a },
 174         { 0x23e, 0x2c66 },
 175         { 0x241, 0x242 },
 176         { 0x243, 0x180 },
 177         { 0x244, 0x289 },
 178         { 0x245, 0x28c },
 179         { 0x3f4, 0x3b8 },
 180         { 0x3f9, 0x3f2 },
 181         { 0x3fd, 0x37b },
 182         { 0x3fe, 0x37c },
 183         { 0x3ff, 0x37d },
 184         { 0x4c0, 0x4cf },
 185
 186         { 0x2126, 0x3c9 },
 187         { 0x212a, 'k' },
 188         { 0x212b, 0xe5 },
 189         { 0x2132, 0x214e },
 190         { 0x2183, 0x2184 },
 191         { 0x2c60, 0x2c61 },
 192         { 0x2c62, 0x26b },
 193         { 0x2c63, 0x1d7d },
 194         { 0x2c64, 0x27d },
 195         { 0x2c6d, 0x251 },
 196         { 0x2c6e, 0x271 },
 197         { 0x2c6f, 0x250 },
 198         { 0x2c70, 0x252 },
 199         { 0x2c72, 0x2c73 },
 200         { 0x2c75, 0x2c76 },
 201         { 0x2c7e, 0x23f },
 202         { 0x2c7f, 0x240 },
 203         { 0x2cf2, 0x2cf3 },
 204
 205         { 0xa77d, 0x1d79 },
 206         { 0xa78b, 0xa78c },
 207         { 0xa78d, 0x265 },
 208         { 0xa7aa, 0x266 },
 209
 210         { 0x10c7, 0x2d27 },
 211         { 0x10cd, 0x2d2d },
 212
 213         /* bogus greek 'symbol' letters */
 214         { 0x376, 0x377 },
 215         { 0x39c, 0xb5 },
 216         { 0x392, 0x3d0 },
 217         { 0x398, 0x3d1 },
 218         { 0x3a6, 0x3d5 },
 219         { 0x3a0, 0x3d6 },
 220         { 0x39a, 0x3f0 },
 221         { 0x3a1, 0x3f1 },
 222         { 0x395, 0x3f5 },
 223         { 0x3cf, 0x3d7 },
 224
 225         { 0xa7ab, 0x25c },
 226         { 0xa7ac, 0x261 },
 227         { 0xa7ad, 0x26c },
 228         { 0xa7ae, 0x26a },
 229         { 0xa7b0, 0x29e },
 230         { 0xa7b1, 0x287 },
 231         { 0xa7b2, 0x29d },
 232         { 0xa7b3, 0xab53 },
 233
 234         /* special cyrillic lowercase forms */
 235         { 0x412, 0x1c80 },
 236         { 0x414, 0x1c81 },
 237         { 0x41e, 0x1c82 },
 238         { 0x421, 0x1c83 },
 239         { 0x422, 0x1c84 },
 240         { 0x422, 0x1c85 },
 241         { 0x42a, 0x1c86 },
 242         { 0x462, 0x1c87 },
 243         { 0xa64a, 0x1c88 },
 244
 245         { 0,0 }
 246 };
 247
 248
 249 static wchar_t __towcase(wchar_t wc, int lower)
 250 {
 251         int i;
 252         int lmul = 2*lower-1;
 253         int lmask = lower-1;
 254         /* no letters with case in these large ranges */
 255         if (!iswalpha(wc)
 256          || (unsigned)wc - 0x0600 <= 0x0fff-0x0600
 257          || (unsigned)wc - 0x2e00 <= 0xa63f-0x2e00
 258          || (unsigned)wc - 0xa800 <= 0xab52-0xa800
 259          || (unsigned)wc - 0xabc0 <= 0xfeff-0xabc0)
 260                 return wc;
 261         /* special case because the diff between upper/lower is too big */
 262         if (lower && (unsigned)wc - 0x10a0 < 0x2e)
 263                 if (wc>0x10c5 && wc != 0x10c7 && wc != 0x10cd) return wc;
 264                 else return wc + 0x2d00 - 0x10a0;
 265         if (!lower && (unsigned)wc - 0x2d00 < 0x26)
 266                 if (wc>0x2d25 && wc != 0x2d27 && wc != 0x2d2d) return wc;
 267                 else return wc + 0x10a0 - 0x2d00;
 268         if (lower && (unsigned)wc - 0x13a0 < 0x50)
 269                 return wc + 0xab70 - 0x13a0;
 270         if (!lower && (unsigned)wc - 0xab70 < 0x50)
 271                 return wc + 0x13a0 - 0xab70;
 272         for (i=0; casemaps[i].len; i++) {
 273                 int base = casemaps[i].upper + (lmask & casemaps[i].lower);
 274                 if ((unsigned)wc-base < casemaps[i].len) {
 275                         if (casemaps[i].lower == 1)
 276                                 return wc + lower - ((wc-casemaps[i].upper)&1);
 277                         return wc + lmul*casemaps[i].lower;
 278                 }
 279         }
 280         for (i=0; pairs[i][1-lower]; i++) {
 281                 if (pairs[i][1-lower] == wc)
 282                         return pairs[i][lower];
 283         }
 284         if ((unsigned)wc - (0x10428 - 0x28*lower) < 0x28)
 285                 return wc - 0x28 + 0x50*lower;
 286         if ((unsigned)wc - (0x104d8 - 0x28*lower) < 0x24)
 287                 return wc - 0x28 + 0x50*lower;
 288         if ((unsigned)wc - (0x10cc0 - 0x40*lower) < 0x33)
 289                 return wc - 0x40 + 0x80*lower;
 290         if ((unsigned)wc - (0x118c0 - 0x20*lower) < 0x20)
 291                 return wc - 0x20 + 0x40*lower;
 292         if ((unsigned)wc - (0x1e922 - 0x22*lower) < 0x22)
 293                 return wc - 0x22 + 0x44*lower;
 294         return wc;
 295 }
 296
 297 wint_t towupper(wint_t wc)
 298 {
 299         return (unsigned)wc < 128 ? toupper(wc) : __towcase(wc, 0);
 300 }
 301
 302 wint_t towlower(wint_t wc)
 303 {
 304         return (unsigned)wc < 128 ? tolower(wc) : __towcase(wc, 1);
 305 }
 306
 307 wint_t __towupper_l(wint_t c, locale_t l)
 308 {
 309         return towupper(c);
 310 }
 311
 312 wint_t __towlower_l(wint_t c, locale_t l)
 313 {
 314         return towlower(c);
 315 }
 316
 317 weak_alias(__towupper_l, towupper_l);
 318 weak_alias(__towlower_l, towlower_l);